CORRUPTION, ILLEGAL TRADE AND COMPLIANCE
WITH THE MONTREAL PROTOCOL
by
Katsiaryna Ivanova
A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF ARTS
(ECONOMICS)
December 2005
Copyright 2005
Katsiaryna Ivanova
UMI Number: 1435117
Copyright 2005 by
Ivanova, Katsiaryna
All rights reserved.
UMI Microform 1435117
Copyright 2006 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, MI 48106-1346
ii
ACKNOWLEDGEMENTS
I am very grateful to Todd Sandler from whom I have learnt a great deal and
whose guidance and recommendations clarified my thinking on the subject; my
thanks also go to him for the care with which he reviewed every draft of the
manuscript. I would like to express my gratitude to John Odell who provided
thoughtful and valuable comments that encouraged me to significantly improve the
thesis. I am very thankful to my sister Anna Ivanova for helpful discussions and
suggestions. I would also like to acknowledge the financial support of the School of
International Relations at the University of Southern California in the preparation of
this thesis.
iii
TABLE OF CONTENTS
Acknowledgements
ii
List of Tables
iv
List of Figures
v
Abstract
vi
Introduction
Literature Review
Theoretical Model
Comparative Statics
Empirical Work
Specification
Data
Results
Conclusion
1
4
9
14
20
20
21
25
38
Bibliography
41
iv
LIST OF TABLES
Table 1: Comparative statics results
19
Table 2: Descriptive statistics
25
Table 3: Panel data regression estimates
26
v
LIST OF FIGURES
Figure 1: Marginal effect of honesty conditional on degree of law
29
Figure 2: Marginal effect of law conditional on degree of honesty
29
Figure 3: Relationship between legal imports and honesty for countries with strong and
weak legal systems in 1998
31
Figure 4: Relationship between legal imports and legal system for countries with high
and low honesty in 1998
31
Figure 5: Marginal effect of honesty conditional on level of tariff
34
Figure 6: Marginal effect of tariff conditional on degree of honesty
34
Figure 7: Relationship between legal imports and honesty for countries with high and
low tariff in 1998
36
Figure 8: Relationship between legal imports and tariff for countries with high and low
honesty in 1998
36
vi
ABSTRACT
This paper develops a theory of illegal trade in ozone-depleting substances
under compliance with the Montreal Protocol, taking into account the level of
corruption, law enforcement and environmental tariffs. The following predictions
emerge: (i) a fall in corruption reduces illegal imports; while it raises legal imports
when law enforcement is weak, but reduces legal imports when law enforcement is
strong; (ii) stronger law enforcement decreases illegal imports; while it increases
legal imports if the level of corruption is high, but decreases legal imports if the
level of corruption is low; and (iii) a higher environmental tariff reduces legal
imports; while its effect on illegal imports depends on the level of corruption and
law enforcement. Using panel data, we find evidence that generally supports the
theoretical conjectures concerning legal imports and, indirectly, allows us to draw
inferences about illegal imports, even though the data on illegal trade do not exist.
1
INTRODUCTION
The Montreal Protocol on Substances that Deplete the Ozone Layer is
considered one of the most successful international agreements to-date. It was
ratified by over 180 countries and the phase-out of ozone-depleting substances
(ODS) detailed in the Protocol is reported to proceed according to schedule.
However, unforeseen at the stage of negotiations and first detected in the mid 1990’s,
illegal trade in ODS has become a cause of serious concern for the future of the
ozone treaty. When the phase-out in industrialized countries began, demand for
chlorofluorocarbons (CFCs) remained high, while expanding production in
developing (Article 5) countries, which were exempted from compliance with the
control measures until 1999, ensured abundant supplies of CFCs to their markets.
CFC-12 could be bought for $1 US per kilogram in China and sold for $16 US in the
UK. Black market in CFCs and halons has been tracked since the mid-1990’s, when
illegal trade in ODS grew to an alarming rate. Since then, ODS smuggling in
developed countries has fallen, but developing (Article 5) countries have experienced
an upsurge in contraband ODS movement, as they recently began implementing the
control measures of the protocol.
According to Shleifer and Vishny (1993), “government officials often collect
bribes for providing permits and licenses, giving passage through customs, or
prohibiting the entry of competitors.” There is evidence from the Environmental
Investigation Agency (EIA) that Pakistan’s Ministry of Commerce has issued import
2
authorization for shipments of CFCs to parties who are not involved in the
refrigeration business, and have never before imported refrigerants. EIA’s report “A
Crime against Nature,” cites examples of illegal traders in China telling EIA
undercover investigators that they will be able to secure the necessary papers from
the local authorities, or when they in fact produce documents for smuggled
shipments stamped by the China Council for the Promotion of International Trade
(Shanghai). This paper analyzes the linkages between corruption, law enforcement,
environmental policy and smuggling under compliance with the Montreal Protocol.
One of the difficulties of studying smuggling is that data on illegal trade is
not available. It is not possible to conduct an empirical analysis of illegal imports of
CFCs, as there is no way to obtain any reliable data on unauthorized shipments of
ODS. Thus, we begin by developing a partial equilibrium model that explains both
changes in legal and illegal imports of CFCs, and proceed by testing our theoretical
predictions about the response of legal imports to changes in the smuggling-related
parameters, using unbalanced panel data for 110 countries. This allows us to
circumvent the data problem regarding illegal trade, as we can obtain insights about
changes in illegal activity by estimating the model describing legal imports.
Our theory builds on the model of illegal trade by Martin and Panagariya
(1983) extended to international trade in waste products in the presence of illegal
disposal by Copeland (2000). These models explicitly incorporate the uncertainty
associated with smuggling into the analysis. The probability of detection depends on
the ratio of illegal to legal imports or illegal to total imports, making the two types of
3
trade ‘joint’ goods. In this paper, the share of illegal imports is determined by the
profit-maximizing decision of the importers of CFCs. We derive the demand for total
imports from the profit-maximizing combination for producers who use CFCs as
inputs in the production process. This analysis allows us to determine how changes
in corruption, or the level of fine or tariff affect legal and illegal quantities of CFCs.
The model yields three main predictions for the empirical analysis of legal
imports of CFCs. First, a tariff imposed to reduce the negative externality tends to
decrease the level of legal trade as smugglers switch to illegal activities so that they
can avoid tariff payments. Second, the effect of a fall in corruption on legal imports
is conditional on the level of fine imposed as a punishment for breaking the law.
Given that the fine is low, a fall in corruption is associated with an increase in legal
imports of CFCs, as it increases the probability of being caught. When the fine is
high, the effect is reversed: less corrupt countries tend to have fewer legal imports of
CFCs. This happens because a fall in corruption decreases total imports and when
the fine is high, this effect dominates. Finally, the effect of the fine is conditional on
the level of corruption. In highly corrupt countries, a rise in the fine increases legal
imports as it raises the expected cost of illegal trade. In countries with low levels of
corruption, the effect is reversed: a higher fine is associated with fewer legal imports.
The intuition is that total imports fall as the fine rises and, when corruption is low,
this effect dominates.
The regression estimates generally support the expectations of the theory.
First, the empirical evidence suggests that the effect of corruption (law enforcement)
4
is conditional on law enforcement (corruption). We find that countries with higher
levels of corruption (lower levels of law enforcement) tend to import fewer CFCs
legally where law enforcement is weak (corruption is strong). Alternatively, where
law enforcement is strong (corruption is weak), the effect is reversed as predicted by
the theory. The empirical evidence only partially supports the theoretical prediction
that higher tariffs are associated with lower legal imports of CFCs: we find that the
tariff decreases legal imports when corruption is high, but the effect is reversed when
corruption is low. Nevertheless, the data, in general, suggests that corruption, law
enforcement and the tariff have a joint influence on legal trade in CFCs, the
substances that are subject to the control measures of the Montreal Protocol.
The rest of this paper is organized as follows. Section 2 provides a literature
review. Section 3 outlines the theoretical model. Section 4 derives the predictions.
Section 5 presents the empirical work. Section 6 concludes.
LITERATURE REVIEW
The case of the ozone treaty is interesting because it makes it possible to
study the observance of environmental regulations on a global level. Previous
literature on this subject offers different views on the effectiveness of international
efforts to curb pollution. Cole et al. (1997) find that environmental performance with
regard to local pollutants (e.g., sulphur dioxide or suspended particulate matter) and
global pollutants (e.g., carbon dioxide or CFCs) may differ. They argue that
5
meaningful environmental Kuznets curves1 exist only for pollutants with local
impact, while global pollutants such as CFCs have the inverted-U relationship with
income only after a multilateral treaty is initiated; otherwise, pollutants with global
impact increase monotonically with income2. Murdoch and Sandler (1997), however,
show that there is a nearly linear relationship between CFC reductions and national
income prior to the Montreal Protocol taking effect and conclude that the Montreal
Protocol only codified the reductions in CFC consumption that nations were
unilaterally willing to accomplish.
Illegal trade in CFCs interferes with the implementation of the ozone treaty,
inhibiting replacement of CFC-using equipment with technologies that can function
on ODS alternatives. In this paper, we examine the factors that increase incentives
for illegal activity. In particular, we focus on three major issues. First, we provide
insights into the relationship between corruption and illegal trade. Second, we
explore the effects of the size of the punishment on illegal activity. Finally, we look
at the consequences of trade restrictions on the scale of illegal trade. A joint
influence of these three individual effects on illegal activities is also considered.
The design of international institutions has been a focus of many studies.
From the pure public good perspective, incorporating selective incentives into the
1
According to the Kuznets curve hypothesis, the relationship between economic growth and
environmental degradation takes an inverted-U shape so that at low levels of income countries value
material well-being more than environmental quality but as their income grows and cleaner
production technologies develop public demand for higher environmental standards increases.
Empirical estimations of the Kuznets curve have been done by Grossman and Krueger (1995), Shafik
(1994), Kaufmann et al. (1998) and others.
2
In some cases turning points could be predicted but they occurred outside of the observed income
range with large standard errors.
6
design of international agreements improves their effectiveness and mitigates the
free-rider problem. Thus, by allowing developing (Article 5) nations to postpone
their compliance with the control measures and offering assistance through a
Multilateral Fund, the Montreal Protocol has achieved a higher level of cooperation
in contrast to other treaties3 (Sandler and Arce, 2003). But from the moral hazard
perspective, the enforcement of new rules and regulations that follows the
introduction of international agreements expands the range of activities through
which self-interested officials can extract bribes. Although such multilateral
agreements as the Convention in International Trade in Endangered Species of Wild
Fauna and Flora (CITES) and the Basel Convention on the Control of Transboundary
Movements of Hazardous Wastes and Their Disposal have met with success, they
have also been threatened by a buoyant illegal trade having opened up possibilities
for corruption in issuing paper certificates and movement documents required for the
traded goods.
Compliance with laws and regulations restricts private economic activity, and
whenever administrative authority is delegated to a self-interested official,
opportunities for corrupt behavior arise. The implementation of environmental
controls can be rather costly to industry owners; bribes to public servants are often
cheaper than complying with the regulations. Shleifer and Vishny (1993) distinguish
3
Sandler and Arce (2003) analyze benefit-cost duality to differentiate between pure public goods and
commons and find that a public good contribution scenario involves positive inducements while a
commons game is characterized by selective punishments, which are more difficult to implement.
They compare the Montreal Protocol (a contribution problem), which resulted in widespread
participation, to the Kyoto Protocol on climate change (a commons problem).
7
two cases of bribery: corruption with theft and corruption without theft. In the former
case, the only price that the buyer pays for the good is the bribe and the transaction is
kept secret from the government. In the latter case, the bribe is paid in excess of the
official price that the government receives for the good. Most of the environmental
problems fall under the first category, which Bardhan (1997) describes as “more
insidious, difficult to detect and therefore more persistent.” Besides, studies of
corruption in an environmental context have shown that policy options to monitor
environmental compliance in the presence of corruption may be quite limited, as in
some cases greater enforcement may lead to stronger incentives to under-provide a
public good and consequently increase the instance or scope of bribery4 (Damania,
2002; Mookherjee and Png, 1995). Lopez and Mitra (2000) analyze the implications
of corruption for the relationship between pollution and growth. They consider both
a cooperative Nash bargaining interaction between the government and the private
firm generating pollution and a non-cooperative Stackelberg model with the firm as
leader, and conclude that in both types of interactions corruption does not preclude
the existence of the environmental Kuznets curve, but the turning point occurs at
income and pollution levels above those corresponding to the social optimum.
Two works provide empirical studies of the effect of corruption on the
determination of environmental policy. Fredriksson and Svensson (2003) analyze the
4
Damania (2002) using a principal-agent framework considers a case when emission tax is imposed
on the firm that emits pollution and concludes that a higher tax creates stronger incentives to
underreport for the inspector charged with monitoring pollution levels by the government, which in
turn requires greater auditing. Mookherjee and Png in a similar setting show that penalties for
corruption and the extent of bribery may reverse direction so that “small increases in penalties may
raise bribery, while larger increases will reduce it.”
8
effect of corruption on the stringency of environmental policy, conditional on the
degree of political instability. Their findings suggest that corruption reduces the
stringency of environmental regulations, but as the degree of political instability
increases, this effect disappears. Since the probability of the incumbent government
remaining in office declines, there are fewer incentives for a producer lobby to
influence environmental policy by offering a bribe. Similarly, Damania et al. (2003)
focus on the interaction between corruption and trade liberalization. They conclude
that while a reduction in corruption is unambiguously associated with a higher
pollution tax, the effect of trade liberalization on environmental regulation depends
on the level of corruption: the greater the level of governmental corruption, ceteris
paribus, the larger the increase in environmental stringency associated with an
increase in openness to trade. However, in countries with the most honest
governments, the effect of trade openness is reversed: more open trade regimes tend
to have less stringent regulations.
The analysis performed in this paper differs from the two studies discussed
above because it explores the link between corruption and the implementation of
environmental policies, as distinct from the formation of environmental policy5. The
distinction is important because there is often a gap between environmental laws and
5
Most of the studies of the Montreal Protocol also focus on the participation or ratification decision
rather than on the actual implementation of the provisions in the Protocol (see Beron, Murdoch and
Vijverberg (2003) and Congleton (1992)).
9
regulations, on the one hand, and their implementation and enforcement, on the
other6. As Desai (1998) points out, this problem is particularly pertinent to
industrializing nations as the “state’s autonomy and its capacity to work its will on
the society often are quite limited in these countries.”
THEORETICAL MODEL
Consider a small, open economy that produces a final consumption good y7
using two intermediate inputs: chlorofluorocarbons (CFCs), x, and environmentally
friendly CFC alternatives, k, where x and k are substitutes. The economy imports all
CFCs used in the production process for y from abroad. The government imposes an
ad valorem tariff rate equal to t on imports of CFCs that cause a negative externality.
However, distributing firms importing CFCs face strong incentives to transport the
good across the border surreptitiously: first, because there are quantitative
restrictions on imports of CFCs as part of the implementation of the Montreal
Protocol and second, because this allows the distributors to avoid paying the tariff.
The quantities of legal and illegal CFCs imported by a typical importer are denoted
by l and s, respectively.
The distributors face a probability λα that the illegal activity is detected and
results in prosecution, where λ is the fraction of illegal imports, i.e. λ = s x , and α is
a measure of honesty in society, which is scaled to range between 0 and 1. As in
6
A simultaneous analysis of both the participation decision and the level of participation in an
environmental treaty can be found in Murdoch, Sandler and Vijverberg (2003), who formulate a twostage game, for which nations first decide whether or not to participate and then choose their level of
participation.
7
For example, y could stand for the manufacture of refrigerators or air-conditioners.
10
Copeland (2001) and Martin and Panagariya (1984), we assume that the higher the
fraction of illegal imports, the more likely it is that smuggling will be detected so
that legal imports may be used to (partially) mask illegal activities. That is, an
increase in the probability of being caught may result from (1) an increase in illegal
trade with legal imports held constant; or (2) a decrease in legal imports with illegal
activity held constant. Since it is most interesting to look at the case where 0 < λ < 1 ,
we focus on the interior solution. As indicated above, the probability of being caught
also depends on corruption. The importers may offer a bribe to government officials
or customs officers to provide them with the needed paperwork or not to report the
contraband if illegal activity is suspected. The less likely it is that the officials will
accept the bribe; the more likely the illegal activity will be discovered. Note that α
here is inversely related to corruption and can be viewed as the proportion of honest
bureaucrats within the political system. Bribe per unit of value of illegal imports is
denoted by b. If caught smuggling, importers pay a fine of f per unit of illegal
imports. The bribe and the fine rates are treated as exogenous8. A further assumption
is that the quantity being smuggled is confiscated if the illegal activity is discovered.
All distributors of CFCs are assumed to be identical and maximize expected
profits. Denoting the price of x in the world and domestic markets by p * and p,
respectively, the importers’ profits denoted by ∏1 and ∏ 2 , when smuggling is and is
not successful, respectively, can be written as
8
Following Copeland (2001), the fine is assumed to be exogenous. This is a reasonable
approximation since endogenizing the fine would require an analysis of the entire legal system.
11
Π 1 = p(s + l ) − [ p * (s + l ) + tp * l + bp * s ]
(1)
Π 2 = pl − [ p * (s + l ) + tp * l + bp * s + fs ] .
(2)
The importers receive a revenue of p(s + l ) if not discovered, but only pl if
discovered, since the quantity smuggled is confiscated. The terms in square brackets
in (1) and (2) indicate the total cost to the distributors of importing (s + l ) units of
CFCs when smuggling is and is not detected, respectively. p * (s + l ) is the amount
the distributors pay to buy s + l units of CFCs at the world market. The amount of
tariff they pay for importing l units equals tp * l . The importers avoid paying the
tariff on s units that are transported across the border illegally. The amount given as
bribes to customs officers or government officials is denoted by bp * s . If caught,
the smugglers are fined an amount fs for the contraband, making the total costs in
(2) greater.
Using (1) and (2), the distributors’ expected-profit function can be written as
(1 − λα ){p(s + l ) − [ p * (s + l ) + tp * l + bp * s ]}
max 
s ,l
+ λα {pl − [ p * (s + l ) + tp * l + bp * s + fs ]}

s.t. s + l = x  .

Note that this function is linear in x and with a change of variables ( s = λ and
l = (1 − λ )x ) can be rewritten as
max(1 − λα ){px − [ p * x + tp * (1 − λ )x + bp * λx ]}
λ,x
+ λα {p(1 − λ )x − [ p * x + tp * (1 − λ )x + bp * λx + fλx ]} .
(3)
The first-order conditions for the optimal choice of total imports and the fraction of
illegal imports satisfy
x:
p − λ2αp = p * + (1 − λ )tp * + λbp * + λ2αf
(4)
12
λ:
- 2λαp = 2λαf + bp * −tp * .
(5)
It is easily verified that (4) implies zero expected profits for the importer9.
Condition (4) simply states that the marginal revenue derived from total trade
equals the marginal cost of total imports. The left-hand side of (4) captures the
revenue from sale of one extra unit of total imports of CFCs, i.e. p, and the expected
loss in revenue per unit of total imports if smuggling is detected, i.e. λ2αp . The
right-hand side of (4) represents the cost of obtaining one extra unit of total imports
consisting of the world price p*, the effective tariff rate (1 − λ )tp * , the bribe λbp *
and the expected fine λ2αf paid per unit of total imports. Condition (5) indicates
that the marginal revenue from importing one extra unit through illegal instead of
legal channels equals its marginal cost. Note that if the probability of detection is
denoted by q = λα , then 2λα = q + λqλ . Thus, the left-hand side of (5) represents
the sum of the direct negative effect on revenue resulting from the confiscation of
illegal imports if smuggling is detected and the indirect negative effect on revenue
through a rise in the probability of detection. In the right-hand side of (5), 2λαf
indicates the direct and indirect positive effect on the cost of one extra unit of illegal,
as opposed to legal, imports; bp * represents the increase in cost in the form of the
bribe paid for one extra unit of illegal imports; and the term tp * captures the direct
savings on tariff payments from importing one extra unit illegally, as opposed to
legally. Condition (5) can be rewritten as
9
Setting the expression in (3) equal to zero, dividing it by x and rearranging, we get (4).
13
2λαp + 2λαf + bp* = tp * ,
where the left-hand side represents the expected loss (consisting of a decrease in
revenue and an increase in cost) from choosing to import one extra unit of CFCs
illegally instead of legally and the right-hand side indicates the expected gain.
Solving (5) for bp * and substituting the resulting expression in the righthand side of (4), we get
(1 + λ α )p + λ αf = (1 + t ) p * .
2
2
(6)
Hence, if λ > 0 , p < (1 + t ) p * . Because traders avoid paying a tariff on the smuggled
quantity of the good, they are able to sell it at a price lower than the tariff inclusive
world price.10
Conditions (4) and (5) determine the domestic price of CFCs p and the
fraction of illegal imports λ . However, the quantities of legal and illegal imports
cannot be determined without specifying the demand for total imports used as inputs
in the production process for y. To do that, we assume that the technology for y
exhibits decreasing returns to scale and is given by
F ( y , x, k ) = 0 ,
which can be inverted to yield
y = f ( x, k ) ,
where f is assumed to be strictly concave and increasing in x and k.
The profit function for producers of good y can be written as
10
Price disparity in the presence of illegal trade was first discussed by Pitt (1981) and emphasized by
Martin and Panagariya (1984).
14
max[ p y f ( x, k ) − px − rk ] ,
x,k
where p y and r denote the domestic prices of good y and CFC alternatives,
respectively. The first-order conditions for the optimal choice of inputs satisfy
x:
p y f x ( x, k ) = p
(7)
k:
p y f k ( x, k ) = r ,
(8)
which set the value of the marginal product of each input equal to its price.
We also assume that in the long run, perfect competition results in zero
profits, i.e.,
p y f ( x, k ) = px + rk .
(9)
COMPARATIVE STATICS
Our primary interest is to find out the effects of corruption, the fine and the
tariff on legal and illegal imports of CFCs to derive predictions for the empirical
work. First, the response of the fraction of illegal imports to changes in the
parameters is considered. Totally differentiating (4), and rearranging, we get11
dp = −
(λt − λb − 1 − t ) dp * + λ2 ( p + f ) dα + (1 − λ ) p * dt + λ2α
2
2
2
2
(1 − λ α )
(1 − λ α )
(1 − λ α )
(1 − λ α ) df .
(10)
Differentiation of (5) yields
dλ =
11
t −b
p*
λ
λ
λ
dp * +
dt −
dp − dα −
df .
α
2α ( p + f )
2α ( p + f )
(p + f )
(p + f )
We used (5) to eliminate
(2λαp − tp * +bp * +2λαf ) dλ
(1 − λ α )
2
from (10).
(11)
15
Combining (10) and (11), we obtain (with dp* = 0 )12
dλ (1 − 2λα + λ2α ) p *
=
dt 2α ( p + f )(1 − λ2α )
dλ
λ
=−
dα
α 1 − λ2α
(
)
dλ
λ
=−
.
df
( p + f ) 1 − λ2α
(
)
Note that (1 − 2λα + λ2α ) is positive for all values of α and λ that lie between 0
and 1. Hence, with corruption and the fine held constant, a rise in tariff increases the
fraction of illegal imports (i.e., dλ dt > 0 ). A higher tariff reduces the expected
profits from legal trade, while leaving the expected profits from illegal trade
unchanged, which results in a higher share of illegal imports. For a given tariff and
fine, a rise in corruption increases the fraction of illegal imports (i.e., dλ dα < 0 ).
Since the probability of detection decreases as corruption rises, importers are willing
to increase the share of smuggled CFCs because they can get higher profits from
illegal trade. An increase in fine, with corruption and the tariff held constant,
decreases the share of illegal imports (i.e., dλ df < 0 ) as a higher fine raises the
expected costs of importing CFCs illegally.
Before we proceed to derive the demand for CFCs, it is important to mention
that although the alternatives for CFCs have proved to be less expensive, the change
12
Later in the empirical analysis, we only need to control for changes in the world price across
periods as it does not change across countries. Thus, for simplicity, we set the change in the world
price of CFCs equal to zero in developing the theoretical model.
16
of technology or adjustment of equipment that use CFCs is rather costly. Since
technological development is slow, we assume that the price of alternatives did not
change much in the 10-year period covered in the empirical analysis and set it equal
to zero in developing the theoretical model. (However, we control for the difference
between developed and developing countries in adapting technology in the empirical
estimation.) To find out how demand for CFCs responds to changes in the
parameters totally differentiate (7), (8), and (9) and rearrange to obtain (with dr = 0 )
dx =
( f k f xk − f x f kk )x +
p
y
(f
xx
f kk − f
2
xk
)
f kk f ( x, k )
dp .
f ( x, k )
(12)
From the second-order conditions, it follows that the sign of the denominator is
positive13. To determine the sign of the numerator, solve (7) and (8) for f x and f k ,
substitute the resulting expressions in (12) and use (8) to obtain
dx =
(xf xk + kf kk )r
(p ) ( f
y 2
xx f kk − f
2
xk
) f ( x, k )
dp = σ
x
dp ,
p
(13)
where σ is the price elasticity of demand for CFCs. Since x and k are substitutes,
f xk is negative. σ is negative.
Substitution of (10) into (13) yields (with dp* = 0 )
dx
x(1 − λ ) p *
=σ
dt
p 1 − λ2α
(
13
The second-order condition requires that
(
)
)
p y f xx dx 2 + 2 f xk dxdk + f kk dk 2 < 0 . One of the
conditions for the quadratic form in the brackets to be negative is that the Hessian determinant
(f
xx
)
f kk − f xk2 should be positive.
17
dx
xλ2 ( p + f )
=σ
dα
p(1 − λ2α )
dx
xλ2α
=σ
.
df
p(1 − λ2α )
Thus, total imports decrease with a rise in tariff (i.e., dx dt < 0 ),an increase in the
fine (i.e., dx df < 0 ) and a fall in corruption (i.e., dx dα < 0) , because producers
substitute alternatives for CFCs due to an increase in the price of CFCs.
To see how corruption, the fine and tariff affect illegal imports, use
ds = λdx + xdλ to get14:
(
)
ds σλ (1 − λ ) 1 − 2λα + λ2α  p * x
=
+

dt 
p
2α ( p + f )  1 − λ2α
ds σλ 2 ( p + f ) 1  λx
=
− 
α  1 − λ2α
dα 
p
(
(
)
)
ds σλ 2α
1  λx
=
−
.

df  p
p + f  1 − λ2α
(
)
A fall in corruption and a rise in fine decrease illegal imports of CFCs (i.e.,
ds dα < 0 and ds df < 0 ) as they increase the expected cost of illegal trade. The
effect of a higher tariff is, however, ambiguous. Although the share of illegal imports
increases with a higher tariff, total imports fall because producers substitute
14
Note that
dx =
dλ =
λ
(1 − 2λα + λ α )p * dt − λ dα −
df
( p + f )(1 − λ α )
α (1 − λ α )
2α ( p + f )(1 − λ α )
2
2
2
σ (1 − λ ) p * x
σλ2 ( p + f )x
σλ2αx
α
dt
+
d
+
(1 − λ2α )p
(1 − λ2α )p
(1 − λ2α )p df
2
.
and
18
alternatives for CFCs; as a result, a higher tariff may lower the absolute quantity of
illegal imports.
Using dl = (1 − λ )dx − xdλ , we get
(
)
2
dl σ (1 − λ )
1 − 2λα + λ2α  p * x
=
−

dt 
p
2α ( p + f )  1 − λ2α
(
)
(14)
dl  (1 − λ )σλ ( p + f ) 1  λx
=
+ 
α  (1 − λ2α )
dα 
p
(15)
dl  (1 − λ )σλα
1  λx
=
+
df 
p
p + f  (1 − λ2α )
(16)
A higher tariff reduces the amount of CFCs imported legally (i.e., dl dt < 0 )
as it raises the expected cost of legal trade. But the effects of corruption and the fine
on the quantity of legal imports are ambiguous. For sufficiently low fine, i.e.,
f <−
(1 + σ (1 − λ )λα ) p , a fall in corruption is positively related to legal imports. As
σ (1 − λ )λα
the fine gets higher, i.e. f ≥ −
(1 + σ (1 − λ )λα ) p , the effect of a fall in corruption
σ (1 − λ )λα
disappears or becomes negative. Although the share of legal imports increases, as
corruption falls, total imports decline because producers substitute alternatives for
CFCs, and when the fine is high, the latter effect is amplified. As a result, lower
corruption may decrease the absolute quantity of legal imports. The effect of fine is
positive when corruption is sufficiently high, i.e., α < −
p
σ (1 − λ )λ ( p + f )
, but when
19
α ≥−
p
σ (1 − λ )λ ( p + f )
, the effect of fine disappears or becomes negative. Similarly,
a rise in fine increases the share of legal imports, but as corruption declines, total
imports fall and the absolute quantity of legal imports declines.
The comparative statics results are summarized in Table 1.
Table 1
Comparative statics results
Comparative
Interpretation
Statics
As the tariff rises, the share of illegal imports increases. However,
ds
>=< 0
total imports fall because producers substitute alternatives for
dt
CFCs. If the latter effect dominates, a raise in the tariff may lower
(
1 − 2λα + λ2α )
illegal imports. If f > − p −
, the latter effect
2σα (1 − λ )λ
dominates.
A fall in corruption decreases illegal imports through a rise in the
ds
<0
probability
of detection.
dα
A rise in the fine decreases illegal imports through a rise in their
ds
<0
marginal cost.
df
A rise in the tariff decreases legal imports through a rise in their
dl
<0
marginal
cost.
dt
As corruption falls, the share of illegal imports declines. However,
dl
>=< 0
total
imports fall because producers substitute alternatives for
dα
CFCs. If the latter effect dominates, a fall in corruption may lower
(1 + σ (1 − λ )λα ) p , the latter effect
legal imports. If f > −
σ (1 − λ )λα
dominates.
As the fine increases, the share of illegal imports declines.
dl
>=< 0
However, total imports fall because producers substitute
df
alternatives for CFCs. If the latter effect dominates, a rise in the
p
fine may lower legal imports. If α > −
, the latter
σ (1 − λ )λ ( p + f )
effect dominates.
20
EMPIRICAL WORK
Specification
Since data on illegal activities are not available, the objective of empirical
analysis is to test implications on the relationship between the level of legal imports
of CFCs, honesty, the tariff and the fine captured in (14), (15) and (16) using
unbalanced panel data on consumption of CFCs for 110 countries from 1992 to
2001.
The hypothesized relationship is:
yit = γ + xit′ β x + β h hit + β f f it + β r rit + β hf hit f it + β hr hit rit + β
fr
f it rit + u t + eit ,(17)
where yit is legal imports of CFCs per capita for country i at time period t; γ is a
constant; xit′ is a vector of controls; hit is the level of honesty in country i at time
period t; f it is the level of fine; rit is the level of tariff; β x is a coefficient vector
and β h , β f , β r , β hf , β hr are coefficient scalars. The time effects ut capture CFC
cutbacks required by the phase out schedule of the Montreal Protocol, as well as the
dynamic changes in global preferences due to technology and education. The error
term eit is an independent, normally distributed random variable with zero mean and
constant variance for all i and t.
This specification allows interaction effects between honesty and law,
between honesty and tariff and between law and tariff implied by the model.
Formally, the predictions of the theory are:
21
Prediction 1.
dy
<0
dr
Prediciton 2. lim
dy
dy
> 0 and lim
<0
f
→
+∞
dh
dh
Prediciton 3. lim
dy
dy
> 0 and lim
<0
α
→
1
df
df
f →0
α →0
If these predictions hold, β h and β
f
will be greater than zero, and β hf and β r will
be less than zero; however, the signs of β hr and β
fr
can only be determined
empirically.
Data
The data on CFC production and consumption are reported by the United
Nations Environmental Program (UNEP), where consumption is defined as
production plus imports minus exports15. In order to get a measure of legal imports
of CFCs, production was subtracted from consumption to get imports minus exports
of CFCs, and the difference was divided by population. This explains why some of
the values are negative, as some countries may export more that they import. The
resulting variable (IMPORTPC) represents net imports of CFCs per capita and is
measured in grams multiplied by ozone-depleting potential (ODP). To control for the
difference between countries domestically producing CFCs, where exports may be
positive, and those importing all their CFCs, a dummy variable (PRODUCTION),
15
Only data on CFC-11, CFC-12, CFC-113, CFC-114 and CFC-115 listed in Annex A of Group I of
the Montreal Protocol were used in the analysis since illegal trade in these substances has been
particularly widespread.
22
equal to 0 if production is zero and 1 if production is greater than zero, was
introduced into the model.
An index of corruption was drawn from the International Country Risk Guide
(ICRG) produced by the Political Risk Services, a private international investment
risk service. The ICRG corruption index captures the degree to which “high
government officials are likely to demand special payments,” but also the extent to
which “illegal payments are generally expected throughout lower levels of
government” in the form of “bribes connected with import and export licenses,
exchange controls, tax assessment, policy protection, or loans” (Knack and Keefer,
1995). In the theoretical model, α is inversely related to corruption representing a
measure of public honesty; thus, the ICRG index was denoted by HONESTY. This
corresponds to the scale of the ICRG index where 0 indicates the highest level of
corruption, or the lowest level of honesty, and 6 the lowest level of corruption, or the
highest level of honesty.
The ICRG dataset also contains data on law and order that capture the
strength and impartiality of the legal system, on the one hand, and popular
observance of the law, on the other. Higher scores indicate “sound political
institutions, a strong court system, and provisions for an orderly succession of
power.” Lower scores indicate “a tradition of depending on physical force or illegal
means to settle claims” (Knack and Keefer, 1995). Since countries with stronger
legal systems are more likely to impose higher fines on illegal CFCs, this index,
23
denoted by LAW, was used as a proxy for the level of fine. LAW ranges from 0 to 6,
with 0 being the lowest level and 6 being the highest level of law enforcement.
Data on environmental tariffs are difficult to obtain. The United Nations
Conference on Trade and Development (UNCTAD) maintains the Trade Analysis
and Information System (TRAINS) database, which contains data on average tariffs
for CFCs from 1996. However, there are only a total of 225 observations from 1996
to 2001 that could be used to estimate (17). An alternative measure of tariff was
taken from the World Bank’s “Trends in Average Tariff Rates for Developing and
Industrial Countries, 1986-2001.” This table provides average tariff rates for all
goods, which can be used as a proxy for tariffs imposed on imports of CFCs.
However, one problem with that approximation is that environmentally conscious
countries may have low tariffs on most of the goods, but high tariffs on
environmental goods (e.g., developed countries), and vice versa. That is why the
regression model was estimated using two different measures of tariff (TARIFF):
data on average tariff for all goods and data on average tariff for CFCs with imputed
missing values16.
The theory predicts that the effect of a tariff is conditional on the level of
honesty and the fine, the effect of honesty is conditional on the fine and tariff, and
the effect of law is conditional on the fine and honesty. Hence, the interaction terms
between honesty and tariff (HONESTYTARIFF), between law and tariff
(LAWTARIFF) and between honesty and law (HONESTYLAW) were added.
16
The non-missing values of average tariff for CFCs were regressed on average tariff for all goods
and the regression function was used to predict values at the missing locations.
24
Since consumption and, consequently, imports of CFCs are subject to the
provisions of the Montreal Protocol, it is necessary to include in vector β a set of
x
control variables that could explain imports of CFCs as part of compliance with the
treaty. First, to test the Kuznets curve hypothesis that there is an inverted-U
relationship between environmental quality and economic growth, GDP per capita in
constant 1995 US$ (GDPPC) and its quadratic term (GDPPCSQUARED) were used.
These variables capture the demand for environmental quality and if the Kuznets
relationship holds, GDPPC will be positively correlated, and GDPPCSQUARED
will be negatively correlated, with IMPORTPC, reflecting the hypothesis that
imports of CFCs will grow at low income levels and decline at high levels. Data on
GDP per capita are from the World Bank’s “World Development Indicators.” Next,
the level of CFC imports should depend on the marginal cost of satisfying the targets
of the Montreal Protocol. The Protocol requires that nations reduce their
consumption of CFCs by a certain percentage of their level of consumption in 1986:
each party should achieve its 1986 level by July 1, 1989, 75% of its 1986 level by
January 1, 1994 and 100% of its 1986 level by January 1, 1996. Countries with high
import levels in 1986 initially incurred greater costs of complying with the Protocol
than countries with low levels. Therefore, 1986 level of CFC imports
(IMPORTPC86) should be positively correlated with IMPORTPC. Finally, since
Article 5 developing countries were allowed to delay implementation of control
provisions in the Montreal Protocol until 1999, a dummy variable ARTICLE5 equal
25
to 1 for Article 5 countries and 0 otherwise was included in the analysis. The data on
the 1986 level of imports and Article 5 countries are from the UNEP.
Summary statistics are reported in Table 2.
Table 2
Descriptive statistics
Variable
IMPORTPC
HONESTY
LAW
TARIFF
CFCTARIFF
GDPPC
ARTICLE5
IMPORTPC86
PRODUCTION
Mean
Standard
deviation
Minimum
Maximum
34.93
3.18
4.00
14.41
7.53
5671.38
0.78
-7.51
0.16
67.96
1.13
1.23
9.12
6.10
9307.43
0.41
1784.47
0.36
-236.82
0.00
0.42
0.00
-1.46
102.24
0.00
-18285.75
0.00
669.76
6.00
6.00
61.10
36.11
46776.51
1.00
2314.87
1.00
Results
Empirical results from estimation of (17) are presented in Table 3. Column 1
contains estimates from the regression model using average tariff data for all goods,
while Column 2 reports estimates using tariff data on CFCs with imputed missing
values. The results are robust to two different measures of tariff. The likelihood ratio
test indicates that the time-specific effects are significant at the p<0.01 level.
26
Table 3
Panel data regression estimates
HONESTY
HONESTYLAW
LAW
LAWTARIFF
TARIFF
HONESTYTARIFF
GDPPC
GDPPCSQUARED
ARTICLE5
IMPORTPC86
PRODUCTION
YEAR93
YEAR94
YEAR95
YEAR96
YEAR97
YEAR98
YEAR99
YEAR00
YEAR01
CONSTANT
Adjusted R2
Observations
Estimation using average tariff on
all goods
CFCs
-5.62
-2.63
(-0.65)
(-0.34)
-2.92*
-2.98*
(-1.81)
(-1.88)
13.82**
14.25**
(2.2)
(2.4)
0.13
0.21
(0.56)
(0.63)
-3.06***
-4.84***
(-3.61)
(-3.68)
0.83***
1.29***
(2.86)
(3.15)
1.01E-02***
9.94E-03***
(11.73)
(11.94)
-2.07E-07***
-2.05E-07***
(-10.91)
(-10.97)
30.27***
30.41***
(4.63)
(4.65)
6.78E-03***
6.89E-03***
(6.04)
(6.15)
-71.73***
-71.39***
(-11.96)
(-11.84)
-12.56
-12.55
(-1.06)
(-1.06)
-31.39***
-31.39***
(-2.7)
(-2.71)
-38.26***
-38.30***
(-3.25)
(-3.27)
-39.54***
-38.98***
(-3.53)
(-3.5)
-43.49***
-43.13***
(-3.95)
(-3.94)
-45.17***
-44.89***
(-4.15)
(-4.16)
-51.62***
-51.73***
(-4.69)
(-4.74)
-53.80***
-53.32***
(-4.88)
(-4.89)
-60.21***
-60.26***
(-5.38)
(-5.43)
36.30
26.68
(1.21)
(0.95)
0.39
0.37
791
792
Notes: Dependent variable is net imports of CFCs per capita in ODP grams. t-statistics in parenthesis
beneath coefficient estimates. ***,**,* Denotes significance at the 1, 5 and 10 percent level,
respectively. All models are significant at the p<0.01 level.
27
The coefficient estimates provided in Table 3 are generally consistent with
the theory. In Columns 1 and 2, the coefficient of TARIFF is negative and significant
at the 1 percent level. This result is consistent with the theoretical prediction that,
when the values of HONESTY and LAW are close to zero, a higher tariff on CFCs is
negatively correlated with legal imports. In other words, for countries with high
corruption and a weak legal system, stringent environmental policy is associated
with a reduction in the legal use of CFCs. LAW enters positively and is significant at
the 5 percent level, supporting the theoretical conclusion that, when the values of
HONESTY and TARIFF are almost zero, stronger legal systems result in an increase
in legal trade. Thus, in highly corrupt countries where environmental standards are
very low, higher fines increase legal imports of CFCs. The empirical results reported
in Columns 1 and 2 do not provide support for the theoretical prediction that when
TARIFF and LAW are low, a fall in corruption is associated with an increase in
legally imported CFCs as the coefficient of HONESTY is negative and insignificant.
However, the empirical analysis corroborates the theoretical predictions that there
are important interaction effects between honesty and tariff and between honesty and
law.
The coefficient of HONESTYLAW is negative, as implied by the theoretical
model, and significant at the 10 percent level, suggesting that changes in corruption
have a greater effect on legal activity in countries with relatively strong judicial
systems, effectively making corruption and judicial strength complements in the
promotion of legal trade. The theory does not make any predictions about the sign of
28
the coefficients on the interaction terms between honesty and the tariff and between
the tariff and fine; thus, our inference about their effects is based on the empirical
estimation. The positive sign of HONESTYTARIFF, which is significant at the 1
percent level, suggests that the effect of corruption on the legal use of CFCs is
greater in countries where environmental tariff is low. This result implies that
corruption and the stringency of environmental standards are substitutes in the
promotion of legal trade. The coefficient of LAWTARIFF is positive but
insignificant.
In summary, the data provide support for the following theoretical
predictions: the effect of the legal system and tariff depends on honesty, while the
effect of honesty is conditional on the level of tariff and strength of the legal system.
The marginal effect of HONESTY on IMPORTPC, conditional on the degree of
LAW, is depicted in Figure 1. It is positive for weak legal systems and negative for
strong legal systems, consistent with the model’s prediction; for weak legal systems,
however, the marginal effect is not statistically different from zero. The relationship
between the marginal effect of LAW and IMPORTPC, conditional on the degree of
honesty, is depicted in Figure 2. Here, the marginal effect is positive for high levels
of corruption and negative for low levels of corruption, consistent with the theory;
for low levels of corruption, however, the marginal effect is not statistically
significant.
d(leagl imports of CFCs)/d(honesty)
29
25
20
15
10
5
0
-5
0
1
2
3
4
5
6
law
-10
-15
-20
-25
d(legal imports of CFCs)/d(law)
Figure 1
Marginal effect of honesty conditional on degree of law. Note: The dotted lines indicate the 95
percent confidence interval. Based on regression 1 in Table 3. Tariff evaluated at the mean.
30
25
20
15
10
5
0
-5
0
1
2
3
4
5
6
honesty
-10
-15
Figure 2
Marginal effect of law conditional on degree of honesty. Note: The dotted lines indicate the 95
percent confidence interval. Based on regression 1 in Table 3. Tariff evaluated at the mean.
30
Figures 3 and 4 depict the total effect of honesty and legal system on legal
imports. Since there is not much variation in the total effects of all relevant variables
on legal imports between different years, we present the results estimated for year
1998, which was randomly picked. Figure 3 shows the total effect of corruption on
legal trade under different legal systems. HONESTY decreases legal imports in
countries with strong legal systems (LAW takes the maximum value of 6) and
increases legal imports in countries with weak legal systems (LAW takes the
minimum value of 117). For most countries in the dataset (the 1998 value of the
honesty index is equal to or less than 5 for approximately 98 percent of the
countries), countries with strong legal systems have higher amounts of legally
imported CFCs than countries with weak legal systems. However, in countries where
corruption is very low, the estimated level of legal imports is about the same. The
total effect of legal system on legally imported CFCs estimated for 1998 is depicted
in Figure 4. The sign of the marginal effect is conditional on the level of corruption:
LAW increases legal imports in highly corrupt countries (HONESTY takes the value
of 0) and decreases legal imports in countries where corruption is very low
(HONESTY takes the value of 6). For the majority of countries (the 1998 value of
the index of legal systems of approximately 93 percent of the countries in the dataset
exceeds 2), more corrupt countries have higher levels of legal trade than less corrupt
ones; but as the strength of the legal system decreases, legal imports become
approximately the same.
17
Based on summary statistics for 1998.
legal imports
31
120
100
80
60
40
20
0
0
1
2
3
4
5
6
honesty
strong legal system
weak legal system
legal imports
Figure 3
Relationship between legal imports and honesty for countries with strong and weak legal systems in
1998. Based on regression 1 in Table 3. All other controls evaluated at the mean.
120
100
80
60
40
20
0
0
1
2
3
high honesty
4
low honesty
5
6
legal system
Figure 4
Relationship between legal imports and legal system for countries with high and low honesty in 1998.
Based on regression 1 in Table 3. All other controls evaluated at the mean.
32
A closer look at the last two figures reveals that knowledge about a country’s
level of corruption and strength of the legal system may help determine the actual
level of compliance with the Montreal Protocol. On the one hand, Figure 4 shows
that countries with strong legal systems and low corruption have lower legal imports
of CFCs than countries with strong legal systems but high corruption. Since the
theory suggests that the former also have lower total and illegal imports, we can
conclude that these countries exhibit better compliance with environmental
regulations that require reducing the consumption of CFCs. In our dataset, these are
generally high-income nations whose environmental performance is traditionally
very strong.18 In general, higher fines reduce the overall use of CFCs in countries
with both high and low corruption; but the size of this effect is larger in countries
with low corruption, as they introduce alternative technologies more quickly. On the
other hand, Figure 3 shows that highly corrupt countries with weak legal systems
also import fewer CFCs legally than highly corrupt countries with strong legal
systems. However, in this case, the theory suggests that total and illegal imports in
the former are much higher, so that lower legal imports in highly corrupt countries
with weak legal systems result in higher illegal activity and deterioration in
environmental performance.
Figure 5 plots the marginal effect of HONESTY on IMPORTPC, conditional
on the level of TARIFF. It is significantly negative for countries with low
environmental tariff and significantly positive for countries with high environmental
18
Countries that were assigned the maximum value of 6 both in the honesty and legal system ratings
in 1998 are Canada, Denmark, Finland, Iceland, Netherlands and Sweden.
33
tariff. With the additional assumption of a negative correlation between the level of
tariff and the absolute value of demand elasticity (i.e., with the assumption that in
countries with elastic demand for CFCs, the levels of tariff are set low19), this result
is consistent with the theory. Since low levels of tariff point to an elastic demand for
CFCs, an increase in honesty in this case will result in a reduction of both legal and
illegal imports,20 as producers switch to alternative technologies in response to an
increase in the price of CFCs. When the levels of tariff are high and the demand for
CFCs is, presumably, inelastic, an increase in honesty will lead to a decrease in
illegal imports, through a rise in the probability of detection, but an increase in legal
imports of CFCs, since alternative technologies are not accessible to producers in
these countries21. One comment, however, is appropriate here. While lower tariffs
are always associated with elastic demand for CFCs, the connection between the
tariff and price elasticity of demand becomes less obvious when the levels of tariff
are high. Rich nations, whose environmental policy has traditionally been more
stringent, may also have high environmental tariffs, even though substitution for
alternatives in these countries is technologically easier.
19
The data support the hypothesis as we found a significant negative correlation of 0.42 between
income per capita (high-income countries, presumably, can easily substitute alternatives for CFCs)
and average tariff for all goods.
20
Formally, from (15)
dl
= −∞ , hence if the absolute value of demand elasticity is high,
σ → −∞ dα
lim
honesty will decrease legal imports of CFCs.
21
Formally, from (15)
dl
1
=
> 0.
σ → 0 dα
α 1 − λ2α
lim
(
)
d(leagl imports of CFCs)/d(honesty)
34
70
60
50
40
30
20
10
0
-10 0
10
20
30
40
50
60
tariff
-20
-30
-40
d(leagl imports of CFCs)/d(tariff)
Figure 5
Marginal effect of honesty conditional on level of tariff. Note: The dotted lines indicate the 95 percent
confidence interval. Based on regression 1 in Table 3. Law evaluated at the mean.
5
4
3
2
1
0
-1
-2
0
1
2
3
4
5
6
honesty
-3
-4
-5
Figure 6
Marginal effect of tariff conditional on degree of honesty. Note: The dotted lines indicate the 95
percent confidence interval. Based on regression 1 in Table 3. Law evaluated at the mean.
35
The marginal effect of LAW on IMPORTPC, conditional on TARIFF (the
graph is not presented in this paper), is positive for all values of TARIFF, but is
significant only around the mean value of TARIFF. The marginal effect of TARIFF
on IMPORTPC, conditional on the level of HONESTY, is plotted in Figure 6. The
marginal effect is significantly negative for high levels of corruption, as predicted by
the model, but also significantly positive for low levels of corruption, which is
inconsistent with the theory. The marginal effect of TARIFF on IMPORTPC,
conditional on LAW (this graph is not presented in this paper), is negative for low
values of LAW and positive for high values of LAW, but is not statistically different
from zero. It is important to note here that the regression model used is a linear
approximation of (14), (15) and (16), which are non-linear functions of the
parameters. This may be one of the reasons why the marginal effects predicted from
the regression estimates do not completely capture the relationships implied by the
model.
The total effect of corruption on legal imports under different environmental
policy regimes is shown in Figure 7. The sign of the marginal effect of HONESTY
again depends on the stringency of environmental regulations: it is positive for high
levels of environmental tariffs (TARIFF takes the maximum value of 3622) and
negative for low levels of environmental tariffs (TARIFF takes the minimum value
of 0). Figure 8 depicts the total effect of environmental policy on legal imports for
different levels of corruption. The marginal effect of environmental tariff is negative
22
Based on summary statistics for 1998.
legal imports
36
120
100
80
60
40
20
0
0
1
2
3
high tariff
4
5
low tariff
6
honesty
legal imports
Figure 7
Relationship between legal imports and honesty for countries with high and low tariff in 1998. Based
on regression 1 in Table 3. All other controls evaluated at the mean.
120
100
80
60
40
20
0
0
5
10
15
high honesty
20
25
low honesty
30
35
tariff
Figure 8
Relationship between legal imports and tariff for countries with high and low honesty in 1998. Based
on regression 1 in Table 3. All other controls evaluated at the mean.
37
for countries with high corruption (HONESTY takes the value of 0) and, as in Figure
6, positive for countries with low corruption (HONESTY takes the value of 6). Both
the figures show that legal imports in highly corrupt countries with weak
environmental policies are higher than in highly corrupt countries with stringent
environmental policies. The data also suggest that, in countries with low corruption,
the relationship is reversed: countries with stringent environmental policies have
higher legal imports of CFCs than countries with weak environmental policies. (The
latter relationship, however, is not consistent with the theory.). From the theoretical
model, we know that highly corrupt countries with weak environmental policies have
higher total imports of CFCs, so that higher legal imports in this case indicate a
lower level of compliance with the Montreal Protocol. If the hypothesis that lower
tariffs are associated with a higher capacity to introduce ODS-free technologies is
correct, then the data suggest that countries with low corruption are more willing to
substitute alternatives for CFCs than highly corrupt countries.
Estimated coefficients of control variables are consistent with our
expectations. GPPPC and GDPPCSQUARED are significant at the p<0.01 level and
support the hypothesis of the inverted-U relationship between income per capita and
demand for environmental quality. GDPPC enters positively and GDPPCSQUARED
enters negatively, implying that countries with low income tend to increase their use
of CFCs, but after they reach a certain point in economic growth, they start to cut
down their consumption of CFCs that damage the environment. ARTICLE5 enters
positively and is significant at the p<0.01 level, suggesting that, conditional on the
38
level of per capita income and the other regressors, Article 5 countries import
approximately 30 more grams of CFCs per capita than non-Article 5 countries. The
coefficient of the 1986 level of imports per capita IMPORTPC86 is positive and
significant, which supports the hypothesis that the level of imports in 1986, used as a
benchmark for verifying compliance with the protocol, should be correlated with the
level of imports in subsequent years. The PRODUCTION dummy is negative and
significant, implying that CFC-producing countries import fewer CFCs. The
coefficients of year dummies enter negatively and significantly in all but one case,
suggesting that the use of CFCs declined, when compared to 1992, the year omitted
from the regression. The F-statistic indicates that the null hypothesis that the
coefficients of year dummies are all zero can be rejected at the 1 percent level of
significance, supporting the conjecture that there are time-specific effects. The
absolute value of the coefficients of year dummies rises with time, indicating that
each year there was an increase in cutbacks of CFCs in compliance with the
Montreal Protocol. For example, net imports decreased by approximately 31 ODP
grams per capita from 1992 to 1994, as compared to approximately 60 ODP grams
per capita from 1992 to 2001.
CONCLUSION
In this paper, we presented and analyzed a partial equilibrium model of
illegal trade in environmental goods. Because of the clandestine nature of smuggling,
it is not possible to test the predictions of the theory about the extent and direction of
response of the illegal activity to changes in parameters. But the theoretical model
39
allowed us to determine changes in legal activity taking place in the presence of
smuggling and test the theoretical predictions using panel data on legal imports of
CFCs.
Both the theoretical and empirical conclusions have important implications
for determining compliance with environmental regulations under an international
treaty when the imposition of quantitative restrictions results in the smuggling of
certain polluting substances across borders. Since illegal trade in these substances
cannot be observed, data on legal imports reported to international monitoring
agencies do not always reflect the actual level of compliance with the requirements
of the treaty. The analysis in this paper shows that knowledge about legal
environment, the level of corruption and the stringency of environmental policies in
different countries may help international observers gain better understanding of the
true level of pollution emitted in those countries. For example, in countries where
corruption is low and penalties for illegal behavior are high or where corruption is
high but environmental policies are stringent, lower legal imports of CFCs reflect
higher environmental performance as they coincide with lower total use of CFCs.
However, in highly corrupt countries where penalties are low, lower legal imports
indicate that higher amounts of CFCs have been transported illegally and the overall
use of CFCs has been increased.
Both the theory and the empirical evidence identify an interaction between
corruption, legal system, the stringency of environmental regulations and illegal
trade. In general, higher penalties for illegal behavior and lower corruption reduce
40
the overall use of CFCs and the best outcome is reached in those countries where the
two complement each other. The results regarding the effect of tariff are rather
mixed. On the one hand, a higher tariff increases the price of CFCs so that total
imports go down. On the other hand, a lower tariff may be associated with a higher
capacity to introduce ODS-free technologies and substitute alternatives for CFCs,
which in combination with lower corruption, may also lead to a reduction in the total
use of CFCs.
This study focuses on compliance with the Montreal Protocol, but could be
extended to other environmental issues, laying the groundwork for a more general
analysis of the role corruption plays in securing adherence to international
agreements.
41
BIBLIOGRAPHY
Bardhan, Pranab, “Corruption and Development: A Review of Issues,” Journal of
Economic Literature, Vol. 35, No. 3 (September, 1997), 1320-1346.
Beron, Kurt J., James C. Murdoch, and Wim P.M. Vijverberg, “Why Cooperate?
Public Goods, Economic Power, and the Montreal Protocol,” The Review of
Economics and Statistics, May 2003, 85 (2), 286-297.
Cole, M.A., A.J. Rayner, and J.M. Bates, “The Environmental Kuznets curve: an
Empirical Analysis,” Environmental and Development Economics, 2 (1997), 401416.
Congleton, Roger D., “Political Institutions and Pollution Control,” The review of
Economics and Statistics, Vol. 74, Issue 3 (Aug., 1992), 412-421.
Copeland, Brian R., “International Trade in Waste Products in the Presence of Illegal
Disposal,” in: A. A. Batabyal, H. Beladi, eds., The Economics of International
Trade and the Environment, Lewis Publishers, Boca Raton, 2001.
Damania, Richard, “Environmental Controls with Corrupt Bureaucrats,”
Environment and Development Economics, 7 (2002), 407-427.
Damania, Richard, Per G. Fredriksson, and John A. List, “Trade Liberalization,
Corruption, and Environmental Policy Formation: Theory and Evidence,” Journal
of Environmental Economics and Management, 46 (2003), 490-512.
Desai, Uday (Ed.), “Ecological Policy and Politics in Developing Countries:
Economic Growth, Democracy and Environment,” State University of New York
Press, Albany (1998).
Fredriksson, Per G., and Jakob Svensson, “Political Instability, Corruption and
Policy Formation: The Case of Environmental Policy,” Journal of Public
Economics, 87 (2003), 1383-1405.
Grossman, Gene M., and Alan B. Krueger, “Economic Growth and the
Environment,” The Quarterly Journal of Economics, Vol. 110, No. 2 (May,
1995), 353-377.
42
Kaufmann, Robert K., Brynhildur Davidsdottir, Sophie Garnham, and Peter Pauly,
“The Determinants of Atmospheric SO2 Concentrations: Reconsidering the
Environmental Kuznets Curve,” Ecological Economics, 25 (1998), 209-220.
Knack, S., and P. Keefer, 1995, “Institutions and economic performance: Crosscountry tests using alternative institutional measures,” Economics and Politics 7
(3), 207-227.
Lopez, Ramon, and Siddhartha Mitra, “Corruption, Pollution, and the Kuznets
Environment Curve,” Journal of Environmental Economics and Management, 40
(2000), 13-150.
Martin, Lawrence, and Arvind Panagariya, “Smuggling, Trade, and Price Disparity:
A Crime Theoretic Approach,” Journal of International Economics, 17 (1984),
201-217.
Mookherjee, Dilip, and I.P.L. Png, “Corruptible Law Enforcers: How Should They
Be Compensated?” The Economic Journal, Vol. 105, No. 428 (January, 1995),
145-159.
Murdoch, James C., and Todd Sandler, “The Voluntary Provision of a Pure Public
Good: The Case of Reduced CFC Emissions and the Montreal Protocol,” Journal
of Public Economics, 63 (1997), 331-349.
Murdoch, James C., Todd Sandler, and Wim P.M. Vijverberg, “The Participation
Decision versus the Level of Participation in an Environmental Treaty: a Spatial
Probit Analysis,” Journal of Public Economics 8 (2003), 337-362.
Pitt, Mark M., “Smuggling and Price Disparity,” Journal of International
Economics, 11 (1981), 44-458.
Sandler, Todd, and Daniel G. Arce M., “Pure Public Goods versus Commons:
Benefit-Cost Duality,” Land Economics, 79 (3), (August, 2003), 355-368.
Shafik, Nemat, “Economic Development and Environmental Quality: An
Econometric Analysis,” Oxford Economic Papers, New Series, Vol. 46, Special
Issue on Environmental Economics (Oct., 1994), 757-773.
Shleifer, Andrei, and Robert W. Vishny, “Corruption,” The Quarterly Journal of
Economics, Vol. 108, No. 3 (August, 1993), 599-617.