Why are some countries more corrupt than others? Many have attempted
to answer this question in recent years, and the subject of political
corruption has rightly captured a rapidly growing share of public and
academic attention, in part because of the recognition of its intrinsic
importance and also because of its close relationship with economic
growth, investment, the rule of law, political accountability, political
institutions, and democratic governance more generally. In fact, few
issues are more important than corruption when it comes to questions of
economic development and effective governance, yet still relatively little
is known about the causes (and cures) of corruption. We focus in this
article on political corruption, which we define as an act by a public
official (or with the acquiescence of a public official) that violates
legal or social norms for private or particularistic gain.1
Several kinds of factors have been used to explain rates of
corruption. We focus on the role of public policies. What kinds of
public policies are likely to be associated with higher or lower rates
of corruption? Which specific policies matter most, and how? While
several public policies may play a role, we focus on the possible
effects of neoliberal economic policies on political corruption. Do
neoliberal policies deter political corruption?2
This article employs a statistical analysis of 181 countries in the
mid- to late 1990s to test the general hypothesis that market-oriented,
neoliberal economic policies are associated with lower levels of
political corruption. Our analysis of political corruption
broadly—but not unanimously—confirms the general neoliberal
argument. More specifically, we find strong evidence of a relationship
between “catch-all” measures of neoliberalism, trade and
investment policies, and regulation on the one hand, and political
corruption on the other. But we find no evidence of a link between the
size of the public sector and political corruption. While the causal
pathways linking neoliberal economic policies and political corruption
are little known and underdeveloped in the literature, this evidence is
strong enough to merit more empirical and theoretical treatment.
In the following sections of the article, we address the broad
neoliberal argument and its three specific components that we test,
present our method of analysis, summarize our specific statistical
specifications and empirical findings, and offer our conclusions.
Hypotheses
There are myriad public policies that might influence levels of
political corruption around the world, but there is little theoretical
work that delineates the ways in which this might occur. Our task in
this article is to identify a series of potentially important policies,
assess their causal logic, and test their explanatory power
empirically. We offer one general hypothesis and three more specific
ones:
H1: Market-oriented, neoliberal economic policies help reduce
political corruption.
H1a: Openness to foreign trade and investment promotes lower levels
of political corruption.
H1b: Lower and more market-friendly regulations inhibit political
corruption.
H1c: Smaller public sectors lead to less corruption than larger ones.
This section presents the causal logic behind each of these hypotheses.
Neoliberalism and Political Corruption
According to the broad neoliberal hypothesis H1, less government
involvement in the economy and in civil society should result in less
corruption. Certainly, there would appear to be less opportunity for
corruption, since political corruption can only occur in a policy area
where government has substantive powers. Fully market-based systems,
almost by definition, should be less prone to governmental graft, all
else being equal. By the same token, widening the scope of the free
market and enhancing its competitiveness should dampen at least some of
the demand for political corruption, since corruption itself is often a
response to blocked market transactions. Despite the intuitive nature
of these arguments, there is little empirical work that tests this
hypothesis in a systematic fashion. Furthermore, there is little
analytical or empirical work that draws distinctions between different
elements of government economic policy or assesses their relative
causal impacts.
For neoliberals, one critical, perhaps fundamental (preceding other
important policies and outcomes) element of sound economic management
is price stability. Without stable and relatively low rates of
inflation, economic actors face tremendous uncertainty over relative
prices, asset values and real wages tend to erode, and prosperity
suffers. In particular, high rates of inflation shorten actors'
time horizons and increase their discount rate, potentially increasing
the expected utility of corruption for politicians. While the private
benefits of corruption will often be garnered in the short term, its
public costs (as seen in slower rates of growth, for example) are borne
out over a longer period of time. By the time the public costs of
corruption become evident, its culprits will likely be long out of
office in a context of high instability.
It is also possible that the policies that generate inflation
themselves promote corruption. Loose, expansionary fiscal policies, for
example, offer greater opportunities for graft by giving government
bureaucrats access to greater volumes of government resources, the
distribution of which may be influenced by bribes, kickbacks, and the
like.3
We address various fiscal factors
in greater detail below, in the discussion of the public
sector.
Trade and Investment Policies
There are good reasons to suspect that trade and investment policies
might be tied to political corruption.4
First, barriers to
international market transactions can create material and bureaucratic
incentives for corrupt behavior. Countries that erect high barriers to
trade (tariff and nontariff) and investment fight the natural
propensity of individuals to truck and barter. By raising the price of
goods above their market price, trade barriers may induce
businesspersons to bribe their way to exemptions or special treatment
(including high levels of effective protection). Similarly,
restrictions on foreign investment may encourage would-be investors and
their domestic allies to seek extra-legal exemptions to prevailing
limits on foreign participation. By the same token, trade and
investment restrictions are complicated to administer, involving an
intricate set of rules and procedures that tend to inflate the size of
the government bureaucracy and that must be delegated to bureaucrats to
enforce. The combination of complexity, particularism, and delegation
can encourage corrupt practices on the part of government officials.
Rule-making powers of this nature, when coupled with the material
resources of would-be foreign traders and investors, can be a recipe
for malfeasance.
A second argument stems from the constituencies formed over time by
closed-market policies. While an open-economy regime is likely to
foster an outward-looking business class, a closed system is likely to
foster an inward-looking, inefficient private sector whose prosperity
rests on close ties to the state and rents provided by protectionist
policies.5
Groups dependent on protection
from foreign producers and investors are unlikely to criticize corrupt
practices by governmental actors upon whom their livelihood depends, or
may even be beneficiaries of those practices themselves. Thus a
protectionist regime can buy political silence or even complicity on
the part of middle-class groups who might otherwise participate in
anticorruption campaigns. Closed-economy policies may hinder the
formation of a constituency for political accountability amongst the
business community and other elites.
Third, an outward trade and investment orientation should enhance
links to the international economy and foreign business groups, and
from thence to “Western” notions of acceptable business and
political practices, for the sake of business expediency if nothing
else. Even if no sort of “cultural diffusion” takes place
between foreign and domestic groups, internationally oriented groups
may find it in their material interest to adopt business and political
practices consistent with those of their primary foreign trade and
investment partners. An inward orientation, by contrast, may give
business and government alike a more domestically oriented perspective,
one relatively insulated from foreign influences. This is not likely a
propitious political environment for fighting domestic corruption.
Finally, to the extent that foreign investors prefer transparent
bureaucracies and strong rule of law, the increase in capital mobility
seen in recent decades would be expected to “discipline”
governments in open economies to combat corruption in an effort to
capture larger shares of increasingly mobile foreign investment
(and— to the extent that foreign trade and investment are linked
via intrafirm trade, vertical integration, and other
mechanisms—trade as well). If firms prefer clean government, and
if they are able to move from one political jurisdiction to another at
a relatively low cost, corrupt governments may have to clean house if
they wish to attract foreign capital.6
Closed economies, on the other hand, (1) face less
pressure from capital mobility (precisely because their capital markets
are more tightly restricted), and (2) are less likely to place a high
value on investors' interest in clean government. Greater openness
should impose greater discipline on countries to reduce corruption.
Regulatory Policy
Governments frequently use regulatory policy to influence the
functioning of markets. Among myriad motivations for such policies,
governments may want to correct market failures, to compensate losers
from other realms of economic policy, or to make indirect transfer
payments to politically influential social actors. The neoliberal
hypothesis argues broadly for minimal regulatory policies that foster
competitive markets and market-clearing prices. This should not be
confused with the absence of regulation. Indeed, recent work has
underlined the vital role governments play in establishing and
enforcing market rules and thereby enhancing trust.7
For example,
several East Asian countries that were long the darlings of the popular
press, international institutions, and envious developing-country
governments around the world learned the hard way the need for
effective financial regulations in their financial crises of
1997–98.
8 The neoliberal hypothesis simply stipulates that
the primary aim of regulatory policy should be to promote effective,
efficient markets rather than to achieve other policy goals (for
example, redistribution).
The logic of market-friendly regulatory policies follows closely on
the logic of the market-friendly foreign trade and investment policies
just reviewed. Unencumbered domestic markets, like openness to
international markets, should enhance clean government. Regulatory
policies vest policymaking power in government bureaucrats to direct
market relations, thus offering incentives to businesspersons to
cultivate special relations with government officials, who in turn may
depend politically on support from key interest groups and actors.9
The classic Stigler-Peltzman theory of
regulation, in which utility-maximizing interest groups offer political
support to regulators in exchange for favorable treatment, is apropos
here. See Stigler 1971; and Peltzman 1976.
The complex nature of such
regulations, and the need for rule interpretation, makes
decision-making transparency difficult to achieve. Moreover, the
concentrated nature of the private good, coupled with the diffuse
nature of the public good, present difficult collective action and
free-rider problems that translate into political incentives that augur
poorly for the protection of the public interest. As with trade and
investment policies, regulatory policies can create powerful political
constituencies for or against corruption. Specific regulatory policies
that might be expected to influence levels of corruption include, among
others, restrictions on financial markets, wage and price controls, and
business licensing requirements.
Public Sector Size
Much contemporary academic work suggests that a large public sector
is more conducive to corruption. Ceteris paribus, the larger a role the
government plays in the market—as producer and/or
consumer—the greater its capacity to engage in corrupt activity.
By this logic, adding functions to government cannot reduce its level
of corruption; by the same token, whatever functions are not entrusted
to government cannot be as easily abused by government. The
semimythical, night-watchman state has few levers by which to pry money
from its citizens. As a rule, proffers Scott, “the larger is the
relative size and scope of the public sector, the greater will be the
proportion of [corrupt] acts.”10
This view implies a positive, perhaps linear relationship between the
size of government and political corruption; improvements are achieved
by cutting the total size of government. But one might also speculate
that leaner government is cleaner government, inch by inch and pound
for pound. Cuts in the size of the public sector may not only lower
corruption by reducing the size of a corrupt government, but also by
making the resulting entity itself less corrupt. When government is
actively engaged in producing economic goods, it must adjudicate claims
among different interests within the public sector. What is a fair
price for goods and services produced by state-run enterprises? (Should
the same price be charged to state and nonstate consumers?) What is an
optimal wage for state employees? Wherever market mechanisms are weak
or nonexistent, the prices set for goods, services, and labor will be,
by definition, “political,” and are likely to reflect the
extant balance of political forces. Losers in these decisions are apt
to consider such decisions corrupt.
Yet there are also reasons to doubt the neoliberal view on the size
of government. Even the most ardent neoliberal likely acknowledges the
need for certain basic functions of government. The literature on the
rule of law, for example, points to the importance of law and contract
enforcement and effective judicial systems for effectively functioning
markets.11
As in the case of regulation,
neoliberals prescribe a minimal but effective role for the state. Although
it may be costly to construct an effective state, especially in a context
of lawlessness such as that exemplified in fractious countries ruled by
warlords, guerrillas, or illegal drug and arms traffickers, such expenses
tend to be nonrecurring over the long term and relatively small in comparison
to entitlement programs and other instruments of the modern state.
Additionally, larger states may be more effective (and therefore less
corrupt). The northern European countries, which typically rank at the
top of most measures of corruption control, come to mind. Citizens may
be more tolerant (or even supportive) of big government where such
government is clean. In that context, governments may become larger in
part precisely because they have less corruption. Relatedly, a state
with greater capacity to intervene in society may also be more able to
withstand pressures emanating from civil society.12
A state with few funds, by contrast, may have
little leverage in the state–civil society relationship and in
combating the forces of corruption.
13 Similarly,
government retrenchment may result in the loss of valued personnel, a
loss that may undermine state capacity and promote corruption. Grindle
notes that public service cutbacks had “important and often
negative consequences for public sector capacity,” since those
with the best opportunities for private sector employment were the most
likely to defect.
14Analysis
To date, most empirical work on these questions is of the case-study
variety. Much of this research centers on neoliberal reforms of the
past two decades. Manzetti and Blake find that recent market reforms
have changed, but not eliminated, the politics of corruption in Latin
America.15
The partial character of neoliberal reform
in most African countries may have enhanced rent-seeking activity on
the part of privileged actors.
16 Market reforms in communist and postcommunist
countries have elicited similar comments from observers of these
cases.
17 By
contrast, other studies demonstrate support for the notion that
neoliberal reforms ultimately reduce political corruption.
18 Moreover, work focused on socialist regimes
almost invariably supports the neoliberal argument.
19 Cross-national studies with larger samples are less common and have
obtained mixed results on these questions. Tanzi and Davoodi find
public investment (as an indicator of government intervention) to be
slightly correlated with corruption, but short of statistical
significance at usual thresholds.20
Looking at
industrial policies, Ades and Di Tella also find a weak connection
between government behavior and aggregate system-level corruption.
21 Sandholtz and
Koetzle find a reasonably strong connection between economic freedom
and clean government.
22 Treisman,
however, with a somewhat larger data set, is agnostic on this
point.
23 This study offers the most comprehensive and thorough test of the
neoliberal hypothesis on political corruption. The indicator of
political corruption that we use was compiled from a wide variety of
international corruption polls by Kaufmann, Kraay, and Zoido-Lobaton,
and later by Kaufmann, Kraay, and Mastruzzi (KKM).24
The advantages of the KKM measure over others are
its enormous breadth of coverage (184 cases, of which we include 181 in
our analysis) and the variety of sources employed in compiling it,
rendering it less susceptible to poll- or question-specific
idiosyncrasies.
25 We excluded three cases
from the KKM data set because of excessive missingness in the
independent variables and/or lack of sovereignty: Puerto Rico,
Samoa, and West Bank and Gaza. Note also that we invert the scale of
the KKM “control of corruption” variable so that higher
scores indicate more corruption.
Our analysis centers on the
mid- to late 1990s and encompasses a wide range of control variables,
some of which we do not report in this study, in addition to several
measures of neoliberal policy.
The KKM index is a “poll of polls,” conducted every two
years, that asks respondents to rate the general level of corruption
among public officials, the effectiveness of anticorruption
initiatives, the frequency of additional payments necessary to
“get things done,” and corruption as an obstacle to foreign
investment and domestic business enterprise. Naturally, questions and
coverage varied from poll to poll.26
Regrettably, it was not possible to obtain disaggregated
survey results (question by question). Most of the data was obtained by
KKM from consulting agencies, which have proprietary interests in
keeping the results in aggregate form.
Respondents were
roughly evenly divided between two groups: (1) businesspeople and
residents of a country, and (2) experts (who were asked to rank
countries on various dimensions). The authors construct the composite
index from these individual surveys using an unobserved components
model, which allows them to include reliable estimates for countries
excluded from other data sets, such as Transparency
International's Corruption Perceptions Index (CPI).
27 See Appendixes A and B at
〈http://www.bu.edu/sthacker/data.html〉 for a
complete list of countries analyzed, their corruption scores,
descriptions of the dependent and independent variables used in the
analysis, and data sources. Tests conducted on Transparency
International's CPI yielded results substantively similar to those
presented here. Because of the KKM index's greater coverage and
the availability of estimates of the uncertainty of its measures for a
large number of cases, we limit our presentation here to the KKM
data.
Much has been written about the validity and utility of
cross-national corruption polls.28
What deserves emphasis, in the final
analysis, is the tremendous advance that such indices offer to current
understandings of the subject. Even studies based on the analysis of
single countries or regions cannot avoid a comparative evaluation of
the dependent variable. (To say that corruption is rife in Country A is
to say, implicitly, that Country A is more corrupt than other
countries.) Analysts have been implicitly judging the extent of
corruption across national borders for some time. Perhaps the safest
interpretation of cross-national indices of corruption is thus as a
codification and systematization of this standard wisdom, allowing one
to plumb the implications of this evidence more systematically. (Many
of the respondents in these corruption surveys are the same
country-experts responsible for the rich set of case studies on these
countries.) In addition, we argue that cross-national indicators should
not be intended to replace case-study research, but rather to
complement it. The proper employment of such indicators is to measure
what can reasonably be measured, and to systematize those judgments
that are broadly comparative in nature.
Because cross-national corruption polls are a relatively recent
phenomenon (earlier polls are less reliable, offer much smaller
samples, and are not generally comparable over time), we conduct a
cross-sectional analysis of corruption data from 1997–98. The
methodological challenge of first order is the issue of specification.
A widely accepted, benchmark equation specifying the causes of
corruption does not yet exist. In order to assess accurately the
hypotheses under consideration in this study, we must take into account
a host of control variables that may influence the dependent variable.
In previous work, we identified six structural-level variables of
causal significance to political corruption.29
These include Protestantism
(percentage of Protestants in a country), English legal origin,
socialist legal origin (current or former), energy imports (net), per
capita gross domestic product (GDP) (natural log), and democracy (years
democratic in the twentieth century).
30 We do not include other political-system variables such as
parliamentarism, unitarism, and electoral system (PR/non-PR), as
these are of questionable significance in the broader sample explored
here, a sample that includes democratic and nondemocratic regimes. The
role of these political-system variables is uncertain where no
semblance of democracy exists. (In regression tests conducted on a
smaller sample composed exclusively of democratic cases, the inclusion
of these political-system variables does not alter the substantive
results presented here.) For a study of the role of such factors in
political corruption in more democratic contexts, see Gerring and Thacker 2004.
We also include
in every estimation a series of unit-level and regional controls that
include population size (natural log), land area, latitude (absolute
value, scaled to 0–1), and dummy variables for Africa, East Asia,
South Asia, Latin America, the Middle East, and the Organization for
Economic Cooperation and Development (OECD).
31 We also tested variables on ethnic, linguistic and
religious fractionalization (and their squares), with no significant
results or changes in the substantive results (not reported).
The inclusion of this wide range of controls creates more difficult
tests for the neoliberalism variables. The core equation displayed in
the first column of
Table 1 serves as the
point of departure for all subsequent specifications.
There is likely some degree of endogeneity between per capita GDP and
corruption. Wealthier countries are likelier to be less corrupt, and
less corrupt nations are also more apt to perform well economically. In
order to minimize this problem, we measure the per capita income
variable as a lagged, twenty-five-year average (1970–95). Even
so, these coefficients may overstate the true causal impact of economic
growth on political corruption. Because we are interested in this
variable only as a control, we may regard the inclusion of per capita
GDP as establishing a most-difficult scenario for our variables of
interest. In addition, the inclusion of the GDP per capita variable
helps establish a more direct causal link between neoliberal policies
and corruption. Neoliberal policies should, according to their
proponents, promote higher rates of economic growth and therefore
higher levels of per capita income over the long term. Those higher
rates of growth, in turn, should also contribute to lower levels of
corruption. By controlling for per capita income, we can better capture
the direct effects of neoliberal policies on political corruption. This
also implies that the regression coefficients presented in Tables 1 and 2 may underestimate
the causal import of the neoliberal policy variables because they do
not capture this putative growth-inducing effect.
The socialism dummy raises interesting specification issues. The
statistical significance of this variable (which enhances corruption,
as expected) may be regarded as fodder for the neoliberal hypothesis.
Yet few would regard this as a contested hypothesis. Socialism is an
extreme case of market intervention, one not seriously supported in
many mainstream academic or policy circles at the present time. This
variable also overlaps conceptually with other variables, which may
raise doubts about its relevance in the present context. Again, we
choose to include it in all specifications to establish a higher
empirical threshold for the causal hypotheses and variables of
interest. If neoliberalism matters while controlling for socialism,
then we have stronger evidence for the free-market argument. If, on the
other hand, the causal effects of neoliberalism disappeared when
socialist cases were controlled for, then we would have an empirical
finding of rather little theoretical interest in the present context.
Thus we err on the side of Type I errors in retaining a wide range of
control variables.32
The exclusion of
these control variables does not alter our substantive
conclusions.
Results and Interpretations
We have striven to operationalize each hypothesis in a variety of
ways, subject only to limitations of data availability.33
We have tested several other possible
indicators of neoliberal policy performance, such as various indicators
of political risk (which have a strongly neoliberal orientation), and
several separate indicators used in some of the aggregate indices
presented below. In each case, results were consistent with those
reported here and the inclusion of additional variables did not disturb
the other results.
Given the likely inefficiency and bias
introduced by missing data, we employ the multiple imputation technique
developed by King and his associates to estimate values for missing
data.
34 The alternative, casewise
deletion of cases with missing data in one or more independent
variables, generates biased results (when compared to those obtained
through multiple imputation) under almost all circumstances that might
be anticipated. See Honaker et al. 2001; and
King et al. 2001. Regression using casewise
deletion generates substantively similar results.
We then use
CLARIFY software (a set of downloadable commands available for use in
Stata) to run regressions on the imputed data that take into account
not only the additional imputed information but also the inherent
uncertainty surrounding the estimates of the missing data.
35 Because of the variable quantity and quality of the polls used to
construct the KKM index (not all polls are available for all countries,
nor do they all agree with each other), we employ weighted least
squares (WLS) estimation, using the inverse of the standard error
provided by KKM for each observation as the analytic weight to give
greater weight to those cases with more reliable data on the dependent
variable. Given the underdeveloped and somewhat ambiguous nature of
theory in this area, as well as the potentially complex relationship
between our dependent and independent variables, we must interpret
results cautiously. Our intent in this study is not to provide
definitive proof of a causal relationship between neoliberalism and
political corruption. Rather, we aim to provide an initial round of
hypothesis testing, to highlight a series of potentially compelling
relationships that may spur further theoretical development and
empirical testing to explore causal pathways more precisely than we are
able to do here. In this regard, the overall results of the estimations
presented in Tables 1 and 2 are encouraging. As measured by the adjusted
R2, the statistical models presented in this study account
for more than 80 percent of the variation in the dependent variable.
The estimations are also all highly significant, as measured by F
tests, all of which are significant at better than the .0001 level.
With respect to our control variables, the Protestantism, socialism,
energy imports, GDP per capita, and democracy variables all emerge
significant and robust across the various specifications, suggesting
that countries with larger proportions of Protestant populations, no
history of socialist rule, higher levels of energy imports (and, hence,
lower levels of energy exports and production), higher incomes, and
longer tenures of democracy tend to exhibit lower levels of perceived
political corruption. Having an English legal origin may reduce
corruption, though this effect is not particularly robust across
different specifications. With respect to geographical variables, only
latitude yields significant results on a consistent basis, with countries
farther away from the equator exhibiting lower levels of corruption.
Countries with larger populations may have more corruption, though the
relationship is not robust.36
This does
not change if land area is omitted from the estimation.
The
regional dummies, while significant collectively in an F-test, were
inconsistent on an individual basis.
37 Because their substantive import is unclear, we conserve
space by not reporting the results for the regional dummy
variables.
With respect to our variables of interest, we use two broad
indicators of “neoliberalism.” The first is KKM's
measure of regulatory quality, which incorporates a broad range of
regulatory factors, including those related to market openness.38
The results for this variable are
highly significant and robust, suggesting that those countries with
more market-friendly policies exhibit less corruption.
39 We found similarly strong results for the
Heritage Foundation and Wall Street Journal's Economic
Freedom Index, but it includes various measures of corruption that
preclude its use as an independent variable to explain corruption. See
Johnson, Holmes, and Kirkpatrick
1999.
Because this variable does not parse out the
various aspects of market-oriented policies, we rely on it solely as an
indicator of a general trend toward neoliberalism in a given country.
(The fact that it overlaps with our measures of trade and foreign
investment polices, among others, makes it impossible to distinguish
between the effects of specific policies with this variable.)
Our second summary indicator of neoliberalism is inflation, as
suggested above in the hypotheses section. We test the relationship
between price levels and corruption with the Heritage
Foundation/Wall Street Journal's multiyear measure of
inflation, in order to flatten out periodic fluctuations and address
the issues of endogeneity and causal direction.40
The measure of inflation compiled by Johnson, Holmes, and Kirkpatrick 1999 and the one
we use here covers 1985–95 and gives greater weight to recent
years. We also ran tests using an unweighted, twenty-five-year
(1970–95) average that produced similar, though less consistent
results (not reported).
We simultaneously tested the impact of
money supply (as a proportion of foreign reserves), government debt (as
a proportion of GDP), and budget balance (as a proportion of GDP), in
an attempt to parse out the possible effects of inflation itself (for
example, time-horizon and discount rate effects) from the potential
effects of its underlying macroeconomic policy causes (for example, the
effects of large fiscal deficits), as discussed above. Regression
results are consistently significant for the inflation variable: high
rates of long-term inflation tend to be associated with higher levels
of political corruption. They are equally consistently weak for the
money supply and fiscal variables, neither of which approach
statistical significance. This suggests that the unstable and uncertain
atmosphere of high inflation may be more important in the relationship
between macroeconomic stability and corruption than a country's
underlying monetary and fiscal policies.
We capture trade openness with imports as a share of GDP, averaged
over the previous two and a half decades (1970–95) to minimize
endogeneity problems (see below).41
A
measure of overall trade (imports plus exports, divided by GDP) yielded
similar results.
Statistically, the results in Tables
1 and
2 are clear: with
greater than a 99 percent degree of confidence, we can conclude that
openness to international trade appears to be associated with lower
levels of political corruption.
We capture foreign investment policy using a measure of net foreign
direct investment (FDI), expressed as a fraction of GDP and averaged
over the 1970–95 time period. The results, while somewhat less
significant than those for trade, suggest that greater openness to FDI
may be linked to lower perceived levels of political corruption.
These results are open to several possible objections. First, one
might object that trade and investment outcomes are not adequate
measures of trade and investment policies per se. Thus a government
might pursue open trade and investment policies but not achieve a great
deal of trade or foreign investment. Conversely, a country may have a
high level of trade, but not a particularly open economy.42
High levels of FDI are probably less likely
under restrictive conditions.
(Statist, outward-oriented
economies like those of the East Asian tigers come to mind.) Using
imports rather than total trade to measure openness avoids some of
these problems. We also tested measures of import duties and mean
tariff rates for trade, and Johnson, Holmes, and Kirkpatrick's
measure of foreign investment restrictions for foreign investment, with
no change in the substantive results presented in this article.
43 Johnson, Holmes, and
Kirkpatrick 1999. The foreign investment restriction variable is
significant, while the measures of import duties and mean tariffs are
not, likely because data coverage is poor for these variables and
because they do not include nontariff barriers to trade, often the most
important policy instruments for restricting imports (not
reported).
Over the long term it is reasonable to assume that
high levels of imports and foreign investment reflect generally open
trade and investment policies, ceteris paribus.
Finally, corruption itself may discourage foreign trade and
investment; hence, the association between openness and noncorrupt
government could be, at least partly, circular.44
Future
work may explore this bidirectional relationship in greater detail. For
now, we employ lagged averages in the independent variables to better
track the direction of causality.
We employ three indicators of regulatory policy, each a component of
the Heritage Foundation/Wall Street Journal's Index
of Economic Freedom. The first, termed regulation, attempts to
capture the overall regulatory environment. Results for this variable
in Table 1 are strongly significant. This
variable, however, includes as one of its six factors corruption within
the bureaucracy. Two narrower variables from the same data set allow us
to capture two critical aspects of regulatory policy. Banking and
finance (financial regulation) captures various elements of
the government's role in regulating banks and the financial sector
more generally. Wages and prices (price regulation) includes
government regulations in labor laws, price controls, and business
subsidies. Each of these variables is statistically significant at the
.05 level and signed in the expected direction, suggesting a link
between heavy government regulation and political corruption.
Public sector size is perhaps the most problematic concept in this
inventory of neoliberal hypotheses. The difficulty stems not from an
absence of viable indicators, but from their abundance and divergence.
Although the concept of the public sector is well known, there is no
generally acknowledged way of measuring it. Given its complexities, we
present five indicators of public sector size. These include (1) public
consumption (as percent of GDP), (2) total central government
expenditure (as percent of GDP), (3) state-run enterprise (as share of
the total economy), (4) government employees (as share of the general
population), and (5) government intervention. The latter is a Heritage
Foundation/Wall Street Journal composite of government
consumption, government ownership of business, and government economic
output.
The results of these various tests do not verify the neoliberal
hypothesis. In no case do any of these variables attain statistical
significance at conventional levels (.10 or better). The general
conclusion is inescapable: public sector size has little or no
relationship with political corruption. This result is not necessarily
surprising. As noted above, there are strong counter-arguments to the
neoliberal proposition on this point, and there is even room within
neoliberalism for a minimal state role (though such a role would likely
not register high on the indicators used in this study).
To this point, we have tested alternate hypotheses by adding
variables individually to the core model—equation (1) presented
in Table 1. In order to see how the primary
causal variables of interest perform when the others are also present
in the specification, we construct a “full” specification
in equation (14) in Table 2. This
specification includes, in addition to our control variables, the
strongest-performing and most-representative variable from each
theoretical hypothesis: inflation (summary measure), imports to GDP
(trade and investment), and price regulation (regulation).45
Results for other variables that reached
statistical significance in narrower specifications, such as
fdi/gdp, regulatory quality, or financial regulations,
were either not significant in the fuller specification (likely because
of inadequate degrees of freedom and/or colinearity problems)
and/or duplicative with other variables.
Results are
consonant with findings reported in previous equations in Tables
1 and
2. Market openness
and low regulatory burdens are associated with lower levels of
political corruption, while the size of the public sector has no
apparent effect.
Conclusion
The findings we present in this article demonstrate a strong,
consistent statistical correlation between trade, investment, and
regulatory policies and political corruption across a sample that
includes most sovereign nation-states in the world. Though the nature
of these relationships is likely far more complex than current
theorizing on the subject, it is certainly plausible to argue that this
statistical relationship reflects an underlying causal relationship.
But the pathways of these causal relationships are complex and
difficult to specify. Many factors are probably operating in tandem, as
discussed above, and current scholarship offers only initial hints as
to how to go about sorting out these pathways. But the patterns that
emerge from the data analyzed in this study call for further research
into these relationships and the potential causality that resides in
them. In the meantime, regardless of the precise nature of these causal
pathways, and the precise weight to assign to each, a general
theoretical argument emerges from the empirical patterns observed in
our study: undue restrictions on foreign and domestic economic exchange
appear to be linked, perhaps causally, to higher levels of perceived
political corruption. Insofar as minimizing political corruption is an
overall policy objective, states may be well advised to consider
neoliberal trade and regulatory policies. The same cannot be said,
however, for limits on the size of the public sector. Cutting back the
economic size of the state may offer benefits (and costs), but a
reduction in corruption does not appear to be among them.
This last result suggests an important qualification of the
neoliberal hypothesis, at least insofar as that paradigm may be viewed
as antistatist. Big government is not necessarily corrupt government. A
brief comparison of regulatory and fiscal policies and their variable
impact on political corruption is instructive. For a variety of
reasons, we suggest that regulatory policy (whether directed at foreign
or domestic activities) is more amenable to corrupt practices than
fiscal policy. It is easier to cloak illicit actions that require no
outlay of funds. Foregone expenditures (as when a bribe induces a
bureaucrat to overlook a violation of a regulatory code) are both less
easily detected and easier to justify, at least in the minds of the
corrupt actors, than outright theft. Regulatory corruption is also more
likely to be the product of bureaucratic malfeasance, while
expenditures generally flow through the legislature—where corrupt
arrangements may be more difficult to engineer, if for no other reason
than that they require the knowledge and acquiescence of a greater
number of actors. Spending decisions are easier to track, as there is a
financial trail. Spending measures, finally, are more likely to address
matters of broad public concern (for example, health, education,
welfare, and public works), and consequently, more likely to be
monitored by watchdog groups and by the general public. Regulatory
measures, by contrast, often address narrow concerns and
constituencies. Social security, in this respect, may be contrasted
with business licenses.
Of course, minimizing corruption is not the sum bonum of good
governance; other goals, such as redistribution, might challenge
neoliberal prescriptions in these policy areas. By the same token,
however, one ought not assume that egalitarian goals necessarily
conflict with market-friendly policies.46
(There is
ample evidence, for example, that high levels of inflation harm the
poor disproportionately.) We leave this matter open for further
investigation.
47 We interpret these findings as neither a vindication nor refutation
of neoliberalism, but rather as a qualification of it. All too
frequently, neoliberal hypotheses have been overaggregated, and too few
distinctions drawn between neoliberal prescriptions. Rather than
adopting sets of policies wholesale, policymakers would be wise to
consider the potentially divergent effects of different types of
measures. Neoliberalism, we suggest, is best understood as a promarket
approach, not an antistate one.48
Our
conclusions thus dovetail with Tanzi, who writes “The way the
state operates and carries out its functions is far more important than
the size of public sector activity.” Tanzi
1998, 566.
It is rightly applied to policies that have
greatest impact on the behavior of markets, such as trade, investment,
and regulatory policies. Its application should be much more tentative
with respect to policies that have ambiguous effects on markets, such
as those related to the size of government.
