I... .a.!r._|lhl:h..,
it .11 :3:

 

 

.
.1

7

 

w.
dawns.“
‘v
.a

._
..

us. _ :.
5.2.2 .
513.95%
7.» I.
23.01.;3 6 \2. .
um

i
r.
1 3‘5.

.
.m... 1.2:: :31

3:2 “up.“ . ;

finial-“a-

: .1
. iivfiax 3:1
31‘, I}. . fix“ .. i... 5‘45”".
l.fi’-¢I!'.‘A3g.-.I1.
a ‘A-"’. —’ '42
4.4....) an: ‘9‘
{a 1.? . 1903.5.

5. It!!! i
hit! ‘3.
. . {I

II...

5?}

:i
«3. a. 21‘. 99.2.
\‘513|)!'!1|;rfll
: 1?. .
b.3513: a.

1591:...
.2...)
‘ 1. :3
34.13.5516
varsivki:

I
,5 311:? .uifisu. o...

53:; y) I I. gauzlltt‘. I

13.32. >I:..1vl&-s}ll‘!

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

ESSAYS IN TRADE AND ENVIRONMENT:
THE ENVIRONMENTAL EFFECTS OF
INTRAINDUSTRY TRADE

presented by

SARMA BINTI ARALAS

has been accepted towards fulfillment
of the requirements for the

Doctoral degree in Aflultural Economics

 

 

 

 

fjor Professor’s Signature
5/23 // 0

Date

MSU is an Affinnative Action/Equal Opportunity Employer

 

..—.--.-._ - ...

 

PLACE IN RETURN BOX to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProjIAoc&Pres/ClRC/Date0ue.indd

 

ESSAYS IN TRADE AND ENVIRONMENT:
THE ENVIRONMENTAL EFFECTS OF INTRAINDUSTRY TRADE

By

Sarma Binti Aralas

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Agricultural Economics

2010

ABSTRACT

ESSAYS IN TRADE AND ENVIRONMENT:
THE ENVIRONMENTAL EFFECTS OF INTRAINDUSTRY TRADE

By

Sarrna B. Aralas

Environmental concerns are at the center stage of current global debates on issues
ranging from climate change to natural resource conservation. Anthropogenic sources of
environmental degradation include development activities undertaken to increase
economic growth and welfare. In the late twentieth century, industrialization
characterized the developmental paths of countries seeking to modernize and to raise per
capita incomes. Countries that engage in international trade expand their potential beyond
domestic borders to reach a global and richer market. As globalization becomes an
important aspect of economic development, countries with accelerated growths in dirty
industries are viewed as contributing to the deterioration of environmental problems such
as global warming, deforestation and resource depletion. Globalization, it is argued, leads
to the expansion of pollution-intensive production which causes harm to the environment.

It is widely recognized that international trade raises economic welfare. The
succeeding question is, if production is pollution intensive, does international trade
necessarily lead to detrimental effects on the environment? Is there evidence to suggest
that trade is beneficial or harmful to the environment?

This dissertation explores the environmental implication of the engagement of

trade in differentiated, dirty goods. The first essay entitled Monopolistic Competition,

Trade and the Environment, presents three analyses. First, it develops a trade-and-
environment framework for an economy that produces differentiated, pollution-intensive
(or dirty) goods. The production of dirty goods is shown to lead to three environmental
effects, namely, the scale, technique and selection effectsfSecond, it presents a
comparative statics analysis of the effects of a change in environmental policy on the
firm’s level of abatement, product price, consumption, the scale of production and the
number of firms in the economy. Third, it examines the relationship between openness to
trade and the environment. The impact of intra-industry trade is shown to be the sum of
the scale, technique and selection effects. In the second essay, Intra-industry Trade and
the Environment, the general framework developed in the first essay is modified to allow
for a constant-elasticity of substitution (CES) utility function. It shows, unambiguously,
that free trade does not lead to detrimental environmental effects. A comparative statics
analysis shows that an increase in the stringency of environmental policy generates a
negative technique effect and neutral scale and selection effects. In the third essay, How
Does Intra-industry Trade A fleet the Environment?, the integrated theoretical predictions
of the pollution models of intra-industry as wells as inter-industry trade are tested using
panel-data methods. Statistical evidence suggests the following. First, the emissions of
sulfur oxides, nitrogen oxides and volatile organic compounds are increasing in the
selection, scale, technique and composition effects. Second, the selection effect is an
important and relevant variable in the estimation of the full impact of international trade
on emissions level. Third, results conform to the realizations of data generated by the
framework of intra-industry as well as of inter-industry trade. Fourth and finally, greater

openness to trade or increased trade liberalization, leads to a decrease in emissions level.

Copyright by
SARMA BINTI ARALAS
2010

To my beloved parents,
Hajah Khadijah Baring Atick and Haji Mohammad Imran Aralas Kua,
and my loving son,

Ibrahem Ghani Wasti

ACKNOWLEDGEMENT

I thank members of my dissertation committee: Professors John P. Hoehn,
Richard D. Horan, Emma Iglesias Vazquez, and Susan C. Zhu.

I am grateful to Professor John Hoehn for his guidance and support throughout
my study. He encouraged me to explore possibilities, to seek answers with fervor and
zeal. Our discussions stimulated the mind and led to new discoveries and new ideas. I
learnt many aspects of the academia from John. I am indebted to him, and I have been
privileged to have him as my major professor and thesis supervisor.

I thank Professor Susan Zhu who challenged me to think beyond what is known
and to ask the right questions. It was the challenge of developing a model for a paper in
her course that I chose the model of monopolistic competition, and constructed the
emission function which became the foundation of the trade-environment models
developed in this dissertation. Professor Zhu’s remarks were not limited to the
dissertation; they included advice on career and professional development. I express my
gratitude to her.

I thank Professor Richard Horan for his meticulous and constructive reviews of
my dissertation. I value the advice he gave, not only on dissertation matters, but also on
academic life, on what it means to have dedication to work and purpose. I am grateful to
him.

I thank Professor Emma Iglesias Vazquez for her reviews and insightful
comments on econometric methods. Her remarks were not restricted to the analyses in the

dissertation; they generated ideas for future research and empirical investigations.

vi

This dissertation would not have been realized without the support and funding
from various sources. I thank J abatan Perkhidmatan Awam (J PA), Malaysia, for the
SLAB scholarship which funded the early stage of my study. I thank Professor Eric
Crawford for his support in ensuring departmental funding during the times I needed
them early in the program. I am grateful to Dr. Eric Scorsone for most needed funding at
critical stages of the doctoral program. I am indebted to him. I thank Professor Scott
Loveridge and Professor Steve Hanson for departmental funding and for their support. I
thank the Department of Agricultural Economics and the Graduate School of Michigan
State University for the following rewards: Glenn and Sandy Johnson Fellowship,
Graduate Summer Retention Fellowship and the Dissertation Completion Fellowship. I
thank the Land Policy Institute and the Kellogg Foundation for the funding they
provided.

I am grateful to my beloved parents, Hajah Khadijah Baring binti Atick and Haji
Mohammad Imran Aralas bin Kua, for instilling in me the appreciation for knowledge
and the wonderment in the wisdom that knowledge brings. My parents taught my siblings
and me the value of the pursuit of enlightenment. Without their encouragement and the
values I Ieamt, I would not have endeavored this far in my quest for knowledge.

I am grateful to my loving son, Ibrahem, whose maturity, beyond his years,
sustained my perseverance to complete my education. He never once complained about
the hectic lifestyle that came with graduate school, nor did he ever present any challenge
during his growing years as a teenager. He has been, and is, the silver lining behind the

clouds, the rainbow after the rain. This has made it possible for me to chase the light at

vii

the end of the tunnel, and reach the destination I sought to arrive at. I am forever grateful
to Ibrahem for his patience and great understanding.

I am grateful to my brothers and sisters: Ali, Tessie, Melinda, (Dr.) Dalia, (Dr.)
Ken, (Dr.) Roseline, Yacob and Sitti, for their understanding of this quest that I started
and now completed. Their love and support sustained my determination to accomplish
the goal I set out to achieve.

I thank friends whose friendships made for a cheerful experience of living life as a
graduate student. I thank Umar for his support and encouragement. Azlini, Vandana,
Abdoul and Eunice provide encouragement at any time, any day; life can never be too
demanding or too mundane with such friends. Makoto, Kwame, Joshua, Mike, Kirimi
and Shaufique are friends one can count on, always. Sri Hayati, Kurnia, Rabiha, Feng and
Zhiying make for merry and endearing company. Irina and Wei are companions in

walking the path of graduate life.

viii

TABLE OF CONTENTS

LIST OF TABLES ............................................................................... xi
LIST OF FIGURES .............................................................................. xii
CHAPTER 1: INTRODUCTION .............................................................. 1
1.1 Trade and Pollution ...................................................................... 1
1.2 Contribution .............................................................................. 12
CHAPTER 2: MONOPOLISTIC COMPETITION, TRADE AND THE
ENVIRONMENT ................................................................................ 14
2.1 Introduction ............................................................................... 14
2.2 Literature Review ........................................................................ 17
2.3 The Model ................................................................................. 25
2.3.1 Demand ............................................................................... 27
2.3.2 Production ............................................................................ 29
2.3.3 Autarky Equilibrium ................................................................ 32
2.3.4 Decomposition of Impact: Scale, Technique and Selection Effects .......... 39
2.4 Effects of Environmental Policy ....................................................... 43
2.5 Trade and Pollution ...................................................................... 53
2.6 Conclusion ................................................................................ 59
CHAPTER 3: INTRA-INDUSTRY TRADE AND THE ENVIRONMENT ............ 61
3.1 Introduction ............................................................................... 61
3.2 The Model ................................................................................. 64
3.2.1 Demand ............................................................................... 64
3.2.2 Production ............................................................................ 65
3.2.3 Autarky Equilibrium ................................................................ 66
3.2.4 Scale, Technique and Selection Effects .......................................... 70
3.3 Comparative Static Effects of Stricter Environmental Policy ..................... 70
3.3.1 Abatement ............................................................................ 71
3.3.2 Emission Intensity ................................................................... 72
3.3.3 Price ................................................................................... 72
3.3.4 Consumption ......................................................................... 73
3.3.5 Net Output ............................................................................ 74
3.3.6 Output ................................................................................. 75
3.3.7 Number of Firms ..................................................................... 75
3.4 Bilateral Trade ........................................................................... 76
3.4.1 Consumption ......................................................................... 77
3.4.2 Production ............................................................................ 79
3.4.3 Pollution Supply and Emission Tax ............................................. 80
3.5 Discussion ................................................................................. 83
3.6 Conclusion ................................................................................ 85

ix

CHAPTER 4: HOW DOES INTRA-INDUSTRY TRADE AFFECT THE

ENVIRONMENT? ........................................................................................................ 87
4.1 Introduction ............................................................................... 87
4.2 Conceptual Framework and Research Hypotheses .................................. 93

4.2.1 Conceptual Framework ............................................................. 94
4.2.2 Research Hypotheses ............................................................... 100
4.3 Pollution and Intra-industry Trade ..................................................... 105
4.3.1 Equilibrium Conditions ............................................................. 107
4.3.2 Scale, Technique and Selection Effects .......................................... 109
4.4 Reduced Form Equations ............................................................... 110
4.4.1 An Empirical Model of Inter-industry and Intra-industry Trade .............. 110
4.5 Empirical Strategy ....................................................................... 115
4.5.1 Measurement and Data Sources ................................................... 115
4.5.2 Method ................................................................................ 119
4.5.3 The Estimating Equation ........................................................... 121
4.5.3.1 Functional Form ................................................................. 121
4.5.3.2 Alternative Functional Form ................................................. 123
4.6 Main Result ............................................................................... 124
4.6.1 SOx and the Scale, Selection, Technique and Composition Effects 126
4. 6. 2 NOx and the Scale, Selection, Technique and Composition Effects ......... 131
4. 6. 3 VOC and the Scale, Selection, Technique and Composition Effects. 134
4.7 Alternative Specifications... 137
4. 8 Discussion ................................................................................. 145
4.9 Conclusion ................................................................................ 151

Appendix A: Monopolistic Competition, Trade and the Environment ................... 155
A.1 Production ................................................................................ 155
A.2 PE and ZE Lines ........................................................................ 161
A3 Comparative Statics Effects of a Change in Environmental Policy .............. 164
A4 Table A.2 Comparative Static Effects of a Change in Emission Tax Rate ...... 172
AS Table A3 The of Impact of Intra-industry Trade on the Economy .............. 174

Appendix B: Intra-industry Trade and the Environment .................................... 175
B.l Production in Trade-Environment Model with CBS Utility Function ............ 175
B2 Pollution Supply ......................................................................... 180
B3 Table B.l Comparative Static Effects of a Change in Emission Tax Rate. 183
B.4 The Impact of Intra-industry Trade on the Economy ............................... 185

Appendix C: How Does Intra-industry Trade Affect the Environment? ................. 186
CI Reduced Form Equation ................................................................ 186
C2 Calculating Real Capital Stock Using Leamer’s Method. . . . . . . . . 189
C3 Calculating Marginal Effects .......................................................... 189
C4 Summary Statistics ...................................................................... 190

BIBLIOGRAPHY ................................................................................ 193

LIST OF TABLES

Table 1.1: Manufacturing Intra-industry Trade as a Percentage of Total

Manufacturing Trade ........................................................................... 89
Table 4.1: Estimation Results of Models A and B - Sulfur Oxides (SOx) ............. 127
Table 4.2: Estimation Results of Models A and B - Nitrogen Oxides (N Ox) ......... 132

Table 4.3: Estimation Results of Models A and B - Volatile Organic Compounds 135
(VOC) ............................................................................................

Table 4.4: Sensitivity Analysis Sulfur Oxides (SOx) ..................................... 138
Table 4.5: Sensitivity Analysis Nitrogen Oxides (NOx) ................................. 140
Table 4.6: Sensitivity Analysis Volatile Organic Compounds (V DC) .................. 142

xi

LIST OF FIGURES

Figure 2.1: Inter-industry and Intra-industry Trade ......................................
Figure 2.2: The Environmental Impact of International Trade .........................
Figure 2.3: The PE and ZE Line ............................................................
Figure 2.4: The Effects of Stricter Environmental Policy ..............................

Figure 2.5: The Effects of Intra—industry Trade ..........................................

xii

26

26

38

50

55

CHAPTER ONE

INTRODUCTION

1.1 Trade and Pollution

Recent events have regenerated interest in global environmental concerns. In
October 2007, the Norwegian Nobel Committee awarded its Peace Prize in recognition of
research on climate change. In the United States of America, California signed into law
two pieces of legislations in September 2007, Chaptered Bill 185 (Senate Bill 97) and
Chaptered Bill 178 (Senate Bill 85). Both bills address the effects of greenhouse gases

(GHG) on climate change.

Environmental disruptions such as the problem of global climactic change are the
results of both man-made and natural causes. Notably, the problem of environmental
pollution is not confined to the effects of greenhouse gases. Other major pollutants such
as lead, nitrogen dioxide, particulate matter and sulfur dioxide can impose physical and
financial costs on the environment. The Institute for Agriculture and Trade Policy (IATP)
and the Minnesota Center for Environmental Advocacy (MCEA) report that pollution
costs Minnesota $1.5 billion annually in childhood diseases (IATP and CEA 2006). The
World Health Organization (WHO) Regional Office for Europe calculates that air
pollution with particulate matter (PM) shortens life expectancy by 8.6 months for every
person in the European Union (EU) and that a reduction in the number of deaths from PM

could save the EU an amount of €5 8-161 billion (WHO Press Release 2005).

Economic development is often cited as a reason for the degradation of the
environment. The two fastest growing economies in the world today, China and India, are
countries experiencing simultaneous growths in international trade. According to the
World Bank (2006), China and India represent a third of global pollution. In the period
between 1992 and 2002, China’s emissions levels increased by 33 percent and India’s
emissions levels increased by 57 percent (World Bank Little Green Data Book 2006).
Between 2000 and 2005, China’s exports increased from 23 percent of GDP in 2000 to
37 percent of GDP in 2005, while India’s exports increased from 13 percent to 20 percent

in the same period (World Bank Statistics 2006).

This dissertation examines the effects of intra-industry trade on the environment.
It investigates the relationship between “new trade theory” and environmental quality in
three essays entitled “Monopolistic Competition, Trade and the Environment ”, “Intra-

industry Trade and the Environment ”, and “How Does Intra-industry'Trade Aflect the

Environment?”.

Literature suggests studies that examine the link between trade and environment
under “new trade” theory are relatively scarce. In contrast, the majority of models that
examine the effects of international trade on the environment are based on the traditional

theory of trade.

Traditional trade theory indicates trade is driven by relative factor differentials
across countries such that a country specializes in the production of goods in which it has

comparative advantage. Under the traditional trade framework, technology yields

constant returns to scale in production and markets are characterized by perfect

competition. In this setting, countries exchange homogenous goods with each other.

The increasing worldwide trend in the trade of differentiated products suggests
intra-industry trade is an important phenomenon in the global exchange of goods. It
explains a significant volume of trade flows for countries that engage in intra-industry
trade or trade in differentiated products. “New trade” theory describes trade as being
driven by the demand for, and the specialization in, differentiated goods where fixed cost
allows increasing returns to scale in production. Under these assumptions, trade may
occur between countries with similar technologies and similar factor endowments, rather

than between countries that are dissimilar in both aspects.

A survey of literature suggests current models in trade-environment literature
have not addressed the following major questions. First, how does intra-industry trade
affect emissions level in the economy? This question has yet to be explored both
theoretically and empirically. Second, in a model of intra-industry trade, does the
stringency in environmental policy affect the monopolistically competitive firm’s
abatement process, its product price, its scale of production, consumption of goods and
the number of firms in the economy? Finally, is free trade in differentiated, final goods

beneficial to the environment?

This dissertation attempts to fill the gaps in literature by addressing the
aforementioned questions. First, it shows that trade in differentiated goods can be
decomposed into in a scale effect, a technique effect, and a selection effect. Notably, the

selection effect distinguishes the effect of intra-industry trade on the environment from

the effects of inter-industry trade on the environment. In addition, the framework
developed in this dissertation assumes firms are engaged in the abatement of pollution.
Previous pollution models of intra-industry trade have not incorporated the firm’s
abatement process in linking new trade to the environment. Second, the framework is the
first to provide theoretical underpinnings to describe how changes in structural factors
impact the emissions level of local pollutants in both the closed and open economy
contexts. Third, this dissertation presents an empirical investigation of how intra-industry
trade affects environmental quality. It considers data from 26 countries in the
Organization of Economic Cooperation and Development (OECD), namely, Australia,
Canada, Czech Republic, Germany, Denmark, Finland, France, Greece, Hungary,
Iceland, Italy, Japan, the Republic of Korea, Mexico, Norway, New Zealand, Poland,
Portugal, Slovak Republic, Spain, Sweden, Switzerland, Turkey, United Kingdom and
United States of America. Three types of pollution are considered: sulfur oxides (SOx),
nitrogen oxides (NOx) and volatile organic compounds (VOC). The study period spans

the years 1995 through 2004.

This chapter provides a condensed overview of three essays in the dissertation.
The first essay is presented in Chapter Two. It proposes a pollution model of intra-
industry trade. The second essay, presented in Chapter Three, modifies the model in
Chapter Two by using a utility function that has constant elasticity of substitution (CES).
The third essay, presented in Chapter Four, is an empirical analysis that tests the

theoretical predictions of the models developed in Chapters Two and Three.

In Chapter Two, I develop a model of pollution and trade which assumes that

firms engage in the production of dirty goods to take advantage of the consumer’s

4

preference for variety. Market structure is imperfectly competitive such that firms are
able to capture the benefits of creating a niche through the uniqueness of their products.
Cost minimization in production is driven by economies of scale which allows the firm to
compete and retain market share. In this setting, I show the following results. First, when
the firm intemalizes the cost of emissions abatement, the quantity of output produced
decreases. One explanation is that the firm incurs higher costs as it complies with
environmental regulation. This reduction in output level seems beneficial to the
environment since less output implies less emissions generated in the joint production of

dirty goods.

Second, the model explicitly and formally makes distinct the environmental
impact of dirty production in a monopolistically competitive market, from that in a
perfectly competitive market. I show that when firms have market power, and entry and
exit of firms prevail, the number of firms in the industry influences the total level of
emissions in the economy. New trade theory refers to the change in the number of firms
as the “selection” effect. In the pollution model, I show an “environmental” selection
effect. Further, I show that openness to trade lead to a smaller number of domestic firms
when the economy faces competition from abroad. This finding recognizes the selection

effect as part of the impact of international trade on the environment.

Third, I show in Chapter Two that under the assumption of monopolistic
competition and increasing returns to scale, the aggregate impact of pollution-intensive
production is the sum of three environmental effects, namely, the scale, technique and
selection effects. A positive scale effect is generated when the firm expands the scale of

production. A negative technique effect is obtained when the firm undertakes abatement

5

activities to lower emissions intensity in production. Finally, a negative selection effect is
generated when the number of firms decreases due to increased competition which leads

to negative profits.

A fourth contribution of Chapter Two is the comparative statics analysis of the
effects of an increase in emission tax rate. I show that when firms engage in pollution-
intensive production and undertake abatement activities to comply with environmental
policy, a higher emissions tax rate has the following consequences: (i) emission intensity
decreases as the firm undertakes greater abatement level to comply with stricter
regulation. This is a policy-induced technique effect; (ii) price level rises as production
cost increases due to, one, increased emissions tax payments, and two, the increased
resources allocated towards abatement activities; (iii) the quantity of output contracts for
two reasons: one, higher production costs imply less supply, and two, quantity demanded
falls as price increases. This is a policy-induced scale effect; (iv) consumption level
decreases due to: one, when firms impose higher prices, demand decreases in quantity,
and two, there is a trade-off between allocating resources towards the production of
goods for consumption versus allocating resources for the purpose of pollution
abatement; (v) the number of firms in the economy increases for the reasons that when a
higher emissions tax rate is imposed, production cost increases due to larger emissions
tax payments and greater amount of resources allocated towards abatement activities.
Consequently, for some firms, the increase in costs is not sustainable as higher price
implies less quantity demanded and thus, less revenue. In this case, firms that earn
negative profits leave the industry. The exit of some firms implies surviving firms take

advantage of economies of scale to expand production by hiring excess labor in the factor

market. Positive economic profits earned by surviving firms attract new firms to enter the
industry. In the long run, zero economic profits imply an increase in the number of firms.

This is a policy-induced selection effect.

In short, the trade-environment framework developed in Chapter Two shows that
the increase in the stringency of environmental policy generates the policy-induced scale,
technique and selection effects. The policy-induced scale effect is negative if a higher
emissions tax rate leads to a fall in the scale of production; the policy-induced selection
effect is positive if the higher emissions tax rate leads to a greater number of firms
engaged in dirty production in the economy; and the policy-induced technique effect is
negative if the stricter policy leads to greater abatement and a subsequent fall in emission

intensity.

The analysis further shows that the full impact of an increase in the stringency of 1
environmental policy is the sum of the policy-induced technique, scale and selection
effects. This result implies that policy implementation needs to take into account not
only the effects of policy on abatement activities (technique effect) and the scale of
production (scale effect), but that it also needs to take into account the number of firms
that are engaged in pollution-intensive production (selection effect). The magnitudes of
each of the policy-induced environmental effects determine the overall impact of policy

change on environmental quality.

A surprising result of the comparative statics analysis in Chapter Two is that
while the total effect of stricter environmental policy may increase, decrease or leave

unchanged aggregate emissions level, its effect on the product price is unambiguous:

more stringent policy leads to higher product price. Therefore, while the effectiveness of
tightening regulation is uncertain in terms of improving environmental quality, its effects
on the cost of production and the level of consumption are definite: it increases product

price and decreases the quantity of goods demanded.

Finally, in Chapter Two, I show that in the open economy case, free trade in
differentiated goods generates trade-induced scale, technique and selection effects. When
trade opens, countries expand production, but the number of firms in the economy falls as
firms face competition from abroad. Consumers demand less of each of the domestic
varieties as they choose to consume more varieties in the forms of both local and
imported brands. In addition, under the assumption environmental quality is a normal
good, I show that when income rises with trade, this implies stricter environmental
regulation which leads to increased pollution abatement and lower emission intensity.
Finally, in the globalized economy, analysis shows that the full impact of international
trade on the environment is the aggregate or sum of the trade-induced scale, technique

and selection effects.

In the subsequent Chapter Three, I propose a trade-environment framework using
the assumption of a constant elasticity of substitution (CES) utility function. This
assumption differentiates the CES model from the previous framework developed in
Chapter Two. The non-homothetic, additively separable utility function in Chapter Two
is replaced by the homothetic CES utility function. The main advantage of using the CES
utility function is that it offers tractability in analysis. The model yields different results:

when there are many firms in the industry (n is large), the CES assumption leads to a

constant markup of price over marginal costs. It also leads to a constant level of output

under the assumption of zero profit in the long run equilibrium.

In the CES framework, comparative statics effects of stricter environmental
policy in the form of higher emissions tax rate leads to an increase in the level of
abatement. This implies a negative technique effect. However, a change in emission tax
rate is found to have neutral effects with respect to scale and selection. On the other hand,
and interestingly, a higher emission tax rate is shown to lead to a decrease in net output,
which is the output allocated for the purpose of consumption. This result is consistent
with the negative technique resulting from higher emissions tax rate. One explanation is
that an increase in the use of output as a resource into abatement activities implies a

decrease in the allocation of output that can be marketed as consumption goods.

In the open economy setting, free trade is shown to have no effect on product
price or on the scale of production. Further, the analysis shows that under both the
autarky and open trade settings, the environmental impact of pollution-intensive

production is neutral with respect to the scale, technique and selection effects.

Furthermore, in the analysis of Chapter Three, I am able to show that in the case
of an endogenous emissions tax rate, an increase in wage rate or income level, raises the
emissions tax rate. This result conforms to the theoretical expectation that an increase in
income leads to more stringent environmental policy. The result is also consistent with
the hypothesized Environmental Kuznets Curve (see Grossman and Krueger 1993) and

the pollution haven hypothesis.

In summary, the two models developed in Chapter Two and Chapter Three show
that, on one hand, different forms of utility functions lead to diverging results in terms of
how intra—industry trade affects the environment. On the other hand, both frameworks
show that the full impact of pollution-intensive production is the sum of three
environmental effects, namely, the scale, technique and selection effects. Further, the
analyses in Chapter Two and Chapter Three show that the aggregated effects of trade
cannot be determined unambiguously. In other words, theoretical analysis does not
provide a definite answer to the question of how international trade affects environmental
quality. However, the analytical results and theoretical predictions can be used to

conceptualize an empirical model of the trade-environment linkage.

In Chapter Four, I undertake an empirical analysis to attempt a quantification of
the effects of international trade on the emissions levels for countries in the Organization
for Economic Cooperation and Development (OECD). Member countries in the OECD
are known to engage intensively in the trade of differentiated goods. I distinguish my
empirical investigation from previous studies by assuming that most countries engage in
both intra— and inter-industry trade. Therefore, the empirical specifications in Chapter
Four include variables that explain both types of international trade, that is, both inter-
and intra-industry trade. Moreover, the analysis recognizes that data considerations are an
important part of the reasons why empirical specifications need to control for variables
that influence the environmental impact of trade in both homogeneous and differentiated

goods.

In the empirical essay, I posit the following hypotheses concerning the trade-

environment relationship: the quality of the environment is increasing in four

10

environmental effects of trade, namely, the scale, technique, selection and composition
effects. To test the hypotheses, I adopt empirical models that can capture the data
generating process of trade that is driven by both cross-country relative factor

differentials and by cross-country preferences for differentiated goods.

Regression results show there is evidence to support the postulated empirical
relationships. In particular, I show that there is strong evidence to suggest that the
selection effect is a relevant variable in the estimation of the full impact of international
trade on environmental quality. Statistically significant results are consistent across six
empirical models for three types of pollutants: sulfur oxides, nitrogen oxides, and volatile
organic compounds. This study is the first to test the environmental effects of intra-
industry trade, and is the first to provide evidence of the importance of controlling for the
environmental selection effect. Further, I am able to show strong evidence to suggest that
the scale and technique effects can be explained by the assumptions of homothetic
production and homothetic consumption functions. These findings are consistent with the
framework of intra-industry trade. Equally interesting, the estimations also indicate
evidence to suggest that the composition effect can be explained by assumptions
consistent with the framework of inter-industry trade. Hence, the analysis support the

empirical proposition that both intra- and inter-industry trade affect environmental

quality.

11

1.2 Contribution

The contributions of this dissertation include the following. In Chapter Two, I
develop a trade-environment model describing the relationship between intra-industry
trade and environmental quality. In the autarky case, I show that the effect of pollution-
intensive production of differentiated goods can be decomposed into scale, technique and
selection effects. Further, I show that stricter environmental policy increases product
price and decreases consumption level, but is ambiguous in terms of its aggregated
effects on environmental quality. The aggregate impact of a change in environmental
policy is shown to be the sum of policy-induced scale, techniques and selection effects. In
the open-economy case, I show that the total impact of intra-industry trade is the sum of
trade-induced positive scale effect, negative technique effect and negative selection

effect.

In Chapter Three, I build a CBS model of pollution and intra-industry trade. In
the closed economy case or autarky, I show that a more stringent environmental policy
leads to greater abatement activities but is neutral with respect to the scale of production
and the number of firms. In the open economy case, I show that the impact of intra-
industry trade on environmental quality is neutral with respect to abatement levels, the
scale of production and the number of firms. This finding implies that in the CES model,

intra-industry trade has no detrimental effects on the environment.

In Chapter Four, I propose an empirical model based on the frameworks
developed in the previous chapters. The model tests the theoretical predictions of the

integrated environmental effects of both intra— and inter-industry trade. Empirical results

12

provide statistically significant evidence of the environmental selection effect. This
finding confirms the importance of controlling for the number of firms in the estimation
of the full impact of international trade on environmental quality. There is also evidence
to support the assumptions of homothetic production and consumption functions,
consistent with the assumptions of the pollution frameworks of infra-industry trade

developed in this dissertation.

Finally, I note that for future research, the proposed pollution models can be
extended to address trade issues such as firm-level heterogeneity. The incorporation of
heterogeneity may be used to analyze the effects of productivity differentials on
environmental factors, including abatement levels and emission intensity, in addition to
the firm’s revenue, profit level and scale of production. Such analysis may be useful in
the investigations of current environmental phenomena including the “race to the bottom”

in environmental policies, and the pollution haven hypothesis.

l3

CHAPTER TWO

MONOPOLISTIC COMPETITION, TRADE AND THE ENVIRONMENT

2.1 Introduction

In this paper, I develop a model to examine the environmental impact of
international trade using new trade theory. The framework explores the trade-
environment relationship by analyzing the structural link between pollution and economic
variables. The analysis delineates an explicit decomposition of the impact of pollution
intensive production into three types of environmental effects, namely the “scale”,
“technique” and “selection” effects. In addition, the paper presents a comparative statics
analysis of the effects of an increase in the stringency of environmental policy which
generate “policy-induced” scale, technique and selection effects. In the open economy,
the impact of intra-industry trade is shown to generate “trade-induced” scale, technique
and selection effects.

The framework developed in this paper explicitly defines and analyzes the
environmental impact of trade in a model of monopolistic competition and increasing
returns to scale. It advances new insights on trade-environment linkage as it seeks to
provide a more definitive answer to the ongoing debate of whether trade is beneficial or
detrimental to the environment. The paper contributes to current literature in the
following way. First, it develops a framework which shows the explicit decomposition of
the impact of new trade on pollution levels. The environmental impact of intra-industry
trade is made distinct and is differentiated from the environmental impact of inter-

industry trade. Second, the paper is the first to show how trade in differentiated, dirty

14

goods generates trade-induced scale, technique and selection effects. Third, the paper
contributes to policy analysis by investigating the effects of an increase in the stringency
of environmental policy on emissions level. More specifically, environmental policy is
shown to have unambiguous effects on economic variables including the level of
abatement, price level, the scale of production and the number of firms or the number of
product varieties.

The results of the analysis of this paper are the following. First, intra-industry
trade is shown to lead to four effects: one, it leads to an expansion in the firm’s scale of
production which in turn leads to an increase in the level of pollution. Thus, holding all
other factors constant, growth in the economy yields a trade-induced environmental scale
effect that is positive; two, trade implies an increase in the level of income which leads to
a negative technique effect. That is, holding other factors equal, income growth leads to
an increase in firm-level abatement activity which lowers emission intensity; three, trade
leads to a fall in the number of domestic firms, which implies a negative trade-induced
selection effect. That is, ceteris paribus, a fall in the number of domestic firm leads to a
fall in emissions level; and finally, four, the total impact of intra-industry trade on the
environment is the sum of the magnitudes of the trade-induced scale, technique and
selection effects. Second, the comparative static effects of an increase in the stringency of
environmental policy shows that an increase in an emission tax rate have the following
unambiguous effects: one, it increases the firrn’s level of abatement activity; two, it
lowers pollution emission intensity; three, it raises product price; four, it increases the
number of firms (or equivalently, the number of product varieties); and five, it decreases

the firm’s scale of production. Further, the analysis shows that the total effect of a change

15

in environmental policy is the sum of the policy-induced scale, technique, and selection
effects.

Third, this paper offers new insights which suggest that the environmental impact
of trade in differentiated goods needs to be accounted for in characterizing and in
measuring the impact of international trade on the environment. Since it is widely
acknowledged that most countries engage in both intra- and inter-industry trade,
economic factors that influence the effect of trade on environmental quality may
characterize not only the equilibrium of the production of inter-industry goods, but also
the equilibrium of the production of intra-industry goods. Trade literature suggests that
trade patterns under new trade theory are distinct from trade patterns under traditional
trade theory. In a similar vein, the environmental impact of international trade under the
“new” or “modern” trade theory needs to be distinguished from the environmental impact
of international trade under the “traditional” framework. Additionally, the current
analysis suggests that trade-induced environmental effects are influenced by economic
factors that characterize trade driven by demand aspects rather than by pricing
differentials across countries. Equally interesting, the analysis shows that policy-induced
environmental effects stemming fiom more stringent environmental regulation are
influenced by determinants that affect market structure and increasing returns rather than
by determinants that affect factor intensity in production. The aforementioned differences
between the environmental impacts of intra- and inter-industry trade suggest the
importance of identifying the distinct effects of new trade from the effects of traditional
trade. This distinction can be particularly useful in improving the precision of the

empirical modeling of trade-environment relationships.

16

This chapter is organized as follows. Section 2.2 presents a literature review,
section 2.3 describes the framework, section 2.4 presents a comparative static analysis of
the effects of changes in environmental policy, section 2.5 presents the open economy

model of trade-and-pollution and section 2.6 concludes.

2.2 Literature Review

In the last few decades, economic globalization has brought increased welfare to
trading nations (Baldwin 1992). In particular, trade liberalization in developing
economies has led to accelerated development and rapid economic growths which have
brought modernization and improvements in standards of living. While global economic
integration and income growths are shown to have increased consumer welfare, issues
have been raised pertaining to the desirability of international trade as it relates to
economic and environmental sustainability (Chichilnisky 1994; Strutt and Anderson
2000). The growing concerns of environmental degradation due to market expansions and
economic activities have led to studies that attempt to answer the question: is
international trade beneficial given the environmental consequences of economic
development and the detrimental effects of pollution-intensive production?

Studies that investigate the relationship between international trade and the
environment began as empirical investigations. The central underlying question
investigated is whether the economic benefits of international trade are counteracted by
the harmful effects of the exchange of dirty goods across borders. Early studies
investigate the costs of environmental abatement (Walter 1973) and the pollution content

of imported goods relative to export goods (Robinson 1988). Subsequent literature

17

suggests that the relationship between international trade and the environment may be
described by a number of empirical observations, including trade-environment
phenomena such as the dirty industry migration (Low and Yeats 1992) or the pollution-
haven hypothesis (Mani and Wheeler 1999), the environmental Kuznets curve (Grossman
and Krueger 1993), and the race-to-the bottom argument (Wilson 1996; Sheldon 2006).

To date, empirical investigations have generated mixed findings on whether trade
increases economic welfare given the damaging effects of pollution (Stern and Common
2001; Wheeler 2001). The relationship between income growth and environmental
quality, now known as the environmental Kuznets curve (EKC), was first described in the
seminal study by Grossman and Krueger (1993). The authors provide evidence that
environmental indicators such as sulfur dioxide concentrations increase in the initial
phase of economic grth but decrease in the later phase of development, with a turning
point at an estimated per capita income of about $8,000. While the interesting findings of
Grossman and Krueger (1993) give impetus to further and expansive research in the area,
new data and more sophisticated statistical methods in subsequent studies do not provide
conclusive findings as to the validity of the environmental Kuznets curve (see Shafik
1994; Harbaugh et al. 2002; Stern 2004).

Subsequent investigations into other trade-environment phenomena such as the
pollution haven and the race to the bottom hypotheses face similar mixed findings. The
phenomenon known as the pollution haven hypothesis postulates that more developed
nations relocate dirty industries to less developed nations (LDCs) to take advantage of lax
environmental protection as they face more stringent environmental policies at home. In

contradiction to the pollution haven hypothesis, Leonard and Duerksen (1980) find that

18

trade and investment data suggest pollution-intensive industries relocate to other
industrial countries instead of to less developed nations (LDCs). Leonard (1988)
concludes that other factors such as labor training, infrastructure and political stability, as
opposed to cost-savings from pollution regulations, play more important roles in the
relocation decisions of multinational firms. However, the recent empirical study by
Antweiler, Copeland and Taylor (2001) suggests that there is evidence to support the
pollution haven hypothesis. Since the empirical study by Antweiler et al. (2001) is based
on the predictions of a formal theoretical framework, the study provides more persuasive
evidence to support the validity of the pollution haven hypothesis.

In the investigations of the “race to the bottom” in environmental protection laws,
the evidence to support the hypothesis remains mixed. The “race to the bottom” concept,
an offshoot of the pollution haven hypothesis, suggests that developing nations compete
in lowering the stringency of environmental regulations in efforts to attract foreign
investments from more developed nations. Developed countries are assumed to seek to
relocation to pollution-intensive industries overseas to avoid stricter environmental
standards at home. Sheldon (2006) concludes that a “race to the bottom” in
environmental policy is unlikely for both small and large countries when standard
analysis of optimal intervention applies. However, the paper suggests that the relaxation
of the assumption of immobile factors implies the possibility that freer trade induces
capital flight from more developed countries with stricter environmental regulations to
less developed countries with weaker regulations. Sheldon (2006) asserts that studies

which provides empirical evidence to support the pollution haven hypothesis lends some

19

credence to the possibility of the race to the bottom in environmental policies amongst
developed nations.

One reason frequently cited for inconclusive evidence in the study of trade-
environment linkages is the lack of theoretical underpinnings to ground empirical
predictions of the trade-environment relationships (Copeland and Taylor 2003). More
recently, formal theoretical frameworks are advanced to provide basis for empirical
hypotheses. In particular, the framework developed by Antweiler et a1. (2001), based on
the Heckscher-Ohlin type trade model, provides an explicit description of the
environmental impact of inter-industry trade. An important contribution of the Antweiler
et al. (2001) framework is the formal decomposition of the environmental impact of inter-
industry trade into the scale, technique and composition effects. Although widely
recognized, the scale, technique and composition effects originally postulated by
Grossman and Krueger (1993, 1994), were never formally shown in earlier literature.
Antweiler et. al (2001) use the predictions of the formal model to derive estimating
equations for their empirical study, and find evidence to support the claim that trade is
good for the environment. More recently, Taylor and Levinson (2004) present both
theory and empirical evidence to “unmask” the pollution haven effect. The authors find
evidence to support the hypothesis that industries that have the most increases in
abatement costs experience the largest increases in net imports. Empirical findings based
on theoretical predictions found in studies such as Antweiler et a1. (2001) and Taylor and
Levinson (2004) are more likely to provide persuasive evidence to validate trade-

environment linkages.

20

While recent development in modeling trade-environment relationships
contributes to providing theoretical underpinnings for empirical tests, most analyses are
based on the traditional theory of trade. The Antweiler, Copeland and Taylor (2001)
framework is based on the traditional framework of trade which assumes constant returns
to scale and perfect competition. In contrast, theoretical frameworks that examine the
impact of trade on the environment using “modern” trade theory are relatively scarce in
literature. There is yet a formal framework to delineate the impact of intra-industry trade
on environmental quality and to examine how structural parameters in a “new” trade
model may unambiguously affect environmental quality. Furthermore, literature shows
that most studies that are based on “new” trade focus on the issue of environmental
policy rather than on describing the effects of international trade on domestic emissions
levels. In particular, strategic behavior pertaining to the effect of domestic environmental
regulation on foreign pollution levels is most typically examined. Markusen et a1. (1995)
use a two-region model to examine plant-level competition in environmental tax behavior
in an imperfectly competitive market with increasing returns to scale. Their theoretical
model shows that competition in environmental taxes may result in either driving the
polluting firm out of the market when the disutility of pollution is high enough, or that
the two regions will undercut each other’s pollution tax rate when the disutility of
pollution is low. Similarly, Haupt (2000) uses a strategic framework to examine the effect
of stricter environmental policy on research and development when consumption is
pollution-intensive. Haupt (2000) shows that when countries set environmental
regulations in a non-cooperative setting, a country benefits from stricter policy abroad

when it has high product standards at home.

21

To date, Gurztgen and Rauscher (2000) is the only study that proposes a general
equilibrium analysis based on new trade theory. The authors develop a framework based
on the Dixit-Stiglitz type model of monopolistic competition to examine the effects of
domestic environmental regulation on trans-boundary pollution. The authors show that
stricter domestic environmental standards may lead to lower emissions levels abroad.
While Gurtzgen and Rauscher (2000) present a new trade-environment model, the study
is similar to Markusen et. al (1995) and Haupt (2000) in that it focuses on the
investigation of the effects of domestic policy on environmental quality abroad. In a
subsequent study, Haupt (2006) analyzes the implications of non-cooperative
environmental policy for local production extemalities and concludes that the impact of
international trade on the environment is ambiguous. The study extends previous
investigations that examine the link between environmental policy and modern
international trade.

A more recent empirical study by Cole and Elliot (2003) offers an investigation
into the relationship between trade patterns and environmental regulations under both the
traditional and the new theories of trade. The authors find that under the Heckscher-
Ohlin-Vanek framework, there is no statistically significant evidence to support an
empirical relationship between environmental regulations and trade patterns. On the other
hand, under a monopolistic competition framework, the authors find statistically
significant evidence to support a relationship between environmental regulation and intra-
as well as inter-industry trade. The empirical equations in the Cole and Elliot (2003)
study are based on Helpman’s (1987) intra-industry trade model. One shortcoming of the

analysis is that the authors do not offer a formal framework to describe the relationship

22

between pollution level and trade variables. In short, the study does not have theoretical
underpinnings in the environmental context.

A most recent theoretical environmental-trade framework based on new trade
theory is advanced by Benarroch and Weder (2006). The authors build a two-country
model of pollution which assumes monopolistic competition and increasing returns. The
framework examines the effects of trade in intermediate products on pollution, output
level and welfare under the conditions of endogenous tax and two pollution functions.
The paper shows that international trade leads either to lower pollution in each country or
lower pollution per unit output in one country. Additionally, the authors show that intra-
industry trade leads countries to import the environmental quality of trading partners. The
framework by Benarroch and Weder (2006) provides interesting insights into the effects
of trade in intermediate goods on foreign emissions level. The analysis does not assume
trade in final goods, the focus of analysis in this paper.

A review of trade-environment literature reveals the following: one, most
investigations into the trade-environment relationships are based on the traditional theory
of trade; two, literature that studies the link between the environment and “new” trade
focus on the issue of environmental policy; three, there is no formal model that explicitly
shows the impact of intra-industry trade on environmental quality in the closed and open
economy contexts. In particular, the environmental impact of trade driven by economies
of scale has not been explicitly differentiated from the environmental impact of trade
driven by factor endowment differentials; and finally, the effects of an increase in the

stringency of environmental policy on domestic local pollutant levels have not been

23

formally shown for a model of monopolistic competition where trade involves the
production of final goods.

The main objective of this paper is to analyze the environmental impact of trade
in horizontally differentiated goods. I develop a model of trade and the environment by
incorporating pollution extemality into a “new” trade framework. The framework departs
from the traditional trade-environment models in that it assumes increasing returns to
scale and imperfect competition as opposed to constant returns to scale and perfect

competition. Further, it departs from existing models under new trade theory by assuming
I . .
pollutants are local rather than trans-boundary and that trade 13 1n final goods rather than

in intermediate goodsz. The framework developed in this paper shows that new trade
theory leads to environmental impact that can be distinguished from the environmental
impact of traditional trade theory. The current analysis derives a decomposition of the
environmental impact of trade under new trade theory which generates “scale”,
“technique” and “selection” effects. This contrasts the results in Antweiler et. a1 (2001)
which show that the decomposition of the environmental impact of trade under the
traditional trade theory yields “scale”, “technique” and “composition” effects. Hence, the
current framework shows that the “selection” effect distinguishes the environmental

impact of “new” trade from the “composition” effect, which is the environmental impact

of “traditional” trade.

 

Gurztgen and Rauscher (2000) developed a theoretical framework that examines how domestic
environmental policy affects trans-boundary pollution level abroad.

Benarroch and Weder (2006) built a model to investigate how infra-industry trade volume affects

pollution emission from the production of intermediate goods, while the current analysis focuses on final
goods.

24

Figure 2.1 and Figure 2.2 show the representations of the distinction between the
environmental impacts of new trade as opposed to traditional trade. The differences in the
theoretical assumptions between “traditional” and “new” trade theories are shown to lead
to differences in the environmental effects of international trade. The impact of trade in
homogenous goods or inter-industry trade, leads, in particular, to a composition effect.
This is distinguished from the selection effect, one effect of trade in differentiated goods,
or intra-industry trade. Figure 2.2 further shows that the combined effects of trade in both
homogenous and differentiated goods yield four environmental effects, namely, the scale,

technique, selection and composition effects.

2.3 The Model

I develop a model of pollution based on the neo-Chamberlinian-Krugman type
model of monopolistic competition and trade. The Chamberlin economy describes a
market structure characterized by the following main features: firms produce similar but
differentiated-and-imperfectly substitutable goods; firms exercise market power; firms
maximize profits; and finally, non-zero profits imply the entry and exit of firms. I extend
the standard model of Krugman (1979) by incorpOrating pollution extemality to examine
the impact of trade on environmental quality.

The analysis in this section proceeds by characterizing the closed economy

equilibrium followed by the open economy equilibrium.

25

 

INTERNATIONAL TRADE THEORY

I

 

 

 

 

 

 

Traditional Trade Theory: New Trade Theory:
Perfect Competition Imperfect Competition
Constant Returns to Scale Increasing Returns to Scale
Factor Differentials Preference for Variety
Comparative Advantage Specialization in Differentiated Goods
Specialization in Homogenous Goods

 

 

 

 

 

, I

Inter-industry Trade lntra-industry Trade

 

 

 

 

 

 

 

 

Figure 2.1: Inter-industry and Intra-industry Trade

 

 

The environmental effects Of

_ . The environmental eflects Of
Inter-Industry trade:

intra-industry trade:

 

 

 

 

SCALE SCALE
TECHNIQUE TECHNIQUE
COMPOSITION SELECTION

 

 

 

The environmental effects of
international trade:

 

 

 

 

SCALE
TECHNIQUE
SELECTION

COMPOSITION

 

 

 

Figure 2.2: The Environmental Impact of International Trade

The economy is composed of consumers, firms, and a regulatory authority.
Market structure is monopolistically competitive and production technology yields
increasing returns to scale. Firms produce differentiated goods which generate pollution
as a by-product, that is, pollution is a joint output of production. Firms have identical

technologies and produce goods with a large number of potential product varieties. Due

26

to the existence of economies of scale, each firm produces one type of product. Producers
are identical except in the design of their products. Firms are able to differentiate their
products without incurring additional costs. A large number of goods are produced such
that there is negligible effect of the price of any one good on the demand of another;
hence, one firm’s behavior is independent of the other, that is, there is no strategic
interdependence between firms.

Labor is the only factor of production that is inelastically supplied in a
competitive labor market. Income in the economy comes from wage earnings. There are
instantaneous adjustments to changes in variables, and there is perfect information.
Finally, countries are identical in size, technology and preference, and there is zero

transportation cost.

2.3.1 Demand

There is L number of consumers with identical preferences in the economy. The
utility function takes the symmetric, additively separable form where love of variety is
assumed with respect to the consumption of goods. Consumers do not derive utility from

leisure. Each consumer receives positive utility from consuming c,- , the consumption of
the ith good, but obtains negative utility (disutility) from pollution, z,-.
Products are horizontally differentiated and enter the utility function

symmetrically3. Social damage from pollution comes from the disutility imposed on

consumers. Consumers have no control over pollution. Therefore, consumers take

 

This is a standard assumption in the Krugman-type (1979/ I 980) model.

27

pollution as given. Thus, the consumer maximizes utility only with respect to the
consumption of goods.

The utility function is as follows:

n 1 n
(2.1) U=ZV(CI)‘ZZI v >0, v <0, u'>0
i=1 i=1

Let the representative consumer maximizes utility with respect to c, subject to a budget
constraint. Total income, y , is equal to wage, w , earned by the individual. Therefore the
consumer’s maximization problem is as follows:

’7
v(c,-)—Zz,- subject to y = w
I i=1

Max U:
' {Ci} i

n

n
where y = 2 pic,- , and p,- denotes the price of the i th good.

i=1
Then, the first order condition for consumer utility maximization is:

(2.2) v'(c,-)=lp,- i=l,...,n

where ,1 is the Lagrange multiplier and the marginal utility of income.

If the number of varieties is large such that the budget share of each variety is
small, the impact of price on the marginal utility of income may be ignored. In that case,
the effect of a change in price implies that the elasticity of demand for variety 1’ is the

following (see Krugman 1979):

(2.3) m..:£’fl.fl=_[l_]>0

H
dpi Ci 01'"

Following Krugman (1979), it is assumed that

(2.4) 1’1"- < 0
dCi

28

so that elasticity is increasing as we move up along the demand curve and consumption is

falling.

2.3.2 Production

Firms have identical technologies where labor is a linear function of output and
takes a particular functional form (Krugman 1979). There is increasing returns to scale
that is internal to firms with positive initial fixed costs, constant marginal costs and thus
declining average costs.

Output, q ,- , is an increasing function of labor, 1,, such that:
(2.5) li=a+flqi a>0,fl>0

where a is the fixed cost of production.

Firms generate pollution jointly and symmetrically with the production of goods.
For simplicity, assume pollution is generated in constant proportion to output production.
Pollutants in the model are local in their manner of dispersion and are uniformly released
into the environment. There is a regulatory authority which imposes emissions tax to
internalize the negative pollution extemality. In this paper, emissions tax is determined
exogenously. In response to the implementation of environmental regulation, firms
undertake emission abatement to avoid costly emission tax payments. In this model,
resources for abatement are drawn from the output that firms produce. Therefore, firms
allocate a portion of output towards abatement activity and allocate the remaining portion
for goods consumption in the market.

F

Pollution emission, z, , is the difference between potential pollution, z , and

pollution abated, 2A . Emission per unit of output or emission intensity denoted e,- , is

29

2,- /q,- . Hence, the following equation specifies the relationship between emission and
emission per unit output:
(2.6) 2" = eiqi

If each firm allocates q,“ of units of output into abatement, then net output is:

(2.7) q?” = q.-(1— 6.)
where 6, = q,“ /q,- is the fraction of output allocated towards emission control.

Since individual consumers are the workers in goods production, total labor force
in the economy is L. Then, supply of output is equal to demand given by the following
relationship:

(2.8) (1- 0,)qi = LC, 0 < 6 <1
where (1- 49,-) is the fraction of output allocated towards consumption.

I specify a functional form for emission per unit to describe the relationship
between pollution emission and output.4 Emission per unit output or emission intensity,
e,- , takes the following form:

(2.9)5 ,- = (l — 6,)“ where O < 6? <1

The parameter 6 measures the responsiveness of a change in emission levels due to a

change in the fraction of output allocated towards consumption. The elasticity is assumed

 

4
This approach allows for the explicit decomposition of the effects of intra-industry trade on

environmental quality. It is also consistent with the approach of specifying a production technology in the
Krugman (1979,1980) model.

This particular functional form for emission per unit output differs from the emissions function of

Antweiler, Copeland and Taylor (2001). The current form implies that emission release is released in
increasing proportion relative to the production of dirty goods, consistent with pollution-intensive
industries including the agricultural sector.

30

to be positive, which means that pollution is emitted in increasing proportion relative to
the production of pollution-intensive or dirty goods. If (5‘ is less than unitary, then this
implies that the percentage change in emission is less than a percentage change in the
fraction of output produced for consumption purpose. Generally, in pollution-intensive
industries such as manufacturing and agriculture, pollution is emitted at an increasing rate
with the increasing allocation of resources towards the production of goods. Therefore,
for the functional form specified above, the assumption that 6 is greater than unitary is
reasonable and has a basis in the practical sense.

The pollution tax, 2' , is taken to be high enough so that firms choose to undertake
abatement activity. It is equal to zero if the regulatory authority does not impose any
environmental regulation. In the absence of environmental regulation or zero emission
tax rate, firms have no incentive to control pollution and the analysis reduces to a model
without the considerations of abatement and pollution extemality. The model without
pollution would be identical to the Krugman (1979) model.

The profit function, denoted 7r , is the firm’s revenue less labor cost, pollution
taxes and abatement cost, is given by the following equation:

(2-10) 7’: =Pi(1—61)qI—Wa‘wflqi‘”i

Since all firms are identical, symmetry across firms implies that:

n
353

for all i

N men
n 11
59:?

II
.5.“

Henceforth, subscripts are suppressed.

31

2.3.3 Autarky Equilibrium
In the subsequent analysis, I refer the reader to Appendix A for detailed
derivations of the results stated in this paper.

The first order conditions for profit maximization with respect to q implies:

(2.11) p'(1-I9)2q+p(1—e)—wp—r(1—I9)‘S =0

By simplification and by substitution for the elasticity of demand, the following equation

is obtained:

p[l--rl;]=(—E%-)-+r(l-l9)5_l

where I] is the price elasticity of demand. Equation (2.12) shows marginal revenue, the

(2.12)

term on the left hand side, is equal to marginal cost, the term on the right hand side.
Marginal cost is the sum of (i) the marginal cost of production from the incremental use
of labor normalized by the fraction of output/resource allocatedtowards goods’
consumption and (ii) the marginal cost of emission per unit normalized by the
consumption fraction.

The first order conditions for profit maximization with respect to 6 is:

I 6—1
(2.13) —qp (1-6)q—qp+r6(l-6’) q=0
Substituting for the elasticity of demand and simplifying, equation (2.13) can be rewritten

as the following:

1 5-1
, l—— =6 l—t9
(2.13) P[ ,7] T( )

Equation (2.13’) shows that marginal revenue is equal to the marginal cost of per unit

emission that is normalized by per unit consumption and multiplied by the factor 6. The

32

parameter 6 is the elasticity of emission per unit with respect to per unit consumption.
Equation (2.13’) implies that the marginal revenue obtained from the incremental sale of
consumption goods is equal to the opportunity cost of resources used to generate
emission per unit. In this model, output is used as resources for two purposes, one, as
consumption goods for the product market, and two, as resources in abating emissions.
Equation (2.13’) shows that the cost of foregone emissions multiplied by a factor of 6
exactly equals the revenue that can be obtained were resources allocated away from
abatement and into goods production for the purpose of consumption.

Second order conditions for profit maximizations are in Appendix A.
Then, assuming an interior solution, the first order conditions can be solved for the
fraction of output allocated towards abatement, 6 , as:

l

__ we 3
(2.14) 6—1 [46—1)]

Equation (2.14) shows that the fraction of output allocated towards abatement is a

fiinction of wage rate, emission tax rate and the predetermined factors or parameters. It is
decreasing in wage W, the labor coefficient ,6, and in the marginal cost of goods
production, wfl. Consistent with theory, the fraction of output allocated towards
abatement is increasing in emission tax rate, r . This means that a more stringent
environmental policy leads to higher level of abatement activity. The equation also shows
that the abatement fraction of output is increasing in the elasticity of per unit emissions
with respect to per unit consumption, 5. The more elastic emission intensity is to a
change in the allocation of output towards consumption, the greater the allocation of

output towards abatement. This implies that higher emission intensity from an expansion

33

of the scale of production necessarily leads to higher abatement levels as the firm
attempts to exert greater control over the production of emission. This is consistent with
the abatement theory that firms choose to reduce emission levels to avoid higher emission
tax cost.

Since the fraction of output used for abatement is a value that is between zero and
one, equation (2.14) implies the following:

I

wfl 3
(2.15) OS[W] SI

Equation (2. 15) indicates that for the condition to hold and be meaningful, it is required

that (6 -1) ¢ 0 , that is, 6 > 1 . Then, as mentioned in the foregoing, 6 > 1 implies that

pollution is emitted at an increasing rate when greater amount of resources are allocated
towards the production of dirty goods.
By substitution of equation (2.14) into equation (2.9), emission per unit output

can be written as (see Appendix A):

(awe—[7516]

Equation (2.16) shows emission intensity decreases with more stringent environmental
policy, but increases with an increase in the marginal cost of production of output. A
more stringent policy implies that firms strive to reduce emissions level by increasing
abatement levels. This results in lower emission intensity per unit output. On the other
hand, higher marginal cost of production implies that there is an expansion of production

which leads to an increase in emission intensity.

34

The equilibrium pricing rule is obtained by substituting equation (2.14) into
(2.12), and is given as the following:

I

(2.17) P[1‘%J=(Wfi)[z(:€1)]-g[fgfil

Equation (2.17) shows that the pricing rule is a function of the elasticity of demand for

 

variety 7] , the emission elasticity parameter 6 , wage rate w , the labor productivity
coefficient )6 , and the emission tax 2' . Unlike the pricing equation in a model without
pollution (see Krugman 1979), the pricing equation in the current model indicates that
price is a function of emission tax rate. The imposition of an emissions tax yields an
equilibrium price that is multiplied by a factor of the emissions tax rate and the elasticity
of emission intensity. Hence, in the model with pollution extemality, product price is
determined not only by the marginal cost of production, but it is also determined by the
emission tax rate and the emission intensity of production.

By substituting for the value of 6 and e , I rewrite equation (2.17) as:

-1 _l
(2.17’) p=[1—;17-] (wfl)e5[ 5 ]

Equation (2.17’) indicates that given the emission elasticity parameter and the elasticity

 

of demand for variety, the pricing rule is a function of emissions per unit output such that
it is decreasing in emission intensity. This implies the higher the emissions intensity, the
lower the price of market goods. One explanation for the lower price is that there is a
trade-off between emission control and the allocation of goods for consumption. Greater

emission intensity means there is less abatement, which in turn implies that there is

35

greater allocation of output for the purpose of consumption. Therefore, holding every
other factor equal, a greater supply of goods induces a lower product price in the market.
Equation (2.17) is one of the two equilibrium conditions that determine the
equilibrium level of consumption of product varieties and the price level for the firm. I
designate equation (2.17) as the PE line. This curve is similar to the PP line in Kugman
(1979), but in the model of pollution, the PE line is a function of emission tax rate. I

rewrite the PE line as:

(2.18) or

The other equilibrium condition is the zero profit condition. Free entry with
positive profits requires that firms earn zero profit in the long run. In the current model,

the zero profit condition is given by the following equation:
(2.19) p(l—6)q—wa—wflq—r(l—6)5q=0
Divide equation (2.19) by net output, (1 — 6)q, and the wage rate, w, and then substitute

total consumption (Lc) for quantity supplied. By equation (2.14) and rearranging, the

following equation is obtained:

H [—11—]

36

Equation (2.20) is the equilibrium condition where price equals the average of the sum of
labor cost and emission cost. Designate this line as the ZE line.

Equations (2.18) and (2.20) form two equations that can be solved for the two

unknowns, fl, and consumption, 0, or alternatively, to solve for quantity of output, q.
w

In this model of pollution extemality, I solve for the quantity of output instead of

consumption for the reason that output is the more relevant variable in analyzing changes

in emission levels6.

Then, given an emission tax rate and fixed parameters in the system, I can graph
equations (2.18) and (2.20) in the price and output space; the equations are designated as
the PE and the ZE curves respectively. This is shown in Figure 2.3.

Following Krugman (1979), I assume that (177, /dc, < 0. Therefore, this means that

in a model of pollution, the PE line that represents equation (2.18) is shown to be upward
sloping, and the ZE line that represents equation (2.20) is downward sloping (see
Appendix A for proofs).

The relationship between the PE and ZE lines with respect to output level is
shown in the following equations (2.21) and (2.22) (see Appendix A for details).
Taking the total differentiation of the PE line with respect to output yields the following

equation:

‘22” M) ={‘(1-(n(c))")_2(n(c))"91-3"-}

dq 6c Oq

 

Equation (2.21) shows the PE line is upward sloping since % < 0 and SS > 0.
C q

This contrasts with the Krugman (1979) trade model which solved for consumption level.

37

Taking the total differentiation of the ZE line with respect to output yields the following:

 

(2.22) d—(ZE—) = —aq‘2M‘1
dq
i
where M =[ wfl )6
1(6 — 1)

Equation (2.22) shows the ZE line is downward sloping.

Figure 2.3 shows the diagrammatic representation of the relationship between the PE and

ZE lines and the levels of output and price.

NW
21:

PE

    

po/w i’ """""""

 

_—-—---

..D
0
£17

 

Figure 2.3: The PE line and the ZE line.

38

Using the full employment equation, I solve for the number of product varieties,
n. Then, by equations (2.8) and (2.14), and assuming symmetry, the following results is
obtained:

L: Z(a+flq,~)= n(a+flq)

i=1

(2.23) or

 

 

L=n a+ 'BLC =n a+ ’BLC
(1-19) _

JL 5

\ [45-1)] I

 

 

Solve for n such that:

__L___ ,.. I
—(a+,8q)’ — ...].
aL—l+flc[ W’B )6

 

(2.23:)

 

r(6—l)

Equation (2.23’) shows the number of product varieties is a function of total labor,
consumption level, the wage rate, emission tax rate and other predetermined variables.
For a given level of consumption, an increase in the labor force increases the number of
product varieties. On the other hand, an increase in emission tax rate decreases the

number of product varieties.

2.3.4 Decomposition of Impact: Scale, Technique and Selection Effects
In this subsection, I show the decomposition of the impact of pollution-intensive
production on total emission.

Use equation (2.6) to obtain the following:

39

n n n
(2.24) Zl =e,-q,- => 2'" =Zeiqi =28]; (LCl/(I—gi))
i=1 i=1 i=1
Note that in the closed economy, total labor L , is fixed, and with symmetry across firms,
these imply:

l

n
ZZ,‘ “zieiqi :>nz=n-e-q
i=1

(2.25) and

n n
22,- =Lzeici/(l—6,-):> nz= L-n.ec/(l—6)

i=1 1
Let nz = Z (total pollution), and rewrite equation (2.25) in differential form (hats denote

percent change) to obtain the following result7:

A

Z =
(2.26) or

+é

m)

+

3)

2=h+L+é+é—0—6)
Thus, equation (2.26) shows that the impact of economic factors on pollution can be

decomposed into the selection, scale, and technique effects in the following way:

(2.26.1) Selection effect, 1%
(2.26.2) Scale effect, S =6 or S = I: + 5 -—(1-— 6)
(2.26.3) Technique effect, é

Hence, in a model of monopolistic competition and increasing returns, the impact
of the production of dirty goods generates three types of effects, namely the scale,

technique and selection effects.

 

See Appendix A for more detailed derivation.

40

Equations (2.26.1), (2.26.2) and (2.26.3) show the environmental effects of
pollution-intensive production, holding other determinants constant. I describe the scale,
technique and selection effects as follows.

The scale effect refers to the change in emissions level due to a change in the
scale of the economy, holding other factors constant. The technique effect is the change
in emissions level due to a change in emission intensity, holding every other factor equal.
The selection effect is the change in emissions level due to a change in the number of
product varieties or in the number of firms, holding other factors constant.

Thus, equation (2.26) shows that growth in total emissions level depends on the
growth of the scale of the economy, the growth in emission intensity of production, and
the growth in the number of firms, or equivalently, in the number of product varieties.

The current model distinguishes the environmental impact of trade driven by the
demand aspect of the economy from the environmental impact of trade driven by the
supply or production aspect of the economy. While the impact of pollution—intensive
production in a perfectly competitive market can be decomposed into scale, technique
and composition effects (Antweiler et a1. 2001), equation (2.26) shows that the impact of
dirty production in a monopolistically competitive market can be decomposed into scale,
technique and selection effects. In other words, the difference between a pollution model
based on new trade theory from one that is based on traditional trade theory is that the
former yields a selection effect while the latter yields a composition effect.

Further, equation (2.26) implies that under new trade theory, the environmental
impact of trade can be broken down into finer structural decomposition. Note that under

monopolistic competition, growth of the scale effect can be fitrther decomposed into the

41

effects of growth in the factor of production, labor, L , growth in the demand or
consumption of products, 0 , and growth in the fraction of output allocated towards

consumption and exports, (1— 6). This differentiates the scale effect in the new trade

model from the scale effect as defined in the traditional model in literature.

While the new trade framework offers a more detailed decomposition, it is noted
that in empirical investigations, difficulty may arise in differentiating the effects of “new”
trade variables from “traditional” trade variables in the data. To facilitate the empirical
measurements of variables, I condense the intra-industry trade decomposition of the scale
and technique effects to parallel the conventional decomposition in inter-industry trade,
hence establishing a link between the new trade theory and traditional trade theory in the
environmental context. I also note that although the parallel decomposition is desirable, it
is not necessary. The environmental impact of trade under new trade theory can be
decomposed in a defined representation that can be explicitly separated from the
environmental decomposition under the traditional trade theory.

In addition, I make a distinction between the scale, technique and selection effects
in autarky and the “trade-induced” scale, technique and selection effects in the open
economy. The scale, technique and selection effects in autarky are driven by changes in
the structural factors in the closed economy, while the trade-induced scale, technique and
selection effects are driven by changes in structural factors caused by the opening of
trade.

Finally, equation (2.26) indicates that the decomposition of the impact of dirty
production shows that the total impact of trade on emission level is the sum of the

magnitude of each of the scale, technique and selection effects. The effects of scale,

42

technique and selection relative to each other will determine whether the joint production

of pollution will raise or lower aggregate emission in the closed economy or autarky.

2.4 Comparative Statics Effects of a Change in Environmental Policy

In this section, I analyze the implications of a change in the stringency of
environmental policy on the main variables of interest in the economy. I present the scale,
selection and technique effects as defined by the relationship between emissions tax and
the level of abatement, emission intensity, price level, level of output, and the number of

firms (or product varieties).

LEMMA 2.1: Consider an economy described by monopolistic competition and
increasing returns in the production of dirty goods. Then, an increase in the stringency of
environmental policy raises the firm ’s level of emission abatement, while a reduction in

the stringency of environmental policy lowers it.

PROOF:

Equation (2.14) shows that the fraction of output allocated towards abatement is a
function of predetermined factors and the exogenously determined emissions tax. Taking
the simple derivative of equation (2.14) with respect to emission tax rate yields the
following result:

1
2.0:- _. we 3
(2.27) fir—(6T) [1054)] >0

43

Therefore, equation (2. 27) shows that an increase in emissions tax rate or environmental

policy increases the level of abatement.

PROPOSITION 2.1: An increase in the stringency of environmental policy lowers the
firm ’s emission intensity while a decrease in the stringency of environmental policy

raises it. This is the policy-induced technique effect.

PROOF:
Equation (2.16) shows that emission intensity is a function of parameters and the
emissions tax. Take the derivative of equation (2.16) with respect to emission tax rate to

obtain the following result:

93* —1 ...izL
(2.28) Or I [t(6-l)]<0

Equation (2.28) indicates that an increase in emission tax rate reduces emission per unit
output. This result can be explained as follows. A more stringent environmental policy
leads the firm to lower emission intensity for the following reason: an increase in
emission tax rate implies that there is a trade-off between using resources for the
production of consumption goods and using resources for abating emissions. When a
higher emission tax rate is imposed, the production of consumption goods becomes a
costlier activity. Higher tax payments contribute to increasing the cost of production. To
mitigate the cost of environmental compliance, the firm reallocates output away from the
production of goods for consumption and allocates it into abating emission. As the firm
undertakes a greater level of abatement, it lowers the emission intensity per unit output.

On the other hand, a reduction in the stringency of environmental policy will have the

44

opposite effect. Lower emission tax rate means emissions abatement becomes a less
costly activity. This provides incentive for the firm to reallocate resources into supplying
goods for consumption to increase revenue. This implies an expansion of the scale of
production. Consequently, less abatement which led to expanded production leads, in

turn, to an increase in emission intensity.

LEMMA 2.2: In an economy where the production technology is increasing returns to
scale and market structure is monopolistically competitive, an increase in the stringency

of environmental policy leads to a decrease in the consumption of product varieties.

PROOF:
Comparative statics effect of a change in emission tax rate on the level of consumption is

given by the following equation (see Appendix A for detail):

I
-l _l I “—1 _|
— (1-(77) ) gr‘s w B
dc
2.29 —-—= <0
I ) dz' 1

-2 _
aLTIC‘”2 —(l—(77)‘1) (tn-'2 %’lr%"a
c

 

I
in __:vfl_5 z _6_
where 6c<0’w_[(6-1)] andB fl[(6—l))‘

Equation (2.29) shows that the numerator on the left hand side is negative while

the denominator is positive. Thus, there is a negative relationship between the

45

consumption of product varieties and the emission tax rate. An increase in emission tax
rate leads to a decrease in the level of consumption.

Intuitively, an increase in emission tax rate leads to the following effects. One, a
higher emission tax rate implies an increase in emission tax payments which increases the
total costs of production. Firms respond to the increase in costs by raising product price
to reflect the increase in resource costs of producing goods for consumption.
Consequently, consumers respond to the increase in higher price levels by reducing the
quantity demanded of each product variety. Two, when environmental policy is
tightened, firms comply with stricter regulation by increasing abatement to reduce
emission intensity. Since output is used for both consumption and abatement purposes, an
increase in emission tax rate implies a trade-off between using output for consumption
and using it for abatement. Increased abatement leads to a reduction of output available

for the goods market, thus leading to a decrease in the level of consumption.

PROPOSITION 2.2: In an economy where the production of dirty differentiated goods
entails economies of scale, an increase in the stringency of environmental policy leads to
a contraction in the firm ’s scale of production, while a decrease in environmental
stringency leads to an expansion of the scale Of production. This is the policy-induced

scale effect.

PROOF:
Comparative static effect of a change in the emission tax rate on the quantity of output is

given by the following equation (see Appendix A for detail):

46

—2
(2.30) gi=—D_l(l—q_l) n-lr—16_lpw"2271£<0
dr 6c 77

Equation (2.30) shows that an increase in the emission tax rate leads to a
contraction in the scale of production. The effect of an increase in the stringency of
environmental policy is twofold. One, the firm faces a higher cost of production due to
increased tax payments. Two, the firm complies with stricter environmental regulation by
increasing abatement levels. Both effects lead to higher production costs which shift the
firm’s marginal cost curve to the left. The firm cuts down on production which leads to a

decrease in the supply of output.

PROPOSITION 2.3: In an economy where the production of dirty differentiated goods
entails economies of scale, an increase in the stringency of environmental policy raises

the price level, while a decrease in environmental stringency lowers the price level.

PROOF:
Comparative statics effect of a change in emission tax rate on the price level is given by

the following equation (see Appendix A for detail):

 

2
__ —l
(2.31) 973: D‘lr6w‘235‘1r"q" an“ a—’73—aq" (1—77“) >0
dr 6c 77
l
6
where w = W’B and B = fl -——6——
(6—1) (6—1)

47

Equation (2.31) shows that the term on the right hand side is positive, which indicates
that a positive change in emission tax rate leads to positive change in the price level and a
negative change in emission tax rate leads to a negative change in the price level.

A positive effect of a change in the emission tax rate on the price of a product
variety is consistent with the prediction of theory. When a higher emission tax rate is
imposed, the firm increases abatement activity to mitigate the rising cost of compliance.
Increased abatement entails higher production cost, which comes from two sources: one,
from increased emission tax payments, and two, from increased resource-cost in
controlling emission level. The latter implies that the firm’s allocation of output towards
abatement activity increases and its allocation of output for consumption goods declines.
Consequently, the firm raises product price to reflect the opportunity cost of meeting

higher environmental standards.

PROPOSITION 2.4: In an economy characterized by increasing returns and
monopolistic competition, an increase in the stringency of environmental policy leads to
an increase in the number of firms, while a decrease in the stringency of policy leads to a

decrease in the number of firms. This is the policy-induced selection effect.

PROOF:
Comparative statics effects of a change in emission tax rate on the number of firms and

the output level are given by the following equations (see Appendix A for details):

dn__ Lfl dq

(2.32) —
dz' (a +ng)2 dr

48

Equation (2.32) shows that the effect of a change in emission tax rate on the number of
firms depends on the effect of a change in emission tax rate on the output level. Since the
effect of a change in environmental policy on the output level is negative, then the effect
of the change in policy on the number of firms is positive. In other words, the effect of a
change in environmental policy on the number for firms is the opposite of the effect of a
change in environmental policy on the output level.

The intuition of the overall impact of an increase in emission tax rate on emission
intensity, the price level, the scale of production and the number of firms, is as follows.
When emission tax rate increases, firms raise product prices to offset the increase in
production cost due to higher tax payments. As price level rises, quantity demanded
decreases such that the firm’s revenue falls. In addition, a higher emission tax rate
implies the firm will raise abatement activities to comply with stricter environmental
policy. Consequently, firms allocate a smaller fraction of output for consumption
purposes. This reduces the quantity of output supplied to the market, thus further raising
the price levels. As product price increases, new firms enter the market motivated by high
price levels. Hence, the number of firms in the industry rises.

Since the amount of labor is fixed, there is tight supply in the factor market.
Existing firms are not able to hire enough labor to take advantage of economies of scale,
as they attempt to lower the price of products to meet falling demand and compete with
newer firms. The increase in production costs, the decrease in revenues and the increase
in competition force the firm to move up along its average cost curves, such that the scale
of production contracts as per unit cost increases. Therefore, higher emission tax rate has

led to lower output levels. For the economy as a whole, higher emission tax rate has led

49

to higher price levels and an increase in the number of firms, or equivalently, in the

number of product varieties available for consumption.

Figure 2.4 depicts the effect of an increase in the stringency of environmental

policy on the price level and output level for the firm. An increase in emission tax rate

shifts the PE line upward and to the left to PE’, and shifts the ZE line upward and to the

right to ZE’. The price level has risen from po/w to pl/w. In this case, firms are not able

p/w

pl/w

{30/ W

ZE’ PE

ZE
PE

 

--—--‘

 

‘11

..Q
C
~91

Figure 2.4 shows that an increase in the stringency of environmental policy shifts
the PE curve upward and to the left to PE’ and it shifts the ZE curve upward and
to the right to ZE’ such that the price level rises and the output level falls.

to take advantage of economies of scale to overcome the offsetting effect of an increase

in price. Consequently, the firm cuts down on production, and the quantity of output

50

decreases from qo to q].

PROPOSITION 2.5: Consider an economy where market structure is monopolistically
competitive, production is increasing returns to scale, and firms undertake abatement in
the pollution-intensive production of differentiated goods. Then, the total effect of a
change in the stringency of environmental policy may increase, or decrease, or leave

unchanged total emissions level in the economy.

PROOF:
Totally differentiate equation (2.24) and take its derivative with respect to emission tax
rate to obtain the following:

d2... .42: 2: a.
(2.33) dr — (eq) dr +(nq) dr +(ne) dr

Equation (2.33) shows that the change in total emission due to a change in the emission

I O I O I O I C de
tax rate 18 the sum of the pol1cy-1nduced changes 1n the emlssmn 1nten51ty, ——d , the
T

number of firms, 5—" , and the scale of production, 51 , which are correspondingly the
r r

policy-induced technique, selection and scale effects, respectively. An increase in
emission tax rate is shown to lead to the following three effects: (i) it raises the firm’s
abatement level and lowers the emission intensity; (ii) it increases the number of firms in

the industry; and (iii) it leads to a contraction in the firm’s scale of production.

51

Therefore, the total impact of the policy-induced change in overall emission is the
sum of the policy-induced decrease in emission intensity, increase in the number of firms
and contraction in the scale of production.

Note that equation (2.28) shows that the policy-induced technique effect is
negatively related to the level of emissions. Equation (2.30) shows that the policy-
induced scale effect is negatively related to emissions level. On the other hand, equation
(2.32) shows that the policy-induced selection effect is positively related to emissions
level. Thus, the aggregate impact of a change in environmental policy on total emission is
contingent on the direction and magnitudes of the policy-induced technique, scale and
selection effects. In other words, the total impact may be positive, negative, or
unchanged, depending on whether the negative effects of technique and scale are larger

or smaller than, or equal to the positive effect of selection.

PROPOSITION 2.6: Consider an economy where market structure is monOpolistically
competitive, production is increasing returns to scale, and firms undertake abatement in
the pollution-intensive production of differentiated goods. Then, while the total effect of
an increase in the stringency of environmental policy may increase, or decrease, or leave
unchanged the total aggregate emission in the economy, the effect of an increase in the

stringency of environmental on the price level is to raise it.

52

PROOF:

Equation (2.31) shows that an increase in emissions tax rate increases the product price,
while equation (2.33) shows it mayincrease, decrease or have no effect on total
emissions level.

Proposition 2.6 presents an interesting result of the effects of stricter
environmental policy. The implementation of more stringent policy has ambiguous
effects on environmental quality because it depends on the aggregate or sum effects of
the three policy-induced scale, technique and selection effects. Theory cannot provide a
definite determination on how more stringent policy can ultimately affect environmental
quality. On the other hand, the cost of a higher emissions tax rate is unambiguous: it
increases product price and decreases the level of consumption of product varieties (see
Lemma 2.2). Therefore, the net effect of tightening regulation may not have the desired
benefit as intended: it imposes costs on consumption and production, and generates
uncertainty with respect to the primary objective of improving environmental quality.

Table A.2 (see Appendix A) summarizes the comparative statics effects of a
change in environmental policy on the variables of interest: abatement level, emission
intensity, price level, output level, the firm’s emission level, the number of firms (or

product varieties) and the total emission in the economy.

2.5 Trade and Pollution
In this section, I consider the environmental effects of trade for two countries with
identical preferences, technologies and factor endowments. Note that in a Heckscher-

Ohlin world, there is no reason to trade when countries are identical in terms of factor

53

abundance. In the current model, trade is driven by existing economies of scale in the
production of differentiated products; products are valued by consumers with love-for-
variety preferences. Since the goods are pollution-intensive, the opening of trade affects
emissions level in the domestic economy.

In the following analysis, the relationship between the price level and the
consumption level of product variety is used to describe the impact of trade in
differentiated goods on environmental quality.

Theoretically, in the integrated world economy, openness to trade influences L ,
the size of labor (Krugman 1979, 1980). While the impact of trade on the economy is
captured by the change in the level of consumption through a shift in the zero-profit
curve, in contrast, the impact of free trade on the quantity of output is not directly
obtained through a shift in the zero-profit curve, when L changes. The effect of trade on

the labor supply, L , can be seen in the following way.

Rewrite the ZE curve as the 22 line (see Appendix A for detail):

 

(2.35) 22: 13: “flm‘ln
W C
where
l
_ (wfl) d _ 5
w-[r(5—l) and B-fl —(6—1)

Then, the slope of the 22 line (with respect to consumption) is:

(31
W —c_2 9- < 0

dc L

Thus the ZZ curve is downward sloping.

54

Then, taking the derivative of the 22 line with respect to fixed labor, L :

a(zz) all?) -7 a

= =—L“—<O
6L 6L c

Therefore, a change in the labor population shifts the 22 curve in a negative direction.

Figure 2.5 shows the 22 and PE curves in the price-consumption space. When
two identical countries trade in differentiated, pollution-intensive goods, it is as if there is

an increase in labor supply.

p/w ll

22’ PE

po/w """"""""

pl/w """"""

 

p-—- ----

 

V

C] C0 C

Figure 2.5: With the opening of trade, it is as if there is an increase in the amount
of labor available for production. Consequently, the ZZ curve shifts down and to
the left, where quantity of consumption decreases from co to c. and the price level
falls from po/w to pl/w. Thus, the gains from trade are achieved through two
sources: one, from the decline in the price level, and two, from the increase in the
number of varieties due to imports. However, since goods are dirty, intra-industry
trade affects the emissions level in the economy.

55

The increase in labor supply shifts the ZZ curve downward and to the left to yield
both a fall in the price level and a fall in the consumption of a variety of goods. The fall
in consumption leads to the exit of firms that earn negative profit as they cannot compete
with foreign firms or products. Firms that survive in the open economy expand their
production by taking advantage of economies of scale and hiring excess labor in the
factor market. As price level falls from po/w to pl/w, there is a gain in real income, which
contributes to an increased level of economic welfare.

Hence, the effects of trade on the structural factors of the economy with pollution
are similar to the effects of trade in an economy without pollution. There is, however, a
distinguishing difference between the impacts of trade on the two economies. In a model
where dirty goods are produced, openness to trade further influences structural factors
which lead to changes in the environmental aspect of the economy.

Trade in differentiated but dirty goods leads to three environmental effects. First,
when countries engage in intra-industry trade of dirty goods, trade acts as if there is an
increase the supply of labor. Thus, open trade implies there is a greater number of
product varieties available for consumption. Imports and the competition from abroad
cause some domestic firms to exit the industry. Consequently, the number of firms in the
open economy falls. Holding everything else constant, a smaller number of firms
generate less emission into the environment. This is the trade-induced selection effect.
Second, surviving firms increase output as they expand production to take advantage of
economies of scale. All other factors equal, a larger scale of production generates a
greater level of pollution emission. This is the trade-induced scale effect. Furthermore, as

output level rises with the opening of trade, the price level falls, and real income rises.

56

Third, higher income level promotes stricter environmental policy since environmental
quality is a normal good. An imposition of a higher emission tax rate leads to greater
emission control, and thus lower emission intensity. This is the trade-induced technique
effect.

Therefore, when production entails a pollution extemality, intra-industry trade
generates the following environmental effects. Holding other factors constant: one, the
trade-induced environmental scale effect is due to the expansion of production and
economies of scale; two, the trade-induced environmental selection effect is due to the
entry and exit of firms due to increased competition from abroad; and three, the trade-
induced environmental technique effect is due to a fall in emission intensity when
environmental regulations become more stringent as income level rises.

The foregoing analysis of the impact of trade on the environment can be stated

more formally as follows:

PROPOSITION 2.7: Consider two identical economies engaged in intra-industry trade.
Then, under the conditions of monopolistic competition market structure and increasing
returns to scale technology, free trade implies there are three trade-induced
environmental effects: a positive scale effect, a negative selection eflect, and a negative

technique effect.

PROPOSITION 2.8: The impact of trade on total emission is the sum of the trade-
induced scale, selection, and technique effects. The magnitude and direction of each

effect determine the overall impact ofintra-industry trade on environmental quality.

57

PROOF:

From equation (2.26) :

(2.26) Z=a+é+q

Equation (2.26) shows that the environmental impact of pollution intensive production
are as follows: (i) it can be decomposed into the selection, technique and scale effects,
and (ii) it is the sum of the scale, selection and technique effects.

Therefore, if international trade expands the firm’s production of differentiated
goods, while the total number of firms has fallen and emission intensity is unchanged,
then the trade-induced scale effect may or may not offset the effects of trade-induced
selection and technique. In this case, total emissions may rise or fall.

In the case where international trade and scale economies allow the firm to
expand production to the extent that the trade-induced scale effect generates an
overwhelming increase in emissions level, then total domestic emissions may rise even if
there are fewer firms in the economy. On the other hand, if trade and economies of scale
allow the firm to expand production but only to a small extent, such that a reduction in
the number of firms, the trade-induced selection effect, is enough to offset the increase in
the scale effect, then total emissions in the open economy may fall. The two alternative
cases mentioned above assume there is either very little or no technique effect.

In the case where trade allows the expansion of production such that an increase
in income is high enough to provide incentive for the regulatory authority to raise the
stringency of pollution policy, then firms will undertake greater abatement level and

induce a technique effect. In the case where the trade-induced technique and selection

58

effects offset the trade-induced scale effect, trade is good for the environment as it leads
to a fall in the level of emissions.

Since there are various alternative possibilities of the combinations of the trade—
induced scale, technique and selection effects, the analysis suggests that the question of
whether intra-industry trade increases or decreases the total level of emissions is an
empirical one.

I will explore the implications of intra-industry trade on environmental quality by
testing the predictions of the pollution model in an empirical analysis in a subsequent
chapter.

Table A.3 (see Appendix A) summarizes the effects of intra-industry trade on the
economic variables of interest: the abatement level, emission intensity, the price level, the
scale of production, the number of firms (product varieties), and the total emission in the

economy.

2.6 Conclusion

The framework developed in this chapter shows that the pollution intensive
production of differentiated goods yields scale, technique and selection effects. The
selection effect distinguishes the environmental impact in a model of monopolistic
competition and increasing returns, from the composition effect in a model of perfect
competition and constant returns.

Comparative statics analysis shows that a change in the stringency of
environmental policy generates unambiguous changes in economic variables. In

particular, stricter regulation raises firm-level abatement activity, lowers emission

59

intensity, raises product prices, lowers output level, and raises the number of firms in the
economy. In addition, it lowers the consumption of product varieties when more stringent
policy leads to higher price levels.

The total effect of a change in environmental policy depends on the “policy-
induced” scale, technique and selection effects. The analysis shows that while a more
stringent regulation increases product price, its effect on emissions level is ambiguous
and depends on the sum of the magnitudes of the policy-induced environmental effects
of scale, technique and selection. Hence, surprisingly, this result implies that the
environmental effect of more stringent environmental policy does not necessarily lead to
a decrease in emissions, but that it explicitly imposes a welfare cost on consumption by
increasing product price and decreasing consumption levels.

In the open economy, analysis shows that the environmental impact of intra-
industry trade can be decomposed into “trade-induced” scale, selection and technique
effects. First, free trade is shown to lead to the expansion of the production of dirty
goods, a positive trade-induced scale effect that increases emissions. Second, competition
from abroad leads to a fall in the number of domestic firms, a negative trade-induced
selection effect which reduces emissions level. Third, if income rises with the increase in
production, this leads to a negative trade-induced technique effect arising from the
income effect of stricter regulation. The negative trade-induced technique effect lowers
emissions level. Therefore, the total impact of intra-industry trade on environmental

quality is shown to be the sum of the trade-induced scale, technique and selection effects.

60

CHAPTER THREE

INTRA-INDUSTRY TRADE AND THE ENVIRONMENT

3.1 Introduction

In the last few decades, globalization has given rise to environmental concerns as
emerging economies expand dirty sectors to attain high income growths. While
international trade augments consumptive welfare, the consequences of pollution
intensive production are said to incur costs on overall trade effects. Current and
vigorously heated debate on how trade affects environmental quality remains ongoing. To
date, literature suggests mixed findings in the determination of whether trade is beneficial
or detrimental to the environment.

Presently, most empirical works that explore the environmental effects of
international trade are based on the traditional trade theory of comparative advantage and
cross-country factor differentials (see Antweiler et al. 2001; Copeland and Taylor 2003;
Frankel and Rose 2005). In contrast, “new trade” theory suggests the demand aspect of
the economy plays an important role in shaping trade patterns (Krugman 1979; Helpman
1987) where consumer preference for product varieties implies specialization in
differentiated goods and market expansions across borders.

Given growing environmental concerns and trade policy issues, an important
question to ask is, how does intra—industry trade in final goods affect the environment?

In this paper, I develop a trade-environment framework based on the Krugman

(1980) model of monopolistic competition and increasing returns to scale. I explore the

61

environmental impact of trade in differentiated goods under the assumption that the
consumer’s utility function has the property of constant elasticity substitution (CES).
The assumption of CES preference in this paper yields different results than those
obtained under the assumption of the more general, non-CES utility function in the
framework developed in Chapter Two.

There are two major motivations for using a CBS utility function to model the
trade-environment relationship. One, trade literature suggests that while the selection
effect of intra-industry trade is empirically significant, the scale effect is very small (see
Head and Ries 1999, 2001). One possible reason for the negligible scale effect is that
goods have constant demand elasticity. Utility functions that exhibit constant elasticity of
substitution give rise to constant demand elasticity, thereby providing an explanation for
a small scale effect. In the context of trade and pollution, the absence of a trade-induced
change in the scale of production implies the absence of a trade-induced environmental
scale effect. Therefore, this result presents an alternative outcome of the impact of intra-
industry trade which contrasts with the result obtained in Chapter Two. Consequently, an
implication of this finding is that divergent empirical results which reflect the different
outcomes can be interpreted based on distinct underlying theoretical assumptions.

The CBS model is the more dominant framework of intra-industry trade. This is a
second and equally important motivation for using a CBS utility function in modeling
pollution and new trade theory. There are at least two reasons for the more mainstream
usage of a CBS framework. One, the CES assumption provides greatertractability in
analytical modeling, and two, the CES framework lends more easily to the development

of extensions to the model. An interesting extension of the CES framework is the

62

incorporation of heterogeneity in firm-level productivity (Melitz 2003). In the context of
a pollution model, an application of firm-level heterogeneity is the investigation into the
firm’s abatement activity with regards to trade and environmental policy concerns. More
specifically, a CBS trade-environment framework may allow the examination into the
question of whether the surviving firm that engages in trade, shown to be the more
efficient firm (Melitz 2003) may also be the cleaner firm. The CBS model developed in
this paper allows extensions to address such heterogeneity issue.

The contributions of this paper are as follows. One, it develops a pollution and,
intra-industry trade framework assuming homothetic preferences. Two, it offers a
comparative statics analysis of the effects of environmental policy changes on economic
variables of interest. More specifically, the effects of a more stringent policy are
examined under the assumption of both ”exogenous and endogenous emission tax rates.

This paper shows the following results. First, in a model of pollution and trade
where preferences are represented by a CBS utility function, there are no trade-induced
scale and selection effects. However, if trade increases income level, then the resulting
trade-induced technique effect implies a reduction in emission intensity. Therefore, an
unambiguous implication of these outcomes is that intra-industry trade does not harm
environmental quality. Second, a change in environmental policy leads to changes in the
level of abatement undertaken by the firm, but does not yield any change in the firm’s
scale of production or in the number of firms in the economy. In other words, while there
is a policy-induced negative technique effect, policy is neutral with respect to scale and
selection effects. An implication of these results is that if the price elasticity of demand is

constant, environmental policy imposed on the production does reduce emission

63

intensity, but does not contract the quantity of dirty output or affect the size of the
industry engaged in pollution-intensive production.

The remainder of this paper is organized in the following way. Section 3.1
describes the theoretical framework and characterizes its equilibrium. Section 3.2
provides a comparative statics analysis of the effects of changes in environmental policy,
section 3.3 presents a bilateral trade model of pollution and trade, section 3.4 presents the

discussion and section 3.5 concludes.

3.2 The Model

The Krugman-Dixit-Stiglitz (1980) trade model is extended to incorporate
pollution extemality: market structure is described by monopolistic competition and
production technology is increasing returns and internal to firms. Pollution is jointly
generated with goods production where pollutants are locally and unifonnly distributed.
For simplicity, in this section, I assume emission tax rate is determined exogenously.
This assumption is relaxed in section 3.3 in the open economy case.

Notations are as defined in Chapter 2 and I omit redefining them in this chapter.

3.2.1 Demand
Consumers face a constant elasticity of substitution (CES) utility function where
the variety of goods and pollution level enter the utility function symmetrically. The

utility function is as follows:

n n
(3.1) U:Zc,P-Z¢z, 0<p<1,(0>0
i=1 i=1

64

The preference parameter ,0 has a value that is between zero and one to reflect the love-
of-variety preference (see Dixit-Stiglitz 1977). The parameter (0 represents the marginal

utility of the damage from pollution.
Consumers have no control over pollution and take pollution as given. Thus, the

consumer’s maximization problem is as follows:

I? n
(3.1’) II{/Ia}x Zcip - Zgoz, subject to y = w
Ci 1:1 1:1

n
where y = 2 pic,-
i=1

The first order condition for consumer utility maximization is:

‘olsa

n
(3.2) Zcf’" = p,- i=1,...,n
i=1

where 2. is the Lagrange multiplier and the marginal utility of income.

3.2.2 Production

Firms maximize profit by taking revenue less factor (labor) costs of production
and less the cost of emissions tax payments made to the government. As in Krugman
(1980), labor is the sole factor of production in the economy and there is fixed cost.
Labor is a linear function of output that takes a particular functional form such that:

where a is the fixed cost of production.
In the presence of a regulatory authority, emissions tax is high enough so that

firms choose to undertake abatement to reduce pollution emission. Firms in the model

65

allocate a fraction, 6 , of output as input into the abatement process. Denoting 2i (61') as
net emission after abatement, the relationship between emission and emission per unit
output is:

(3-4) 21 = €in

Net output is:

(3.5) q?“ = q.-(1—6,-)

Thus, the supply and demand of goods is given by the relationship:

(36) (L—@)qp=Lq

where L is total labor force, and (l — 6,) is the portion of output allocated towards

consumption.

The functional form for emission per unit output, e,- , is specified as the following:

07) q=0-qf 5>1

Denote It as profit, which is the firm’s revenue less labor cost, pollution taxes and
abatement cost such that:

(3.8) 7r,- =p(l—6,-)qi—wa—w,6q,--rz,-

All firms are identical. Henceforth, subscripts are suppressed.

3.2.3 Autarky Equilibrium
The first order condition with respect to output can be written as the following

(see Appendix B for details):

__l__ _ 64: Wfl
as) p@ alto 0) (Le)

66

 

where 77 is the price elasticity of demand for net output.

The first order condition of profit maximization with respect to 6 , the fraction of
output allocated towards abatement, is given by (see Appendix B for details):

l

(3.10) 6=1—[ wfl )5

 

r (6 — I)
The interpretation of equation (3.10) is identical to the interpretation in the previous
Chapter Two and is omitted it here.

Solve for emission intensity by substituting equation (3.10) into equation (3.7) to '

obtain the equilibrium emission intensity, given by the following:

(3.11) ”[4:60]

Substitution of equation (3.10) into equation (3.9) yields the equilibrium pricing

 

rule, which is (see Appendix B for details):

1
1 6

(3.12) Fifi-31.)“ [05—11(54)]5

The pricing rule is determined by the preference parameter p, the emission elasticity
parameters, 6 , the wage rate w, the labor productivity coefficient ,6, and the emission
tax 2', imposed by the government. Equation (3.12) shows that price is increasing in the

marginal cost of production of goods and increasing in the environmental tax rate.
From the zero profit condition, the firm’s output level can be determined. Free
entry and exit of firms implies that the zero profit condition is given by the following

equation:

67

(3.13) p(l—6)q—wa—wflq-—r(l—6)6q=0
Then, by substitution of the equilibrium value of 6 , and rearranging, the quantity of

output is obtained as follows (see Appendix B for details):

6 —1
(3.14) q 2 $14
I35 (1 - p)
Equation (3.14) shows that quantity produced depends only on the fixed parameters in the
system. Output is an increasing fimction in fixed cost, the preference parameter and the

per unit consumption elasticity of emission per unit; it is decreasing in the inverse of the

labor coefficient.

Then, net output, (1— 6)q , is the following (see Appendix B):

(3 15) net : (Kg); 000(6-01—1/6
' q 1 [35(1- p)

 

From the demand and supply equation (3.6), the equilibrium level of consumption
demand is obtained. Substituting for the equilibrium level of abatement indicated by
equation (3.10) gives the following equation:

1

Equation (3.16) shows that consumption level is decreasing in labor, L. An expansion in
the population force implies a reduction in consumption as the number of goods available
for consumption decreases for each consumer. Equation (3.16) also shows that there is an
inverse relationship between consumption level and emission tax rate; consumption is

decreasing in the stringency of an emissions tax.

68

The full-employment condition is:
n
0-”) L=Z(a+flq,~)
i=1
Use the full employment condition (3.17) to solve for n , the number of product varieties,

to obtain the following (see Appendix B):

L6(1—,O)

(”8’ "insomniac—111

 

Equation (3.18) indicates that the number of firms or product varieties is increasing in
labor, L. An expansion in the labor force implies an increase in resources or factors of
production. This raises the scale of production and the number of product varieties, or

equivalently, the number of firms operating in the economy.

The foregoing results in equations (3.10), (3.14) and (3.18) are summarized by the

following proposition.

Proposition 3.1: Consider an economy where market structure is monopolistic
competition, production is increasing returns to scale and consumer preference is
represented by a constant elasticity of substitution (CES) utility function. Given an
emission tax rate, and an elasticity of emission intensity per unit consumption larger than
unitary such that 5 > 1, then, (i) the firm ’s abatement level is a fixed proportion of
output, and (ii) the scale of production and the number of firms in the industry are

independent of price and abatement levels.

69

3.2.4 Scale, Technique and Selection Effects:
Following the analysis is the previous Chapter Two and from equation (3.4), I

define the scale, technique and selection effect as follows:

A

(3.19) Z=h+é+q

where 2 is the percent change in total emission, h is the percent change in the number of
firms (or equivalently, in the number of product varieties), e is the percent change in

emission per unit output, and q is the percent change in the level of output.

Equation (3.19) shows that growth in total emission is the sum of growths in the
number of firms, emission intensity and scale of production, which are, respectively, the

selection effect, the technique effect, and the scale effect.

3.3 Comparative Statics Effects of Stricter Environmental Policy

Changes in environmental policy can have wide ranging implications on
consumer welfare as well as on economic performance. If greater environmental quality,
for example, in the form of cleaner air is desirable, then, stricter environmental regulation
can be implemented to attain it. However, the imposition of more stringent policy affects
the firm’s production decisions in the consideration of rising compliance costs. Hence, a
trade-off exists between the achievements of environmental objectives and economic
objectives. In this section, I analyze the effects of an increase in the stringency of
environmental policy. More specifically, the comparative statics effects of an increase in
emission tax rate are examined for the following variables: the fraction of output
allocated towards abatement, the price level, the levels of output and net output, the

consumption level, and the number of firms in the industry.

70

3.3.1 Abatement

Theoretically, an increase in pollution tax is expected to increase the fraction of
output used for abatement. For the cost-minimizing firm, it is efficient to equate marginal
abatement cost to the tax rate imposed on emissions (see Baumol and Oates 1988;
Hanley, Shogren and White 1997). Thus, if emission tax rates were to increase, then the
marginal benefit of reducing emissions level increases.

From equation (3.10), I can show the effect of a change in emission tax rate on the

level of abatement level as the following:

SE = [[53] [i]; > 0

Or (6—1)

The result shows that the direction of the relationship between a positive change in
emission tax rate and the fraction of output allocated for abatement is positive. Hence, an
increase in emission tax rate increases the fraction of output allocated for abatement.
Intuitively, in the CES model of intra-industry trade and pollution, the increase in output
allocation for abatement implies that the opportunity cost of allocating output towards
producing goods for consumption has increased. The increase in the environmental tax
rate forces the firm to make a decision on whether to increase the allocation of output for
pollution abatement and reduce emission tax payments, or to increase the allocation of
output for consumption to increase revenue. The firm will increase the level of abatement
as long as the marginal benefit of the incremental abatement level equal the marginal coSt
of an incremental increase in emission tax rate. When the emission tax rate is raised,
equating the marginal benefit of reducing emission equal to the marginal abatement cost

means it pays to increase abatement activity; hence, the allocation of output for

71

abatement purposes increases. In other words, more stringent pollution regulation acts as

an incentive for the firm to allocate greater amount of resources for abatement purposes.

3.3.2 Emission Intensity
From equation (3.11), the effect of a change in emission tax rate on emission

intensity is given by the following:

§3__ -2 W6
cf T [(5-1)]<0

This result shows that an increase in the stringency of environmental policy leads to a

 

lower emission per unit output. An increase in emission tax rate gives firms the incentive
to allocate more resources towards reducing pollution. This result is consistent with the
theoretical expectation that a more stringent environmental policy leads to the firm’s
decision to increase its abatement capacity. Further, since the level of abatement is
increasing in emission intensity, any factor that increases the level of abatement implies a
reduction in per unit emission. Notably, an increase in wage rate or income which leads
to an increase in abatement levels will in turn lead to a decrease in emission intensity.

This is the policy—induced technique effect.

3.3.3 Price
Equation (3.12) shows that a change in emission tax rate on the product price is:
a 1:5 1
—p = r 5 (aw-1))"l [(WWH (5-1)}3 > 0
That is, a change in the environmental tax rate affects the price level in a positive

direction. Hence, an increase in emission tax rate leads to an increase in the price level.

72

When firms have to make tax payments to internalize the costs of pollution, the profit
maximizing firm will equate marginal revenue to marginal cost that is now inclusive of
the marginal tax cost (see Baumol and Oates 1988). Hence, the increase in marginal cost
is reflected as an increase in the price of goods.

In the current model, the increase in the price level due to an increase in emission
tax rate can be explained in the following way. Am emission tax rate increase implies that
the firm increases abatement activity to lower emission intensity and to reduce tax
payments. Consequently, there is a trade-off between using resources for the production
of consumption goods versus using the same resources for undertaking abatement. An
increase in pollution which entails stricter environmental regulation constitutes a resource
cost — more abatement is required to deal with the emission increase which in turn diverts
resource use from the purpose of consumption goods’ production. The resulting scarcity
of resources in the goods production is reflected in an increase in the price level. The
greater the change in emission intensity due to greater abatement, or in other words, the
greater the change in the fraction of output allocated towards abatement, the higher is the

increase in the price level.

3.3.4 Consumption
Consumption level decreases with the increase in emission tax rate. Equation
(3.16) shows the effect of a change in emission tax rate on consumption level as follows:
I

6 (1+6) ,6 g
1:- _ a _W__ -l
at 1' [(6—1)] L q<0.

 

73

The result shows that a stricter environmental policy induces a reduction in the
consumption of goods. Intuitively, there are two reasons for the resulting decrease in
consumption. One, a higher emission tax rate implies it pays the firm to allocate a greater
amount of resources (output) towards the abatement of emissions. Since total output is
used for both abatement and consumption, then the trade-off between allocating a
fraction of output towards abatement and the remaining fraction towards consumption
implies that the more stringent the environmental policy, the less the amount of output
allocated for consumption purposes. Thus, this means a higher emission tax rate leads to
lower consumption levels. A second reason for decrease in consumption is that higher tax
payment implies higher marginal cost of production, which leads to an increase in
product price. The increase in the price level leads to a decrease in the quantity of goods

demanded.

3.3.5 Net Output
Equation (3.14) indicates that equilibrium net output is an inverse function of
emission tax rate such that:

1+6 1 _
all]: = 4(7) (Wfl)6 ap(5 — 1)1 V5

 

The result implies that an increase in emission tax rate leads to a reduction in net output
produced for consumption purposes, while a decrease in emissions tax rate leads to a rise
in the level of net output. There is a trade-off between using output as resource for
abatement versus using it for increasing revenue from sales of consumption goods.

Therefore, the higher the emissions tax rate, the more resource or output is required for

74

abatement purposes, consequently leaving less net output produced for the purpose of
goods consumption. In contrast, if tax rate were lowered, a greater fraction of output
would be allocated for the purpose of filling market demand for consumption, hence net

output increases with a decrease in emission tax rate.

3.3.6 Output
Equation (3.16) shows that total output produced in the economy is not a function

of the emissions tax rates.

Therefore, 91 = 0.
at

In this model, a change in the stringency of environmental regulation has no effect on the
total level of output produced. This outcome is consistent with the assumptions of a CBS
utility function and a constant elasticity of demand which leads to the result that the scale
of production that is independent of price and abatement levels. Hence, notably, the zero
effect of emission tax rate on the scale of production means environmental policy is

neutral with respect to total output. This is the policy-neutral scale effect.

3.3.7 Number of Firms
The number of firms operating in the economy is independent of emission tax rate
as shown by equation (3.16). A change in emission tax rate on the number of firms is

given by the following:

75

In the model with CBS utility function, the equilibrium number of firms is not affected
not a function of emission tax rate. Thus, a change in the stringency of environmental
policy has no effect on the number of firms, or equivalently, on the number of product
varieties. Therefore, the pollution model of monopolistic competition generates a policy-

neutral selection effect.

The results of the foregoing comparative statics analysis of the effects of environmental
policy on the firm’s abatement level, product price, emission intensity, the level of
consumption, net output, total output and the number of firms are summarized by the

following proposition.

Proposition 3.2: Consider an economy that engages in the production of differentiated,
dirty goods. Preferences are homothetic with constant elasticity of substitution. Then, an
increase in the stringency of exogenous environmental policy leads to (i) an increase in
the firm ’s abatement level; (ii) an increase in the price of product varieties; (ii) a
decrease in emission intensity; (iv) a decrease in the level Of consumption; (v) a decrease
in net output, (vi) no effect on total output; and, (vii) no effect on the number of firms in
the economy. Notably, a stricter environmental policy induces a negative technique

effect but generate policy—neutral scale and selection effects.

3.4 Bilateral Trade
In this section, I consider a two-country world where both economies are identical

in production technology and preferences. Constant elasticity of substitution between

76

goods is assumed. The model shows the following results. One, the number of domestic
firms, the level of consumption, the level of output, the level of abatement and the price
level are shown to be functions of predetermined factors. Two, welfare gain from trade
comes from the increase in the number of product varieties available to consumers
through imports. Three, there are no trade-induced scale, technique or selection effects.
The world consists of two countries, Home (H) and Foreign (F), engaged in the
trade of dirty goods. Monopolistic competition market structure prevails, characterized by
technology that is increasing returns to scale and is internal to firms (Krugman 1980).
Consumers have Dixit-Stiglitz type preferences, where a CBS utility function is
maximized subject to a budget constraint. Pollution that is jointly generated with dirty
goods production is locally and uniformly released into the environment. A regulatory
authority levies an emissions tax to internalize the negative extemality of pollution. Firms

maximize profits taking into account the emission tax rate.

3.4.1 Consumption

Let i =1,...,n index the number of varieties produced domestically, and

I = 1,..., n‘ index the number of varieties produced abroad. Small letters denote Home

variables, while letters with asterisks denote Foreign variables, with the exception of the

index of the number of product varieties, where small letter denotes Home and capital

letter denotes Foreign. Demand or consumption at Home (Foreign) is denoted by C(C*),
abatement level by 6(6* )output level by q (q... ) , price by p( p* ) , imports by m (m* ) ,

total income by y(y*) , wage by w(w* ) , and emissions tax by r (r. ).

77

The representative consumer’s utility function for the Home country (and

similarly for the Foreign country) is a function of the consumption of product varieties

produced at home (c,) and imported varieties (m I) produced abroad, and a disutility of

pollution extemality associated with production of domestic products (2,- ). The

preference parameter, p, and the elasticity of substitution between any two goods,

0' = (1- p)_] >1, are assumed identical for both countries.

The utility functions for Home and Foreign are as follows:

n n* n
(3.20) U” = cp+ mp- (02- O<p<l
I I I
i=1 [:1 i=1
F n* ... n 1: n"‘ *
(3.21) U =Zcf+2mf— (pz,
[:1 i=1 [:1

where (p is the disutility of emission.

The budget constraint is total income composed of wage earned, w such that

w = y (and similarly for Foreign). Then, the budget constraint implies:

n n*
(3.22) y = Epic.- + 2 10 1m 1
i=1 [:1
at: "I“ III 11: n
(3.23) y = 2 p [c 1 +Zp1m.
[:1 i=1

Henceforth, I will write results for the Home country, keeping in mind that the Foreign
country will have similar results.
Following Dixit and Stiglitz (1977), and as in Gurtzgen and Rauscher (2000), the

levels of consumption are given by the following equations:

78

 

 

[’1’
(3.24) c- = -
’ "pp/(p-l)+n*p*e/(e-l) y
* I/(p—I)
(3.25) m, — p ’

._ ,5 - _ cy
"pp/(p l)+,,*p*et(p 1)

3.4.2 Production

Since Home and Foreign are identical in terms of endowments and technology,
supply side conditions for the Foreign country mirror domestic supply (as in Home
autarky conditions). Home country equilibrium conditions are characterized in the
previous Chapter Two, while Foreign country equilibrium conditions are given below:

I

t

*_ — W3" ,5
(3.26) e -1 “1*(531)

 

(327) pt =_1T[6*5:1][(Wtfl*)5 -IT¢ (6¢_1):l5*

 

 

 

 

(3.28) q =fl,5,(l_p,)
L . . . 1-1/3*
(3.29) q*ne1=(w‘fl*]a* " p (5 ‘1)
_ ra- fl*6* (l-p*)
(3.30) n"' = L5 (l-p )

Proposition 3.3: Consider two identical economies engaged in bilateral intra-industry
trade. Then, in the open economy, when the price elasticity of demand is constant, firm-
level abatement, scale of production and the number of firms are unaffected by free trade.
Consequently, there are no trade-induced environmental scale, technique or selection
effects. However, fiee trade implies as if there is an increase in domestic labor force,

which increases the number Of product varieties available for consumption.

PROOF:

Equation (3.18) implies fin/6L > 0 so that an increase in the labor force leads to an

increase in the number of product varieties. In a two-economy world, the integrated

world economy implies as if there is a doubling in the labor force (Krugman 1980), such

that the total number of varieties available for consumption is N = n + n*. This leads to a
trade-induced selection effect in terms of consumption. Further, equation (3.16) implies

that consumption of domestic product varieties decreases with an increase in labor force,
. 6c . . . . . .
that 13, 5-1: < 0 and the doubling of labor forces 1mp11es a reductlon 1n consumptlon of

domestic product by half. However, without an actual change in the labor force so that
dL = O , there is no change in the scale of production or in the number of firms in the
economy. Therefore, there are no trade-induced environmental scale, technique or

selection effects.

3.4.3 Pollution Supply and Emission Tax
In this section, I solve for pollution supply as determined by the price of pollution

emission (Antweiler et al. 2001). I assume that the regulatory authority levies an

80

emissions tax to internalize pollution extemality. Following Gurtzgen and Rauscher
(2000), the consumer’s utility function is optimized with respect to emission tax rate.
Substituting for equilibrium values, welfare is maximized given the production constraint
and the full employment condition.

I note that policy instruments such as an emission tax rate which intemalizes
pollution damage, may not correct for the allocative inefficiency associated with market
power. An emission tax rate may exacerbate the social cost of a suboptimal level of
production due to monopoly pricing (Carraro 1998). On the other hand, in models of
contestable market with preferences for differentiated goods, economies of scale in
production may imply more efficient use of inputs. Therefore, greater efficiency in the
employment of factors of production and a suboptimal level of output may contribute to
reducing pollution level. In addition, consumers may associate these aspects as “green”
qualities of product differentiation. Hence, both the increase in utility from the
consumption of environmental-friendly goods and the decrease in pollution due to the
technology of economies of scale may offset some of the welfare loss due to suboptimal

levels of production and consumption.

. . . . . 8
To solve for the consumer-welfare max1m121ng emrsS1on tax rate , the demand
equations in (3.22) and (3.23) are substituted into equation (3.20) to obtain the indirect

utility function:

 

(3.31) V” = —ngoz

 

As noted, the consumer-welfare maximizing emission tax rate in a model of monopolistic competition is
a second best tax rate since it does not correct for the welfare loss due to market power and thus does not
correct for allocative inefficiency.

81

where y = w. Then, p , p *, n , and n * are as given in equations (3.12), (3.27), (3.18)

and (3.30) respectively.

Subsequently, identical preference and technology across countries implies (3.31) can be

written as:

,, p ,, p/(p-l) * *e/(e-l)
(3.32) V” =(T z) ( p M p )—ngor“[——W'B ]q
("pp/(p4) + "*p*p/(p—l))p (5 ‘1)

 

Solve for environmental policy by maximizing equation (3.32) with respect to emission

tax rate, r . The first order condition is:

 

 

((5—11

p—l p p/(e-ll * *P/(P-l)
CV” —,02' (n2) (np +np (0 _2[ wfl )q:

Or — (npp/(p-l)+n*p*p//(P“))p

Solving for emission tax rate yields the following equation:

l

(3.33) T:_[[il¢p-l[(;_fl1)]q];;

where

 

("pp/(e4) + n“ p*e/(p-1))

("pp/(e4) + "*p*e/(e-1) )p

 

Equation (3.33) is the Samuelson rule where the term on the right hand side of the

equation indicates the marginal damage of pollution for each consumer. The equation

82

shows the emission tax rate is a function of the fixed level of output (see equations
(3.13)), the marginal disutility for pollution, the numbers of product varieties produced at
home and abroad (see equations (3.18) and (3.30) respectively), the world price level, and
the wage rate.

Notably, equation (3.33) implies that an increase in the wage rate, or income,
leads to an increase in emission tax rate so that:

1

Thus, equation (3.34) shows that an increase income or the wage rate leads to an increase
in the stringency of environmental regulation. This result provides a basis for the
technique effect: higher income levels, which leads to a more stringent environmental
policy, induces greater abatement level undertaken by the firm and hence lower emission
intensity (see section 3.2). In other words, given that environmental quality is a normal
good, an increase in income implies a decrease in emission intensity through an increase

in the stringency of environmental policy.

3.5 Discussion

In a pollution model of monopolistic competition and increasing returns to scale
with CBS utility function, constant price elasticity of demand leads to the result that for a
given exogenous emission tax rate, abatement is a fixed proportion of output and is
decreasing in emissions tax rate. In addition, the firrn’s output level and the number of
firms in the economy are shown to be independent of the price and abatement levels.

Consequently, autarky equilibrium implies a policy-induced technique effect, but does

83

not generate scale and selection effects. Environmental policy is neutral with respect to
the scale of production and the size of the industry. Hence, in a model where preferences
exhibit CES, environmental policy addresses the efficiency in the abatement aspect of
pollution emission solely. These results are distinct from the results obtained under the
assumption of non-CES and non-homothetic preferences shown in Chapter Two, where a
more stringent environmental policy generates policy-induced scale, technique and
selection effects.

An important implication of the outcomes of the foregoing analysis is that
environmental policy implementation needs to take into account not only the structure of
the market, but also the elasticity of demand for goods. Policy effects in markets where
constant returns to scale and perfect competition prevail yield policy-induced scale,
technique and composition effects (Antweiler et al. 2001). On the other hand, in markets
where increasing returns and monopolistic competition prevail, policy effects yield scale,
technique and selection effects (Chapter Two). However, in the latter where the
additional restriction on demand elasticity is assumed to be constant, policy effect is
limited to the technique effect, as shown in the current analysis.

In the open economy analysis, the CES model implies the complete absence of
trade-induced environmental effects. Free trade does not yield any change in the
production side of the economy, leaving the environment unaffected by trade activities.
Therefore, when preferences for goods have constant elasticity of substitution, the result
is unambiguous: infra-industry trade does not harm the environment. This result contrasts
the findings in Chapter Two where the analysis shows that a non-CES utility function

generates trade-induced positive scale, negative technique and positive selection effects.

84

In the non-CES, more general case, the impact of intra-industry trade is shown to be the
sum of the trade-induced scale, technique and selection effects.

Finally, the foregoing analysis suggests that the question of whether international
trade in dirty, differentiated goods raises or lowers emissions level is an empirical one.
The realization of data and responses to the scale, technique and selection effects of intra-
industry trade will depend on economic factors that determine how pollution-intensive
production influences environmental quality. I explore the answer to this question in the
next chapter by undertaking an empirical analysis to attempt to find evidence in the data

and estimate the impact of intra-industry trade on emission levels.

3.6 Conclusion

The model developed in this paper offers a number of insights into the trade-
environment relationship for countries that trade in differentiated goods. First, in a
pollution model of trade where preference is represented by a CBS utility function, there
are no trade-induced scale, technique or selection effects. This result suggests that free
trade does not necessarily contribute to raising emissions level when countries engage in
pollution-intensive production. Hence, if demand elasticity is constant, the CES model of
pollution implies that intra-industry trade is good for the environment as it increases the
number of product varieties available for consumption without imposing negative
environmental effects. Dirty production is confined within domestic borders. Second,
under the assumptions of increasing returns and CBS utility functions, an increase in the
stringency of environmental policy in a monopolistically competitive industry does not

lead to any change in the scale of production or in the number of firms. There is,

85

however, a reduction in emission intensity when firms face stricter regulation. Therefore,
greater stringency leads to a policy-induced technique effect but is neutral with respect to
scale and selection effects. The important implication of this result is that the
implementation of changes in environmental policy needs to be market specific - the
effects of policy may differ across industries in accordance with differences in production

technology, market structure and consumer preference.

86

CHAPTER FOUR

HOW DOES INTRA—INDUSTRY TRADE AFFECT THE ENVIRONMENT?

4.1 Introduction

In this paper, I investigate the empirical relationship between international trade
and environmental quality for countries that engage in both intra- and inter-industry
trade. The analysis uses pollution data from the Organization for Economic Co-operation
and Development (OECD) for the years 1995-2004. Three types of pollutants are
considered, namely, sulfur oxides (SOx), nitrogen oxides (N Ox) and volatile organic
compounds (V OC). Estimating equations are based on the predictions of two main
theoretical frameworks. The first framework is the pollution model of intra-industry trade
developed in the previous chapters (Two and Three), which generates the environmental
effects known as the “scale”, “technique” and “selection” effects. The second framework
is the pollution model of inter-industry trade developed by Antweiler, Copeland and
Taylor (2001) which generates the “scale”, “technique” and “composition” effects.

This is the first study to integrate the trade-induced environmental impacts of both
intra- and inter-industry trade. In particular, it examines the selection effect as an
important factor to be accounted for in the estimation of the environmental impact of
international trade.

Investigations into the trade-environment linkage have mainly sought to address

the welfare effect of international trade. Studies attempt to answer the heatedly debated

question of whether dirty trade contributes to increasing emission levels and thus negate

87

the benefits of cross-border exchanges of goods. While it is recognized that free trade
increases consumptive welfare, concerns are raised about the environmental cost of dirty
production and whether globalization can lead to net welfare gains. A number of
empirical studies indicate that trade liberalization is good for the environment (see
Antweiler et al. 2001 ). Notably, the majority of empirical studies investigating the trade-
environment linkage are based on the traditional theory of trade.

Table 1.1 shows data on manufacturing intra-industry trade as a percentage of
total manufacturing trade. The statistics indicate that the volume of trade in differentiated
goods has risen for many developing and newly-emerging markets. However, although
intra-industry trade explains a significant volume of trade patterns, currently there is no
empirical study that investigates the impact of “new” trade on domestic local pollutants.
Further, theory suggests that the environmental impact of infra-industry trade is distinct

from the environmental impact of inter-industry trade (see Chapters Two and Three).

Therefore, if countries engage in both inter— and intra-industry trade9, an empirical
assessment of how international trade affects environmental quality should account for
the impact of both types of trade. There has not been a study that assesses the full impact
of trade under the assumption that countries engage in both intra— and inter-industry trade.
This paper analyzes the effects of scale, technique, selection and composition on

environmental quality. Additionally, it examines the effect of openness to trade on

¥

9
The World Bank (2009) calculates and makes available data on lntra-industry Trade (IIT) Index and the

Revealed Comparative Advantage (RCA) Index. The IIT index can be computed for any country and is
generally computed for the manufactured goods traded Standard Industrial Trade Classification (SITC)
three-digit level.

88

Table 1.1: Manufacturing intra-industry trade as a percentage of total manufacturing trade

1988-91 1992-95 1996-2000 Change

 

High and increasing intro-industry trade n.a. 66.3 77.4 11.1
Czech Republic n.a. 69.8 76.0 62.2
Slovak Republic 62.5 74.4 73.4 10.9
Mexico 54.9 64.3 72.1 17.2
Hungary 67.1 72.0 72.0 5.0
Germany 63.5 65.3 68.5 5.0
United States 56.4 61.7 62.6 6.2
Poland 52.4 56.3 61.3 8.9
Portugal

High and stable intro-industry trade

France 75.9 77.6 77.5 1.6
Canada 73.5 74.7 76.2 2.7
Austria 71.8 74.3 74.2 2.4
United Kingdom 70.1 73.1 73.7 3.6
Switzerland 69.8 71.8 72.0 2.2
Belgium/Luxembourg 77.6 77.7 71.4 -6.2
Spain 68.2 72.1 71.2 3.0
Netherland 69.2 70.4 68.9 -0.3
Sweden 64.2 64.6 66.6 2.4
Denmark 61.6 63 .4 64.8 3 .2
Italy 61.6 64.0 64.7 3.1
Ireland 58.6 57.2 54.6 -4.0
Finland 53.8 53.2 53.9 0.1
Low and increasing intro-industry trade

Korea 41.4 50.6 57.5 16.1
Japan 37.6 40.8 47.6 10.0
Low and stable intro-industry trade

New Zealand 37.2 38.4 40.6 3.4
Turkey 36.7 36.2 40.0 3.3
Norway 40.0 37.5 37.1 -2.9
Greece 42.8 39.5 36.9 -5.9
Australia 28.6 29.8 29.8 1.2
Iceland 19.0 19.1 20.1 1.1

 

Source: OECD International Trade Statistics, OECD Economic Outlook 71, OECD 2002

emissions level. The firll impact of international trade on the environment is estimated as

. . 10 . .
the sum of the magnrtudes of four env1ronmental effects , that rs, as the aggregatron of

10
The empirical study by Antweiler et al. (2001) investigated three environmental effects, namely the

scale, technique and composition effects using the framework based on the traditional theory of trade.

89

the responses to the scale, technique, selection and composition effects. In accounting for
the selection effect, this paper controls for an important variable that describes the data
generating process of a trade-environment relationship for countries engaged in both
intra- and inter-industry trade. In other words, the inclusion of the selection effect
variable in the model addresses the omitted variable problem in estimation.

In addition, the analysis in this paper recognizes two major concerns in translating
theory into the application of data. One concern is whether to assume, on one hand, that
data generates the realization that a country engages solely in either intra- or inter-
industry trade; or on the other hand, that data generates the realization that countries
engage in both intra- and inter-industry trade. The difference in assuming one assumption
instead of the other is that each may yield different interpretations of the results of data
analysis. More specifically, if countries are assumed to engage solely in intra-industry
trade, then the environmental consequences of international trade may be described by
three environmental effects, namely the scale, technique and selection effects.
Conversely, if countries are assumed to engage solely in inter-industry trade, then the
impact of international trade may be described by the scale, technique and composition
effects. On the other hand, if countries are assumed to engage in both intra- and inter-
industry trade and that the trade flows for any one country are comprised of both
differentiated and homogeneous goods, then the impact of trade on environmental quality
requires that the consequences of international trade be described by the four
environmental effects, namely, the scale, technique, selection and composition effects.

The second concern involves the definition and measurement of the scale and

technique effects in the data. Data used to measure the scale of production is normally

9O

represented by the Gross Domestic Product (GDP). Since GDP does not distinguish
between the change in the scale of production due to a change in the supply of
differentiated goods from that which is due to a change in the supply of homogeneous
goods, it is reasonable to assrune that the environmental scale effect represented by
changes in GDP is the combined effect of the changes from both the production of

differentiated and of homogenous goods. In a similar reasoning, the technique effect as

measured by the change in income or Gross National Income11 should constitute the
combined effects of the change in emission intensity that is due to both the abatement of
differentiated goods and the abatement of homogenous goods. In this paper, it is
recognized that the scale and technique effects as represented by GDP and Income,
respectively, measure the effects of both inter- and intra-industry trade.

The empirical approach in this paper addresses the foregoing issues in the
following ways. First, I specify an empirical model which takes into account the distinct
effects of both intra- and inter-industry trade. Therefore, the analysis assumes that
countries in the study sample engage in the production of both differentiated goods (intra-
industry trade) and homogeneous goods (inter-industry trade). Consequently, the analysis
controls for both the composition effect arising from comparative advantage due to cross
country factor differentials and the selection effect arising from the specialization due to

economies of scale.

 

The technique effect is the change in emission intensity as firms undertake changes in the abatement of
pollution to meet environmental regulation standards. Since environmental quality is a normal good, the
higher the income level the more stringent environmental regulation will be. Thus the change in emission
intensity can be represented by the change in income level (see ACT, 2001) where an increase in income
level is associated with a decrease in emission intensity.

91

Second, since data for the scale and technique effects measure both inter- and
intra-industry production effects, I interpret the estimates for scale and technique effects
as estimates that represent the integrated environmental effects of an economy that
engages in both inter- and intra-industry trade. In other words, the analysis recognizes
that there is no distinction or separation based on types of trade in the measurements of
the scale and technique effects.

The contributions of this paper are as follows. One, it builds an empirical model
of the trade-and-environment relationship which integrates the effects of both intra- and
inter-industry trade. Two, it tests the theoretical predictions of the environmental effects
of trade measured by the scale, technique, selection and composition variables. In
particular, it tests the response to the selection effect on emissions level. Three, it
investigates the impact of trade on three types of pollutants, namely, sulfur oxides (SOx),
nitrogen oxides (N Ox) and volatile organic compounds (VOC). Four, the study
contributes a new and original dataset that spans a recent time period, from the year 1995
to year 2004. Five, it provides evidence to attempt to answer the central question, is trade
beneficial to the environment?

The findings of this paper are the following. First, evidence suggests that in
addition to the scale, technique and composition effects, the selection effect is an
important factor in describing the empirical relationship between trade and pollution.
Estimation results suggest that the exclusion of the selection effect variable poses a
specification error in the form of omitted variable bias in estimation. Second, the signs or
directions of the effects of the four environmental variables conform to theoretical

expectations. Estimates show that emission levels are increasing in the selection, scale

92

effect, composition and technique effects. The latter implies that emission is decreasing
in the stringency of environmental regulation. Third, data on OECD countries suggests
that the scale and technique effects can be described by homothetic production or
consumption fimctions, consistent with the underlying theoretical assumptions of intra-
industry trade frameworks. Fourth, there is evidence to suggest that greater openness to
trade leads to a decrease in emission levels, that is, freer trade contributes to improving
environmental quality.

This paper is organized as follows. Section 4.2 discusses the conceptual
framework and research hypotheses. Section 4.3 presents the economic model of
monopolistic competition, trade and environment which provides the basis for the
predictions of the effects of intra-industry trade on emissions level. Section 4.4 derives
the reduced form equations for an econometric model of trade-and-environment. Section
4.5 discusses the empirical strategy, section 4.6 presents the main results section while
section 4.7 presents the results of alternative model specifications. Section 4.8 is

discussion and section 4.9 concludes.

4.2 Conceptual Framework and Research Hypotheses
Theory suggests that the benefits and costs of both the production and the
consumption of pollution-intensive goods determine the demand and supply of pollution
levels. Economic factors that influence the demand and supply of dirty goods include the
price of goods, the factor intensity used in production, the output level of production and
the price of environmental regulation (see Antweiler et al. 2001). In Chapter Three of this

dissertation, I show that the demand and supply of pollution depends on the number of

93

firms in the industry in addition to other economic factors such as the environmental tax
rate and the level of output. Therefore, observable variables in an empirical model that
relates pollution levels to economic factors may include the aforementioned variables,
including goods or product prices, the factor intensity of production, the level of output
and the number of firms.

In the following discussion, I define concepts of environmental variables and their
relationships to pollution levels. Empirical hypotheses to be tested are based on the
theoretical predictions derived in previous frameworks (see Chapters Two and Three;

Antweiler et al. 2001).

4.2.1 Conceptual Framework

Theory suggests intra- and inter-industry trade generates four environmental
effects, namely the scale, technique, selection and composition effects. In the following, I
relate each environmental variable to pollution.

The first environmental effect considered is the scale effect. In pollution-intensive
industries, the expansion of the scale of production implies a simultaneous increase in the
joint production of pollution. For example, economic growth that leads to growth in the
transportation industry implies growths or increases in emission levels of nitrogen oxides,
carbon monoxide, carbon dioxide and non-methane hydrocarbon. The scale effect
measures the environmental impact of the growth in the scale of production given there
are no changes in emission intensity, in the number of firms in the industry, and in the

factor intensity of production. Therefore, if the engagement of trade expands economic

94

production by five percent, then the scale effect implies there will be a five percent
increase in the level of pollution, holding all other factors constant.

The seminal paper by Grossman and Krueger (1991, p. 3) defines the scale effect
as the increase in pollution level when trade and investment liberalization expands
economic activity “if the nature of that activity remains unchanged”. Antweiler et al.
(2001) defines the scale effect as the increase in the level of pollution due to an increase
in the grth of production, holding the technique and composition effects constant.
Antweiler et al. (2001) measure the scale effect as the value of GDP at base-period world
prices; their study provides evidence to suggest a positive relationship between grth in
the scale of production and growth in emissions level.

In this paper, I distinguish the scale effect of intra-industry trade from the scale
effect of inter-industry trade. The former arises from the growth or expansion in the
production of differentiated goods while the latter arises from the expansion of the
production of homogenous, inter-industry goods. Notably, this distinction cannot be
discerned in the data when the scale effect is measured as the value of gross domestic
product (GDP) at world prices since GDP reflects the value of total national production
of all goods. Therefore, when the scale effect is represented by GDP, the analysis in this
paper recognizes that the scale effect constitutes the effects from both the production of
differentiated and the production of homogeneous goods. This recognition emphasizes
the explicit distinction that should be made between the scale effect arising from intra-
industry trade and the scale effect arising from inter-industry trade.

The second environmental effect included as an explanatory variable in the

current analysis is the technique effect. In this paper, the technique effect is defined as the

95

effect of a change in emission intensity on environmental quality, holding the scale,
selection and composition effects constant. There are two major ways to generate the
technique effect. One, the technique effect is related to the level of abatement that a firm
undertakes. Holding every other factor equal, greater abatement levels will reduce
emission per unit output, thereby allowing for a decrease in total emissions level in the
economy. Two, the technique effect may also be sourced from the cross-country diffusion
of efficient abatement techniques. Multinational firms that relocate to countries with lax
environmental regulations may introduce more sophisticated abatement technology into
host countries.

The technique effect arising from international trade occurs not only for inter-
industry trade goods but also for infra-industry goods. Assuming environmental quality is
a normal good, when trade leads to higher income levels, it in turn leads to more stringent
environmental regulation. Consequently, firms increase abatement activities to comply
with stricter environmental standards. The response to policy requirements applies in
any, or both industries engaged in the production of differentiated or homogeneous
goods.

Grossman and Krueger (1991, p. 4) define the technique effect as a change in the
per unit output of pollution subsequent to trade liberalization and foreign investment. In
their study, Grossman and Krueger (1991, 1993) find evidence to suggest that income per
capita and pollution levels can be described by the phenomenon now known as the
Environmental Kuznets Curve (EKC). The environmental Kuznets curve is the
hypothesis which states that pollution levels rise with per capita income, reach a peak,

and then fall with the continued increase in per capita income. This finding seems to

96

suggest that as countries become more developed and income level rises, environmental
quality as a normal good improves with higher income levels, which subsequently leads
to a decrease in per unit output of pollution. Antweiler et al. (2001) define the technique
effect as growth in emission intensity, holding everything else constant. The authors use a
one-period-lagged, three-year moving average of per capita gross national product (GNP)
to represent the technique effect. The authors find evidence to suggest that there is a
negative relationship between per capita income levels and total emissions.

The third environmental effect and explanatory variable in the current analysis is
the selection effect. The framework I developed in the previous chapters shows that the
selection effect distinguishes the environmental impact of trade driven by cross-country
demand factors from the environmental impact of trade driven by cross-country factor
abundance differentials, namely the composition effect. In the model of monopolistic
competition, I define the selection effect as the change in the level of emissions due to a
change in the number of firms engaged in the production of differentiated products,
holding the scale and technique effects constant. Everything else equal, including the
scale effect, fewer numbers of firms leads to lower levels of emission.

The selection effect arises from the assumptions that market structure is
imperfectly competitive, that production is increasing returns, and that equilibrium is
conditional on the achievement of long run zero economic profits. In this setting, free
trade implies that foreign competition leads to a decrease in the number of domestic firms
as consumers choose to consume both domestic and foreign products under a constrained

budget. Therefore, a trade-induced selection effect implies that holding all other factors

constant, the fall in the number of firms leads to a fall in aggregate emissions level. That

97

is, there is a positive relationship between environmental quality and the number of firms,
ceteris paribus.

I note that the environmental effect of selection varies in accordance with
assumptions imposed on consumer preferences. In a model which assumes non-CES,
non-homothetic utility function, international trade in differentiated goods yields a
change in the number of firms (or equivalently, in the number of product varieties) which
gives rise to a change in environmental quality. In contrast, in a model which assumes a
CBS utility function, trade in differentiated goods does not affect the number of firms in
the economy. This means that openness to trade is neutral with respect to the selection
effect.

Literature suggests there is evidence to support a selection effect (Head and Rice
1999). Free trade leads to a reduction in the number of domestic firms as competition
from abroad implies some firms earn negative profit and leave the industry. In the context
of the environment, a smaller number of firms lead to lower pollution levels, holding
other factors constant. In this case, consumers are better off as trade allows an increase in
the number of product varieties available for consumption while the smaller number of
surviving firms implies less pollution is generated, everything else equal. The empirical
analysis in this paper attempts to provide evidence on whether the selection effect leads
to a neutral or a positive selection effect.

The fourth environmental variable included as an explanatory variable in the
current analysis is the composition effect. Grossman and Krueger (1991, p. 4) define the
composition effect as the change in the composition of sectors due to trade liberalization

derived from two sources: one, from the competitive advantage in environmental

98

regulation in the pollution-intensive sector, and two, from comparative advantage in
cross-country differentials in factor abundance and technology in pollution-intensive
activities. Antweiler et al. (2001) define the composition effect as the change in
emissions level due to a change in the factor intensity employed in the production in dirty
goods, holding the scale and technique effect constant.

The composition effect contrasts with the selection effect of intra-industry trade
as the former is generated in the production of inter—industry, homogenous goods, while
the latter is generated in the production of intra-industry, differentiated goods. In the
Heckscher-Ohlin trade framework where markets are perfectly competitive, changes in
goods’ prices yield changes in relative factor intensities employed in the production of
specialized goods which provides comparative advantage. Therefore, if the economy
holds a comparative advantage in the production of dirty goods, this leads to an increase
in factor intensity when the dirty sector expands. Consequently, a trade-induced
composition effect implies that an increase in the factor intensity employed in the dirty
sector leads to an increase in the level of emissions.

A final variable of interest is the trade intensity variable, which measures
openness to trade and the degree of trade liberation. Openness to trade is defined by the
absence of restrictions that would hinder free trade between trading nations, while trade
liberalization is defined as the removal or reduction of trade barriers that restrict the free
flow of goods and services across countries. Barriers to free trade include tariff and non-
tariff barriers.

In translating the assumption of free trade in the theoretical model into data, most

empirical models control for cross-country differences in trade restrictions by including

99

the trade intensity variable as a measure of trade openness. Similarly, in quantifying the
effects of trade liberalization, the trade intensity variable is used to measure the degree to
which countries may engage in the reduction of trade barriers (see Antweiler et al. 2001).
In the current analysis, the trade intensity variable is used to measure two kinds of
effects. One, it is used in interaction forms to measure the responses to trade-induced
scale, technique, selection and composition effects. Two, it measures the effect of free
trade and trade liberalization on environmental quality. On the former, the aggregate
effect of trade on environmental quality is the sum of the effects of each of the four
environmental effects. On the latter, the effect of trade openness on environmental quality
is the estimated marginal effect of a change in emissions level with respect to a unit

change in the trade intensity variable.

4.2.2 Research Hypotheses

The main objective of this paper is to examine the impact of international trade on
the level of pollution. More specifically, this paper investigates the empirical relationship
between emission levels and the scale, technique, composition and selection effects for
countries known to engage in the trade of both differentiated and homogeneous goods.
Additionally, it examines the effect of trade intensity on environmental quality.

The exchange of differentiated goods, or intra-industry trade, is driven by factors
that explain market power and scale economies. There has not been an empirical study
that specifically addresses factors that influence the environmental effects of intra-
industry trade, namely the selection, scale and technique effects. Notably, the selection

effect is specific to intra-industry trade and has not been estimated in any study.

100

Additionally, the scale and technique effects arising from the production of differentiated
goods have not been explicitly shown to generate similar effects as the scale and
technique effects arising from the production of homogenous goods. In this paper, the
environmental consequences of intra-industry trade are integrated with the environmental
effects of inter-industry trade. I propose the following hypotheses about the trade-

environment relationship which describe the impact of both intra- and inter-industry on

environmental quality.

Hypothesis 1 .' A positive scale effect contributes positively to the level of emissions, that
is, emission is increasing in the level of the scale of production.

Consider an economy with pollution-intensive sectors that produces both
homogeneous and differentiated goods. Then the joint-production of pollution implies
that the scale of economic activities determines emissions level in the economy. Given
technology, an increase (decrease) in the production of dirty goods leads to an increase
(decrease) in pollution levels. This is known as the environmental scale effect (Grossman
and Krueger 1992; Antweiler et al. 2001). Theory suggests that the scale effect implies a
positive relationship between environmental quality and the production of dirty goods.
This prediction has been shown to hold in empirical studies (Antweiler et al. 2001). In the
current analysis, the direction of the effect of the scale on emissions level is similarly
CXpected to be positive. However, in contrast to past studies, the current study assumes
that the trade-induced scale effect represents the expansion of dirty production due to

both intra- and inter-industry trade. Hence, the scale effect in the current analysis

assumes that an increase (decrease) in the scale of production is not only due to the

101

expansion (contraction) of the production of specialized homogeneous goods with
comparative advantage, but that it is also due to an expansion (contraction) of the
production of differentiated goods with scale economies. Consistent with theory, the
aggregate scale effect of both intra- and inter-industry trade is assumed to be increasing
in emissions level.

Note that theory suggests if preferences are CES, then the trade-induced scale
effect of intra-industry trade may yield little or no effect on the level of total emission in
the domestic economy (Chapter Three). On the other hand, if preferences are represented
by a utility function that is non-CES, then the scale effect yields changes in the level of
total emission as openness to free trade leads to an expansion in the scale of production
(see Chapter Three). The current analysis includes the possibilities of the theoretical

predictions of both the non-CES and the CES frameworks.

Hypothesis 2: A negative technique effect contributes negatively to the level of emissions,
that is, emission is increasing in emission intensity.

Holding the scale effect and other determinants constant, the technique effect
refers to the change in pollution levels due to a change in emission intensity (see
Antweiler et al. 2001). In an economy where environmental policy necessitates firms to
engage in the reduction of pollution emissions, emission intensity is influenced by
abatement technology and the stringency of environmental regulation. Literature suggests
that assuming environmental quality is a normal good, income level is tied to emission
intensity through the income effect on environmental regulation (see equation (3.34) in

Chapter Three). Stricter environmental regulation implies firms will choose to exert

102

greater pollution control which consequently lowers emission intensity. Therefore, the
technique effect has a positive relationship with pollution levels. A more stringent (less
stringent) environmental regulation leads to lower (higher) is emission intensity and thus
lower (higher) pollution emissions. On the other hand, income level which influences
environmental policy, is negatively related to pollution levels. An increase (decrease) in
income level implies stricter (more lax) policy which leads to lower emission intensity.
This implies that holding the scale, selection and composition effects constant, higher
income levels leads to lower emission intensity which leads to lower pollution levels.
Thus, the current analysis hypothesizes a negative relationship between pollution and

income level.

Hypothesis 3: A negative selection effect contributes negatively to the level of emission,
that is, emissions level is increasing in the number of firms.

Holding the scale, technique, composition effects and other determinants constant,
the selection effect refers to the change in emissions level due to a change in the number
of firms that produces differentiated goods. The infra-industry model of pollution
suggests a positive empirical relationship between the number of firms and pollution
emissions. In the model of monopolistic competition, firms are identical in size and
production technology. With trade, consumers choose to consume foreign products
resulting in a decrease in the consumption of domestic varieties. The fall in the number of
domestic product varieties leads to the exit of unprofitable firms. Assuming the scale and
techniques of production remains unchanged, a smaller industry implies less pollution-

intensive activities, which means less emission. Hence, everything else equal, openness

103

to trade implies a fall in the number of firms which leads to a fall in emissions level. In
other words, emissions level is increasing in the number of firms, ceteris paribus.

However, if preferences are CES, then the trade-induced selection effect is absent.
In the CES case, the number of firms is a function of predetermined factors or parameters
such that n is fixed (see Chapter Three). The opening of trade does not lead to a change
in the number of domestic firms, thus, there is no trade-induced selection effect.

This paper statistically tests for the possibilities of the theoretical predictions of
both the general and the CES frameworks. More specifically, if preferences are non-CES,
then the selection effect is predicted to have a positive or increasing impact on emissions
level. Conversely, if preferences are CES, then the selection effect is neutral, so that the

null hypothesis holds.

Hypothesis 4: The composition effect is positively related to the level of emissions, that is,
emission is increasing in the intensity of the pollution-intensive factor used in production.

The composition effect is the change in emissions level due to a change in factor
intensity used in the production of dirty goods, holding the scale, technique, selection and
other determinants constant. Factor intensity in the production of dirty goods is
influenced by factors such as the price of factors of production, as well as output and
product prices. An increase (decrease) in the factor intensity used in pollution-intensive
production should therefore increase (decrease) the level of pollution emissions. In the
current study, the composition effect is predicted to have a positive relationship with
emissions level (see Antweiler et al. (2001) for evidence to support a positive

composition effect). In the open economy context, trade that expands the production of

104

A—

export goods in the pollution-intensive sector leads to a trade-induced composition effect
as factors of production shift from other sectors to the export sector. In this case, a
positive trade-induced composition effect generates an increase in the aggregate level of

emissions.

Hypothesis 5: The level of emissions is decreasing in trade intensity or Openness to trade.
The impact of free trade on environmental quality is measured by the variable
defined as the ratio of trade volume to GDP, namely, trade intensity. In this analysis, it is
postulated that freer trade leads to lower emissions levels. This hypothesis is consistent
with findings in Antweiler et al. (2001). There are many reasons why openness to trade
may contribute to the improvement of environmental quality. One reason is that greater
openness to trade implies an increase in trade flows which leads to an expansion in the
scale of overall production. Consequently, income level rises, which implies stricter
environmental policy that leads to increased abatement and lower emissions intensity.
Further, openness to trade may lead to an inward flow or diffusion of more efficient
abatement technology which contributes to reducing emission intensity. Therefore,
holding other factors constant, greater openness to trade leads to a negative growth in

emissions level.

4.3 Pollution and Intra-industry Trade
In this section, I expand on the decomposition of the environmental effects of
intra-industry trade to show the relationship between emissions level and economic

factors that define intra-industry trade. I show how the theoretical results in Chapter

105

Three form the basis of the estimating equations in the empirical analysis. The results in
this section complement the decomposition of the environmental effects of inter-industry
trade as shown in Antweiler et al. (2001) which I will not replicate in this paper (see
section 4.2 on the integration of the environmental concepts of intra- and inter-industry
trade).

Although the theory presented in this section is based on the CES framework
developed in Chapter Three, the interpretation of the empirical results in this paper will
be based on the predictions of both the more general (non-CES) and CBS frameworks of
infra-industry trade, in addition to the framework of inter-industry trade.

The major advantage of using the CES framework is that it provides tractability in
computation and yields closed form solutions. However, one drawback of using the CES
assumption is that it suggests fixed output level and fixed number of firms in the open
economy. While there is empirical evidence to support the former, there is no clear
evidence to support the latter (Head and Rice 1999, 2001). Therefore, literature on intra-
industry trade suggests the following. One, the scale of production and the number of
firms are fixed and openness of trade does not yield trade-induced scale and selection
effects. This result is consistent with the assumption of a CBS utility function. Two, and
alternatively, trade generates changes in the scale of production and in the number of
firms, which lead to trade-induced environmental scale and selection effects. This result
is consistent with the assumption of a non- CES, non-homothetic utility function (see
Chapter Two). Subsequently, while the analysis adopts the more mainstream CES

framework to derive reduced form estimating equations of the intra-industry component,

106

 

I relax the assumption of CES preferences in interpreting empirical results in order to
account for responses that are better explained by different underlying assumptions.
Consider a world which consists of two countries, Home and Foreign (or
equivalently, the rest of the world) engaged in the intra-industry trade of dirty goods.
Monopolistic competition market structure prevails, characterized by increasing returns
technology that is internal to firms (Krugman 1980). Consumers have Dixit-Stiglitz type
preferences, where a CES utility function is maximized subject to a budget constraint.
Pollution is jointly generated with goods’ production. A regulatory authority levies an
emission tax to internalize the negative extemality associated with the production of
pollution. Firms maximize profits taking into account the emission tax imposed by the

environmental authority.

4.3.1 Equilibrium Conditions
Generally, a monopolistically competitive industry achieves equilibrium through
adjustments in the market price, in the firm’s scale of production, and in the number of

firms in the industry. Specifically, equilibrium is characterized by two conditions: one,
that marginal revenue be equal to (long run) marginal cost(MR = MC) ; and two, that
price equal average revenue which equal average total cost or long run average cost
(p=AR=ATC=LRAC).

In the following, I summarize the equilibrium conditions of a CES model of intra-
industry trade- and-environment (see Appendix B for details). I omit redefining variable
notations in this section. Notations used to denote variables are defined in Chapter Two

and Chapter Three, with foreign variables superscripted with asterisks.

107

The fraction of output allocated towards abatement is given by equation (4.1):

l

_ _ we 3
(4.1) 6—1 (464))

 

and can be written as
(4.1’) 6=6(w,r,,6,6)

The pricing equation is given by equation (4.2):

_l 5 5-1 2}
(4.2) p—;[3:][(wfl) 45—1)]

or (4.2’) p=p(w,r,,6,5,p)

The level of output is given by equation (4.3):

ap(6—l)
4.3 =———
( ) q [35(1-10)
so that (4.3’) q = q(a,,6,6,p)

The equilibrium number of firms is:

 

LOO-p)
4.4 = ,
‘ ’ " a[e(1-p1+p(e-1)I
or (4.4’) n = n(L,a,6,p)

Consumption of good i in the domestic country is given by:

 

piI/(p-I)
(4.5) c. = , . . y
’ ”pp/(e-1)+,,*p* .i/(e-ll
that is, (4.5’) c,- =ci(p,n,p*,n*;y,p)

108

The import of foreign good I is given by equation (4.6):

 

* l/(p—l)
P I
(4.6) m = , , y
I "pp/(e4)+n*p*e/(p-l)
or (4.6’) m/ = m(e. 11. pinhy, p)

Emission tax rate is given by the following equation (4.7):

(4.7) T = {6)q,-1 [(61423]qu

1-
e/(e-l) + "*p‘e/(e-ll) ’0

 

where (a = (np

so that (4.7’) 2' = r(1,u;w,a,,6,5,p)

4.3.2 Scale, Technique and Selection Effects
From equation (3.19) in Chapter Three, the relationship between emission level
and the scale, technique and selection effects is given by:

(4.8) 2=e+é+q

where 2 is the percent change in total emission, 13 is the percent change in the number of
firms (or equivalently, in the number of product varieties), é is the percent change in

emission per unit output, and q is the percent change in the level of output. Equation
(4.8) shows that the impact of intra-industry trade on the environment is measured as the

sum of the scale (a) , technique (é) and selection (13) effects. The scale effect describes

how a change in the scale of production affects emission levels in the economy, holding

the technique and selection effects constant. The selection effect describes the change in

109

 

emission due to a change in the number of product varieties consumed, holding the scale
and technique effects constant. The technique effect is the change in emission levels due

to a change in emission intensity, holding the scale and selection effects constant.

4.4 Reduced Form Equations

The private demand for pollution is given by equation (4.8) while the supply of
pollution is given by equation (4.7). Equation (4.7) can be decomposed into its primitive
determinants (see Appendix B), and combining the supply and demand for pollution

generates a parsimonious reduced form equation given by:
(4.9) 2 = n1fi+n25+n3mn41f +115)“; +n6£+n7p+ngfi+n95+nloa+nlla
Equation (4.9) relates emission levels to economic variables. Total emissions is

influenced by the number of domestic firms (n ), the level of output produced for the

purpose of consumption (S), wage which is net income (w ), the factor of production in
the economy (L), imported product varieties (nit ), the world price level ( if ), the

preference parameter ( p ), the productivity of labor parameter (,6 ), the elasticity of

emission with respect to the fraction of output allocated towards consumption (6 ), the

marginal disutility of pollution ( (a ), and fixed cost ( a ).

4.4.1 An Empirical Model of Inter-industry and Intra-industry Trade

In this paper, the estimating equation will take into account the effects of inter-
industry trade, in addition to the effects of infra—industry presented in the previous
section. OECD countries in the sample are assumed to engage in both intra- and inter-

industry trade. This assumption recognizes that countries tend to engage in intra—industry

110

trade with trading partners that are in close geographical proximity and to engage in inter-
industry trade with countries that are distant to their geographical borders.

Antweiler et al. (2001) derive a reduced form equation that relates total emissions
to economic factors composed of scale, capital intensity or the capital labor ratio (in a
two-factor economy), income, the trade friction parameter, world price level, and country
type. Their empirical study shows that trade-induced technique and scale effects result in
a net reduction in pollution when international trade changes the composition of national
output (a composition effect). The significance of the study by Antweiler et al. (2001) is

that it develops a formal trade-environment framework which provides theoretical

underpinnings of a trade-environment relationship

The analysis in this paper departs from Antweiler et al. (2001) in that it takes into
account the environmental effects of both inter- and infra-industry trade. This paper links
the effects of intra-industry trade to the effects of inter-industry by integrating the
reduced form variables shown in the Antweiler e al. (2001) with the reduced form
variables described in the intra-industry trade model in Chapter Three. Hence, the
empirical model is a function of four environmental effects, namely, the scale, technique,
composition, and selection effects. Note that the inter-industry and intra-industry models
overlap in terms of the scale and technique effects and thus these effects will be
represented by the same measures in the estimating equation.

To my knowledge, there has not been an empirical study that combines the effects

of both intra-industry and inter-industry trade on environmental quality. By controlling

 

12
Antweiler et al. (2001) conclude that freer trade is good for the environment. Their formal framework is

based on the traditional trade model where trade is driven by cross-country differentials in relative factor
endowments.

lll

for the environmental effects of intra-industry trade, more precise estimation is obtained
by preventing an omitted variable, which, if excluded, may produce biased estimates.
Moreover, if the selection effect is a significant factor in the data generating process, then
its inclusion implies an improvement over, if not a correction to, current econometric
specifications which analyzes the relationship between pollution and economic factors
without accounting for the selection effect.

Based on the foregoing arguments, the composition effect factor from the
Antweiler et al. (2001) trade model is added to the reduced-form equation in equation
(4.9). For the purpose of reference, the reduced form derived in Antweiler et al. (2001) is

reproduced in following equation (4.10), using the authors’ original notations. Equation
(4.10) links emission levels :2 , to the scale effect, denoted S , the capital-labor ratio 1?,
income I , trade frictions ,6 , the world price levels )3” , and country type parameter T

such that:

(4.10) 2=”13+fi2k—7I3i'i'fl'4fi'l-fl'5fiw-7I6f

A comparison of equation (4.10) to equation (4.9) indicates that the following

economic factors distinguish intra-industry effect from inter-industry trade effects: the
number of firms or the number of product varieties available for consumption (n, n!“ ) , the
preference parameter p , the emission abatement parameter 5 , the labor productivity

parameter fl , and the price of factor of production w. On the other hand, the two

equations overlap in terms of economic factors that represent trade barriers, world price,

and income levels.

112

 

Combining the reduced form equation obtained in this section with the reduced

form equation obtained in Antweiler et a1. (2001) yields the following:

(411) 2 =H1fi+II2S+TI3w+TI4IT +1151“; +TI6L+II7p+I186+Tlgd+
H10¢+Hlld+I'Ilzpw+H13£+H14O+H156+H16T

where pw denotes world price, K denotes capital intensity, to denotes income, .9 denotes

trade friction, and T denotes country type13

The selection effect, denoted n , describes the change in emission level due to a
change in the number of product varieties consumed, everything else constant. A trade-
induced selection effect occurs when the country is exposed to foreign competition which
affects a change in the number of domestic firms, or equivalently, in the number of
varieties produced domestically. In the open economy, consumers with love-for-variety
preferences choose to demand foreign varieties in addition to domestic varieties. Theory
suggests a trade-induced increase (decrease) in the number of domestic product varieties
or firms leads an increase (decrease) in the level of emission (see Chapter Three).

The scale effect is denoted S in (4.11). It describes the change in emissions level
due to a change in the scale of production, holding everything else equal. In this paper,
the realization of a trade-induced scale effect comes from two sectors, one, from the
intra-industry trade sector and two, from the inter-industry trade sector. The scale effect
from each sector is indistinguishable in terms of data as represented by the change in

GDP. An expansion in the scale of total production leads to an increase in emissions.

 

l3
Notations in the Antweiler et al. (2001) may differ from the ones used here.

113

The technique effect is the change in emission levels due to a change in emission
intensity, holding the scale, selection, composition effects and other factors constant. In
equation (4.11), the technique effect is represented by a change in wage, w, which is net
income in the intra-industry trade sector and in a) , which is income in the inter-industry
trade sector. Since the realization of the technique effect from each trade sector is
indistinguishable when measured as country-level income, I combine w and a) in

equation (4.11) as total income, denoted y. Under the assumption that environmental

quality is a normal good, higher income levels imply stricter environmental regulation
which leads to lower emission intensity.

The composition effect is the change in emissions level resulting from a change in
the capital intensity used in the production. In equation (4.11), capital intensity is denoted '
as K. A trade-induced composition effect stems from a change in goods’ price which
leads to a change in the composition of factor intensity in the inter-industry sector (see
Antweiler et al. 2001). Assuming capital is the factor of production used in the pollution-
intensive sector, an increase in the level of capital intensity leads to an increase in
emissions level.

Note that the preference for product varieties, p , the labor productivity, ,6 , the

elasticity of emission with respect to the fraction of output allocated towards

consumption, 6 , fixed cost, a , and the marginal disutility of pollution, (a , are fixed

factors in the model. Measures and data for these country-specific effects, namely,

a , p, ,6, 6 and (a , are not readily available, therefore they are considered as unobserved

country parameters in the model.

114

It is noted that while the scale, technique, selection and composition factors are

endogenous variables, emission levels are determined endogenously but recursively”. In
the model, the scale, technique and selection effects are determined by basic economic
factors, predetermined parameters and exogenous variables. Pollution or emissions level
itself does not cause any change in the factors that affect the reduced-form variables. For
example, a change in the number of firms is determined by changes in factors such as the
output level and the level of employment of the labor force but is not determined by a
change in emissions level. In other words, there is no simultaneity or feedback between
the number of firms and the level of emissions. A change in emissions level does not
cause second-order changes in the number of firms, or in the measures for scale and
technique effects. Thus, while the variables of scale, the number of firms, real income

and pollution tax are set simultaneously in the model, emissions level is set recursively.

4.5 Empirical Strategy
To obtain the estimating equation, theoretical relationships need to be translated
into an empirical specification using measures that can be estimated. To do so, data and

their sources are discussed in this section.

4.5.1 Measurement and Data Sources
The dependent variable in the proposed estimating equation is the level of
emission that prevails in the economy. In this paper, I consider three types of air pollutant

that may be used as separate measures of pollution emission: sulfur oxides (SOx),

 

14
This assumption is similar to Antweiler et al. (2001).

115

nitrogen oxides (NOx) and volatile organic compounds (VOC). SOx, NOx and VOC are
measured in metric tonnes.

Sulfur oxides (SOx) are produced mainly through the burning of fossil fuels.
Sulfur oxides gases are formed when fuels containing sulfur such as coal and crude oil
are burned or when gasoline is extracted from oil. Since sulfur is prevalent in ore that
contains metals such as aluminum, copper, zinc, lead, and iron, sulfur oxides are also
produced in the extraction of metal from ore. Reductions in sulfur oxides emissions are
achieved through the implementation of air quality standards, through sulfur dioxide
emissions trading programs, from abatement efforts which include the installations of
pollution control equipments, and through other control programs that would reduce, for
example, the average sulfur content of burned fuels. Sulfur oxides (SOx) are the more
widely regulated pollutant among the three pollutants.

Nitrogen oxides or NOx include the highly reactive gasses nitrogen dioxide
(N 02), nitrous acid and nitric acid. In particular, N02 is formed from emissions produced
in transportation such as cars, trucks and buses, from power plants and off-road
equipments. It contributes to the formation of ground-level ozone and fine particle
pollution. Nitrogen oxides emissions are mainly regulated through air quality standards
set by regulatory authorities.

The third pollutant, volatile organic compounds (VOC), is defined as "any organic
compound that participates in atmospheric photochemical reactions except those
designated by EPA as having negligible photochemical reactivity" (EPA 2009). Volatile
organic compounds (VOC) are emitted from the burning of coal, oil and gasoline. Other

sources of VOC include cleaners, paints and solvents. Volatile organic compounds may

116

also be found to contaminate ground and drinking water. While outdoor VOC may be
regulated, indoor VOC may not be easily regulated due to the lack of authority in
regulating indoor air quality and in collecting information on household products.

The pollutants, sulfur oxides, nitrogen oxides and volatile organic compounds are
ofien used in studies as indicators of environmental quality. Their effects are local in
nature, a characteristic that is consistent with the theoretical assrunption in this study.
Data on the three pollutants, SOx, NOx and VOC, are sourced from OECD
Environmental Data for the years 1995-2004. I use the logarithmic transformation of each
pollutant to provide a lognonnal distribution of the annual country level data as
recommended in previous work in this area (see WHO (1984) and Antweiler et al. (2001)
on the appropriateness of lognonnal distributions).

Data on GDP, GNP in 2000 dollars and other macroeconomic variables is sourced
from the Penn World Tables (PWT 6.2) as described by Summers, Heston and Aten

(2006), for the years 1995-2004. Gross domestic product per square kilometer

(GDP/kmz) measures the scale effect. Gross national product per capita (GNP/L) earned

both domestically and abroad15 measures the technique effect. The technique effect is the
effect on emissions level resulting from environmental compliance induced by a change
in income level. To avoid high or perfect correlation between GDP and GNP, the
difference between the two variables will be used to separate the technique effect from

the scale effect. To prevent contemporaneous correlation with GDP, I construct a three-

Antweiler et al. (2001) use income to measure the technique effect which is replicated in the current
analysis.

117

 

year moving average of per capita GNP that is lagged one period (see Antweiler et al.
2001)

In equation (4.11), the selection effect of intra-industry trade is represented by the
number of firms, n. In theory, the selection effect comes from the change in the number
of product varieties or in the number of firms engaged in the production of differentiated

goods. Thus, in the data, the selection effect can be represented by two measures: one, by

the change in the number of product varieties16 produced domestically, and two, by the
change in the number of domestic firms. Theory suggests the selection effect contributes
to a positive (negative) change in pollution levels when there is a trade-induced
expansion (contraction) of labor force which leads to an increase (decrease) in the
number of domestic firms. Thus, the pollution-trade model suggests a definition of a

selection effect that is best defined by firm-level production, that is, by the number of

firms that are producing dirty products.l7 Accordingly, I use the number of listed

domestic companies to represent the selection effect, sourced from the World
Development Indicators (WDI) for the years 1995-2004. Consistent with other measures

in the model, the measure for n is in intensive form, that is, it is the number of listed

domestic companies per kilometer squared (Companies/kmz).

 

16
Product varieties have been measured in various ways in trade literature, depending on the theoretical

context in which varieties are defined. Theoretical tractability imposes restrictive assumptions on which
variety may be defined; for example, to allow for the aggregation of price indices or utility functions, it is
necessary to use CES or Cobb-Douglas functions in modeling how consumers value variety.

A literal approach of translating theory to data implies that the production of one firm comprises of a
single variety. In reality, a firm may produce or export more than one variety in its product line. However,
since the objective of the current model is to measure the firm’s production of pollution emissions, then, for
the purpose of estimating “n”, it is sufficient to represent the selection effect by the number of firms
engaged in the production of dirty goods or product varieties.

118

 

The composition effect is the change in emissions level due to a change in the
capital intensity used in production, holding all other factors constant. In the Antweiler et
al. (2001) framework, capital is the factor of production used as input in the pollution-
intensive industry. Since capital stock data was unavailable for the years of interest in
PWT, I construct a capital stock dataset using data on investment sourced from WDI for
the years 1995-2004, and based on the method outlined in Learner (1984 p. 233; see
Appendix C for detailed method). Labor data is acquired from PWT so that capital
intensity is the ratio of capital stock to labor force (K/L).

Openness to trade is measured by the variable trade intensity, the ratio of imports
plus exports to GDP. Trade intensity data is sourced from PWT for the years 1995-2004.

Freer trade or trade liberalization is hypothesized to lead to lower emission intensity.

4.5.2 Method

Panel data method is used to estimate the empirical model. The advantage of
using panel data is that it provides increased precision in estimation as it allows the
analysis of both the spatial and temporal dimensions of units of observation. In contrast
to a single cross section analysis, panel data analysis has the advantage of allowing for
the dynamics of individual country behavior to be taken into account. In addition, panel
data provides consistent estimation of a fixed effects model, where unobserved individual
heterogeneity which may be correlated with regressors, can be accounted for. In the case
where the unobserved individual heterogeneity is assumed to be distributed
independently of the regressors, random effects method of estimation may be used. In

this paper, both random and fixed effects estimations are reported. Further, in order to

119

 

obtain panel-robust statistical inferences, the estimation corrects for both the correlation
of model errors over time for given countries and for any heteroskedasticity across
countries. A Sargan-Hansen test is performed to determine whether fixed effects are

present.

Unobservable Variables
Unobservable parameters such as the parameter of the elasticity of emission
intensity with respect to per unit consumption, the disutility of emissions parameter and

the preference parameter can be considered as time-invariant country-specific effects
represented by the unobserved heterogeneity, denoted as 97,, . It is also noted, as in

Antweiler et al. (2001), that common-to-all-countries effects such as changes in the

relative price of goods and changes in abatement technologies, may be considered as

time-specific effects, denoted as 5,. Human error in calculation or tabulation, and

machine error in reading pollutant concentrations constitute the idiosyncratic error, Ukr-

To account for the unobservables, an individual effects model for the error term

”kt is specified as the following:

where 5, is a time-specific effect, gk is a country specific effect, and Ukt is an

idiosyncratic measurement error for country k at time t.

While both fixed effects and random effects estimates are presented, the Sargan-

Hansen test allows the determination of the more appropriate estimator between the two.

120

 

If country-specific effects contained in the error term are correlated with regressors, then
the fixed effects estimator allows for consistent estimation of the model. On the other
hand, if the unobserved country-specific heterogeneity is distributed independently of the
regressors, then the random effects estimator allows for consistent and efficient
estimation of the model. Note that in the case where the fixed effects estimator is
appropriate, pooled ordinary least squares (OLS) estimator is inconsistent, while if the

random effects estimator is appropriate, then pooled OLS is less efficient.

To capture the effects of time specific, common-to-all-countries variables that are
excluded in the model and left unobserved in the error term, I include a set of unrestricted

time dummies in the estimating equation.

4.5.3 The Estimating Equation

4.5.3.1 Functional form

A linear representation of pollutant emissions in metric tonnes per squared

kilometer at time t and country k is the following model:
(4.13) Zf, = a0 + a1 FIRM,“ + azSCALEk, + a3INCk, + a4 (KL)k, + a5TIk, + GL6 + uk,

where FIRM is the ecuntry specific number of listed companies per squared kilometer,

COM/kmz, SCALE is country-specific GDP/ kmz, INC is GNP/L, TI is trade intensity

measured as the import-export ratio to GDP or (X+M)/GDP, and G contains country-

specific variables and physical characteristics. Note that the world price and country-type

variables are captured in the time-specific error term, 4?, , and unmeasured economic and

121

 

physical variables such as the disutility of emissions parameter are captured in the

country-specific error term gk .

To account for trade-induced environmental effects, I include interacted terms of
the variables representing the scale, technique, selection and composition effects with the
trade intensity variable.

I specify the following linear model:

(4.14)

Z]?! = a0 +a1F1RMk,+ 025CALEk1+a3INth +a4(KL)/“ +a5TIk, +
a6FIRMk, ~T1k, +a7SCALEk, -le, +a81NCk, -TI,,, +a9(KL)k, T1,, +G},,,/I + uk,

where FIRM - TI is country k ’s trade intensity interacted with the number of listed
domestic companies per squared kilometer, SCALE - T I is country k ’s trade intensity
interacted with gross domestic product per squared kilometer, INC -T I is country k ’s
trade intensity interacted with real income per capita and KL -TI is country k ’s trade
intensity interacted with its capital-to-labor ratio. The vector G contains variables that
describes country characteristic which includes the following: population density, POP;
country k ’s real income measured relative to the world average, denoted REL. INC;
country k ’5 number of listed domestic companies per squared kilometer interacted with
real income denoted FIRM .INC ; and time dummies that account for time-specific

effects. Equation (4.14) is referred to as Model A in the analysis.

122

 

4.5.4 Alternative Functional Form

To account for the possibilities of nonlinearities in responses to the scale,
technique and composition effects, I consider an alternative specification by adding
squared terms to the linear-in-variables representation of the model in equation (4.14).
Non-homotheticities in production or consumption is one reason for a nonlinear response
to the environmental effects. Differences in producer prices brought about by cross-
country differences in income and in techniques of production may imply that the
composition effect should be modeled as a nonlinear function of the capital intensity
variable, K/L. These possibilities are consistent with the assumptions of a traditional
framework of trade. Therefore, to model nonlinear responses to the environmental
effects, I include quadratic measures of the technique and composition effects.

On the other hand, the CES model of pollution and infra-industry trade assumes
homotheticity in consumption. This assumption suggests linearity in the response to the
scale effect. Further, I maintain that the response to the selection effect is linear in the
number of firms. The intra-industry trade framework assumes identical size, preference
and production technology across countries. Hence, these imply similarities in income
levels. If income levels are similar, then the response to the scale, technique effect and
selection variables can be represented by linearity in the data of GDP, income and
number of firms to reflect similarities in prices and techniques of production.

To account for the aforementioned possibilities, the following functional form is

designated as Model B.

123

 

Model B:

2,5, = a0 + alFIRMk, + azSCALEk, + a3INCk, + a4INC2k, + a,(KL),,,
(4.15) +01,,(1<L)2 k, + a7TIk, + +a8FIRMk, TI,“ + agSCALEk, -TI,,,

+a101NCk, -TI,,, + a, 1(1(2),, -T1k, + 0),,3 + uk,
Models A and B sufficiently capture the essential responses to the environmental effects

of production and of trade variables in parsimonious specifications.

4.6 Result
4.6.1 Main Result

Tables 4.1, 4.2 and 4.3 present the results of fixed and random effect estimations
for three dependent variables, namely, the pollutants sulfur oxides (SOx), nitrogen oxides
(N Ox) and volatile organic compounds (VOC), respectively.

For each estimation of the 80x, NOx and VOC equations, two econometric
models are considered: Models A and B. Model A assumes linearity in all variables and
Model B assumes non-linearities in the technique and composition effects variables. I
examine the results for each pollutant in turn.

In the next section, sensitivity analysis of the estimations are presented in Tables
4.4, 4.5 and 4.6, each comprising of three estimating models, namely, Model C, Model D

and Model B.

Three main results in the analysis warrant attention. One, for each pollutant and
for every econometric model A through E, F-tests statistics show that at the 1 percent
level of significance, there is evidence to reject the null hypothesis that the variance due

to cross-country differences is not different from zero. This result implies the presence of

124

significant country-level effects. Hence, using pooled OLS estimator in the analyses will
not be appropriate.

Two, the correlation coefficients between the explanatory variables and
unobserved country-level effects range from -0.95 to -0.97 in the fixed effects models of
sulfur oxides (SOx), 0.64 to 0.69 in the fixed-effects models of nitrogen oxides (N Ox),
and -0.51 to -0.57 in the models of volatile organic compounds (VOC). These results
suggest high correlation between the regressors and the country level effects. On the
other hand, statistically significant coefficient estimates obtained under fixed effects (FE)
estimations and under random effects (RE) estimations are nearly identical. Hence, I use
the Sargan-Hansen test to determine the appropriateness of using fixed effect (FE)
estimator versus random effect (RE) estimator, with cluster-robust standard errors. The
Sargan-Hansen test statistics, shown in Tables 4.1 through 4.6, are statistically significant
at the 1 percent level of significance which imply we can reject the null hypothesis that
the additional orthogonality condition imposed by the RE estimator is valid. These results
imply strong evidence to suggest that unobserved country-level heterogeneity is
correlated with regressors. Hence, in all models considered, there is evidence to suggest
that estimates generated by FE estimators are consistent estimates, whereas estimates
generated in the RE models are not. Therefore, in describing the regression results in the
following section, I focus only on estimates based on the FE models.

Finally, the interaction terms of each of the environmental variable and the trade
intensity variable use demeaned data which subtract the sample mean of each variable
from its respective data values. Therefore, the interpretation of the coefficient estimates

of the trade-induced scale, technique, selection and composition effects variables are

125

 

. . . I8 .
made at the mean value of the trade 1ntensrty vanable . In other words, coefficient
estimates of trade-induced environmental effects are calculated for the “average trading
country” defined as the country whose level of trade intensity is equal to the sample

mean.

4.6.1 Sulfur Oxides (SOx) and the Selection, Scale, Technique and Composition
Effects
Table 4.1 shows that in the Models A an B, there is a positive relationship between the

selection effect variable, represented by the number of listed companies per squared

kilometer (Companies/ kmz), and the dependent variable, sulfur oxides emissions (SOx).

Estimates are statistically significant for the coefficients on the selection effect variables
which do not vary greatly in magnitudes in the estimates of both Model A and Model B.
In Model B, evidence suggests that holding other factors constant, a unit change in the
number of domestic companies per squared kilometer results in approximately 73 percent
change in the level of sulfur oxides per squared kilometer (see Appendix C for

calculation details). In Model A, the percentage change in sulfur oxides is 57 percent.

The d1rect1on of the effect of Companres/ km on sulfur oxrdes em1ssrons rs posrtrve,

consistent with the theoretical prediction that an increase (decrease) in the number of
firms leads to an increase (decrease) in the level of emissions.
Next, the environmental effect of scale is represented by gross domestic product

per kilometer square (GDP/kmz) in Model A and Model B. Estimates of the semi-

 

8
See Wooldridge (2003) pp. 194-195.
126

 

Table 4.1: Estimation Results of Models A and B - Sulfur Oxides (SOx)

 

 

Dependent Variable: Log of Sulfur Oxides per sq km

 

Estimation Method:

Fixed Effects

 

Random Effects

 

 

 

Model Specification: A B A B
Levels Only Levels Only
Variable/Column: ( I ) (2) (3) (4)
Companies/ km 0.501 * 0.548“ 0.587" 0.605"
2 (0.277) (0.260) (0.248) (0.251)
0.320“ 0.318" -0.124 0.108
GDP/km (0.129) (0.125) (0.121) (0.120)
Lagged per capita income -0.184** -0.41 1“ -0.1 17* -0.056
(INC) (0.073) (0.187) (0.069) (0.200)
Lagged per capita income 0.059 -0.013
squared (0.044) (0.046)
Capital abundance (K/L) -0.333 0.066 0385*" -0.262
(0.197) (0.259) (0.139) (0.184)
(K/L) Squared -0.036*** -0.010
(0.013) (0.011)
Trade Intensity -0.026*** -0.026*** -0.021*** -0.021***
(TI=X+M/GDP) (0.007) (0.006) (0.006) (0.006)
Relative Income 0.009 0.009 0.009 0.010"
(0.007) (0.006) (0.006) (0.006)
Population Density -0.333*** -0.379*** 0.032 0.036
(0.096) (0.102) (0.025) (0.027)
Tl x Companies / km -0.013** -0.012** -0.007 -0.006
2 (0.005) (0.005) (0.005) (0.005)
-0.0001 0.0002 -0.002* -0.002*
TI x GDP] km (0.002) (0.002) (0.001) (0.0001)
TI x Per capita Income -0.007** -0.008*** -0.006*** -0.006*"”"
(0.003) (0.002) (0.002) (0.002)
TI x (K/L) -0.001 -0.0001 -0.001 -0.0009
(0.002) (0.002) (0.002) (0.002)
Companies x Per capita -0.203"' -0.23 l ** 0.181" -0.l97**
Income (0.103) (0.095) (0.088) (0.091)
Intercept -0.689 -0.919 -4.714*** -5.1 17*"
(0.722) (1.916) (0.973) (1.040)
Observations 211 211 211 211
Group 23 23 23 23
R2 . . 0.7731 0.7826 0.7559 0.7603
- wrthm
R2 _ between 0.2290 0.2616 0.7785 0.7588
R2 0.2109 0.2412 0.8054 0.7915
- overall
Sargan-Hansen T651] 12 (df) 1609*“: 5721111114:
LR test/ 12 (d1) 8.93....

 

*** Significance at the 99-percent confidence level.
“Significance at the 95-percent confidence level.
*Significance at the 90-percent confidence level.

Note: Standard errors in parentheses are cluster-robust.

127

elasticities of scale in both models show there is a statistically significant positive

relationship between GDP/km2 and the dependent variable, sulfiir oxides. Coefficient

estrmates Indicate that holding other factors equal, a unrt 1ncrease 1n GDP/km leads to a

38 percent increase in sulfur oxides emissions for Model A and a 37 percent increase in
sulfur oxides for Model B.

The technique effect is represented by the lagged per capita income (INC)

variable and the lagged per capita income squared (INC 2) 19. Regression estimates for

Model A and Model B suggest that there is a negative relationship between income per
capita and sulfur oxides emissions. In Model B, a 1 unit increase in income level leads to
a 51 percent decrease in sulfur oxides emissions level, while in Model A, it leads to a 20
percent decrease, holding other factors constant. The coefficient estimates for the
quadratic term of lagged per capita income in Model A and Model B are not statistically
significant.

There are two observations to make of the effect of income on sulfur oxides. First,
OECD countries are comprised of mostly developed nations whose incomes per capita
are greater than the per capita incomes of developing countries. The inverted-U shape of
the Environmental Kuznets Curve (EKC) has a turning point of approximately $8,000 to
$10,000 per capita income (Grossman and Krueger 1993) which is below the sample
mean of per capita income of OECD countries in this analysis (see summary statistics of
data in Appendix C). The EKC is the postulated relationship between environmental

quality and income per capita; the inverted-U shape of the EKC indicates that at the early

 

This functional form is consistent with the Environmental Kuznets Curve (EKC) hypothesis, postulated
by Grossman and Krueger (1993).

128

 

stage of development, pollution levels increases with the increase in income per capita.
The curve reaches a peak and descends after the turning point, indicating that in the later
stages of development, as income per capita continues to rise, pollution levels decrease as
countries impose stricter environmental regulations to obtain higher environmental
quality, a normal good. Therefore, for the more affluent and developed economies, an
increase in income leads to stricter environmental policy which then leads to lower
emission intensity. Thus, the technique effect is manifested as a negative relationship
between income per capita and emissions levels, as indicated by the negative direction of
the effect of income on sulfur oxides . A second observation is that the coefficient

estimate of the level-fonn of INC is statistically significant but the coefficient estimate of

the squared INC rs not statistrcally srgmficant. These results 1nd1cate that linearity 1n the

income variable generates statistically significant estimates while non-linearity generates
statistically insignificant estimates. The former is consistent with the assumption of a
homothetic utility function where the shares of income spent on consumption goods are
constant. Further, in the theoretical framework, a fixed fraction of output or resources is
allocated towards abatement activity where pollution production is constant returns to
scale. Thus, a change in the stringency of environmental policy implies a linear change in
abatement levels. This provides an explanation of the statistically significant estimates of
the technique effect in the linear or level form of the income variable. Thus, the
realization of income data seems to suggest that the assumption of homotheticity,
consistent with intra-industry trade, is more relevant in capturing responses to the

technique effect, as opposed to an assumption of non-homotheticity.

129

 

The composition effect variable is represented by capital intensity. In Model A
which assumes linearity in all variables, the coefficient estimate of capital intensity is not
statistically significant at the conventional levels of significance. In Model B, which
assumes a quadratic functional form of the composition effect variable, the semi elasticity
estimates of composition effect is statistically significant at the 5 percent levels. This
result indicates strong evidence to suggest that a 1 unit increase in capital intensity leads
to a 38 percent increase in sulfur oxides emissions, calculated at the mean value of capital
intensity.

Interestingly, in both Models A and B, the coefficient estimates of the capital
intensity variables measured in level forms are not statistically significant. A test of the
joint significance of the variables capital intensity (K/L) and capital intensity-squared
(K/L-squared) generates an F-test statistic that is statistically significant at the 1 percent
significance level. This result suggests we can reject the null hypothesis that the model
which excludes the variables K/L and K/L-squared is correctly specified relative to the
full model. Therefore, the evidence suggests that a quadratic functional form captures the
responses to the composition effect on pollution. This is consistent with the assumption
of non-homothetic production, under the theoretical proposition that the impact of capital
accumulation on the emission of sulfur oxides depends on the techniques of productions
with varying income levels across countries (see Antweiler et al. 2001).

The impact of freer trade or openness to trade on environmental quality is
estimated by the effect of a change in the trade intensity variable on the change in
emissions level. In both Models A and B, coefficient estimates on the trade intensity

variables are statistically significant at less than 1 percent level of significance. These

130

 

 

results indicate evidence to suggest that a 1 unit change in the ratio of the volume of trade
to GDP leads to approximately 2.6 percent decrease in the levels of sulfur oxides
emissions. The negative direction of the effect of trade intensity on emissions level
conforms to theoretical prediction which suggests that greater openness to trade or
increased trade liberalization is beneficial to the environment.

I use a likelihood ratio (LR) test to evaluate the goodness of fit of the nested
models A and B. Results show that the linear-in-variables and more restricted model A
has lesser fit compared to the more general Model B. A significant LR-test statistic shows
that the restrictions on Model A are rejected by the data. One explanation for this result is
that for the countries in the sample, consumption or production is non-homothetic such
that the response to the composition effect implies that capital accumulation generates
non-linear effects on the composition of the economy. Hence, for the pollutant sulfur
oxides, the specification (LR-test) result indicates that greater emphasis should be given

to estimates generated in Model B.

4.6.2 Nitrogen Oxides (NOx) and the Selection, Scale, Technique and Composition

Effects

Table 4.2 shows the responses of changes in the selection, scale, technique and
composition effects on the level of nitrogen oxides (N Ox). Fixed effects estimations
indicate that for the Models A and B, there is a positive relationship between the selection

effect variable, represented by the number of listed companies per squared kilometer

(Companies/ km2), and the dependent variable, nitrogen oxides emissions.

131

Table 4.2: Estimation Results of Models A and B - Nitrogen Oxides (NOx)

 

 

 

 

Estimation Method: Fixed Effects Random Effects
Model Specification: A B A B
Levels Only Levels Only
Variable/Column: (I) (2) (3) (4)
Companies/ km 0.225" 0257*“ 0382*" 0374*"
2 (0.085) (0.090) (0.090) (0.090)
0.107 0.087 0.1 10* 0.084
(mm km (0.072) (0.068) (0.061) (0.053)
Lagged per capita income -0.060 -0.01 1 -0.047 0.034
(0.049) (0.1 14) (0.040) (0.089)
Lagged per capita income -0.01 1 -0.019
squared (0.021) (0.019)
Capital abundance (K/L) 0.006 0.122 -0.048 0.070
(0.067) (0.084) (0.045) (0.065)
(K/L) Squared -0.010 -0.009
(0.006) (0.006)
Trade Intensity -0.004 -0.004 -0.002 -0.002
(TI=X+M/GDP) (0.003) (0.003) (0.003) (0.003)
Relative Income -0.006** -0.006** 0.002 0.0009
(0.003) (0.003) (0.004) (0.003)
Population Density 0.006 0.010 0045*" 0052*“
(0.057) (0.059) (0.011) (0.010)
Tl x Companies / km -0.005"‘"‘ -0.004 -0.001 -0.0005
2 (0.002) (0.003) (0.002) (0.002)
0.0002 0.0003 -0.0003 -0.0002
T] x GDP/ km (0.0008) (0.001) (0.0007) (0.001)
TI x Per capita Income -2.55e-06 -0.0004 -0.00002 -0.0002
(0.002) (0.002) (0.001) (0.001)
TI x (K/L) -0.001 -0.0004 -0.0004 -0.0004
(0.001) (0.001) (0.001) (0.001)
Companies x Per capita 0.032 0.011 0.020 0005
Income (0.035) (0.035) (0.026) (0.027)
Intercept 6473*" -6.843"”“* -6.98l*** -7.369***
(0.784) (0.848) (0.382) (0.336)
Observations 21 l 21 1 211 211
Group 23 23 23 23
R2 . . 0.6035 0.6170 0.5745 0.6001
- wrthm
R2 b 0.6952 0.7047 0.7915 0.7833
- etween
R2 0.7397 0.7685 0.8037 0.8007
- overall
Sargan-Hansen TCSU [2 ((11) 2783*** 3924“”.
LR Test/ 12 (df) 7.33:” 0.02

 

 

 

" *Significance at the 99-percent confidence level.
"Significance at the 95-percent confidence level.
*Slgnificance at the 90-percent confidence level.
Note: Standard errors in parentheses are cluster-robust.

132

Statistlcally srgnrficant coefficient estrmates of Compames/km at the 99-percent

confidence levels provide strong evidence to suggest that, holding other factors constant,
a 1 unit change in the number of domestic companies per squared kilometer leads to an
approximately 25 percent change in the level of nitrogen oxides per squared kilometer in
Model A and approximately 29 percent in Model B. The direction of effect of the
selection variable on emissions level is positive, consistent with the theoretical prediction
that an increase (decrease) in the number of firms leads to an increase (decrease) in the
level of emissions.

Coefficient estimates of the other environmental effects namely, the scale,
technique and composition effects are not statistically significant in models A or B.
Similarly, estimates of the effects of trade intensity and other country-specific variables
are not statistically significant where p-values indicate we fail to reject the null
hypothesis at the conventional levels of significance.

Interestingly, except for the coefficient estimate of interacted term between the

selection effect variable and the trade intensity variable (that is, GDP/km2 x TI), evidence

suggests that all other coefficient estimates of interacted terms in the three models are
statistically not different from zero. Meanwhile, the Relative Income variable in both
models A and B is statistically significant at the 5 percent level of significance. These
results seem to suggest that for the pollutant nitrogen oxides, the scale of production, the
emission intensity of pollution and the composition of capital to labor ratio do not affect
emissions level. While the selection effect is robust in the two models, the neutrality of
the scale, technique and composition effects may be explained by the proposition in new

or intra-industry trade theory that assumes CES preferences (see Chapter Two). That is,

133

 

the findings: one, that the selection effect is statistically significant in the level form and
in the interaction forms, and, two, that the scale and technique effects are not statistically
significant; would suggest that trade-induced environmental-selection effect exists but
that environmental- scale effect and technique effects are absent. These results are
consistent with evidence found in trade literature which supports a trade-selection effect,

but not a trade-scale effect (see Head and Rice 1999, 2001).

4.6.3 Volatile Organic Compounds (VOC) and the Selection, Scale, Technique and
Composition Effects

Table 4.3 shows the estimates of the selection and scale effects which are statistically

significant in Models A and B at the 1 percent level of significance. Both the selection

and scale effects have positive relationship to the emissions levels of volatile organic

compounds (VOC). A 1 unit increase in the number of listed companies per square

kilometer (Companies/ kmz) leads to 64 percent increase in volatile organic compounds

emissions in Model A, and 68 percent in Model B. Coefficient estimates of the scale

effect in the linear-in- variables Model A and the main Model B indicate strong evidence

to suggest that a 1 unit increase in GDP/km2 leads to 14 and 12 percent increase in the

level of volatile organic compounds emissions, respectively.

Estimates for both the technique and the composition effects in the models A and
B are not statistically significant at conventional levels. Thus, we fail to reject the null
hypotheses that the coefficient estimates of the income (INC) and capital intensity (K/L)
variables are equal to zero. One possible explanation for the absence of a technique effect

is that a major proportion of volatile organic compounds emissions may not be subjected

134

Table 4.3: Estimation Results of Models A and B - Volatile Organic Compounds (V CC)

 

 

 

 

Estimation Method: Fixed Effects Random Effects
Model Specification: A B A B
. Levels Only Levels Only
Variable/Column: (I) (2) 43) (4)
Companies/ km 0497*" 0517*" 0520*" 0537*"
2 (0.054) (0.063) (0.055) (0.065)
0129*" 0112*" 0145*“ 0131*"
GDP/ km (0.032) (0.033) (0.040) (0.044)
(GDP/ kmz) Squared
Lagged per capita income -0.008 0.086 0.0003 0.057
(0.044) (0.1 16) (0.040) (0.097)
Lagged per capita income 0019 -0.013
squared (0.022) (0.019)
Capital abundance (K/L) -0.037 0.003 -0.046 0.015
(0.041) (0.098) (0.043) (0.081)
(K/L) Squared -0.003 -0.005**
(0.007) (0.006)
Trade Intensity (TI=X+M/GDP) -0.004** -0.004* -0.004*** -0.004
(0.002) (0.002) (0.002) (0.001)
Relative Income 0.001 -0.002 0.001 0.002
(0.004) (0.004) (0.003) (0.003)
Population Density 0.067 0.077 0.031” 0035*"
(0.055) (0.061) (0.013) (0.012)
Tl x Companies / km 0.001 0.002 0.001 0.002
2 (0.002) (0.002) (0.002) (0.002)
-0.0005 -0.0004 -0.0003 -0.0003
U x CD” km (0.001) (0.001) (0.001) (0.0006)
Tl x Per capita Income 0003* -0.002 -0.002 0.002
(0.001) (0.002) (0.001) (0.001)
Tl x (K/L) -0.003*** -0.002*** -0.002*** 0002*"
(0.001) (0.001) (0.001) (0.0008)
Companies x Per capita Income 0090'” 0.076 0.081 ** 0.068
(0.040) (0.047) (0.040) (0.046)
Intercept -7.538*** -7.821*** -7.115*** -7.366***
(0.717) (0.699) (0.252) (0.246)
Observations 21 l 21 1 21 1 21 1
Group 23 23 23 23
R2 . . 0.8220 0.8249 0.8209 0.8238
- wrthrn
R2 _ between 0.6507 0.6495 0.6796 0.6836
R2 0.6510 0.6500 0.6804 0.6862
- overall
Sargan-Hansen Test/ 12 (C10 23 7011””! 8704'”
LR Test] 12 (dt) 3.46....

 

 

***Significance at the 99-percent confidence level.
"Significance at the 95-percent confidence level.
*Significance at the 90-percent confidence level.

Note: Standard errors in parentheses are cluster-robust.

135

to environmental regulation. Indoor VOCs, for example, are outside the purview of the
regulatory authority; hence, regulation compliance is not enforceable. In the United
States of America, sources of volatile organic compounds (V DC) emissions comes
mainly from road vehicles, solvent use, fires and non-road equipment, which in the year
2002 makes up seventy five percent of total volatile organic compounds emissions (EPA
2009). The absence of a composition effect on volatile organic compounds emissions
may be explained by the possibility that changes in the level of capital intensity in
pollution intensive production do not affect volatile organic compounds. This may be the
case if the production of volatile organic compounds is not capital—intensive. That is,
shifts in the level of the use of capital in the VOC-intensive sectors have no effect on the
levels of volatile organic compounds emissions.

For the effect of openness to trade, coefficient estimates of the variable trade
intensity are statistically significant at the 5 percent level of significance in Models A and
at the 10 percent level in Model B. There is evidence suggests that a 1 unit increase in
trade intensity leads to 0.4 percent decrease in volatile organic compounds emissions in
Models A and B. These findings substantiate the prediction of a negative relationship
between trade intensity and emissions level. The results are consistent with findings in
the cases of sulfur oxides and nitrogen oxides.

Estimates for the trade-induced environmental effects for volatile organic
compounds, namely, the trade-induced scale, technique, composition and selection
effects are represented by the respective variables interacted with the trade-intensity
variable. In the linear-in-variables Model A, with the exception of the selection effect,

estimates of the trade-induced technique, composition and selection effects are

136

 

 

statistically significant at the conventional level of significance. In Model B, the sole
variable with statistically significant coefficient estimate is the trade-induced composition
effect. These results suggest that the one common statistically significant trade-induced
effect in the two models is the composition effect. However, when calculated for the
average country in the sample, the composition effect does not lead to any change in the
emissions level of volatile organic compounds. This result is consistent with the result of

neutral linear-form composition effect in models A and B.

4.7 Alternative Specification

In this section, I present three alternative econometric models to the main Model
B to test the robustness of the results obtained in the previous section. The three
alternative specifications are: Model C which specifies an estimating equation that is
absent of the selection effect, Model D which does not include any interaction terms, and
Model E which includes foreign direct investment as an additional trade variable. I
discuss the results in Tables 4.3, 4.4 and 4.5 for each Model C, D and E in turn. I begin
with estimates of Model C, followed by Model D, and finally with Model B.

Regressions of the Model C, which excludes the selection effect variable

Companres/ km , generate mostly statrstlcally 1ns1gn1ficant estrmates for the three

pollutants sulfur oxides, nitrogen oxides and volatile organic compounds.

For the pollutant sulfur oxides, fixed effects estimation of Model C shows one
statistically significant coefficient estimate at the 1 percent level, for the variable trade
intensity. In the cases of nitrogen oxides and volatile organic compounds, the sole

significant estimate is for the variable population density. An explanation for the lack of

137

Table 4.4: Sensitivity Analysis for Sulfur Oxides (SOx)

 

Dependent Variable: Log of Sulfur Oxides per sq km

 

 

 

 

 

Estimation Method: Fixed Effects Random Effects
Model Specification: C D E C D E
No No FDI No No FDI
Selection Interaction Selection Interaction
Variable/Column: (I) (2) (3) (4) (5) (6)
Companies/ km 0808*" 0439* 0688*” 0.591“
(0.102) (0.145) (0.230) (0.264)
GDP/ m2 —0.282 0.177 0305*“ -0.160 -0.l30 0.100
(0.130) (0.123) (0.1 17) (0.238) (0.089) (0.1 16)
Lagged per capita -0.004 -0.426* -0.464** 0.038 -0.030 -0.033
income (0.226) (0.188) (0.177) (0.259) (0.21 1) (0.207)
Lagged per capita 0.005 0.1 12 0.069 -0.009 0.008 -0.019
income squared (0.054) (0.046) (0.042) (0.060) (0.054) (0.048)
Capital abundance -0.062 0462* 0.1 10 -0.221 -0. 101 -0.307*
(K/L) (0.206) (0.158) (0.157) (0.218) (0.004) (0.173)
(K/L) Squared -0.026 -0.066** -0.039**"‘ ~0.013 -0.017 -0.007
(0.017) (0.013) (0.013) (0.014) (0.020) (0.012)
Trade Intensity -0.018*** -0.017*** -0.027*** - -0.011*** -
(TI=X+M/GDP) (0.003) (0.002) (0.003) 0018*" (0.040) 0.021 ***
(0.007) (0.007)
Relative Income 0.017 -0.023 0.011 0.019” 0027*“ 0010*
(0.017) (0.015) (0.013) (0.009) (0.010) (0.006)
Population Density 0.206 -0.293** -0.401*** 0.112 0.090*** 0.039
(0.134) (0.127) (0.1 13) (0.069) (0.036) (0.026)
TIxCompanies/ km '0'015 * * ‘0'007
2 (0.004) (0.006)
-0.001 0.001 0.001 0002*
T1 x GDP/ km (0.001) (0.001) (0.002) (0.001)
TI x Per capita -0.001 -0.008"‘** -0.001 -0.006*
Income (0.002) (0.002) (0.002) (0.002)
Tl x (K/L) -0.002 0.0001 -0.002 -0.001
(0.001) (0.001) (0.002) (0.002)
Companies x Per -0.239** -0.203**
capita Income (0.068) (0.095)
Foreign Direct -0.009*** -0.007**
lnvestrnent (0.002) (0.003)
Intercept -6.56** -3.67* -0.54 -5.34*** -6.63*** -5.02***
(1.609) (1.372) (1.333) (1.217) (0.639) (1.025)
Observations 211 211 210 211 211 210
Group 23 23 23 23 23 23
R2 . . 0.6312 0.7021 0.8002 0.6276 0.6643 0.7702
- wrth1n
R2 b 0.6203 0.2446 0.2779 0.6729 0.7120 0.7586
- etween
R2 0.6255 0.2200 0.2610 0.681 1 0.7366 0.7894
g - overall

 

138

 

 

Table 4.4: Sensitivity Analysis for Sulfur Oxides (SOx) Continued

 

Sargan-Hansen Test/ 525*“ 714*“ 7626*"
2
2 (d0

 

LR Test/ 12 ((11) ”1.45:”:- 66.44**"

 

*“Significance at the 99-percent confidence level.
MSignificance at the 95-percent confidence level.
*Significance at the 90-percent confidence level.

Note: Standard errors in parentheses are cluster-robust.

evidence to support the prediction of theory in these cases is that, in excluding the
selection variable, the estimation is not controlling for a theoretically relevant and
important variable that belongs in the estimating equation. The omission of the selection
variable leads to misspecification and omitted variable biases. The effects of
misspecification and variable omission are shown by comparing the results obtained in

Model C with the results obtained in the main Model B. Model B includes the selection

effect variable, Companies/ kmz.

. . . 2 . . .
When the selectron variable, Companres/ km , rs added, thus generatrng the marn

model that is Model B, all other coefficient estimates of the theoretically relevant
variables, namely, the scale, technique and composition variables, which were previously
not statistically significant in Model C, are statistically significant in Model B. In
addition, the estimates in Model B have the correct signs in terms of the direction of
effects on the dependent variables, thus conforming to theoretical predictions postulated
in both the inter-and infra-industry trade frameworks. These results are consistent with

findings in past empirical studies.

139

Table 4.5: Sensitivity Analysis for Nitrogen Oxides (N Ox)

 

 

Dependent Variable: Log of Nitrogen Oxides per sq km

 

 

 

 

 

Estimation Method: Fixed Effects Random Effects
Model Specification: C D E C D E
No No FDI No No F Dl
Selection Interaction Selection Interaction
Variable/Column: (l) (2) (3) 44) (5) (6)
Companies/ km 0360*" 0.246" 0365*" 0321‘"
(0.039) (0.066) (0.051) (0.090)
mm km2 -0. 185 0074* 0.086 -0113 0049*" 0.074
(0.054) (0.047) (0.053) (0.126) (0.018) (0.050)
Lagged per capita 0.122 -0.009 -0.016 0.066 0.015 0.025
income (0.095) (0.072) (0.080) (0.1 17) (0.078) (0.090)
Lagged per capita -0.039 -0.008 -0.010 -0.027 -0.014 -0.017
income squared (0.023) (0.018) (0.019) (0.030) (0.022) (0.019)
Capital abundance 0.098 0.200“ 0.130 0.1 15 0.146 0.089
(K/L) (0.086) (0.061) (0.071) (0.1 13) (0.096) (0.065)
(K/L) Squared -0.008 -0.017 -0.011"‘ -0.010 -0.014 -0.009
(0.007) (0.005) (0.006) (0.009) (0.009) (0.006)
Trade Intensity -0.0003 -0.001 -0.004 -0.001 -0.001 -0.003
(TI=X+M/GDP) (0.001) (0.001) (0.001) (0.004) (0.001) (0.003)
Relative Income 0002 -0.003 -0.005* 0.002 0.001 -0.001
(0.007) (0.006) (0.006) (0.004) (0.003) (0.003)
Population Density 0.226“ 0.024 0.008 0.107*** 0060*" 0054*"
(0.056) (0.049) (0.052) (0.042) (0.013) (0.011)
. -0.004 -0.002
TI x Companrezs/ km (0.002) (0.003)
0.0004 0.0004 0.001 -0.00004
n x GDP/ km (0.0006) (0.001) (0.001) (0.0007)
Tl x Per capita Income 0.0001 -0.0004 -0.00005 -0.0002
(0.0008) (0.001) (0.002) (0.002)
Tl x (K/L) -0.0006 -0.0004 -0.001 -0.0004
(0.0005) (0.0004) (0.001) (0.001)
Companies x Per capita 0.01 1 0.009
Income (0.03 1) (0.029)
Foreign Direct -0.001* «0001
Investment (0.001) (0.001)
Intercept -8.76**"‘ -7.42*"'* -6.82*** -7.46*** -7.67"* -7.36***
(0.675) (0.528) (0.606) (0.509) (0.321) (0.377)
Observations 211 211 210 211 211 210
Group 23 23 23 23 23 23
R . . 0.4037 0.5948 0.6191 0.3869 0.5896 0.6103
- wrthrn
R2 b 0.6489 0.7424 0.6783 0.6483 0.7715 0.7619
- etween
R2 0.6508 0.7895 0.7535 0.6529 0.7931 0.7841
- overall
Sargan-Hansen
2 2066*“ 2127'“M l.8e+06*
Test/ 2' ((11) n-
LR Test/ [2((10 9344:": I 1.93’”

 

140

Table 4.5: Sensitivity Analysis for Nitrogen Oxides (N Ox) Continued

"*Significance at the 99—percent confidence level.
"Significance at the 95-percent confidence level.
*Significance at the 90-percent confidence level.
Note: Standard errors in parentheses are cluster-robust.

Therefore, the results imply that the omission of a relevant variable, Companies/

km2 which is the selection effect variable, leads to the serious problem of wrongly

specified regression models. The consequences of misspecification errors due to the
correlation between the explanatory variable and the error term are biased estimates of
coefficients.

In Model D, interaction terms are dropped from the main model specification. In
the model for sulfur oxides (SOx), responses to the scale, technique, selection and trade
intensity effects are statistically significant at conventional levels. However, the response
to the composition effect is not statistically significant. In the case of nitrogen oxides
(N Ox), coefficient estimates of the scale, selection and composition effects are
statistically significant, but the coefficient estimate of the technique effect is not
statistically significant. Finally, in the case of volatile organic compounds (VOC), only
the estimates of the scale, selection and relative income effects are statistically
significant.

Because Model D excludes the interaction terms, these results may be interpreted
as estimates of the environmental effects on pollution in the autarky case. Hence, in
model D, trade-induced effects are not estimated. Consequently, responses to
environmental effects that are fully or partly explained by trade are not captured in the

coefficient estimates at autarky levels.

141

 

Table 4.6: Sensitivity Analysis for Volatile Organic Compounds (V CC)

 

 

Dependent Variable: Log of Volatile Organic Compounds per sq km

 

Estimation Method:

Fixed Effects

Random Effects

 

 

 

 

Model Specification: C D E C D E
No No F DI No No F DI
Selection Interaction Selection Interaction
Variable/Column: (l) (2) (3) (4) (5) (6)
Companies/ km 0470*" 0524*" 0.478" 0539*"
(0.055) (0.063) (0.046) (0.065)
GDP, ka -0.266 0.066 0112*" -0.158 0.066" 0132*"
(0.242) (0.052) (0.033) (0.219) (0.033) (0.045)
Lagged per capita 0.175 -0.106 0.089 0.067 -0.093 0.062
income (0.166) (0.098) (0.1 16) (0.143) (0.096) (0.097)
Lagged per capita -0.047 0.017 -0.020 -0.025 0.013 -0.014
income squared (0.040) (0.025) (0.022) (0.035) (0.025) (0.019)
Capital abundance -0.017 0.259 0.001 0.045 0.217 0.124
(K/L) (0.165) (0.164) (0.099) (0.153) (0.143) (0.084)
(K/L) Squared -0.001 -0.023M -0.003 -0.006 -0.019** -0.005
(0.013) (0.010) (0.008) (0.01 1) (0.009) (0.006)
Trade Intensity -0.001 0.001 -0.003* -0.002 0.001 -0.004"”"
(TI=X+M/GDP) (0.003) (0.002) (0.002) (0.002) (0.001) (0.002)
Relative Income 0.006 0.007“ -0.002 0.007 0.008" 0.002
(0.007) (0.004) (0.004) (0.005) (0.004) (0.003)
Population Density 0.324" 0.050 0.079 0120* 0054*” 0.037"
2 (0.149) (0.059) (0.061) (0.066) (0.021) (0.012)
. 0.002 0.002
Tl x Companrezs/ km (0002) (0.002)
0.001 -0.0005 0.001 -0.0003
TI x 60W km (0.001) (0.0008) (0.001) (0.0006)
Tl x Per capita Income 0.001 -0.002 0.001 0.002
(0.002) (0.0002) (0.002) (0.002)
Tl x (K/L) -0.002 - -0.002 -
(0.001) 0002*“ (0.001) 0002*"
(0.001) (0.001)
Companies x Per -0.077 0.070
capita Income (0.047) (0.047)
Foreign Direct -0.0005 0.0005
Investment (0.0008) (0.0008)
Intercept -9.54*** ~8.06*** -7.86*** -7.35*** -8.02*** -7.39***
(1.447) (0.556) (0.702) (0.546) (0.461) (0.238)
Observations 211 211 210 211 211 210
Group 23 23 23 23 23 23
R . . 0.6295 0.7577 0.8252 0.6048 0.7569 0.8242
- wrthrn
R2 - between 0.5351 0.6728 0.6496 0.5105 0.6840 0.6809
R2 0.5276 0.6876 0.6514 0.5049 0.6940 0.6849
- overall

 

142

 

Table 4.6: Sensitivity Analysis for Volatile Organic Compounds (V OC) Continued

 

Sargan-Hansen Test/ 1520*” 2732*“ 1.5e+06*
2 18*
1 (d0

LR Test/ 12 (C10 1580“" 68.5””

 

 

*"Significance at the 99-percent confidence level.
"Significance at the 95-percent confidence level.
*Significance at the 90-percent confidence level.

Note: Standard errors in parentheses are cluster-robust.

However, and interestingly, note that for all the three pollutants, coefficient
estimates of the selection effect variable are statistically significant at the 1 percent level.

The above findings suggest the following. One, there is evidence to substantiate
the claim that the selection effect is a relevant and important variable in modeling the
impact of dirty production and of trade on environmental quality. Two, the findings
provide evidence of the robustness of the earlier results with respect to estimates of the
environmental effects of scale, selection, technique, and composition.

Finally, in Model B, the trade variable foreign direct investment (F D1) is added to
the main Model B to examine the environmental impact of the flow of factor movement

across borders. Results show that for the pollutants sulfur oxides and nitrogen oxides,
adding the FDI variable worsens the R2 -statistic when compared to the main Model B.
For sulfur oxides, overall R2 is 0.2722 and 0.2610 for Models B and E, respectively,
while for nitrogen oxides, overall R2 is 0.7685 and 0.7535, respectively. In the case of I

volatile organic compounds, the R2 -statistic in Model B is 0.6500, and it is slightly

improved in Model E at 0.6514.

143

Coefficient estimates of the effect of the F DI variable are statistically significant
at the 1 percent level for sulfur oxides and at the 10 percent level for nitrogen oxides, but

are not statistically significant at the conventional levels for volatile organic compounds.

A 1 unit 1ncrease m FDI/ km leads to a 0.9 percent decrease 1n sulfur ox1des emissmns

and a 0.1 percent decrease in nitrogen oxides. Therefore, in the model for sulfur oxides,
there is strong evidence to suggest that inward flow of foreign direct investment mitigate
the emissions level of sulfur oxides, and some evidence to suggest that foreign direct
investment lowers the emissions of nitrogen oxides.

Note that the estimates of the coefficients of selection effect variable, Companies/

km , are statlstlcally Significant for all three pollutants, sulfur ox1des, nitrogen ox1des and

volatile organic compounds at the 10, 5 and 1 percent levels of significance, respectively.
These results indicate the robustness of earlier findings with respect to coefficient
estimates for the selection effect variable.

Estimates of the scale, technique, composition, trade and population density are
statistically significant at 5 percent level or less for sulfur oxides (SOx). In the case of
interaction variables between trade-intensity and the selection and technique effects,
coefficient estimates are found to be statistically significant at the 5 percent level or less.

In the model for nitrogen oxides (N Ox), coefficient estimates of the composition
effect variable and the relative-income variable are significant at the conventional levels.

Finally, for volatile organic compounds (VOC), estimates of the scale, trade
intensity and trade-induced composition effects are significant at the conventional levels

of significance.

144

 

 

 

Therefore, the results show that in Model E, the signs and magnitudes of
statistically significant coefficient estimates for the three pollutants: sulfur oxides,
nitrogen oxides and volatile organic compounds, conform to and are consistently nearly
identical to the signs and magnitudes of significant estimates in the main Model B. The
exceptions are the mixed responses to the environmental effects of scale, technique and
composition in Model E for nitrogen oxides and volatile organic compounds.

Comparatively, the coefficient estimates of the effects of selection and relative-
income variables for nitrogen oxides emissions (NOx) are, respectively, 0.257 and -0.006
in Model B, and 0.246 and -0.005 in Model B. Meanwhile, the coefficient estimates of
the statistically significant effects of the selection, scale, trade intensity and trade-induced
composition effects on volatile organic compounds (V DC) are, respectively, 0.517,
0.112, -0.004 and -0.002 in Model B, and 0.524, 0.112, -0.003 and -0.002 in Model B.
Thus, the results suggest that adding the foreign direct investment variable in Model B in
the case of volatile organic compounds does nothing to improve or worsen the
estimations of variable effects.

Furthermore, since evidence fails to reject the null hypotheses that the coefficient
estimate of the foreign direct investment (F DI) effect is zero, dropping the FDI as a

Variable will not affect the model specification for volatile organic compounds.

4.8 Discussion
The impact of intra-industry trade on environmental quality can be decomposed
into scale, selection and technique effects. Previous empirical studies based on the

traditional trade framework shows that the environmental effects of inter-industry trade

145

 

can be decomposed into scale, composition and technique effects. Most, if not all
countries engage in both intra- and inter-industry trade. The premise of the analysis in
this paper is that international trade is composed of both types of trade, namely, trade in
homogenous goods described by the theoretical framework of traditional trade, and trade
in differentiated goods described by the framework of new trade theory. The integrated
environmental effects of both inter- and intra-industry trade can therefore be decomposed
into four environmental effects namely the scale, technique, selection and composition
effects.

The selection effect distinguishes the impact of trade driven by market structure
and increasing returns from the impact of trade that is driven by comparative advantage
based on cross-country differences in factor abundance. If countries engage in the
pollution-intensive production of homogenous and differentiated goods in an integrated
economy, then an empirical estimation of the total impact of trade on the environment
needs to account for a selection effect in addition to scale, technique and composition
effects. In this paper, findings show there is strong evidence of the statistical significance
of the selection effect in the trade-environment linkage.

Results show that the ex-ante prediction of an increasing relationship between the
selection effect and emissions level is borne out in the data. Statistically significant
coefficient estimates are found have positive signs, a result that is robust across the six
empirical models and across all three pollutants, sulfur oxides, nitrogen oxides and
volatile organic compounds. These results conform to the prediction of a nonnegative
trade-induced selection effect postulated by the intra-industry trade framework under the

assumption of both non-CES and non-homothetic utility function. In other words, there

146

is strong evidence to support the hypothesis that the level of emissions is increasing in the
selection effect. A theoretical explanation is that, trade in differentiated goods induces a
negative selection effect: openness to trade implies increased competition from abroad
which leads to a reduction in the number of domestic firms or in the number of product
varieties. Then, holding the scale and technique effects constant, a decrease in the number
of firms leads to a decrease in emissions level.

Furthermore, the current analysis suggests there is a linear and positive
relationship between selection and emissions level. This finding provides evidence for
the intra-industry trade framework which assumes that firms have identical technology
where a fall in the number of firms implies a proportionate fall in emissions level,
holding other factors constant.

Estimates of responses to the second environmental effect, the scale effect, are as
follows. First, in the main model of the current analysis, Model B, statistically significant
estimates of a positive relationship between the scale of production and emissions level
are consistent with theoretical predictions. In the models for sulfur oxides and volatile
organic compounds, coefficient estimates suggests that an increase (decrease) the level of
scale of production raises (lowers) the emissions levels of sulfur oxides and volatile
organic compounds, holding other factors equal. Second, findings show that responses to
the scale effect variable are statistically significant for the functional form that specifies
linearity in the scale variable. This result provides evidence to indicate that for countries
in the sample, the assumption of homothetic production or consumption prevails over the
assumption of non-homothetic functions. In the context of international trade effects,

since homotheticity is a feature consistent with trade in intra- rather than inter-industry

147

 

goods, the evidence suggests that the realization of data is more consistent with trade
driven by market structure and increasing returns rather than by cross-country
differentials in factor endowments.

For the third environmental effect, the technique effect, coefficient estimates are
negatively signed and statistically significant for sulfur oxides estimations. However,
estimates are not significant at the conventional levels for nitrogen oxides and volatile
organic compounds. This finding suggests that for sulfur oxides, there is evidence to
support the hypothesis that emissions level decreases in abatement activities or in the
stringency of environmental regulations. On the other hand, in the cases of nitrogen and
volatile organic compounds, one reason for the lack of evidence for the technique effect
is that nitrogen oxides and volatile organic compounds are not subjected to strict
environmental enforcement. The pollutants are difficult to regulate: nitrogen oxides gases
are formed from the emissions generated in transportation vehicles, and they contribute to
ozone formations and fine particle pollution. Volatile organic compounds emissions are
mainly sourced from transportation and indoor products. In contrast, sulfur oxides
emissions are subjected to rigorous environmental laws and standards at plant and firm
sites, and in the abating process which includes regulation on equipment requirements
and production methods.

Findings also suggest that similar to the scale effect, the technique effect is
statistically significant when measured as level-form income per capita but not
statistically significant when measured in non-linear or quadratic form. This result

provides evidence to suggest that the assumption of a homothetic consumption function is

148

 

 

relevant for countries in the sample. The assumption is consistent with a CES utility
function in a pollution model of intra-industry trade.

For the fourth environmental effect, the composition effect, the analysis finds
evidence to support the theoretical prediction of a positive and increasing relationship
between emissions level and capital intensity. Similar to the technique effect, the
composition effect variable is statistically significant in the models for sulfur oxides, but
is not significant in the models for nitrogen oxides and volatile organic compounds. One
possible explanation for these findings is that sulfur oxides are generated in highly capital
intensive production processes but that nitrogen oxides and volatile organic compounds
emission methods are not dependent on capital-intensive production sectors.

In the open trade context, statistically significant coefficient estimates of the
interacted terms between capital intensity and trade intensity (K/L x TI) provide strong
evidence to suggest that trade-induced composition effects lead to increased sulfur oxides
emissions.

Interestingly, in contrast to the scale and technique effects, coefficient estimates
of the composition effect variables are statistically significant in the quadratic or non-
linear form, but are not statistically significant in the level-form. This result conforms to
the theoretical assumption of inter-industry trade, where the impact of capital
accumulation on pollution is linked to non-homothetic production or consumption, owing
to cross-country differentials in income and/or production techniques.

Finally, in the case of the environmental effect of openness to trade or trade
liberalization, coefficient estimates are statistically significant in the models for sulfur

oxides and volatile organic compounds models. For sulfiir oxides, coefficient estimates of

149

the trade intensity variable are consistently significant and negative in signs for the five
models considered, Models A through E. For volatile organic compounds, in four out of
the five model specifications, estimations yield statistically significant trade intensity
coefficient estimates. These results suggest strong evidence for the prediction that greater
openness to trade or greater trade liberalization leads to lower emissions level. One
explanation for the findings is that when pollutants are confined to domestic borders,
imports of cleaner goods produced abroad substitute for dirty goods produced
domestically, thus mitigating emissions level in the home market. A second explanation
is that openness to trade brings in cleaner and more efficient abatement technology that is
imbedded in imported goods or imported production technology. The diffusion of cleaner
techniques of production in the domestic market alleviates emissions production.
Interestingly, this possibility is consistent with the result of negative coefficient estimates
on foreign direct investment, which are statistically significant in the models for sulfur
oxides.

I note that the parsimonious model specifications adopted in the current analysis
captures the essential aspects of the trade-environment relationship as postulated by
theory. Statistically significant semi-elasticity estimates of variables do not differ in any
great manner across the fixed effects and random effects estimations. In the fixed effects
Models A and B, coefficient estimates are invariably found to be identical or nearly
identical in their magnitudes. The signs or direction of effects of statistically significant
variables are consistently identical across models and across estimators. Furthermore, and
more importantly, statistically significant results conform to theoretical predictions and

can be explained in terms of the underlying assumptions of the frameworks which form

150

 

 

the basis of the econometric models specified. Findings are supportive of theoretical
expectations consistent with intra-industry trade theory — this is not smprising
considering our data describes countries that are known to engage in the trade of
differentiated goods. But the findings also support theoretical predictions of inter-
industry trade as evidenced in the estimates of the composition effect. Statistically
significant capital intensity coefficients conform to the assumption that the trade-induced
composition effect is generated by non-linearity in the techniques of production, brought
about by income and price differentials across countries. On the other hand, the
composition effect estimates do not conform to assumptions that can be explained by
intra-industry trade and homothetic production. This result is significant and interesting

since new trade theory does not generate a composition effect.

4.9 Conclusion

International trade is comprised of trade in both homogenous and differentiated
goods. Theory suggests the environmental effects of an integrated open economy can be
explained by factors that drive inter-industry as well as intra-industry trade. In this paper,
an analysis of panel data from OECD countries provides evidence of the following
results. First, strong empirical evidence supports the hypotheses postulated by theory that
emission levels are increasing in the scale of production, in emission intensity, in the
number of firms, and in the composition of the economy. Second, the estimation of the
impact of international trade on environmental quality needs to control for the selection
effect, in addition to the widely recognized scale, technique and composition effects.

Statistically significant estimates of the selection effect are shown to be robust across six

151

 

 

different model specifications and across three types of pollutant. Third, in consonant
with the study by Antweiler et al. (2001), the findings in this paper show that responses to
the scale, technique and composition effects are statistically significant for sulfur oxides.
However, in contrast to findings in Antweiler et al. (2001), the results show that for the
countries in the sample, estimates of the scale and technique effects can be explained by
the assumptions of homothetic production or consumption, consistent with the intra-
industry or new trade framework rather than with inter-industry or traditional trade
framework. On the other hand, the estimates of the composition effect can be described
by the assumption of non-homothetic production, consistent with the inter-industry trade
framework. This result replicates the finding in Antweiler et al. (2001). Estimations of the
scale, technique and composition effects for the pollutants nitrogen oxides and volatile
organic compounds generate mixed findings, which can be explained by the inherent
characteristics, processes and sources of the respective pollutants. Fourth, the analysis
suggests that openness to trade or trade liberalization is beneficial to the environment in
that there is strong evidence to indicate that greater openness to trade or increased
liberalization of trade frictions leads to lower levels of emissions.

The findings in this paper build on previous findings in literature which seek to
resolve the heated debate of whether international trade is good or bad for the
environment. This paper departs from previous work by showing that the estimation of
the environmental impact of international trade needs to account for a selection effect, a
variable that helps explain the total volume and integrated patterns of the exchange of

goods across borders. The omission of a relevant variable can lead to specification error

152

 

 

 

and biased estimates. The significance of unbiased estimation lies in the effect it has on
resolving an important issue which may have implications on policy recommendations.
Finally, further empirical investigations based on the model presented in this
paper may include testing the theoretical predictions on a larger sample to include a
greater number of countries in the world economy, and on groups of countries defined by
developmental stages such as the developing countries including China and India, and by

geographical proximities such as the North and South countries.

 

153

 

APPENDICES

 

154

Appendix A: Monopolistic Competition, Trade and the Environment
A.1: Production

The profit function is the firm’s revenue less labor cost and pollution taxes such that:
(A.1) 7:,- = [)0 — 6)q,- — wa - wflq; - H:
where
P = Pi ((1—91)qi)
By the substitution of equations (1.3) and (1.7) into (A.1), the following profit function is

obtained:

(A-Z) ”i = Pi ((1 ‘91)61i)(1 ‘91")(11 ‘ W“ ”W561i ‘ 7(1—9ilach
First Order Conditions

The first order conditions for profit maximization with respect to q implies:
(A.3) p'(1—a)2q+p(1—0)—wfl—r(1—6)5=0
Divide equation (A.3) through by (1—- t9) and rearrange to obtain:

(A.4) p'(1-9)q+p—r(1—19)5" = wfl(l—0)_1

The first order conditions for profit maximization with respect to 0 is:

(A.5) —qp'(l—6)q—qp+6r(l—6)6—lq=0

Divide (A.5) by q and rearranging, equation (A.5) can be rewritten as:

(A6) p'(1—6l)q+p—(57(1—9)§_1 =0

Second Order Conditions

The second order condition for profit maximization with respect to quantity of output is:

155

 

 

622?

H l 2 l
(A.7) '57:!) (1—6)q-(1-6)q+p(1—6) +p(l-t9)
‘1
6p 62p 62”
Since p' = —— < 0 and assuming p" = ——2— < 0 , then —-—2— < 0 and satisfy the second
661 aq aq

order condition for a maximum.

The second order condition for profit maximization with respect to quantity of abatement
is:

62” (5—2
a—gf-qp"(-q).(1—e)q+.,,,.(1-9),.(_1),.p.(_q).(5_.)5,(,_,) -(—1)
(M) g:— = P"(1-9)q3 - p'(1-6)q3 -p'q—(6—1)5r(1—9)‘5“2

2 2
Since p' = 353% > 0 and assuming p" = 2—3 < 0 , then 2—7: < 0 and satisfy the second

a92 662
order condition for a maximum.

Solve for the abatement fraction, emission per unit output, the pricing equation, and
the number of firms/product varieties
Equate equations (A.4) and (A6) to obtain:

61(1—6)6_l—r(1—9)6_1= w,8(1—t9)_l

Divide through by (1— 9)_I , collect terms and rearrange to obtain:

r(I—t9)6 (5-1): wfl
Solve for theta:

l

_ _ Wfl 35
(.9. [W]

 

156

 

 

Thus, the fraction of output allocated towards abatement, 0, is fixed and is decreasing in
wage w, and the labor coefficient ,6 ; it is increasing in the emission tax 1' , and the
elasticity of emission with respect to the consumption fraction of output 6 .

Then, substituting into the equation for emission per unit (1.9), emission intensity can be

written as:

 

 

(A.10) —
_ Wfl 5 _ Wfl
8‘ "l“w-ul mire—1)]

Marginal revenue in elasticity term
Let C = Lc = (I —6)q
Then,
P = p(C) = PHI-(9)61)
Taking the derivative of price with respect to quantity of output:

6p 6p 6C 6;) ,
—=—.——=—. 1—9 = 1—9
aq ac aq ac( ) p( )

Then, from equation (1.1 1), the first order condition of profit maximization with respect

to output:

1911-4502q+p(1—6)—w/3—r(1—0)‘S =0

Rewrite as:
6p 6
56(1’6)q(l—6)+p(l—6)= Wfl+T(l—6)

Substituting for demand, C = (1— 6)q ,

157

 

 

a
3%C(1~6)+p(1—0)=wfl+r(l—6)6

Divide through by (l — 6):

 

0p wfl 6—1
—C = 1—9
ac "LP (1—6)+T( )

Multiply the first term on the LHS by p / p to obtain:

 

p933,” ”’3 +r(1—6)5_l

6C p (1—6)

pl“%l=(:[;)+’("6)5_l

where 77 = _[§&Q]/[§g] is the elasticity of demand.
p

 

Substituting for the price elasticity of demand, we can rewrite equation (A.4) as:

 

__1_ = W,8 _ 5—1
(A.11) p[1 77) (1_6)+r(1 9)

where 77 is the price elasticity of demand.

Then by substitution of equation (A.9) into equation (A.11), the pricing equation can be

obtained as follows:

 

 

1— has)":

Simplifying, write the equation as:

158

 

 

1 “ 52:1
6 (5
p 1—1- = wfl w’6 + r —W’B—-
77 r (6 — l) r(6 — 1)
Divide the LHS and the RHS by wage rate, then rearrange and collect terms to obtain:

(m) fill—i)” lag/inf flag/fol?

Equation (A.12) is the pricing rule.

 

 

 

Rearrange to obtain:

1

p034” fil‘ (5 1)) [IT'- 1>il=(wfl)lr(;€1)l6l(757)l
p11-%1=<wfl>e'i1(£»1

That is,

—1
, E: __1_ a 5
(A.12) w [l 77] e lfl[(-6—-_ 1)]

Equation (A.12’) is the PE line.

 

 

From the Zero-profit condition, obtain P = AC in the following way:
p(1—0)q = wa + wflq + r(1-6)5 q

Divide by quantity of output:

p(1—6)= waq'l + wfl+ r(l —6)6

Divide through by wage rate, rearrange and collect terms to obtain:

159

 

 

~52((1-6))_1[aq"1+fl+%(l—-6)§]

Substituting for 6 , rewrite the above as:

—if 6
£__ wfl 6 £1— _2'_ wfl 6
w- [46-1)] q+fl+w [T(5-l)]

or

p: wfl E g 1

W [ms—1)) lq+flll+(5-l)]l
Simplifying, obtain:

g=[,(;e.,)‘$[aqw[w—inll

Equation (A. 13) is the ZE or zero-profit line.

 

 

 

 

 

 

 

To solve for the number of firms, use the Full-employment condition:

L=§(a+flqi)=n(a+6q)=n(a+fl%_6))=n

Solve for the number of firms, n:

r v1

(A.14) n L a+ =

 

/

 

K

a+

,6Lc

 

 

17.2/i.)

l

15.

 

 

 

 

 

160

1
W6 5 —1
\ [705—01 ) aL +flc[

 

1:24.13

 

 

Growth in Emission level and the Scale, Technique and Selection Effect:

From equation (1.25):

n n

(1.25) Zz,=Zeiqi:>nz=n-e-q
i=1 i

Let Z = nz and rewrite (1.25) as:

(A.15) Z = neq

Take a logarithmic transformation of equation (A. 15) and write as:

(A.15’) logZ = logn+loge+logq

Totally differentiate:

(A.15”) l-dZ=-l—dn+lde+la'q 0r £=fl+§5+§q
Z n e q Z n 9 q

Then, take the percent change (multiply through by 100%) to obtain:

+6}

N)

(A. 16) 2 = a +
Equation (A.16) shows that the change or growth in total emission can be decomposed
into, or is the sum of, the selection, technique and scale effects (respectively).

A.2 PE AND ZE LINES

Use the FCC with respect to q from the profit maximization problem and the Zero-profit

condition to obtain the PE and ZE curves.

The equilibrium conditions can be written as:

=[1_%]“[T(;{1)]_3fl[3%31

(A.13) €=(,(:€1)]—g jag—l ”8 [751-13)]

161

 

(A.12’)

31‘s

 

 

 

 

Rewrite as:

:=ll-a‘:sllri[é?—illiflffil

 

Z” Tflifililg [...—1.11 1651101

Let M {4:61)}; and B = fl[(g‘§—6]

I

 

 

Then, rewrite the PE line as:

 

—I
(A.12”) E=[1— 1 ] M'IB

W ”(C(60)

Rewrite the ZE line as:

(A.13’) £=M"[aq" +13]
W

Slopes of the PE and ZE curves

To find the slopes of the PE and ZE curves, totally differentiate the two equations with
respect to output level. Assume no changes in the parameters, wage rate and emission tax
rate.

Total differentiation of the PE equation implies:

P

iéqflzljl—(MY')_2("(C))-2%3'gil

The PE curve is upward sloping since ii < 0 and EC- > 0.
c

661
Total differentiation of the ZE equation implies:

I62

 

dlgl
w = —M_latq_2 < 0

Therefore, ZE curve is downward sloping.
Shifts in PE and ZE due to a change in emission tax rate:

Assume no changes in output level, wage rate and the parameters.

1

3
Let (I! = [i] , B = ,6 [—13—] and rewrite the PE and ZE curves as the following:

(6—1) (6-1)

I
_] _
PE: E=(l—(77)_l) r5w-IB
w

I

ZE: =r5w_l[aq-1+B:l

ale

Totally differentiate the PE curve with respect to emission tax rate:

1 1-1 1 "
d(£)— ——r5 “1“] 6‘13] dr=0
w 6 77

Therefore,

d(PE)=d[€)= lré'l[[1_1]_lw—13] >0 QED

dr dr 6 r]

 

Therefore, a change in emissions tax shifts the PE curve in a positive direction.

Totally differentiate the ZE curve:

163

 

 

 

w
Thus,
p
d — 1
d ZE ( j T4
( )z W =-1—r(> w“[aq"+B]>o QED
dz' dr

Therefore, a change in emissions tax shifts the ZE in a positive direction.

A.3 Comparative Statics of the Effect of Change in Emission Tax on Abatement
Level, Price Level, Emission Intensity, Output Level and the Number of Product
Varieties (Comparative Statics of Change in Emissions Tax on PE and ZE Curves)

Assume no changes in the parameters and the wage rate.

(i) From the first order condition of profit maximization with respect to output, q,

and by substitution for the fraction of output allocated towards abatement in
equation (A.7), and simplifying, the PE curve is:

(Al.12’)20 £=[1_l]-l 12i[_(“_’€)_]-%fl[_§_]

w '7 (6 ' 1) (5 ’ 1)
where 77 is the elasticity of demand such that 77 = 77(c(q,T))

[—11—]

Then rewrite equation (A1 .12’) as the following:

I

-1 _
(141.12”) 3:04] r5w-IB
w 77

‘

Note that the first order condition of profit maximization with respect to the fraction of output allocated
towards abatement, 6, which is equation (1.13) yields the same equation as (Al.12) after substitution of

the equilibrium value of 6.

164

 

 

 

From the zero profit condition and after substituting for the fraction of output allocated
towards abatement in equation (A.7) and simplifying, we rewrite equation (1.17) as:

(Al.13) ‘Ezréuwfi—W l—ll:aq +£[fgfijl

1‘” [FET 1)] “"dB ifs—"(3]

Then (Al.13) can be written as:

 

'00!—

l
(A1.I3’) —p—=r5w-l[aq'l+B:|
w

In the following, we evaluate the effects of an increase in the emissions tax, 2' , on total

P

production, q , real price of consumption, — , consumption, c, and other endogenous
W

variables.

 

The equilibrium system of equations can be simplified by substitutions (see dissertation
equations 2.18 and 2.19) to two equations that are functions are production, price, and the
emissions tax,

I

-1 _
(Al.14) F‘ =(1-77"') r5w‘13—an
W

l

(Al.15) F2=r6w"'[aq“+3]—an
W

where w is a constant.

The negative of elasticity of demand, 77, is a function of consumption, c, as described by
equations 2.3 in the dissertation,

(Al.16) 77=77(c)
with g < 0 as indicated by dissertation equation 2.4.
0

Equations 2.8 and 2.14 show that c is, in turn, a function of q and r . Using equation
(2.24), consumption is written as:
l

(Al.17) c=c(q,T)=L-le-—gq

165

with derivatives

6c_

(Al.18) 6—- —- [lair—g
q
and
l
ic: —6_lz' 11:le gq
(Al.19) 5?
_ _ 66
-51 1q_
661

Equations (A1 .14) may be written as
I
-1 .—
(A1.20) F' ={1—q“[c(q,r),p]} ”(1743—350
w
Derive the partial derivatives of F l and F 2 with respect to the production, price, and the

environmental tax. The partial derivatives of F l are

-7 l

_ ~ _ 67766 ’ -1
A12] F'=—1— ‘ 2———r5w B>O
( ) q ( 77 j 77 acaq
(A122) F), = —w“‘ < o

l

l
-2 _ __ -1
FT] 2—(1‘77—1) 77—2flfiraw—l8+6-lT-lr(sw—IB(1_n—l)
6c 62'

(A123) 1

= (1—77—1 )_2 rgw—lB[—77—2 %E§E+(l—7fl )6_lr—l:l:0

The partial derivatives of F 2 are:
I

(A124) 71,2 =—r5w"aq‘2 <0
(A125) FIE =—w" <0

.1 l _1
(A126) 1”,2 =6_lr—lr5w_l [ozq—l +3] 2 6_lr-lr5w_lB(l—77’l) > 0

where the second line of equation (Al .26) comes from using equations (A1 . 14) and

—1
(Al.15) to solve for (1—77‘4) B = ozq"l +8.

166

 

 

F 1 is the PE curve and describes the equilibrium in the consumption market. The slope

p
d —— F1
of the PE curve is 7‘31 = -—C1L w > 0. F 2 is the ZE curve and describes the zero profit
‘1 F p
d 3 F2
equilibrium in production. The slope of the ZE curve is —d-l—V— = ——%w < 0.
q F
p

Equations (Al.14) and (A1 . 15) give a non-linear two equation system that determines the
equilibrium values of production and the price of consumption. Derive the system of
differential equations from equations (Al .14) and (Al.15):

ngq + ngp+ ngr = 0
(A127)
[7(1qu + Fjdp + Ffdr = 0

Take the total derivatives ? and 38 based on (Al .27) by dividing each equation by
r r

dr. In matrix form, the result of dividing by the differential of the emission tax is

dq l
— -F
(A128) A d’ = T
d_p —FT2
dr
where
F‘ F'
(A129) A: Z Z
Fq Fp

Solve the system for the total derivatives by finding the inverse of A,

I 1

F F
(Al.30) A“: q p D“1

F2 F2

q P
where

_ l 2 2 I

(A130) D _ Fqu —Fq 17,, < 0

The sign of D‘1 follows from equations (A121), (Al .22), (A124), and (A125).
167

 

 

Multiplying both sides of equation (A1 .30) by A—1 gives

271‘ F2 F. ,
— -F

-- q q r

-dr_

 

 

Performing the matrix multiplication gives

d61_ —1 1 2 2 1
(A132) E’D (FpFr—FPFT)
and

dp__ —1 2 1 1 2
(A133) d—T—D (FqFr—FqFr)

Substituting equations (A121), (A123), (A125), and (A126) into equation (Al .32)
gives

(A1.34)

d ’1“ ‘1 ‘2 6 6 l
—q-=—D_lw—l6—]r’lr5w—IB(1—77‘l) —D’lw_l(l—77_1) 77411757343
dr 6c 6r

1

—, —1
+D—lw—l6'lr—lr9w—IB(1 -77—])

The first term in the second line of equation (A1.33) cancels with the fourth line, so

I

dq —1 —1 —1 ‘2 —2 577 56' — -1
Al.35 —=—D w l- ——r5w B
( ) d1 ( 77 ) 77 6c 61
Thus, since €1<0, c>0 and 77>0,theseimply 22£<0,sothat£11<0.
6c 60 77 (12'

Analysis of equation (A1.33) shows that consumption price rises in response to an
increase in the emission tax,

2
_ —I
(AI.36) SE = D—lr5c77'2136_lr_lq'l [877—] a—ni—afl (1-77'4) ]> 0
2'

Since 0'1 <0 and a—Tl£<0.
6c 77

168

 

 

Next, solving for dc/dtau, we use the market clearing condition:

In equilibrium, total net output (supply) is equal to total consumption (demand) as shown
in equation (2.8). Then, take the total differential and solve for the effect of a change in
emissions tax rate on consumption level as the following.

(1—6)q = Lc

By substitution for the equilibrium value of the fraction of output allocated towards
abatement, 6 and rearranging, this implies,

l l
_. 1 _
—1 —1 W73 5 —1 ’" WP 5
c = L l— 6 = L -——— = L r 5 ——
( ), giro—1)] q [(3-1)]
Taking the total differential with respect to consumption, output and emissions tax rate,
we obtain the following:

 

 

_ ‘4 _.
arc—L" ”’3 dq+L" [—11 6 “”6 dr
1 (6 — 1) 6 (6 — l)
Dividing both sides by d r , and recognizing that e =( (:fl 1)], this implies:
T .—

1 1
51:: [1,3 fl_ie}§

(A137)
dr dr L6r
. dq . . . . . . . . dc
S1nce 718 shown to be negatlve 1n the foregomg analy31s, this 1mp11es that 6— < 0.
z' I

Find the change in firm-level emission due to a change in emission tax rate:

Emission is:
z = eq

Total differentiation implies:

dz = qde + edq
Then,

dz de dq

._ _—_ q_+e_

dr dr 612'

169

 

 

Therefore, the change in emission at the firm level depends on the change in emission
intensity due to a change in emission tax rate (tax-induced technique effect) and the
change in output due to a change in emission tax rate (tax-induced scale effect). There is
no tax-induced selection effect at the firm level.

The effect of a change in emission tax on the number of firms or product varieties is
given by the following:

dn _dn d_q= d_n_ d_q

dr dr dq dq dr

Note that from full employment equation we have:

L
=(01+,6q):L(a+'Bq)l

Taking the total differential :
dn = {—L (a + fig).2 ,6} dq
This implies:

Then,

. dn . . . .
S1nce :1— < 0 , the change in the number of firms or product varieties moves 1n the
q
opposite direction of the change in the output level when there is a change in emissions

tax rate.

Note, the policy-induced selection effect can be written as:

:—:— 1——] >1.——.:7.1———< 1.7657112:

where

__1;_
(a+,6q)

n:

170

 

 

Trade and the Environment

The ZE line is:

l
ZE: £=r6

117—~71

The market clearing condition is:

 

(l—6)q=Lc or q:

 

471-9):
1

Then, substituting for the equilibrium level of abatement, 6, gives:

Then, rewrite the ZE curve as the 22 line:

-1.

 

 

 

_1. +4

 

 

 

Total differentiation of the 22 curve is:

(7({31 = ((4)62 5%} dc = —c‘2 %dc

Therefore, the slope of the 22 curve is:

w = —c-2 3
dc L

<0

which implies that the ZL curve is downward sloping.

Given consumption, the wage and emission tax rates and the parameters, a change in the
labor population, L , shifts the 22 curve in a negative direction. This is shown as the
following:

7131
6(ZZ)= w =—LT2£<O
6L 6L c

 

Table A.2: Comparative Statics Effects of a Change in Emission Tax Rate

 

 

—- 1
ea 1 -1 w,6 6 -1 _
6t 6t [T(5-l)] (T) e >

An increase in emissions tax rate increases the level of abatement undertaken by

 

firms.
59.6.-- -2 fl =_ —1_£ =1
2' 6r— T {(64)} T [t(6—1)] r<0

An increase in emissions tax rate increases lowers emission intensity. This is the
policy-induced technique effect.

 

 

172

 

 

Table A.2: Comparative Statics Effects of a Change in Emission Tax Rate

 

 

 

 

 

 

 

Continued
dz de d
3. _=(q_+e.21<o
dz' dr dr
or
£=é+q
A change in firm-level emission due to a change in emission tax rate depends on
e d
-—- , the policy-induced technique effect and —q , the policy-induced scale effect.
dr 612'
d 2 a 1
_ '— _ _ _ _ _ C _ _ '—
4. £=D lr5w 23612“qu 8771—1—-aq1(1—771) >0
dr 6c 77
An increase in emissions tax rate raises the price level.
d _ , _ ‘2 _ _ _ .. 6 6 c
5. i=—D 1(1—77 I) 771116'pw 2 _77£+_77_ <0
dr 6p 77 6c 77
An increase in emissions tax rate leads to lower output level. This is the policy-
induced scale effect.
1 1
dc _ - d "
6. —= L ‘e5 i-l—eé <0
dr dz L6z'
A change in emissions tax rate leads to a negative change in consumption level.
dn L d
T (a + ,6q) 7
An increase in emissions tax rate depends on the scale effect, that is, on whether the
effect of emissions tax leads to a contraction or expansion of output production.
Since the scale effect is negative, a higher emissions tax leads to an increase in the
number of firms. This is the policy-induced selection eflect.
dZ dn de a' >
8. —- = [(eq —+(nq)— +(ne)—1)—0
d 1 dr dr dr <
or
~ . . . >
1 Z = (n + e + q) — 0

 

An increase in emissions tax rate may increase or decrease the total level of
emissions in the economy depending on the sum of the policy-induced scale,
selection and technique effects.

 

173

 

 

 

Table A.3: The Impact of Intra-lndustry Trade on the Economy

 

Effects 0f1ntra-industty Trade
on:

Note: All countries are identical

 

I. Abatement (6)

Openness to trade does not lead to a direct change on the
level of the firm’s abatement activity. However, a trade-
induced technique eflect occurs if trade leads to an
expansion or contraction of the scale of production
which leads to a rise or fall in income level. This in turn
leads to either an increase or decrease in the stringency
of environmental policy respectively. Firms react
accordingly by increasing or reducing abatement
activity to mitigate the cost of environmental
compliance. This leads to a trade-induced technique
effect if abatement affects a change in emission
intensity.

 

2. Emission per unit output

(9)

Free trade does not lead to a direct change in emission
intensity. However, a trade-induced technique effect
occurs if trade leads to an expansion or contraction of
production that affects income level, which in turn leads
to a change in the stringency of environmental policy.
Firms lower or raise emission intensity by undertaking
more or less abatement as a response to a change in
environmental policy.

 

3. Price (p/w)

Trade implies as if there is a doubling of labor, thus the
22 curve shifts downward which leads to a fall in
price.

 

4. Consumption (c)

The 22 curve shifts downward, leading to a fall in the
consumption of each product variety.

 

5. Quantity of output (q)

Entry and exit of firms with free trade imply surviving
firms take advantage of scale economies, increasing
output. This is the trade-induced scale eflect.

 

6. Number of Firms or
Varieties (77)

As output increases and labor is fixed, the number of
product varieties or firms necessarily falls with trade.
This is the trade-induced selection eflect.

 

7. Total emission (Z)

 

 

The total effect of intra-industry trade on total emission
level is the sum of the trade-induced scale, technique
and selection effects:

d2 = (eq)dn +(nq)de + (ne)dq

or write in percent change:

A

Z=h+é+q

 

 

174

 

 

Appendix B: Intra-Industry Trade and the Environment
B.l: Production in Trade-Environment Model with CES Utility Function
The profit function is the firm’s revenue less labor cost, pollution taxes and abatement

cost such that:
(8.1) 7r,- =p(1—6)q,~—wa—w,6q,——rz,-
where
p = p, ((1-QM.)
By the substitution of equations (2.3) and (2.7) into (B.l), the following profit function is

obtained:

(B2) 7r,- = p,((1—e,)q,.)(1—e,.)q, —Wa —qu,. -r(1—6,-)6 q,-

The first order conditions for profit maximization with respect to q implies:
(3.3) p'(1—e)2q+p(1—e)—wp-r(1—e)5=0

Divide equation (B.3) through by (1— 6) and rearrange to obtain:

_1_ wfl
-(1-9)

 

7 6
(B.4) p (1—6)q+p—r(l—6)
The first order conditions for profit maximization with respect to 6 is:

(3.5) —qp'(l—6)q—qp+r6(l—6)6_1q=0

Divide (B.5) by q and rearranging, equation (B.5) can be rewritten as:

(B.6) p'(l—6)q+p—r6(1—6)6—1=0

Using (B.4), this implies:

175

 

 

l

__ we 3
(3.7) 6—1 (454))

 

The fraction of output allocated towards abatement, 6, is decreasing in wage w, and the
labor coefficient ,6 ; it is increasing in the emission tax 2' , and the elasticity of emission

with respect to the consumption fraction of output 6 .

Substituting for elasticity of demand, we can rewrite equation (B.4) as:

(3'8) pl“il"(“9)6_l=11‘:/;)

where 77 is the price elasticity of demand for net output.

 

Then by substitution of equation (B.7) into equation (B.8), the pricing equation is

obtained:

711—:112111W1

Since price elasticity of demand is equal to 1/(1— p), equation (B.8) can be rewritten as:

l

(3.10) p-p=(gé_-l-)[(WP)6" r(6—1)]5

l
I 6 6—1 ]“
01‘ = — -——— z- 6 _1 6
p p(,_,)[14) 1 )
Then, price is decreasing in p , and increasing in 6 , w (wage), ,6 (labor productivity

coefficient), and 2' (emission tax imposed by the government).
Free entry of firms implies that the zero profit condition is given by the following

equation:

176

 

 

 

(3.11) p(1—6)q—wa-wpq—r(1—9)5q=0
By substitution of equation (8.7) into equation (8.1 l) we solve for output level, as
follows:

The zero profit condition is:
(13.11) p(1—6)q—wa—wpq—r(1—6)5q=0

But price is:

 

p[1—-:;)—r(1_9)6—1:(1t:flg)

Rewrite as:

 

p-,0=r(1-t9)5—l + W'B

 

(I-o
Thus,
_ 1 - wfl
p—;{r(l-6)6 l+(l—6)}

Substitute into the Zero Profit Condition:
p(1—6l)q—war—w,8q-z'(l—(9)(S q =0

i{r(1-—9)5"+ Wfl }(l—6)q-wa-wflq-r(l-9)5q=0 l

 

 

p (1-9)

Simplify by cancelling (1 - 6) :
1 5 6
—{r(1—t9) +wfl}q—wa—wflq—r(l—t9) q=0
p

Divide by q:

6

%{r(1—6’)(S +w,6}—wozq_1 -w,B—r(l—6) = 0

Collecting terms:

177

 

1(1—9)5 [i-l]+wfl[—;——l:l—waq-l =0

p

Substituting for 6 from (8.7):

l 5
_ _ Wfl :5: _1__ W i- -W _1=
H [1 (ta—6)] ] [p I} flip 1] W 0

 

Simplifying:

(

1 5
MB 3 -1-— w i— —w _l-
T (ta—6)) ] [p I} flip 1] “q ‘0

K
( wfl l 1 _1_
Ttr<I-6>)[3'li+wfiiZ‘li'waq ’0

Cancel r and collect terms:

lé-Illla‘iilwfll-W“=0

 

 

 

 

Divide by w:

 

 

Simplify:
[LIM [3 J+fl(1-6)}=aq_1
P (1-5) (1-5)
lfllll”““““”ll=aq"‘
p (1-5)
[till—Mlle“
p (1-6)
Solve for q:

178

 

 

lflllfi‘ialr

(_____1- p)!” _
pu— 6) "’

Thus,
ap(l—6)
3.12 =-————
( ) fl5(1-p)

Then, net output, (1—6)q , is the following:

(8.13) qnet_ [Wfl)5[ap(5-l 011/6 ]

 

r ,66(1-p)

To solve for the number of product varieties, we use the full-employment condition:

 

(3.14) L: i( a(+pq,.)

i=1
Then, solving for n , the number of varieties is:

L

(8.15) n=a+flq

 

And by substitution of equation (B.11) into equation (8.14), obtain:

L6(1— p)

(8.16)n:(1—a[5p)(+p(—6 1)]

 

Equation (B. 1 5) implies that n is a constant, a function of all fixed parameters. It
indicates that the number of firms or varieties is increasing in labor, L . With free trade,

the consumer’s range of varieties available for consumption will be n + N .

179

B.2: Pollution Supply
Government and Emissions Tax

The regulatory authority levies an emissions tax on firms to internalize the
extemality associated with the production of pollution.

Then, to solve for tax, insert the demand equations in (2.3) and (2.3’) into (2.1) to

obtain the indirect utility function:

(817) VH yp(”pp/(p WWM ))
. = —n(pz

("pp/(P-1)+ "*p*P/(P 1))p

 

where y = w+ s and s = 12 z, = rnz by symmetry, and is a lump sum transfer.
i=1

Then, assume symmetry and write (3.11) as:

V” (rnz)p(npp/(p_W)+npp/(p 1))
= —ngoz
("pp/(P‘1)+n*p"'P//(P 1))p

 

(rnz)p np'D/(p—l )+n p Mp/(p 1)
VH = ( p )—ngoeq
(”pp/(P- )+n* p*P/(P-1))

 

(B.18)

 

_ (ritzy) (”Pp/(p—l )+n P Wp/(p 0). __‘V_V.B_
_ _1 17¢ T(5-1) q
(”pp/(P-H)” p "-p/(P l)”

n /<—) n /( —1)
< w + lw-‘l—(W L

V” =
("pp/(w )+,, ,, p/(p 1)) 6 -1)

 

Then, environmental policy will be given by maximizing equation (3.12) with respect to

emission tax, 2' . The first order condition is:

180

 

 

 

 

 

6VH prp (nz)p(np (’0 )+n p Mp/(p 1)) fl
— +ngor 2[ W ]q=0
5T ( n/pp/(p— I)“, p *p/(p-|))p (5‘1)
Then,

6V” pr-l (nz)p(npp/(p_ )+n p Wp/(p 1)) ,5
: =—n¢T-2 (_w“ ]q
at (”pp/(r )+,, p p/(p— 1)) (5‘1)

(np p/(p 1”)“,1, PAP-1))

 

 

Let w:
("pp/(p- )+n p "/P (P- 1))p

Then,

 

 

 

(3.19)-'7 ziiflwp—lh; 160%}:

Equation (B. 1 9) indicates that the emission tax is a function of the marginal disutility for
pollution, the number of product varieties produced at home and abroad, price, wage and
the fixed parameters. Since wage is income from working, then emission tax is a function
of income. Equation (B. 1 9) shows that an increase in the wage rate increases the emission

tax rate:

181

 

 

 

 

:37; = (1 + p)w_(2+p) Haw" {(39:5qu; > 0

Thus, equation (8.20) shows an increase in the wage rate which increases income, leads

to a rise in the stringency of environmental regulation.

Equation (B. l 9) shows the Samuelson rule, which can be written as:

Tz’ll‘llu'lw4ldfnlqlfi;

The term on the right hand side of the equation indicates the marginal damage of

 

pollution for each consumer. Note that the number of product varieties and the level of
output are functions of fixed parameters, including labor. Thus, emission tax rate varies

with economic conditions and can be written to be as:

(8.21) z'=t(n,n*,p,p*;go,w,L,,6,6,p,a)

182

 

 

Table 8.]: Comparative Statics Effects of a Change in Emission Tax Rate
in the CES Model of lntra-industry Trade and Environment

 

1

~ 1
(39 l -l wfl 5 _1 -
at 6T [45-1)] (T) e >

An increase in emissions tax rate increases the level of abatement undertaken by firms.

An increase in emissions tax rate increases lowers emission intensity. This is the policy-
induced technique effect.

3. dz {fie—”flaw

 

 

 

3? dr dr

or
2=é+é

. . . . . . e
A change m firm-level emISSIon due to a change In emISSIon tax rate depends on — ,
T

d
the policy-induced technique effect and —dq , the policy-induced scale effect.
T

 

l

F6 _
4. :—:—=r—5—(p(§-l))_l[(wfl)6_l(5—l)]5>0

 

An increase in emissions tax rate raises the price level.

 

92:0
62'

An increase in emissions tax rate does not affect the scale of production. The policy-
ina’uced scale effect is neutral.

 

 

 

183

 

Table 8.]: Comparative Statics Effects of a Change in Emission Tax Rate
in the CES Model of Intra-industry Trade and Environment
Continued

 

_ 1:2 1 l—l"6
6. fl=—r L 5 )(wfl)5 ap(6—1) / <0

6r 6 ,B6(l—p)

An increase in emissions tax rate leads to a decrease in net output level.

 

 

6 (1+6) fl 2;—
_€=_ _ a _”’_ -1
7. 67 2' [(6—1)] L q<0

A change in emissions tax rate leads to a negative change in consumption level.

 

 

s. 91:0
a:

An increase in emissions tax rate does not affect the equilibrium number of firms in the
economy. The policy-induced selection eflect is neutral.

 

dZ dn de dq >
9. —= — — _ _0

dz' ((eq) dr +(nq) dr +(ne) d1) <

01'

Z=(n+é+ “)30

An increase in emissions tax rate may increase or decrease the total level of emissions
in the economy depending on the sum of the policy-induced scale, selection and
technique effects.

 

 

 

 

184

Table B.2:The Impact of Intra-Industry Trade on the Economy
with CES utility function.

 

Effects OfIntra-industry Trade

on

(Note: All countries are identical)

 

l.

Abatement (6’)

Openness to trade does not lead to a direct change on the
level of the firm’s abatement activity. The trade-induced
technigue effect is neutral.

 

Emission per unit output

(6)

Free trade does not lead to a direct change in emission
intensity. However, a trade-induced technique eflect
occurs if trade leads to a change in the stringency of
environmental policy. Firms lower or raise emission
intensity by undertaking more or less abatement as a
response to a change in environmental policy.

 

Price (p/w)

In the CES model of trade and environment, product
price is unchanged under the free trade regime.

 

Consumption (c)

The opening of trade in the integrated world economy
leads to imports and exports which implies there is a
greater number of product varieties available for
consumption. However, the scale of production, the level
of abatement and the price level is unchanged with open
trade, thus there is no change in the level of consumption
of domestic product varieties.

 

Quantity of output (q)

Constant elasticity of substitution of the utility function
leads to fixed equilibrium gross output level. This is the
neutral trade—induced scale eflect.

 

Number of
FirmsNarieties (n)

As output is unchanged, the number of product varieties
or firms is also unchanged with trade. This is the neutral
trade-induced selection eflect.

 

 

Total emission (Z)

 

The total effect of intra-industry trade on total emission
level is the sum of the trade-induced scale, technique
and selection eflects:

d2 = (eq)dn + (nq)a’e + (ne)dq
or write in percent change:
Z=n+é+é

 

185

 

 

 

Appendix C: How Does Intra-industry Trade Affect the Environment?
C.1 Reduced Form Equation

The scale effect in the model can be further decomposed into three effects: (i) the
effect of labor force, L, (ii) the effect of consumption or demand, c, and (iii) the effect of

the fraction of output allocated towards production of goods rather than towards

abatement, (1 - 6). Equation (3.9) indicates that:

t #
C=C(n,n ,p,y.t.r;p ,p.5,W)
Then, the scale effect as represented by demand/consumption is given by:
A A At A A A A* A A "
(Cl) C = ecnn+ean +ecpp +ecyy‘+ec,r + .9ch +ecww+ecpp+£c55

From equation (3.6), producer price is a function of emission tax and the fixed parameters

including wage. Then, price can be reduced in terms of its primitive factors:

(C.2) [3:8 f+e .f)*+£

p, pp ww+eppp+ep56

p

Lastly, pollution demand and supply are combined to obtain the final reduced form
equation where the technique effect will be represented by basic determinants obtained
through the efficient tax equation. From equation (3.15), efficient tax is a function wage
(income), the number of product varieties, producer and world prices, and the fixed

parameters for preferences and emissions, such that:
(03) r=1(n.n*.p,p*;<o.w,L,fl.6,p,a)

From equation (3.9) and (2.10), emission per unit output can be solved for and written as:
(C.4) e=e(2';,6,6, w)

Then, the effect of per unit emission can be written as:

186

(C.5) é =eerf+eewfii+eeflfi+ee5c§
From equation (3.5), the effect of abatement can be written as:
(C6) (5 = £9tf+50wiii+£0flfi+eg5c§

Thus, from equation (3. 10), the pollution demand equation is:

(C7) 2 = a + f. + 5 + e — (1 — 6) Pollution demand equation

Let the Scale effect be defined by total domestic consumption:

A

S = i + 5 - (1 - 61)
Then, rewrite (C.7) as:
(cs) 2=a+s+é
Substituting for emission intensity using equation (C.5) to obtain:
(C9) 2 = h+ S + {emf + eewfiz+eeflfi+eegc§}

Pollution supply from equation (3.15) implies:

(CJm

r = erpn+emm +erpp+erptp +Er¢¢+8rww+5rLL+5rflfl+5r65+8ma 4'5er

Substituting (C.10) into (C.9) implies:

187

 

 

(011)

Z =n+S+eeTr+eeww+eeflfl+8856
‘+ ‘*+ ‘+ ‘*+ ‘+
A A 8 n 8 m 8 p 6‘ *p 8 (0 x A
_ .. rp tn rp rp 27p ..
Z—n+S+eeT . +eeww+eeflfl+ea56

etwfit+ SILL + ewfl + 6,565 + grad + amp
Open bracket and collecting terms:
A A A A A* A A * A A
Z = n + S + eererpn + eaemm + swamp + eflgrpatp + screwy) + sereTLL
+eeremoi + swamp + enemy} + sewn) + eflawfl + 685,6 + 8a8T55 + .9956

Rearranging :

A A A* A A* A A
Z=(l+ee,e,p)n+S+eeTem*n +eeterpp+eererptp +£etewgo+eeretLL

+eeremo'i + emerpf) + (eererw + sew) 132+ (swam + eefl )8 + (5915,55 + £35 )5
Substitute for producer price in equation (CH) and obtain the reduced form equation:
(012) 2 = n1a+n2s+n3w+n4tf +1151“; +H6£+H7p+n83+n93+n10¢3+n11é
We adopt a parsimonious reduced form to exclude the effects of fixed parameters,

namely, p,a, ,6,§, go , and assume these as country-specific effects. We note that the

effect of the number of imported product varieties multiplied by foreign price level is the
value of imports. We sum the two effects to indicate the amount of imported goods and
denote the sum as M . Further, we note that the effect of labor and wage can be summed
as the effect of total income, and denote it as Y . Then, the following reduced form

equation is obtained:

((3.13) 2 = nlfi+n25+n3f+n41f4

188

C.2 Calculating Real Capital Stock Using Method in Learner (1984, pp. 233):

Let
I, = gross domestic investment in year t in units of home currency;

P,” = implicit gross domestic investment deflator at time t with base year b, Pbb = 1.0;
e, = exchange rate in time period t, dollars per unit of home currency;
6 = rate of depreciation

The real capital stock at the end of year t in year is calculated by converting “investment

flows year by year into dollars and use the US. gross domestic investment (GDI) deflator
to convert to constant dollars:

I .
K, = P,”($)Z (1—5)"/(1jej /ij($))
j=0
Asset life of 15 years is selected. The corresponding rate of depreciation commensurate
with double declining balance method is 13.3%.”
C3 Calculation of Marginal Effects

The partial effect of a change in the selection effect on SOx in Model B is (see
Wooldridge (2003) pp.):

%A 5?). = 100[exp(afi,m -1) — 1] = 100[exp(0.548) — 1] = 72.97%

The partial effect of a change in the composition effect (capital intensity) on SOx in
Model B is (see Wooldridge (2003) pp. 192-193):

A
%A SOx =100[0 — 2(-—0.036)5.275] = {1oo[2(0.036)5.275]} (6.275 — 5.275) = 37.98%.

189

 

C.4 SUMMARY STATISTICS

FIXED EFFECTS (FE) AND RANDOM EFFECTS (RE) ESTIMATION USING STATA:
log: C:\Users\Sarma\Documents\EMPIRICAL ANALYSIS\DOFILES\OECD\NOV
2009\OECD Trade-Environ 11.05.09.smc1

use "C:\Users\Sarma\Documents\EMPIRICAL ANALYSIS\DATA\OECD
Pollution\OECD Data INTERACTIONS 8.12.09.dta"

***SUMMARY STATISTICS***

sum office countrycode country year company avewgnp fdi sox nox voc
1and2km pop cgnp rgdpl rgdptt openk ppp labor k POPDNSTY SOXKM NOXKM
VOCKM logSOXKM logNOXKM logVOCKM _Iyear_1996 _Iyear_1997 _Iyear_1998
_Iyear_1999 _Iyear_2000 _Iyear_2001 _Iyear_2002 _Iyear_2003
_Iyear_2004 GNPCAP GNP RELINC KLINC TIRINC COKM ATRANSKM COKMSQ ATRANSP
LANDKM GrDP GrDPKM GrDPKMSQ KapL KapLSQ GNPcap GNPcapMA INCapL INCapLSQ
AWGNP GrNP KapLINC COM COMKM FDIKM POP POPDNSY TID INCAPD COMD TICOMD
TINCD INCSQD TICSQD COMKMSQ GDPINCD

Variable | Obs Mean Std. Dev Min Max

............. 1------------------_------------_-__---_-_-____------_-_-
office I 0

countrycode | 259 15.57143 8.866239 1 30
country | 0

year | 259 1999.517 2.869697 1995 2004

company | 259 886.3205 1502.627 18 8851

_____________ +_-___-__-__-_-----_----___-_--_---___-_---------_-_-_---

avewgnp | 259 2.26e+10 2.09e+09 1.96e+10 2.61e+10

fdi | 257 2.928871 3.147977 -3.597555 23.10155

sex | 253 1327.166 2967.182 7.4 17182

nox | 253 1607.198 3904.295 25.9 22405

voc | 253 1337.149 3102.822 7.1 19577

............. +-__-------_--------------------------_-------_----------

1and2km | 259 1283477 2698198 39770 9161923

pop | 259 42241.79 58511.56 267.478 295409.6

cgnp | 259 98.93482 2.437064 92.53037 108.3746

rgdpl | 259 20314.07 7608.587 5080.958 36100.44

rgdptt I 259 20316.29 7513.61 5097.181 36196.11

............. +-_-__--_---__---------_------_--__-__-__-_----_---------

openk I 259 70.31149 30.51214 17.065 186.7788

ppp | 259 14891.72 97543.08 .6384608 911798.4

labor I 259 2.01e+07 2.97e+07 152790.9 1.52e+08

k I 220 1.33e+12 2.45e+12 7.55e+09 1.16e+13

POPDNSTY | 259 .1151543 .1097338 .0023551 .4852282

------------- +—--—-.------——-——-——-—-—-------—-——-—-—-—--—-—-—--—---—-—

SOXKM | 253 .0023544 .0023263 .0000738 .0141053

NOXKM | 253 .0029997 .0026903 .0001768 .0138711

VOCKM | 253 .0025026 .0019346 .0000708 .0081169

logSOXKM | 253 —6.750883 1.394456 -9.513943 -4.261206

logNOXKM | 253 -6.274315 1.081451 -8.64038 -4.27795

............. +-__-----_-_----------------__---_-_-__--___---__--__----

logVOCKM | 253 —6.451888 1.160987 —9.555327 -4.813805

_Iyear_1996 | 259 .1003861 .3010959 0 1

_Iyear_1997 | 259 .1003861 .3010959 0 1

190

191

_Iyear_1998 | 259 .1003861 .3010959 0 1
_Iyear_1999 | 259 .1003861 .3010959 0 1
............. 1---_____----__-_-__-_-_-_____-_-__-__-__-__-___-___-----
_Iyear_2000 | 259 .1003861 .3010959 0 1
_Iyear_2001 | 259 .1003861 .3010959 0 1
_Iyear_2002 | 259 .1003861 .3010959 0 1
_Iyear_2003 | 259 .1003861 .3010959 0 1
_Iyear_2004 | 259 .1003861 .3010959 0 1
............. +-_----_--_-__------------__-_-_-____-_-------___--_-__--
GNPCAP | 259 2014072 768892.9 514145.9 3623461

GNP | 259 9.39e+10 1.85e+11 5.45e+08 1.07e+12

RELINC | 259 4.159284 8.167973 .0278396 51.06895
KLINC | 220 6.95e+15 1.45e+16 2.68e+13 8.19e+16
TIRINC | 259 176.3229. 215.8261 1.811234 1358.609
_____________ +--__---__-____-_-_------___-__---____-__----_------_---_
COKM | 259 .0028064 .0039876 .000079 .0211579
ATRANSKM | 259 1.146915 1.294471 .0290214 7.267563
COKMSQ | 259 .0000237 .0000616 6.25e-09 .0004477
ATRANSP | 259 5.849271 15.67781 .01817 95.66226
LANDKM | 259 12.83477 26.98198 .3977 91.61923
_____________ +_-_____-_-_--_--_-___-_--_---____--______----__---___--_
GrDP | 259 9.39556 18.36956 .0561082 106.6441

GrDPKM | 259 2.241202 2.339727 .0533382 8.938198
GrDPKMSQ | 259 10.47617 17.76185 .002845 79.89139
KapL | 220 5.274884 2.3053 1.079551 11.80081

KapLSQ | 220 33.11465 29.33276 1.165431 139.259
............. 1----------------_-_--_----_--_---__-----_--_-------___-_
GNPcap | 259 2.014072 .7688928 .5141459 3.623461
GNPcapMA | 257 2.007256 .7240291 .5400245 3.521008
INCapL | 256 2.001342 .7192028 .5400245 3.467017
INCapLSQ | 256 4.520604 2.788162 .2916265 12.0202
AWGNP | 259 2.264583 .2085505 1.957 2.612
_____________ 1---___-----___----___-_-_-_-_-_---------__---__--_--___-
GrNP | 259 9.388832 18.46268 .0544821 107.0405
KapLINC | 220 69.50029 144.6858 .2681757 819.055

COM | 253 6.353096 18.23276 .0222 108.929

COMKM | 253 .8179843 .5549844 .0221446 2.604412

FDIKM | 257 2.390672 5.230807 -8.486001 53.06591
............. +_----------------_-_____--__-_----__-_-_---___---------_
POP | 259 42.24179 58.51156 .267478 295.4096

POPDNSY | 259 11.51543 10.97339 .2355064 48.52282

TID | 259 2.55e-06 30.51214 -53.24649 116.4673

INCAPD | 256 4.30e-07 .7192028 -1.461318 1.465675

COMD | 253 3.786-08 .5549844 -.7958397 1.786428
............. +_-__-------___-_----_---------____-____-__-_----_-_---_-
TICOMD | 253 —3.177935 9.697517 -41.88528 23.58792
TINCD | 256 -5 952525 25.15625 —112.6293 71.36241
INCSQD | 256 -3.01e-07 2.788162 -4.228978 7.4996
TICSQD | 256 -22.54219 96.23718 -401.9096 208.7931
COMKMSQ | 253 .9758886 1.140806 .0004904 6.782964
_____________ +-_------------_______-_-_---__----------_---_-_-__-_-___
GDPINCD | 256 .3720973 1.303684 -3 169328 3.26564

BIBLIOGRAPHY

192

 

Bibliography

Anderson, K. "The Standard Welfare Economics of Policies Affecting Trade and the
Environment," K. Anderson and R. Blackhurst, The Greening of World Trade
Issues. Ann Arbor: University of Michigan Press, 1992, 25-48.

Antweiler, Werner; Copeland, Brian R. and Taylor, M. Scott. "Is Free Trade Good
for the Environment?" The American Economic Review, 2001, 91(4), pp. 877-908.

Balassa, B. and Bauwens, L. "Intra-Industry Specialization in a Multi-Country and
Multi-Industry Framework." Economic journal, 1987, 97, pp. 923-39.

Baldwin, R.E. "Measurable Dynamic Gains from Trade." Journal of Political Economy,
1992, 100(1), pp. 162—74.

Barnett, A.H. "The Pigouvian Tax Rule under Monopoly." American Economic Review,
1980, 70, pp. 1037-41.

Barrett, S. "Strategic Environmental Policy and International Trade." Journal of Public
Economics, 1994, 54, pp. 325-38.

Baumol, William J. and Oates, Wallace E. The Theory of Environmental Policy.
Cambridge: Cambridge University Press, 1988.

Bommer, R. and Schulze, Gunther G. "Environmental Improvement with Trade
Liberalization." European journal of political economy, 1999, 15, pp. 639-61.

Broda, Christian and Weinstein, David E. "Globalization and the Gains from Variety."
Quarterly Journal of Economics, 2006, 121(2), pp. 541-85.

Cameron, A. Colin and Trivedi, Pravin K. Microeconomics: Methods and
Applications. Cambridge: Cambridge University Press, 2006.

Chamberlin, Edward Hastings. The Theory of Monopolistic Competition. London:
University Press, 1962.

Chichilnisky, G. "North South and the Global Environment." American Economic
Review, 1994, 84(4), pp. 851-74.

Conrad, K. "Taxes and Subsidies for Pollution-Intensive Industries as Trade Policy."
Journal of Environmental Economics and Management, 1993, 25, pp. 121-35.

Conrad, K. "Choosing Emission Taxes under International Price Competition," C.

Carraro, Y. Katsoulacos and A. Xepapadeas, Environmental Policy and Market
Structure. Dordrecht: Kluwer, 1996, 85-98.

193

Copeland, Brian R. "International Trade and the Environment: Policy Reform in a
Polluted Small Economy." Journal of Environmental Economics and
Management, 1994, 26, pp. 44-65.

Copeland, B. R. and Taylor, M. S. "North-South Trade and the Environment."
Quarterly Journal of Economics, 1994, 109, pp. 755-87.

Copeland, B. R. and Taylor, M.S. "A Simple Model of Capital Mobility, Trade and the
Environment," NBER Working Paper Series No. 5898. 1997.

Copeland, Brian Richard; Taylor, M. Scott and Princeton University. International
Economics Section. Trade and the Environment: Theory and Evidence.
Princeton, N.J.: Princeton University Press, 2003.

Copeland, B. R. and Taylor, M.S. "Trade and Transboundary Pollution." American
Economic Review, 1995, 85, pp. 715-37.

Cole, Matthew A. and Elliot, Robert J.R. "Do Environmental Regulations Influence
Trade Patterns? Testing Old and New Trade Theories." World Economy, 2003,
26(8), pp. 1163-86.

Dasgupta, S.; Mody, A.; Roy, S. and Wheeler, D. "Environmental Regulation and
Development: A Cross-Country Empirical Analysis," Working paper, World
Bank, Policy Research Department. 1995.

 

Dean, Judith M. "Testing the Impact of Trade Liberalization on the Environment:
Theory and Evidence." Canadian Journal of Economics, 2002, 35(4), pp. 819-42.

Dixit, A. and Stiglitz, J.E. "Monopolistic Competition and Optimum Product Diversity."
American Economic Review, 1977, 67, pp. 297-306.

Frankel, Jeffrey A. and Rose, Andrew K. "Is Trade Good or Bad for the Environment?
Sorting Out the Causality," NBER Working Paper Series. Cambridge, 2002.

Freeman III, Myrick A. The Measurement of Environmental and Resource Economics.
Washington, D.C.: Resources for the Future, 2003.

Gamper-Rabindran, Shanti. "NAFTA and the Environment: What Can the Data Tell
Us?" Economic Development and Cultural Change, 2006, 54(3), pp. 605-33.

Grossman, Gene and Krueger, Alan B. "Environmental Impacts of a North American

Free Trade Agreement," NBER Working Paper Series, No. 3914. Cambridge,
1991.

194

Grossman, Gene and Krueger, Alan B. "Environmental Impacts of a North American
Free Trade Agreement," P. M. Garber, The (IS-Mexico Free Trade Agreement.
Cambridge, MA: MIT Press, 1993, 13-56.

Grossman, G. M. and Krueger, Alan B. "Economic Growth and the Environment."
Quarterly Journal of Economics, 1995, May 1995, pp. 353-77.

Gurtzgen, Nicole and Rauscher, Michael. "Environmental Policy, Intra-Industry Trade
and Transfrontier Pollution." Environmental and Resource Economics, 2000, 17,
pp. 59-71.

Hanley, Nick; Shogren, Jason F. and White, Ben. Environmental Economics: In
Theory and Practice. New York: Oxford University Press, 1997.

Harbaugh, William T.; Levinson, Arik and Wilson, David Molloy. "Reexamining the
Empirical Evidence for an Environmental Kuznets Curve." The Review of
Economics and Statistics, 2002, 84(3), pp. 541-51.

Haupt, A. "Environmental Product Standards, International Trade and Monopolistic
Competition," Mimeo. Department of Economics, European University Viandrina,
Frankfmt (Oder), 1999.

Haupt. "Environmental Policy in Open Economies and Monopolistic Competition."
Environmental and Resource Economics, 2006, 33, pp. 143-67.

Hausman, Jerry A. "Specification Tests in Econometrics." Econometrica, 1978, 46, pp.
1251-72.

Heil, Mark T. and Selden, Thomas M. "International Trade Intensity and Carbon
Emissions: A Cross-Country Econometric Analysis." Journal of Environment &
Development, 2001, 10(1), pp. 35-49.

Helpman, Elhanan and Krugman, P.R. Market Structure and Foreign Trade. London:
The MIT Press, 1985.

Helpman, Elhanan and Krugman, P.R. Trade Policy and Market Structure.
Cambridge: The MIT Press, 1989.

Hettige, Hemamala; Lucas, Robert EB. and Wheeler, David. "The Toxic Intensity of
Industrial Production: Global Patterns, Trends and Trade Policy." American
Economic Review, 1992, 82(2), pp. 478-81.

Hettige, Hemamala; Mani, Muthukumara and Wheeler, David. "Industrial Pollution

in Economic Development: The Environmental Kuznets Curve Revisited."
Journal of Development Economics, 2000, 62, pp. 445-76.

195

 

 

Hilton, EC. and Levinson, Arik. "Factoring the Environmental Kuznets Curve:
Evidence from Automotive Lead Emissions." Journal of Environmental
Economics and Management, 1998, 35(2), pp. 126-41.

Kennedy, P.W. "Equilibrium Pollution Taxes in Open Economies with Imperfect
Competition." Journal of Environmental Economics and Management, 1994, 27,
pp. 49-63.

Krutilla, K. "Environmental Regulation in an Open Economy." Journal of
Environmental Economics and Management, 1991, 20, pp. 127-42.

Krugman, P.R. "Increasing Returns, Monopolistic Competition, and International
Trade." Journal of international economics, 1979, 9(4), pp. 469-79.

Krugman, P.R. "Scale Economies, Product Differentiation, and the Pattern of Trade."
American Economic Review, 1980, 70(5), pp. 950-59.

Krugman, P.R. "Intra-Industry Specialization and the Gains from Trade." Journal of
Political Economy, 1981, 89(5), pp. 959-73.

Lange, Andreas and Requate, Till. "Emission Taxes for Price-Setting Firms:
Differentiated Commodities and Monopolistic Competition," E. Petrakis, E. S.
Sartzetakis and A. Xepapadeas, Environmental Regulation and Market Power.
Cheltenham: Edward Elgar, 1999, 1-26.

Lai, Huiwen and Zhu, Susan Chun. "The Determinants of Bilateral Trade." Canadian
Journal of Economics, 2004, 37(2), pp. 459-83.

Lai, Huiwen and Zhu, Susan Chun. "U.S. Exports and Multinational Production." The
Review of Economics and Statistics, 2006, 88(3), pp. 531-48.

Learner, Edward E. Sources of International Comparative Advantage. Cambridge,
MassachusettszThe MIT Press, 1984.

Levinson, Arik. "Environmental Regulations and Manufacturers' Location Choices:
Evidence from the Census of Manufacturers." Journal of Public Economics, 1996,
62, pp. 5-29.

Leonard, H.J. and Duerksen, C.J. "Environmental Regulations and the Location of
Industry - an International Perspective." COLUMBIA JOURNAL OF WORLD
BUSINESS, 1980, 15(2), pp. 52-68.

List, John A. and Co, Catherine Y. "The Effects of Environmental Regulations on
Foreign Direct Investment." Journal of Environmental Economics and
Management, 2000, 40(1), pp. 1-20.

196

Low, Patrick and Yeats, Alexander. "Do 'Dirty' Industries Migrate?," P. Low, ed.,
International Trade and Environment. Washington, DC: The World Bank, 1992.

Lucas Jr, Robert E. "Why Doesn't Capital Flow from Rich to Poor Countries?" The
American Economic Review, 1990, 80(2), pp. 92-96.

Lucas, Robert E. 8.; Wheeler, David and Hettige, Hemamala. "Economic
Development, Environmental Regulation and the International Migration of Toxic
Industrial Pollution," P. Low, International Trade and the Environment.
Washington, DC: World Bank Discussion Paper 159, 1992, 67-89.

Mani, Muthukumara and Wheeler, David. "In Search of Pollution Havens? Dirty
Industry in the World Economy, 1960-1995." Journal of Environment and
Development, 1998, 7(3), pp. 215-47.

Markusen, J. "International Extemalities and Optimal Tax Structures." Journal of
International Economics, 1975a, 5(15-29).

Markusen, J. "Cooperative Control of International Pollution and Common Property
Resources." Quarterly Journal of Economics, 1975b, 89, pp. 618-32.

McGuire, M. "Regulation, Factor Rewards, and International Trade." Journal of Public
Economics, 1982, 17, pp. 335-54.

Melitz, Marc J. "The Impact of Trade on Intra-Industry Reallocations and Aggregate
Industry Productivity." Econometrica, 2003, 71(6), pp. 1695-725.

Neary, J. Peter. "International Trade and the Environment: Theoretical and Policy
Linkages." Environmental and Resource Economics, 2006, 33(1), pp. 95-118.

Petrakis, Emmanuel; Sartzetakis, Eftichios S. and Xepapadeas, Anastasios.
Environmental Regulation and Market Power. Cheltenham: Edward Elgar, 1999.

Rauscher, M. "National Environmental Policies and the Effects of Economic
Integration." European journal of political economy, 1991, 7, pp. 313-29.

Rauscher, M. "On Ecological Dumping." Oxford Economic Papers, 1994, 46, pp. 822-
40.

Rosen, Sherwin. "Wage-Based Indexes of Urban Quality of Life," P. Mieszkowski and
M. Strashaim, Current Issues in Urban Economics. Baltimore: John Hopkins
University, 1979.

Selden, Thomas M. and Song, Daqing. "Environmental Quality and Development: Is

There a Kuznets Curve for Air Pollution Emission?" Journal of Environmental
Economics and Management, 1994, 27(2), pp. 147-62.

197

Shafik, Nemat. "Economic Development and Environmental Quality: An Econometric
Analysis." Oxford Economic Papers, 1994, 46(0), pp. 757-73.

Schulze, Gunther G. and Ursprung, Heinrich W. International Environmental
Economics. Oxford: Oxford University Press, 2001.

Sheldon, I. "Trade and Environmental Policy: A Race to the Bottom?" Journal of
Agricultural Economics, 2006, 57(3), pp. 365-92.

Siebert, H.; Eichberger, J.; Gronych, R. and Pethig, R. Trade and the Environment: A
Theoretical Enquiry. Amsterdam: Elsevier/North Holland Press, 1980.

Stern, David I. "The Rise and Fall of the Environmental Kuznets Curve." World
Development, 2004, 32(8), pp. 419-39.

Strutt, A. and Anderson K. "Will Trade Liberalization Harm the Environment? The
Case of Indonesia to 2020." Environmental and Resource Economics, 2000,
1 7(3), pp. 203-32.

Taylor, M. Scott and Levinson, Arik. "Unmasking the Pollution Haven Effect," NBER
Working Paper W10629. Cambridge, 2004.

Ulph, A. "Environmental Policy and International Trade When Governments and
Producer Act Strategically." Journal of Environmental Economics and
Management, 1996, 30, pp. 265-81.

Wilson, John D. "Capital Mobility and Environmental Standards: Is There a Theoretical
Basis for a Race to the Bottom?," J. N. Bhagwati and R. P. Hudec, Fair Trade
and Harmonization: Prerequisites for Free Trade. Vol. 1. Cambridge and
London: MIT Press, 1996.

Yusuf, Shahid; Nabeshima, Kaoru and Ha, Wei. "What Makes Cities Healthy?,"
World Bank Policy Research Working Paper 4107. Washington, DC, 2007.

Walter, I. "Pollution Content of American Trade." Western Economic Journal, 1973,
11(1), pp. 61-70.

Wooldridge, Jeffrey M. Econometric Analysis of Cross Section and Panel Data.
Cambridge: MIT Press, 2002.

Wooldridge, Jeffrey M. Introductory Econometrics: A Modern Approach. Cincinnati:
South-Westem Thomson, 2006.

http://vwvw.epa.gov/airs/aewin/

198

http://wwwmfe. govt.nz/publications/air/gems/ gems-2005/gems-report-2005-part1 .pdf

Alan Heston, Robert Summers and Bettina Aten, Penn World Table Version 6.2, Center
for International Comparisons of Production, Income and Prices at the University
of Pennsylvania, September 2006.

World Bank’s GGDC and the National Institute of Economic and Social Research,
Manufacturing Productivity and Unit Labour Cost Database at
http://www.ggdc.net/icop.html

.lntemational Yearbook of Industrial Statistics at http://www.unido.org/en/doc/3 544

http://wwwworldbank.ogz/depweb/english/modules/social/life/index02.html

http://raob.fsl.noaa.gov/

http://www.wwf.orgflk/filelibrarv/pdf/brazil.pdf

http://www.panda.or news facts/newsroom/indcx.cfm?uNewsID=58760

199

3 1293 03063162

 

u
"I
"I
n
"I
“I
“
Ellilllll
Tl
A“
”I
$11".
"I'll