DOES PROTECTION BEGET CORRUPTION?

*




Pushan Dutt
Department of Economics
7-11 Tory, University of Alberta
Edmonton, AB T6G 2H4
pdutt@ualberta.ca
July 22, 2002




Abstract

This paper develops and tests a model, which predicts that protectionist policies
on the part of the government leads to increased corruption on part of the bureaucracy. A
one-sector small open economy model is presented. Corruption is shown to be increasing
in tariffs and subsidies, and decreasing in civil sector wages. Using multiple measures of
corruption and protection, the predictions of the model are tested. We find evidence
suggesting that corruption is indeed higher in countries pursuing active trade and
industrial policies. The empirical results are checked for robustness and policy
implications are discussed.

Key Words: Corruption, Protection
JEL Codes: F10, F13, F14, D72, D73

*
I would like to thank Andres Velasco, Raquel Fernandez, Giovanni Maggi, Giorgio Topa, Debraj Ray and
Devashish Mitra for invaluable suggestions and comments. I thank seminar participants at New York
University, University of Alberta, University of Georgia and the Southeast International Economics
Conference for very useful discussions. I would also like to thank Francesco Rodriguez, Daniel Lederman,
Paulo Mauro, Caroline Van Rijckeghem and Beatrice Weder for sharing their data with me. The usual
disclaimer applies.
1 Introduction
Are countries that are more protectionist also more corrupt? The paper addresses this question,
rst within a theoretical framework that delineates the channels through which protectionist
trade policies aect corruption. It then veries the model predictions empirically. Traditionally,
a number of arguments have been advanced to justify trade protection: (i) The terms of trade
argument - this requires that a country possess some degree of market power in world markets.
(ii) Strategic trade policy arguments - in the presence of strategic interactions, governments can
use appropriate policies (taris or subsidies) to shift rents from foreign to domestic rms. (iii)
The production eciency eect - protection can induce, under some circumstances, an increase
in domestic production and improve welfare. (iv) The external economies argument - the most
popular version is the infant industry argument widely cited by policy makers in developing
nations in the post World War period. (v) Factor market distortions provide another justication
for trade policies. All such arguments sketch a variety of scenarios where trade protection can
enhance welfare.
The purpose of this paper is not to take up any particular position on the relative merits of
free trade vs. protectionism. Rather, it is to point to a side eect of trade protection that could
temper or even compromise the goals of trade policy. We show that activist trade and industrial
policies lead to an increase in corrupt activities on the part of the bureaucracy. In particular,
trade and subsidy protection transfer rents to rms, and bureaucrats who we assume, control the
rights to licensing of production, extract these rents from rms in the form of bribes. If we believe
that protectionist policies are welfare enhancing, the calculus of such policy needs to factor the
eect of corruption that such policies generate and which, in turn, can be welfare reducing.
1
In a
nutshell, there exists a trade o between the direct welfare enhancing eects of trade and industrial
policies and their indirect welfare reducing eects operating through the channel of corruption.
1
Mauro (1995) for instance, shows that corruption has a negative impact upon economic growth.
1
The trade literature has addressed this issue briey in the form of the Kruegers (1974) analysis
of rent seeking activities.
2
She recognizes that government regulations are pervasive, and give
rise to rents and rent-seeking, which may take the form of bribery and corruption. However, her
analysis is mainly concerned with showing that welfare losses with import quotas (that give rise
to rent-seeking) are greater than losses under an equivalent tari. This paper focuses on the fact
that any of a plethora of trade restrictions can transfer rents to domestic rms and provide greater
incentives to the bureaucrats at the margin to be corrupt.
While corruption has recently attracted increasing attention,
3
empirical studies on the factors
that promote corruption are limited. A majority of the studies, such as Becker and Stigler (1975),
Rose-Ackerman (1975), and Klitgaard (1988, 1990), analyze bureaucratic corruption as a principal-
agent problem. The focus is on devising ways and means that enforce honesty on part of the
bureaucrats. Others seek to explain the persistence of corruption in terms of frequency dependent
models that generate multiple equilibria. Lui (1986), Cadot (1987), and Andvig and Moene (1990)
present models in this vein. Another stream of research analyzes the consequences of governmental
corruption. Mauro (1995) looks at the eect of corruption on investment and growth; Wei (1997)
analyzes its impact on foreign direct investment and Mauro (1998) examines whether corruption
distorts the composition of government expenditure.
To the best of the authors knowledge there exist a handful of papers that try to explain cross-
country variations in corruption. Ades and Di Tella (1997), show that corruption is associated
with active industrial policies and Van Rijckeghem and Weder (1997) show that there is a negative
relationship between corruption and civil sector wages. While the former concentrates purely on
industrial policies in terms of examining the eects of favorable domestic procurement and scal
policies on corruption, the latter concentrates on testing eciency wage models of corruption. Nei-
ther paper examines the eect of trade policies and both test their models only for a small sample
2
See also Bhagwati (1982) for a general concept of directly unproductive activities and the welfare consequences
of such activities.
3
See Bardhan(1997) for a useful survey.
2
of countries (32 countries in Ades and Di Tella and 25 developing countries in Van Rijckeghem and
Weder). More recently, Treisman (2000) nds that countries with Protestant traditions, histories
of British rule, more developed economies, and higher imports were less corrupt. Ades and Di
Tella (1999) focus on how competition aects corruption and nd empirically that rents, proxied
by lack of import competition, fuel and mineral exports, and trade distance, increase corruption.
However, none of these papers systematically investigate the eect of activist trade policies upon
corruption.
4
This paper, while emphasizing the eect of protectionist policies on corruption, also
attempts to explain dierences in corruption levels across a broad cross-section of countries. Ac-
cordingly, this paper identies the channels through which corruption operates, the inducements
for bureaucrats to engage in dishonest behavior, and the various checks on such malfeasant behav-
ior. It then proceeds to test whether the empirical counterparts to these explanatory variables,
do in fact, explain the cross-sectional variation in levels of corruption.
We set up a simple small open economy model within which corruption and bribes are deter-
mined. Corruption in this model takes the form of petty bureaucratic corruption. The bureaucrats
are the agents of the government and are entrusted with carrying out a task by the government
but engages in malfeasance that is dicult to monitor. Bribes take the form of a price for license
(whose market price is set to zero, viz., there are no licensing fees), and corruption manifests itself
in the form of a transfer of surplus from rm owners to bureaucrats. Bribes are assumed to be
determined via a bargaining process between the bureaucrat and the rm. We assume that the de-
tection probability and the wages of bureaucrats are exogenously given, thereby abstracting from
issues of multiple equilibria and the design of incentive compatibility contracts by the government
to deter corruption. In our model we show that a larger surplus provides greater inducement to be
corrupt, and protectionist policies add to the surplus earned by domestic rms. The model, thus
predicts that trade and industrial policies lead to higher bribes and a greater incidence of corrup-
4
Most empirical papers on the determinants of corruption throw in

X+M
GDP

or
M
GDP
as a control for openness.
However, this is an indirect measure of trade restrictions and notoriously unreliable.
3
tion. In the model, corruption results arises out of the interaction of the discretionary power of the
bureaucrats and from the monopoly power enjoyed by rms. Corruption indices from a variety of
survey measures, is regressed upon a variety of measures of trade and industrial policy, licensing
regulations, proxies for the detection probability, civil sector wages and relevant controls. We nd
strong support that protectionism engenders corruption - a result which is robust to the treatment
of corruption as an ordinal measure, to endogeneity issues, and to variations in corruption over
time and space.
The paper is organized as follows. Section 2 sets out the model; Section 3 solves for the
equilibrium and performs some comparative statics; Section 4 lists the data sources and provides
descriptive statistics for the variables to be used in the empirical analysis; Section 5 discusses the
empirical ndings; Section 6 concludes.
2 The Model
2.1 The Consumer and Firms Problem
This section develops a basic monopolistic competition model with dierentiated products and
intra-industry trade. Consider a small open economy model where following convention, we label
the two countries as Home and Foreign with the analysis being conned to the home country
and where the label Foreign will be reinterpreted to mean the rest of the world.
5
It is assumed that there are a continuum of dierentiated products, each produced under in-
creasing returns to scale by a monopolist for that particular variety. The representative individual
has a Dixit-Stiglitz constant elasticity utility function,
U =

Z
n
0
x
1

hi
di +
Z
n

0
x
1

fi
di
!
1
, > 1 (1)
where x
hi
is the consumption of the domestically produced variety i and x
fi
is the consumption of
the imported variety i. n and n

are the total number of domestic varieties and foreign varieties

5
In a two country framework, the analysis will remain unchanged.
4
respectively, and is the elasticity of substitution. This structure yields the following domestic
demand functions for the dierentiated products.
6
x
hi
=

p
hi
P

E
P
; x
fi
=

p
fi
P

E
P
(2)
where E is the countrys total spending which we assume to equal its income, p
hi
is the consumer
price of the domestically produced variety, p
fi
is the domestic consumer price of the imported
variety and
P =

Z
n
0
p
1
hi
di +
Z
n

0
p
1
fi
di
! 1
1
(3)
is an index of the prices of the dierentiated goods. Foreign demand for domestic varieties is
derived in an analogous manner
x

hi
=

p

hi
P

(4)
where E

is the foreign spending and income, p

hi
is the foreign consumer price of the variety i
produced at home and
P

Z
n
0
p
1
hi
di +
Z
n

0
p
1
fi
di
! 1
1
Under the small open economy assumption, the country faces a given foreign price and a given
number of the dierentiated products, and a given foreign spending on dierentiated products.
The small open economy imposes a tari t on the imports of dierentiated products. This implies:
p
fi
= (1 +t)p

fi
where p

fi
is the foreign price of the foreign variety i. Despite the small open economy assumption,
domestic rms posess monopoly power and face a downward sloping demand schedule. We further
assume that the home country provides export subsidies at the rate e on home dierentiated goods
so that
p

hi
=
p
hi
(1 +e)
6
denotes the variable refers to price or demand in the foreign country and the subscript, h, f denotes whether
the good in question is of the home country or foreign country.
5
We also assume that foreign varieties are equally priced with units conveniently chosen so that
p

fi
= 1, i. The above price relationships imply
x
hi
=

p
hi
P

E
P
; x
fi
=

p

fi
(1 +t)
P

E
P
; x

hi
=

p
hi
(1+e)
P

(5)
P =
Z
n
0
p
1
hi
di +n

[1 +t]
1
1
1
; P

Z
n
0

p
hi
1 +e

1
di +n

! 1
1
(6)
The home producer of variety i faces the demand function of x
hi
+ x

hi
. Given the continuum
of varieties each producer is negligible with respect to the entire market and therefore faces a
constant elasticity of demand, .
We now assume that there are two factors of production in xed supply: unskilled labor that
we denote as L and entrepreneurs whom we denote as

K. Entrepreneurs provide managerial
and organizational skills and we assume that 1 unit of entrepreneurs is required to commence
production. Therefore, production has a constant unit labor requirement of c and a xed cost of 1
unit of entrepreneurs. The latter is independent of the quantity produced. w
h
is the remuneration
to the entrepreneur and w is the wage of unskilled labor. w is chosen as the numeraire with units
chosen such that w = 1. Firms engage in monopolistic competition and obtain an output subsidy
at the rate of s per unit produced. Total costs of producing x units of a variety is therefore
w
h
+wcx sx = w
h
+ (c s)x (7)
The pricing rule derived from prot maximization is
p
hi
= p
i
=

1
(c s) (8)
which implies that all home varieties are equally priced with p
i
= p and produced in the same
quantity, x
hi
+x

hi
= x
h
+x

h
= x. The xed endowment of entrepreneurs, K determines, n = K
as the number of varieties. Finally, the wages of the entrepreneur is simply the prot or surplus
left over after payment to unskilled labor. Therefore w
h
must satisfy
w
h
= px (c s)x (9)
6
There are L units of unskilled labor each of whom earn w = 1 and K entrepreneurs each of whom
earn w
h
. National income is therefore given as
Y = E = L +w
h
K +T (10)
where T is the net rebate from the governments policy. We assume that subsidies are nanced
and taris are rebated by the government in a lump sum fashion. The government also pays the
wages of bureaucrats w
b
who are of measure 1. Therefore, the budget constraint of the government
is as follows
w
b
+eKp

h
x

h
+sK(x

h
+x
h
) +T = tn

x
f
(11)
where w
b
is the wage of each bureaucrat. T adjusts so that the above equality always holds.
Lemma 1 In the prot maximizing equilibrium, a small increase in the tari rate t, in the export
subsidy s, or in the output subsidy e, raises the wages or the operating surplus of the entrepreneur..
Proof: Operating surplus can be written as
w
h
= (p c +s)x
By the envelope theorem
w
h
s
= x > 0
Total revenue may written using equations (6)-(9) as
px = p(x
hi
+x

hi
)
=
p
1
np
1
+n

(1 +t)
1
Y + (1 +e)
h
p
1+e
i
1
n
h
p
1+e
i
1
+n
E

Now prots may be written as a constant share of revenue using the mark-up pricing equation.
That is
w
h
=
px

Since > 1, from the above equations it follows that

w
h
t
> 0
w
h
e
> 0
7
Therefore, within this framework any form of government intervention serves to increase the return
to the skilled factor owner or the entrepreneur
7
.
2.2 The Bureaucrats Problem
Each entrepreneur in order to commence production has to obtain a license from the government.
Licenses are approved by bureaucrats who act as agents of the government, thereby providing them
with discretionary power. There are a continuum of bureaucrats over the interval [0, 1] and every
entrepreneur is matched at random with a particular bureaucrat. Licenses are assumed to be cost-
less. Bureaucrats might, however, choose to be corrupt and demand bribes from the entrepreneur
in return for issuing the license. In case they choose to demand a bribe, the bribe amount is
determined through a Nash bargaining game between the bureaucrat and the entrepreneur.
Each bureaucrat faces a two stage decision problem: in stage 1 they choose whether to be
honest or to be corrupt, in stage 2 they choose the bribe conditional on the decision in stage 1. Of
course, if they choose to be honest they set the bribe to zero in stage 2. Bureaucrats are assumed
to be risk neutral and maximize their expected income from corruption.
Stage 1: In stage 1 bureaucrats choose between being honest or being corrupt. If they are
honest they merely receive their wages w
b
which is assumed to be exogenously given. If they
choose to be corrupt, then their income consists of their wage income w
b
and bribe income, b so
that their total income from corruption equals w
b
+b. However, there is an exogenous determined
probability of being caught (b) in corrupt transactions, in which case they are simply red and
lose both their wage as well as bribe income. With probability (1 (b)) they are not caught so
that the expected income from being corrupt equals (1(b))(w
b
+b)+(b)0 = (1(b))(w
b
+b).
We assume that
(b) =
_

_
0 for b = 0
for b > 0
(12)
7
Given that entrepreneurs also consume the dierentiated goods, subsidies are simply a transfer from consumers
(who are presumably taxed in a lumpsum fashion) to entrepreneurs so that Y is unaected. Taris on the other
hand generate tari revenues which is repatriated to consumers so Y would increase.
8
This assumption implies that every bureaucrat who chooses to be corrupt and sets b = 0 in eect
chooses to be honest.
In addition, we assume that bureaucrats have certain moral costs from being corrupt and these
moral costs, denoted as c
i
for the i
th
bureaucrat, are heterogenous across bureaucrats. These costs
are distributed over the interval [0, c

] with the cumulative density F(c) such that F(0) = 0 and

F(c

) = 1. Income and costs are measured in terms of the numeraire. Therefore, bureaucrat i
will choose to be corrupt as long as
c
i
< (w
b
+b) (1 ) w
b
(13)
= b(1 ) w
b
(14)
Since bureaucrats are identical in all other respects, conditional on choosing to be corrupt they all
demand the same proportion g as bribes. This implies that the proportion of corrupt bureaucrats
or the incidence of corruption, given the values of , g, w
b
and w
h
equals
=
_

_
F (max(0, (b(1 ) w
b
)) if c

> (b(1 ) w
b
)
1 otherwise
(15)
The assumption that
c

> (w
h
(1 ) w
b
)
is sucient to guarantee that = 1 cannot be an equilibrium. A low enough w
b
and and a high
enough w
h
will guarantee that in equilibrium
b(1 ) w
b
> 0
so that = F(b(1 ) w
b
) > 0.
8
Stage 2: In this stage bureaucrats, conditional on choosing to be corrupt, choose b - the
bribe that they will demand which will come out of the entrepreneurs operating surplus. The
bribe amount originates as a Nash bargaining exercise between the corrupt bureaucrat and the
8
Note that if the government could freely choose either w
b
or it could ensure that = 0. However, budget
constraints may prevent it from enforcing a zero corruption outcome.
9
entrepreneur. The disagreement point in this exercise is (0, 0), i.e., if there is no agreement, the
bureaucrat receives no payment as bribe, and the entrepreneur is who is denied the license cannot
commence production and therefore earns zero surplus.
9
The entrepreneur and the bureaucrat
are both assumed to be risk neutral. Under the assumption of symmetrical bargaining power, the
equilibrium level of the bribe is determined as the solution to
arg max
b
[b (w
h
b)]
The Nash bargaining solution is
b =
w
h
2
(16)
With asymmetric bargaining powers, bribes will continue to be a fraction of the surplus and
therefore increasing in the surplus.
10
Therefore, we have the following lemma.
Lemma 2 The proportion of operating surplus demanded as bribe by the bureaucrat is increasing
in the operating surplus.
In order to nd the incidence of corruption, we solve the bureaucrats problem backwards.
Stage 2 gives b =
w
h
2
which we use to obtain an expression for corruption as
= F

max
h
0,
w
h
2
(1 ) w
b

i
if c

>

w
h
2
(1 ) w
b

= 1 otherwise (17)
We will get an interior equilibrium with (0, 1) as long as 0 <
w
h
2
(1 ) w
b
< c

in which
case we may rewrite a simplied expression for .
= F

w
h
2
(1 ) w
b

Lemma 3 The incidence of corruption is increasing (weakly) in: (a) The operating surplus of
the entrepreneur w
h
; and (b) weakly decreasing in the exogenously given detection probability
and in the wages of the bureaucrat w
b
9
We could very well assume that the entrepreneur if denied the license, works as an unskilled worker and earns
wages. This will not change any of the results. In this case the higher the wages of unskilled labor, the better is
the entrepreneurs outside option, and thus lower will be corruption.
10
Here we set up the entreprenreur-bureaucrat interaction as a bargaining game. Alternatively, we could set it
up as a non-cooperative game between the bureaucrats and entrepreneurs. The results remain unchanged.
10
Proof: Since F(.) is a cumulative density function it follows that
F
w
h
= F
0
(.)
(1 )
2
0
F

= F
0
(.)

w
h
2
w
b

0
F
w
b
= F
0
(.) () 0
The weak inequality signs are used since may take the values 0 or 1. In the case of an interior
equilibrium, each of the signs above will be a strict inequality, provided F
0
> 0.
3 The Equilibrium
We now try and determine the equilibrium levels of bribe and the operating surplus. Combining
Lemmas 1, 2 and 3 we can write the following proposition.
Proposition 4 The equilibrium level of the bribe g as a proportion of the surplus, and the inci-
dence of corruption in equilibrium, is increasing in the import tari t, the export subsidy s, and
the output subsidy s.
Proof:
With proportion of bureaucrats being corrupt. The entrepreneur will maximize her expected
income as
max
x
(w
h
b) + (1 ) (w
h
)
If the bureaucrat demands a proportion g as bribe so that b = gw
h
where g (0, 1) depends on
the bargaining power of the respective parties then, this is equivalent to maximizing
w
h
(1 g)
Since g < 1 the entrepreneur simply maximizes her operating surplus w
h
. Moreover, each corrupt
bureaucrat takes the operating surplus to be equal to this value and chooses whether to be corrupt
or honest to maximize her expected income. From Lemma 2 and 3, we have that
db
dw
h
> 0;
d
dw
h
0
11
Finally from Lemma 1 we have that the Nash level of the operating surplus is such that:
dw
h
dt
> 0,
dw
h
de
> 0 and
dw
h
ds
> 0
Therefore
dg
dt
> 0,
dg
de
> 0 and
dg
ds
> 0
and
d
dt
0,
d
de
0 and
d
ds
0
The intuition behind this results is simple: greater protection yield greater rents or surplus to
the entrepreneurs thereby providing greater incentives for corruption. However, this is not the only
channel via which higher taris lead to a higher incidence of corruption. Where taris exist, there
is an incentive to evade them via smuggling, whether overt or in the form of invoicing changes.
11
Both of these can manifest themselves in the form of corruption - the former as bribes paid to law
enforcement ocials and the latter as bribes to customs ocials. These incentives increase with
the tari rate so that importers are willing to pay higher bribes as protection increases and this
will lead to higher corruption.
A second channel that may also lead to rent seeking and corruption, is tari destroying lobbying
with the aid of bribes to politicians (See Bhagwati, 1982). Again, the higher the taris the higher
will be the willingness-to-pay of rms upon whom the tari is imposed to reduce or remove the
tari. Finally, as Bhagwati and Srinivasan (1980) point out, there may be a competition for
securing a share in the disbursement of tari revenue. While they model the process as a legal
directly unproductive activity, the shares may be distributed in accordance to bribes paid by
various interest groups. This provides an additional channel through which protectionist policies
may impact corruption. Here, whether corruption is increasing in protectionist policies depends
crucially upon the elasticity of import demand.
11
Bhagwati and Hansen (1973), Bhagwati and Srinivasan (1973), Kemp (1976), Ray (1978) analyze smuggling
as a directly unproductive activity and the relationship between commercial policy and smuggling.
12
4 Data Sources and Some Basic Statistics
Our dependent variable is corruption and our main independent variables are taris and subsidies.
In addition, we will be using independent variables that are likely to aect the detection probability
, that control for alternative channels of corruption (in addition to taris and subsidies) and that
control for certain region specic eects. We will also use a variable that captures the level of
regulatory discretion present in the economy. It is the existence of such discretion from which
bureaucrats derive their power to extract rents.
We use three sources for data on the indices of corruption, all of which are based on surveys.
The rst measure is the International Country Risk Guide (ICRG), available for 129 countries,
used previously and described in detail in Knack and Keefer (1993).
12
Corruption is measured
on a zero to six scale, where low scores indicate high levels of corruption. The second measure
is compiled by Transparency International (TI). The TI index ranges from one (least corrupt) to
ten (most corrupt) and is itself an average of a number of survey results. The averaging procedure
should reduce measurement error if the errors in dierent surveys are independent. Table 2 in
the appendix shows the corruption ratings from Transparency International for 86 countries. We
use the 1998 TI index of corruption. Third, we use the German exporter corruption index (GCI)
which measures the total proportion of deals involving kickbacks, according to German exporters
(See Neumann (1994).
13
To avoid awkwardness in the interpretation of the coecients, the ICRG
and TI measures were recoded as six minus the original corruption index, and as ten minus the
original corruption index respectively, so that now higher numbers indicate higher corruption. As
shown in table 1.2, these three indices of corruption are highly correlated with the correlation
coecient exceeding 0.8 for any pair of measures.
12
This measures the extent to which bribery is present within the political system. Forms of corruption given
in the ICRG are related to bribes in areas of exchange controls, tax assessments, police protection, loans and
licensing of exports and imports. Low scores on the corruption index indicate that government ocials are likely
to demand special payments and illegal payments are generally expected throughout lower levels of government
in the form of bribes.
13
This measure has the advantage of being an easily interpretable cardinal measure and accordingly we only
report OLS and IV results for this measure. However, data is available for only 43 countries.
13
To test for the robustness of our results, we use a variety of trade policy measures: total
import duties collected as a percentage of total imports (IMPORT DUTY), an average tari
rate calculated by weighing each import category by the fraction of world trade in that category
(TARIFF),
14
a coverage ratio for non-tari barriers to trade (QUOTA). The last two measures,
taris and quotas are used in conjunction since they better capture trade restrictions when tari
and non-tari barriers are substitutes. We also use the Sachs-Warner openness indicator - an
indicator that attempts to solve the measurement error problem by constructing an index that
combines several aspects of trade policy. Although this is a coarse partition of economies into open
and closed economies, it will allow us to evaluate if the economies that are regarded as closed also
have a tendency to be more corrupt.
15
For subsidies, there is very little cross-country data on
export subsidies. As a result, we are compelled to use a measure for subsidies obtained from the
World Development Indicators that measures all transfers from the Central government to private
and public enterprises as well as consumption subsidies.
We use two measures of civil sector wages from two dierent sources. First, a new data
set on civil sector wages has recently become available for 86 countries from the World Bank.
From here we use the real wage rate per government employee in constant 1997 dollars in our
analysis. The second wage concept used is the government wages relative to manufacturing wages.
Manufacturing wages may be thought as a proxy for outside option for bureaucrats (in case they
are caught taking bribes and red) since the skill-content in the manufacturing sector is probably
lower than in the government sector, making it a suitable candidate for representing outside
opportunities for the bureaucrat. High manufacturing wages make corruption more attractive to
14
The variable is referred to as taris, although it includes all import charges, such as duties and customs fees.
15
The Sachs-Warner dummy takes the value of zero if the economy is closed according to any one of the following
criteria: 1. Average tari rates exceeded 40%; 2. Non-Tari barriers covered more than 40% of imports; 3. The
country had a socialist economy system; 4. It had a state monopoly of major exports. 5. The black market premium
on the exchange rate exceeded 20% during either the 1970s or 1980s. This measure has been criticized on various
grounds. For instance, the black market premia is more an indicator of macroeconomic imbalances and/or balance
of payment crisis. Further, as Rodrik and Rodriguez (1999) point out, the state monopoly of exports variable
is virtually indistinguishable from a sub-Saharan African dummy because the World Bank study on the basis of
which it was constructed covered African countries that were under structural adjustment programs. Therefore,
the empirical results that use the Sachs-Warner measure need to be interpreted with some caution.
14
bureaucrats as they become more willing to assume the risks of detection. This data is from van
Rijckeghem and Weder who have compiled this data for 28 developing over the period 1982-94.
We take the average relative wage for the 1980s. In addition, data for six OECD for exactly the
same variable was added from the OECD database.
The probability of detection cannot be observed directly so in its stead we use a number
of variables that we feel impact monitoring and through it the probability of detection. The
rst variable that we use is democracy. Democratic countries are likely to exercise pressure
on politicians to keep corruption in check with the electorate punishing corrupt regimes at the
voting booth. A dictatorial regime on the other hand is likely to be less sensitive to demands for
controlling corruption through monitoring mechanisms and punishment schemes. For a measure
of democracy, we use the Freedom House measure of democracy, which derives from the work of
Gastil and his followers. The Gastil index provides a subjective classication of countries on a
scale of 1 to 7 on political rights, with higher ratings signifying less freedom. The average for
the 1980s is used. The other measures of checks that might aect the probability of detection
are schooling and the number of newspapers per thousand of the population. Both are likely to
increase the awareness of the public to illicit activities on part of the bureaucrat creating social
pressures against corruption. The fourth estate also acts as a guardian for the government and its
agents. The data on these measures are obtained from the World Development Indicators (2000).
A necessary condition for corruption to be present in the model is the existence of licensing.
This provides discretionary power to bureaucrats and thereby the ability to extract rents. More-
over, alternative channels could exist that promote corruption, such as distortion of wages and
prices through price controls. Therefore, we use two other variables - an index of regulation that
measures the degree to which licensing is prevalent in the economy and an index of wage and
price regulation. The former measures the degree of licensing requirements, as well as prevalent
levels of labor, environmental, consumer safety and worker health regulations. Wage and price
regulation, measures the degree to which the government controls wages and prices, through min-
15
imum wage laws and price controls. The data on these two measures is derived from the Heritage
Foundation, 1998. To control for regional eects we use three dummies: an East Asia dummy, an
oil dummy and a Sub-Saharan African dummy. Finally, in our instrumental variables estimation
we use ideology and ideology multiplied by the capital-labor ratio as instruments. The data on
political orientation are obtained from the Database of Political Institutions (DPI) (Beck et al,
2001) and measures whether the political party or ruler in power can be classied as left-wing,
centrist or right-wing. The data on capital-labor ratios are obtained from Easterly and Levine
who use aggregate investment and depreciation data to construct capital per worker series for 138
countries.
Table 1.1 provides summary statistics for these independent variables.
5 Results
5.1 OLS and Ordered Probit
Figures 1 and 2 plot the data for corruption and measures of trade protection (taris and import
duties respectively). The plots seem to indicate a positive relationship as hypothesized in this
paper with a few outliers.
16
Next, I use both ordinary least squares and ordered probit methods
to estimate the model. The surveys upon which our corruption measure is based, ask how likely it
is that the respondent will encounter corruption. We can interpret the response in a probabilistic
fashion, thus giving it a cardinal interpretation and justifying the use of OLS methods. However,
if we believe that the corruption measures are simply a ranking and cannot be given a cardinal
interpretation, using an ordered probit model seems more appropriate.
Table 3 presents the base regression where we regress each of the three measures of corruption
on dierent measures of trade restrictions. All measures of trade policy save the quota coverage
ratio are signicant (at the 99% level of condence) and with the predicted sign. At the rst run,
therefore, our theoretical prediction is strongly supported. A one standard deviation increase in
16
The outliers on the basis of the outlier tests of Hadi (1992) are Italy, Japan and Korea. However, our results
are robust to their deletion.
16
taris raises the TI and ICRG measures of corruption by one point and 0.59 points respectively.
Similarly, a one standard deviation increase in import duty raises the ICRG and TI measures
by 0.8 and 1.1 points respectively. From table 2, we can see that the latter is equivalent to an
improvement from the corruption levels in Zimbabwe to that in Belgium.
In tables 4-7, we present the full regression results where all the regressors are included. Each
of the tables corresponds to a particular measure of protection. In table 4 we use taris and quotas,
in table 5 we use import duty and in table 6 we use the Sachs-Warner measure of protection.
17
In each table we report regression results for all three measures of corruption - both OLS as well
as the ordered probit results. In table 4-7, we regress corruption on all the independent variables,
apart from wages. The reason for doing so is the small sample of countries for which consistent
relative wage data are available.
18
In table 4 we see that the data supports the positive relationship between corruption and
tari rates. This relationship holds across corruption measures and regardless of whether we
give corruption a cardinal or an ordinal interpretation. Quotas, however, are not signicant and
even have the wrong sign. The quota coverage ratio, as such suers from measurement error
problems due to smuggling, coding problems and weaknesses in the underlying data. It also does
not distinguish between barriers that are highly restrictive and barriers that are not binding and
have little eect. Nor is it possible to measure the impact of relaxing quotas on trade ows.
The coverage ratio only suggests that barriers to trade exist, but cannot measure their eect. If
non-tari barriers are ineectively implemented they will not shift demand to domestic producers
so that the postulated channel through which protection impacts the incidence of corruption -
shifting demand leading to an increasing surplus which increases the benets of being corrupt
at the margin and thus the incidence of corruption - will be ineectual. Harrigan (1993) has
found that for OECD countries in 1983 both price and quantity NTB coverage ratios are, in most
17
We have recoded the dummy variable so that 0 implies that the country is open and 1 implies protected
according to the criteria listed above
18
Note also that the R
2
reported in the tables for the ordered probit estimates are the pseudo R
2
.
17
cases, not associated with lower imports. He points out that these coverage ratios are the noisiest
indicators of trade policy as there are severe problems with their construction procedure and are
not even conceptually what would be desired.
19
Therefore, not surprisingly, in all our regressions,
we fail to nd any signicant results for the quota coverage ratio.
In tables 5 and 6, import duty and the Sachs-Warner index of protection are also strongly
signicant for all measures of corruption. Countries that are classied as closed by the Sachs-
Warner measure jump approximately 10 spots in the corruption rankings. Similarly, a one standard
deviation increase in taris (import duty) raises the ICRG measure by 0.18 (0.25) points and a
similar increase raises the TI measure by 0.4 (0.24) points. While the magnitude of change is
smaller after including controls, it is by no means unsubstantial.
20
Subsidies are strongly signicant in all the regressions and have the predicted sign. Again, a
one standard deviation increase in subsidies raises the TI measure on average by approximately
0.5 points, and the ICRG measure by 0.2 points. This variable measures the degree of support to
domestic economic agents - both consumers and producers and for this reason it does not coincide
with the exact denition of subsidy - production subsidy and export subsidy - that we used in our
model above. Unfortunately, data on subsidies that makes such distinctions are not available at a
national level for a cross national sample of countries. However, we can argue that consumption
subsidies add to the consumers income and therefore the demand for dierentiated products. This
increases the operating surplus and thus they should have a similar eect upon corruption. We
nd that this conclusion is supported in all regressions.
Thus the empirical ndings support our prediction that protection from foreigners through
taris and promotion of national rms via subsidies tend to increase corruption. The fact that
corruption is measured via surveys of rms and entrepreneurs (so that these indices measure cor-
19
For a detailed discussion of the problems with quantity and price NTB coverage ratios, see Leamer (1990).
20
In contrast Treisman (2000) nds that openness reduces corruption but the size of the eect is very small.
However, he uses the import-GDP ratio rather than any direct measures of trade restrictions. The share of imports
in GDP is not an accurate measure of trade restrictions, it depends on a variety of factors such as country size,
and comparative statics of corruption with respect to this measure is not very meaningful.
18
ruption and/or bribery demands that these rms face in conducting business activities) supports
the proposition that protection increases the operating surplus thereby providing bureaucrats
more incentives to be corrupt at the margin. The fact that licensing regulations exercise a strong
inuence on corruption lends further support to this contention. We address this point in some
more detail in sub-section 5.4.
Among the other independent variables, licensing is strongly signicant while wage price regula-
tion fails to be signicant. This suggests that while licensing does in fact contribute to corruption,
wage and price regulation (another channel through which corruption might arise) is not a signif-
icant source of corruption. While schooling has the predicted sign, it is signicant only in two of
the regressions (table 5). Political liberty and the number of newspapers do better suggesting that
the discipline of the ballot and the media does act as a check on corruption. We also aggregated
these three measures into a single linear composite, using principle component analysis, to con-
struct a single variable that proxies the detection probability. This variable is highly signicant
in all of the regressions.
21
Political rights, a vibrant media and an educated public go a long way
in mitigating corruption.
22
Among the country dummies, no clear pattern is discernible though
the dummy on East Asian countries has a positive and signicant coecient in a majority of the
regressions. Perhaps this simply reects the accusations of crony capitalism often made in the
context of East Asian countries. Surprisingly, we obtain a negative coecient on sub-Saharan
Africa.
In table 7 we add civil sector wages to our regressions with ICRG as the dependent variable.
23
In addition, we use the rst principle component to save on degrees of freedom.
24
For both
21
Using the principal component also corrects for the high collinearity between schooling, newspapers and
political rights. These results are available from the author upon request.
22
It might be plausible that developed countries have better institutions and schooling, political rights and the
number of newspapers simply act as an index of development. To address this, we added per capita GDP as an
independent regressor - it does not enter signicantly in a majority of the regression results.
23
Results with respect to other corruption measures are similar. However, these regressions suer from low
degrees of freedom.
24
The principal component is constructed as 0.37 dictatorship + 0.40 schooling + 0.43 newspapers
19
measures of bureaucrat wages (absolute wages and relative to manufacturing sector wages) we
nd a negative relationship between civil sector wages and corruption. This nding is in line with
the results of Van Rijckeghem and Weder (1995). In a majority of the regressions, protection
measures, continue to be strongly signicant as do subsidy rates and the proxy for detection
probability. The small sample size, especially when we use relative wages, prevents us from
interpreting these results with much condence.
The model(s) as a whole explains a signicant proportion with an adjusted R
2
in excess of
0.7. In each of the models we are able to reject the hypotheses that the coecients as a whole are
jointly insignicant at the 1% level of signicance. To sum up, protectionist policies and subsidies
to domestic rms, the relative wages in the civil sector, and proxies for the detection probability
help explain a signicant proportion of the variation in corruption across nations.
5.2 Two-Stage Least Squares
In a dynamic context, the trade restrictions may be endogenous with respect to corruption violat-
ing the orthogonality assumption. It is possible that the government chooses an optimal amount
of protection for the domestic industry through some welfare maximization exercise and recognizes
that increased protection leads to an increase in corruption. In such a case the government would
trade o the benets of protection against the costs of corruption and both protection and cor-
ruption would be determined simultaneously. Moreover, protection may be endogenous, if bribes
accrue to policy makers as well - the latter would then have an incentive to provide protection,
since this provides opportunities for rent-seeking activities. Such endogeneity problems imply that
the OLS estimates may be biased.
Therefore, we perform a two-stage least square estimation where we instrument the trade
restrictions by political ideology, and ideology*capital-labor ratio. The choice of the instruments
are guided by Dutt & Mitra (2002) who show that a partisan, ideology-based model predicts that
left-wing governments will adopt more protectionist trade policies in capital rich countries, but
20
adopt more pro-trade policies in labor rich economies. They also nd strong and robust support
for their theoretical prediction using cross-country data.
25
For the instrumental variables results
to be valid, we also require that ideology as well as the capital-labor ratio be uncorrelated with
corruption, since these are in eect, our instrumental variables. The correlation between these
variables range from -0.05 (for ideology*capital-labor ratio) to 0.16 (for ideology). While the
correlation seems low enough to warrant the use of ideology and ideology*capital-labor ratio as
instruments, we report the Basmann test for overidentifying restrictions as well. We are unable to
reject those restrictions, for all but one regression. Thus, we can be assured that our instruments
are valid and of good quality.
As table 8 shows,
26
across all measures of corruption and trade protection, the model pre-
dictions are supported. Trade restrictions and subsidies, exacerbate corruption as do licensing
regulations, weak political rights, and a weak media. As before the degree of schooling and wage-
price regulations do not seem to exercise any signicant inuence.
5.3 Panel Regression
We also test our model using cross-sectional time series data available for import duty covering
the period 1982-1997. We use both a xed-eects model with country specic xed eects and
a random-eects model with country-specic xed eects. The reason for doing so is that a xed
eects model with country-specic eects will not be able to identify the estimates for some of our
variables that do not vary within groups - for instance, the regional dummies, and our measures
of licensing, wage-price regulation and real wages.
27
Even though one would expect that trade
policies would take time to aect the level of corruption, our predictions are strongly borne out
here as well in the case of the import duty measure. This is reassuring especially since import
25
In their estimation, ideology measures the degree of leftist orientation of the government. They predict and
obtain a negative sign on ideology and a positive sign on ideology interacted with the capital-labor ratio, in the
regressions with trade policy as the dependent variable.
26
For the regressions where tari is the measure of trade restrictions, we have dropped the quota coverage ratio.
This is purely for ease of exposition and does not aect the results at all.
27
For these three variables we have data only for a single year.
21
duty is the only direct measure of protection for which we have data over time. We obtain the
following result for the xed-eects regression
Corruption = 3.8
(0.737)

+ 0.023
(0.009)

import duty + 0.02

(0.01)

subsidy + 0.14
(0.043)

(dictator)
+ 0.005
(0.007)
school 0.004
(0.002)

newspapers +fixed effects

R
2
= 0.16, N = 752, No. of countries = 47
The asterisks represent signicance levels as in the tables with the standard errors in parentheses.
A random-eects regression allows us to add the rest of the variables as well. We obtain
Corruption = 6.13
(1.66)

+ 0.021
(0.01)

import duty + 0.017

(0.008)

subsidy + 0.106
(0.05)

(dictator)
0.018

(0.009)
school 0.005
(0.002)

newspapers + 0.518
(0.362)

licensing
0.695

wage-price regulation 0.042

(0.028)

bureaucrat wage
R
2
= 0.5, N = 496, No. of countries = 31
Here we nd as well that trade restrictions and subsidies encourage corruption, that the proxies for
detection probabilities reduce corruption, and that higher bureaucrat wages discourage corruption
(though this is only marginally signicant).
5.4 The Importance of Licensing
In our model a necessary condition for the existence of corruption was the presence of some form
of licensing regulations. We have already seen that the licensing variable is highly signicant in
the majority of the regressions. Next, we investigate whether the model t is better in countries
that have a greater degree of regulatory controls in place. We do this by generating residuals from
our main regressions with import duty as the measure of trade restriction, and then regressing
the absolute values of these residuals on the licensing variable. As we can see from the regression
results below, for both the ICRG and TI measure of corruption the absolute residuals are lower
22
in countries with more stringent licensing requirements.
28
For the GCI measure we are unable to
conrm this result.
|ICRG Residual |=1.264
(0.272)

0.163
(0.087)

(licensing measure),
|TI Residual |= 2.026
(0.409)

0.307
(0.131)

(licensing measure)
Next, we generated predicted values of protection from our regressions without controls and
found their correlation with the actual values separately for the dictatorship sample (Gastil mea-
sure above 4) and the democracy sample (the rest). The correlation coecients shown in the
matrix below indicate that the t is superior for the sample of countries that can be classied as
democracies in the regressions for the ICRG and TI measures.
_

_
ICRG TI GCI
Licensing (high) 0.8 0.86 0.83
Licensing (low) 0.68 0.76 0.82
_

_
We obtain similar results for the other measures of trade restrictions as well. These results suggest
that the model ts countries better that have a greater degree of licensing regulations.
29
6 Conclusion
This paper sets forth a theoretical model that predicts that the incidence of corruption is increasing
in the level of protection oered to the domestic industry both in terms of protection from foreign
competition as well as through the subsidization of production. The model also predicts that
corruption will be decreasing in the remuneration of bureaucrats and in the detection probability.
Using a sample of 40-60 countries, we nd that these predictions are robustly supported across
a range of measures of trade protection and alternative measures of corruption. Countries with
high civil service wages, a more educated and aware general populace, and high levels of political
liberty, experience lesser corruption.
28
Similar results obtain when we use the square of the residuals instead of the absolute values.
29
Additionally, we interacted the licensing measure with the trade protection measures. This did not yield any
meaningful or signicant results.
23
These ndings oer several policy implications. First, the literature on determination of opti-
mal taris and subsidies, and infant industry protection arguments has not investigated empirically
that there may be a ip side to the arguments for transferring rents from foreigners to domestic
producers - namely, that it encourages corruption on part of the bureaucracy. Any assessment
of optimal levels of protection that ignores these eects would fail to maximize welfare since
corruption itself can be welfare reducing.
Second, corruption in our theoretical model arises from the discretionary power granted to the
bureaucracy. It is reinforced by and feeds upon the rents generated through further interventions
by the state in the domestic economy. This provides some validation to Gary Beckers prescription
of abolish the state and you shall abolish corruption.
30
While interventionist policies are desir-
able in many situations, this paper establishes that such policies generate additional detrimental
eects that manifest themselves in the form of corruption. The designing of policies should take
such eects into account.
Third, given our nding that corruption is decreasing in civil sector wages and the detection
probability, perhaps the best way to tackle corruption is through providing better remuneration
to bureaucrats and by improving methods of detection of corrupt acts. Improvement in the
institutional and legal system, greater political rights and education of the public create greater
accountability on the part of the government and its agents and as such act as a deterrent to
corrupt activities.
Finally, competition amongst domestic rms eliminates supernormal prots and in the ab-
sence of such prots corruption (as dened in the context of our paper) may be negligible and
perhaps impossible. Encouraging competition through pro-competitive and anti-trust practices
by eliminating such rents would benet consumers as well as mitigate corruption.
30
This is a positive, rather than a normative statement.
24
References
Ades, Alberto and Rafael Di Tella. 1999. Rents, Competition, and Corruption. American
Economic Review 89, 982-993.
Ades, Alberto and Di Tella, Rafael (1997). National champions and corruption: Some
unpleasant interventionist arithmetic, Economic Journal, 107, pp. 1023-1042.
Andvig, Jens C. and Moene, Karl O. (1990). How corruption may corrupt, Journal of
Economic Behavior and Organization, 13, pp. 63-76.
Bardhan, Pranab (1997) Corruption and Development: A review of issues. Journal of Eco-
nomic Literature, Vol XXXV (September).
Bhagwati, Jagdish and Hansen, Bent (1973). A theoretical analysis of smuggling, Quarterly
Journal of Economics, 87, pp. 172-187.
Bhagwati, Jagdish and Srinivasan, T.N. (1973). Smuggling and trade policy, Journal of
Public Economics, 2, pp. 377-389.
(1980). Revenue seeking: A generalization of the theory of taris, Journal of
Poilitical Economy, 88, pp. 1069-1087.
Bhagwati, Jagdish (1982). Directly unproductive, prot-seeking (DUP) activities, Journal
of Poilitical Economy, 90, pp. 988-1002.
Becker, Gary and Stigler, George J. (1974) Law enforcement, malfeasance and the compen-
sation of enforcers, Journal of Legal Studies, 5, pp. 1-19.
Bliss, Christopher and Di Tella, Rafael (1997). Does Competition Kill Corruption? Journal
of Political Economy, 105(5), pp. 1001-23.
Cadot, Olivier (1987). Corruption as a gamble, Journal of Public Economics, 33, pp. 223-
244.
Dutt, Pushan and Mitra, Devashish (2002). Political Ideology and Endogenous Trade Policy:
An Empirical Investigation:, mimeo.
Hadi, A.S. (1992). Identifying multiple outliers in multivariate data, Journal of the Royal
Statistical Society, Series B 56, pp. 393-396.
Harrigan, James. OECD Imports and Trade Barriers in 1983. Journal of International
Economics 35(1), August 1993, pp. 91-111.
Kemp, Murray (1976). Smuggling and optimal commercial policy, Journal of Public Eco-
nomics, 5, pp. 381-384.
Knack, S. and P. Keefer (1995): Institutions and Economic Performance: Cross-Country
Tests Using Alternative Institutional Measures, Economics and Politics, 7, pp. 207-227.
Knack, S. and P. Keefer (1997): Does Social Capital have an Economic
Krueger, Anne O. (1974). The political economy of the rent-seeking society, American
Economic Review, 64, pp. 291-303.
Klitgaard, Robert (1988). Controlling Corruption. Berkeley: University of California Press.
25
Klitgaard, Robert (1990). Gifts and bribes, in Richard Zeckhauser, ed. Strategy and Choice,
Cambridge, MA: MIT Press.
Leamer, Edward. The Structure and Eects of Tari and Non-tari Barriers in 1983, in R.
Jones and A. Krueger, eds., The Political Economy Economy of International Trade: Essays
In Honor of Robert E. Baldwin, pp. 224-260, Cambridge, MA: Basil Blackwell, 1990.
Lui, Francis T (1986) A dynamic model of corruption deterrence, Journal of Public Eco-
nomics, 3, pp. 1-22.
Mauro, Paolo (1995). Corruption and growth, Quarterly Journal of Economics, 60, pp.
681-712.
Mauro, Paolo (1998). Corruption and the composition of government expenditure, Journal
of Public Economics, 69, pp. 263-279.
Neumann, Peter (1994). Bose: Fast Alle Bestechen. Impulse, pp. 12-6.
Pritchett, Lant. Measuring outward orientation in LDCs: Can it be done? Journal of
Development Economics 49, 1996, pp. 307-335.
Ray, Alok. (1978). Smuggling, import objectives and optimum tax structure, Quarterly
Journal of Economics, 92, pp. 509-514.
Rodrik, Dani and Rodriguez, Francisco. Trade policy and economic growth: A skeptics
guide to the cross-national evidence. NBER Working Paper No. W7081, April 1999.
Rose-Ackerman, S (1975). The economics of corruption, Journal of Political Economy, 4,
pp. 187-203.
Shleifer, Andrei and Vishny, Robert W. (1993). Corruption, Quarterly Journal of Eco-
nomics, 108, pp. 599-617.
Treisman, Daniel. 2000. The Determinants of Corruption. Journal of Public Economics,
76, 3, June 2000, pp.399-457.
Van Rijckeghem and Weder, Beatrice (1997). Corruption and the rate of temptation: Do
low wages in the civil service cause corruption? IMF Working Paper WP/97/73.
Wei, Shang-Jin (1997). How taxing is corruption on international investors? Working Paper
6030, NBER.
26

27


Figure 2: Corruption and Tariffs
0
1
2
3
4
5
6
7
8
9
10
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Tariff
C
o
r
r
u
p
t
i
o
n

Figure 1: Corruption and Import Duty
0
1
2
3
4
5
6
7
8
9
10
0 10 20 30 40 50
Import Duty
C
o
r
r
u
p
t
i
o
n
Table 1.1: Summary Statistics
Variable Obs Mean Std. Dev. Min Max
ICRG 129 2.67 1.53 0 6
TI 86 5.08 2.41 0 8.6
GCI 42 3.52 3.44 0 10
tariff 93 0.16 0.12 0.00 0.48
quota coverage ratio 92 0.20 0.24 0.00 0.87
import duty (1980-90, avg.) 91 12.01 8.29 0.01 35.68
Sachs-Warner measure 110 0.71 0.46 0.00 1.00
subsidy (% of GDP) 126 0.17 0.52 0.00 5.26
licensing 162 3.36 0.92 1.00 5.00
wage-price controls 162 2.83 0.89 1.00 5.00
licensing 162 3.36 0.92 1.00 5.00
dictatorship (Gastil political rights) 166 4.28 2.17 1.00 7.00
schooling 138 59.72 25.43 4.33 99.74
newspapers per 1000 people 161 122.47 153.04 0.07 823.44
real wage rate in civil sector 86 10.82 13.82 0.13 62.18
government wages relative to manuf. sector 34 1.15 0.54 0.49 3.49
ideology (1=Right, 2=Center, 3=Left) 92 2.11 0.83 1 3
ideology*capital-labor ratio 92 19.21 7.30 7.34 33.75
log of capital-labor ratio 92 9.36 1.54 5.71 11.43
Table 1.2: Correlation Matrix for measures of corruption
ICRG TI GCOR
ICRG 1
TI 0.86 1
GCOR 0.8 0.89 1
Table 2: Index of Corruption: Transparency International (1-10)
Quartile1 Index Quartile2 Index Quartile3 Index Quartile4 Index
Denmark 10 Spain 6.1 Zimbabwe 4.2 Nicaragua 3
Finland 9.6 Botswana 6.1 Zambia 3.5 India 2.9
Sweden 9.5 Japan 5.8 Yugoslavia 3 Egypt 2.9
New Zeala 9.4 Estonia 5.7 Turkey 3.4 Bulgaria 2.9
Iceland 9.3 Costa Rica 5.6 Thailand 3 Ukraine 2.8
Cananda 9.2 Belgium 5.4 Slovak Rep. 3.9 Bolivia 2.8
Singapore 9.1 Namibia 5.3 Senegal 3.3 Latvia 2.7
Norway 9 Malaysia 5.3 Romania 3 Pakistan 2.7
Netherland 9 Taiwan 5.3 Philippines 3.3 Uganda 2.6
Switzerland 8.9 South Africa 5.2 Morocco 3.7 Vietnam 2.5
Luxembour 8.7 Hungary 5 Mexico 3.3 Kenya 2.5
Australia 8.7 Tunisia 5 Malawi 4.1 Russia 2.4
United King 8.7 Mauritius 5 Jamaica 3.8 Ecuador 2.3
Ireland 8.2 Greece 4.9 Guatemala 3.1 Venezuela 2.3
Germany 7.9 Czech Rep. 4.8 Ghana 3.3 Columbia 2.2
Hong Kong 7.8 Jordan 4.7 El Salvador 3.6 Indonesia 2
Austria 7.5 Poland 4.6 Cote d'Ivoire 3.1 Tanzania 1.9
United Stat 7.5 Italy 4.6 China 3.5 Nigeria 1.9
Israel 7.1 Peru 4.5 Brazil 4 Honduras 1.7
Chile 6.8 Uruguay 4.3 Belarus 3.9 Paraguay 1.5
France 6.7 Korea, South 4.2 Argentina 3 Cameroon 1.4
Portugal 6.5
Table 3: Regression of Corruption on Trade Protection
Transparency International
Tariif and Quota Import Duty Sachs-Warner
tariff 6.08*** import duty 0.631*** open 3.3***
(2.957) (0.124) (0.461)
quota 0.927
(0.815)
constant 3.843*** constant 4.041*** constant 3.04***
(0.572) (0.345) (0.391)
No. of observations 63 64 73
R
2
0.26 0.26 0.44
F-statistic 11*** 21.3*** 56.78***
International Country Risk Guide
Tariif and Quota Import Duty Sachs-Warner
tariff 3.613*** import duty 0.466*** open 2.076***
(1.698) (0.068) (0.274)
quota 0.185
(0.571)
constant 1.97*** constant 1.86*** constant 1.117***
(0.325) (0.189) (0.229)
No. of observations 84 103 100
R
2
0.17 0.26 0.37
F-statistic 8.26*** 47.52*** 57.42***
German Corruption Index
Tariif and Quota Import Duty Sachs-Warner
tariff 8.92*** import duty 0.693*** open 3.391***
(3.46) (0.149) (0.922)
quota -1.776
(2.098)
constant 2.561*** constant 2.813*** constant 1.91***
(0.713) (0.473) (0.727)
No. of observations 38 42 42
R
2
0.27 0.24 0.25
F-statistic 3.45*** 21.72*** 13.53***
***- significant at 5%; ** - significant at 10%, * - significant at 15%. Standard errors in parantheses
Table 4: Corruption Indices on Tariff and Quota
ICRG TI GCOR ICRG TI
tariff 1.053** 2.117** 3.257** 1.043** 2.098***
(0.645) (1.074) (1.926) (0.58) (0.903)
quota -0.858 -1.112 -1.837 -0.772 -1.184**
(0.635) (0.796) (1.883) (0.575) (0.652)
subsidy 0.308*** 0.854*** 1.166*** 0.494*** 0.611***
(0.065) (0.255) (0.437) (0.108) (0.225)
licensing 0.503*** 1.024*** 1.49*** 0.51*** 0.932***
(0.196) (0.298) (0.46) (0.221) (0.26)
wage-price regulation -0.149 0.24 -0.075 -0.034 0.208
(0.236) (0.347) (0.617) (0.245) (0.287)
political rights 0.24*** 0.255*** 0.264 0.35*** 0.206***
(0.113) (0.123) (0.304) (0.098) (0.103)
schooling -0.007 -0.017 -0.025 0.005 -0.013
(0.009) (0.013) (0.031) (0.01) (0.011)
newspapers -0.005*** -0.007*** -0.01*** -0.009*** -0.006***
(0.001) (0.002) (0.003) (0.002) (0.002)
sub-saharan africa -1.308*** -0.884 -1.591 -1.09** -0.632
(0.606) (0.673) (1.49) (0.574) (0.555)
east asia 0.475 0.596 3.314* 1.103** 0.527
(0.724) (0.587) (2.03) (0.589) (0.476)
oil 0.573 0.222 -2.939 0.997** 0.325
(0.462) (0.748) (2.23) (0.573) (0.616)
constant 1.866** 2.37 1.643
(0.965) (1.737) (4.58)
No. of observations 61 53 38 61 53
R
2
0.67 0.79 0.77 0.28 0.18
F-statistic 22.58*** 14.63*** 38.2***
Wald-statistic 87.5*** 72.4***
Standard errors in parantheses; *** - significant at 5% level, ** - significant at 10% level * - significant at 15% level
Ordinary Least Squares Ordered Probit
Table 5: Corruption Indices on Import Duty
ICRG TI GCOR ICRG TI
import duty 0.166*** 0.156* 0.255*** 0.201*** 0.126**
(0.052) (0.107) (0.121) (0.062) (0.068)
subsidy 0.281*** 0.925*** 1.854*** 0.447*** 0.644***
(0.082) (0.099) (0.516) (0.085) (0.12)
licensing 0.261 0.991*** 1.418*** 0.195 0.867***
(0.236) (0.256) (0.455) (0.227) (0.249)
wage-price regulation -0.054 0.22 0.307 -0.079 0.199
(0.192) (0.234) (0.515) (0.18) (0.216)
political rights 0.11* 0.116 0.098 0.114* 0.074
(0.074) (0.145) (0.399) (0.071) (0.114)
schooling -0.002 -0.018* -0.04** 0.001 -0.012
(0.009) (0.012) (0.023) (0.008) (0.009)
newspapers -0.006*** -0.008*** -0.008*** -0.008*** -0.006***
(0.001) (0.001) (0.003) (0.001) (0.001)
sub-saharan africa -0.87** -1.061* -2.499** -0.952*** -0.763
(0.471) (0.642) (1.431) (0.446) (0.568)
east asia 0.645 0.764 3.135** 0.86 0.677**
(0.621) (0.552) (1.756) (0.618) (0.411)
oil 0.659*** 0.106 -3.534 0.619** 0.247
(0.322) (0.961) (2.722) (0.343) (0.788)
constant 2.072*** 3.121*** 2.001
(0.989) (1.406) (3.484)
No. of observations 80 61 42 80 61
R
2
0.6 0.77 0.73 0.21 0.17
F-statistic 37.63*** 29.87*** 23.85***
Wald-statistic 110.26*** 95.41***
Standard errors in parantheses; *** - significant at 5% level, ** - significant at 10% level * - significant at 15% level
Ordinary Least Squares Ordered Probit
Table 6: Corruption Indices on Sachs-Warner Measure
ICRG TI GCOR ICRG TI
sachs-warner dummy 0.986*** 1.525*** 1.821* 1.37*** 1.409***
(0.301) (0.544) (1.166) (0.389) (0.471)
subsidy 0.357*** 0.903*** 1.913** 0.64*** 0.73***
(0.075) (0.078) (1.095) (0.094) (0.122)
licensing 0.413*** 0.962*** 1.519*** 0.404*** 0.974***
(0.169) (0.254) (0.517) (0.199) (0.276)
wage-price regulation -0.053 0.4** 0.534 -0.062 0.385*
(0.21) (0.235) (0.539) (0.235) (0.244)
political rights 0.095 0.16 0.298 0.103 0.129
(0.09) (0.131) (0.261) (0.087) (0.126)
schooling -0.006 -0.009 -0.018 -0.001 -0.01
(0.008) (0.011) (0.027) (0.008) (0.01)
newspapers -0.004*** -0.006*** -0.006*** -0.008*** -0.005***
(0.001) (0.001) (0.003) (0.002) (0.001)
sub-saharan africa -0.918** -0.852** -2.571*** -1.063*** -0.754**
(0.481) (0.437) (1.153) (0.516) (0.429)
east asia 1.241*** 1.484*** 5.075*** 2.078*** 1.431***
(0.564) (0.483) (1.76) (0.648) (0.479)
oil 0.88*** 0.629 -4.167*** 1.131*** 0.654
(0.299) (0.783) (1.841) (0.367) (0.673)
constant 1.221 0.813 -1.899
(1.04) (1.271) (3.678)
No. of observations 72 59 42 72 59
R
2
0.71 0.82 0.76 0.3 0.2
F-statistic 37.53*** 36.06*** 9.87*** 92.15*** 107.2***
Wald-statistic
Standard errors in parantheses; *** - significant at 5% level, ** - significant at 10% level * - significant at 15% level
Ordinary Least Squares Ordered Probit
Table 7: Wages and the ICRG Corruption Index
tariff import duty sachs-warner tariff import duty sachs-warner
trade protection 0.416 0.169** 0.975*** 0.816*** 0.023 0.043
(0.57) (0.093) (0.366) (0.315) (0.094) (0.559)
subsidy 0.539* 0.601*** 0.761*** 0.471 -0.393 1.037
(0.353) (0.281) (0.279) (0.803) (1.274) (1.245)
licensing 0.316 0.088 0.419** 0.22 0.289 0.538***
(0.238) (0.311) (0.217) (0.198) (0.244) (0.179)
wage-price regulation -0.323 -0.191 -0.229 -0.184 -0.104 -0.272
(0.275) (0.215) (0.251) (0.204) (0.261) (0.202)
principal component -0.936*** -0.725*** -0.847*** -1.363*** -1.289*** -1.459***
(0.172) (0.237) (0.206) (0.171) (0.207) (0.306)
sub-saharan africa -0.669 -0.75 -0.794** -0.535** -0.534 -1.315***
(0.494) (0.466) (0.434) (0.294) (0.347) (0.295)
east asia 0.618 0.504 1.158*** 3.017*** 3.684*** 2.9**
(0.524) (0.553) (0.425) (1.166) (1.318) (1.537)
oil 0.758 0.714*** 1.053*** -0.044 -0.073 -0.126
(0.554) (0.352) (0.37) (0.321) (0.327) (0.332)
wages -0.03*** -0.028*** -0.008 -1.441*** -1.573*** -1.107
(0.011) (0.012) (0.01) (0.532) (0.568) (0.808)
constant 2.621*** 2.757*** 0.999* 3.751*** 3.589*** 2.723**
(0.602) (0.964) (0.615) (0.776) (0.985) (1.359)
No. of observations 44 58 51 27 31 30
R
2
0.8 0.6 0.78 0.9 0.86 0.86
F-statistic 27.91*** 16.18*** 50.31*** 82.75*** 50.4*** 138.21***
Wald-statistic
Standard errors in parantheses; *** - significant at 5% level, ** - significant at 10% level * - significant at 15% level
Real Civil Sector Wage Relative Civil Sector Wage
Table 8: Instrumental Variables Regression
ICRG ICRG ICRG TI TI TI GCOR GCOR GCOR
tariff import duty sachs-warner tariff import duty sachs-warner tariff import duty sachs-warner
trade protection 6.942* 0.53** 3.215*** 8.273** 0.356* 2.746*** 7.894 0.31* 4.304**
(4.477) (0.292) (1.254) (4.598) (0.226) (1.257) (8.516) (0.211) (2.52)
subsidy 0.284*** 0.296 0.396*** 0.908*** 0.95*** 0.966*** 1.475*** 1.966*** 2.486***
(0.114) (0.238) (0.185) (0.112) (0.124) (0.121) (0.719) (0.523) (0.865)
licensing 0.501** 0.368 0.364 1.022*** 0.946*** 0.883*** 1.378*** 1.414*** 1.185***
(0.251) (0.276) (0.298) (0.351) (0.341) (0.295) (0.499) (0.378) (0.542)
wage-price regulation -0.117 -0.028 0.049 0.498*** 0.269 0.351 0.334 0.98* 1.169*
(0.344) (0.268) (0.399) (0.281) (0.297) (0.267) (0.661) (0.592) (0.778)
political rights 0.262*** 0.058 -0.1 0.306*** 0.26** 0.082 0.281 0.431* 0.068
(0.113) (0.105) (0.173) (0.153) (0.153) (0.197) (0.264) (0.268) (0.345)
schooling 0.008 0.011 0.009 -0.002 -0.005 0.005 -0.011 -0.02 -0.007
(0.011) (0.013) (0.014) (0.014) (0.016) (0.014) (0.029) (0.021) (0.028)
newspapers -0.003*** -0.005*** -0.002 -0.006*** -0.008*** -0.006*** -0.008*** -0.007*** -0.004
(0.001) (0.001) (0.002) (0.001) (0.002) (0.001) (0.003) (0.002) (0.004)
sub-saharan africa -0.939 -0.752 -0.652 -1.42*** -1.86*** -1.174*** -2.232 -2.897*** -2.595***
(0.932) (0.564) (0.633) (0.448) (0.612) (0.369) (1.569) (1.273) (1.384)
east asia 0.38 0.713 3.178 0.839 1.603*** 3.55*** 3.641*** 3.759*** 7.389***
(0.89) (0.766) (0.985) (0.668) (0.668) (0.936) (1.26) (1.054) (2.337)
oil 1.249 1.089** 0.199 3.209*** 2.992*** 2.446***
(0.961) (0.542) (0.741) (0.61) (0.574) (0.858)
constant -0.912 0.206 -1.118 -0.643 1.755*** -0.229 -1.1 -2.047 -4.235
(1.659) (1.459) (2.169) (1.686) (1.821) (1.691) (3.965) (3.304) (3.998)
No. of observations 49 60 56 44 50 49 34 38 38
R
2
0.64 0.4 0.44 0.84 0.76 0.81 0.79 0.77 0.75
F-statistic 15.57*** 7.69*** 18.52*** 102.98*** 43.41*** 161.77*** 24.63*** 50.65*** 41.01***
Basmann test 2.823 2.793 0.389 0.24 3.55 0.269 5.3** 3.91 1.402
Standard errors in parantheses; *** - significant at 5% level, ** - significant at 10% level * - significant at 15% level