首页 | 农村土地改革 | 法经济学茶座 | 法经济学论坛
     网站首页≯ 刑法≯
文/Roger Bowles


Roger Bowles
School of Social Sciences
University of Bath
© Copyright 1999 Roger Bowles

This chapter reviews the literature on corruption, which is both vast and
diverse. Analytically, corruption has been of great interest to economic theorists
in recent years because it is an archetypal problem of asymmetric information
involving collusion between agents and ‘outsiders’ against principals, or
between supervisors and agents against principals. Most of the models up to
now have been essentially static, but recent papers have shown that there is a
real prospect of developing dynamic models which can explain the persistence
of corruption.
JEL classification: K14, K42
Keywords: Bureaucratic Corruption, Private Sector Corruption, Illegal

1. Introduction

Corruption has been around for just about as long as institutions of any sort. In
the times of the Greek and Roman empires, for example, corruption was
widespread (MacMullen, 1988). Indeed, the corruption of tax officials was so
common at that time that it was found expedient to ‘privatize’ tax collection by
auctioning rights to collect taxes (Webber and Wildavsky, 1986). Modern
democratic government emerged from earlier structures which were ‘rotten to
the core’, and the professionalisation of the civil service in the nineteenth
century in England was aimed in part at eliminating the sapping effects of
corruption on public servants. In the present day, corruption ‘scandals’ are
widely reported in the press from almost all walks of life including politics,
government and business. The sums involved can range from tiny payments or
even small personal favours to bribes running into millions of dollars.

Corruption can be distinguished from ‘fraud’, ‘embezzlement’ and
‘extortion’. The essence of corruption is that two individuals or groups act in
concert to further their own interests at the expense of a third party. Fraud, or
extortion, by contrast involves an individual or group acting unilaterally to
further their own interests at the expense of others. In the language of
principal-agent theory this entails collusion between the agent and supervisor
against the principal. There are, nevertheless, many everday situations in which
such collusion occurs, even when it is at the expense of the principal, where
there would be no (legal) presumption that corruption has occurred. A patient
and doctor, for example, may ‘collude’ against the insurer responsible for the
costs of treatment by agreeing for the patient to have more costly medicine than
is necessary. This might be considered corrupt if the doctor has a financial
interest in the provision of medicine, but otherwise it would probably be called
just another ‘agency cost’ (Bowles and Rickman, 1998). A good deal of
‘corporate hospitality’ involves the provision of private benefits to individuals
in the hope that the closer personal relationships it fosters will prove a
‘worthwhile investment’.

Another everday instance of the ‘grey area’ between honest and corrupt
transactions is where a consumer can benefit by paying ‘over the odds’ for
additional service which directly benefits the employee of a firm but not the
firm itself. In some service sectors, such as hairdressing or hotels, gratuities are
regarded as normal and acceptable. In other sectors, such as refuse collection,
gratuities may be discouraged or even prohibited.

In many jurisdictions corruption is a criminal offence, although it is clear
from the relevant legislation that there are difficulties in defining it and in
setting standards of evidence for it as an offence. In England and Wales, for
example, the Prevention of Corruption Act 1916 (the act which governs public
sector corruption in Britain today) coverage extends to members, officers or
servants of a public body. Any ‘gift, loan, fee, reward, or advantage whatever’
will count as corruption if it is in consideration of any act or admission by the
public servant. The penalties for the offence can amount to seven years
imprisonment or a fine or both, in addition to repaying the amount involved.
On a second offence a person stands to lose various rights to vote and may lose
any compensation or pension rights to which a public position might otherwise
entitle them. As a measure of the response of the legislature to difficulties in
seeking successful convictions for the offence, in some circumstances the
burden of proof lies on a public employee who has received a payment to show
that a suspicious transaction was not corrupt.

Under US law the position is a little different. There is not an outright ban
on receiving rewards. The US Department of Defense directive 55007 allows
gratuities when they are a ‘part of a customary exchange of social amenities
between personal friends and relatives when motivated by such relationships
and extended on a personal basis’ (Adams, 1981).

Though there are many countries in which bribing officials is illegal, it was
only in 1997 that 34 countries signed an agreement aimed at eradicating
bribery of foreign officials by making such payments a criminal offence and
ceasing the practice of making them tax deductible (Tanzi and Davoodi, 1998).
In many countries there may be conflict between social custom and practice
on the one hand and the letter of the law on the other. Many cultures rely on the
giving of gifts as a symbol of a mutuality of interest. The dividing line between
recognising a mutual obligation of support and doing favours expressly in
exchange for a reward can be a fine one (Klitgaard, 1991). It is in part because
of this ambiguity that some authors have argued that the definition of
corruption should be refined to include a wider morality and allow for
obligations to others (Palmier, 1983; Lewis, 1996).

In everday use corruption is a term which conveys an element of moral
disapproval. It represents an unwelcome deviation from some desired state of
the world. Officials or politicians taking bribes are portrayed as ruthless and
selfish people exploiting their office for their own private interests at the
expense of the wider public good. It is implicit in most discussion of this kind
that elimination of corrupt transactions would self-evidently improve welfare.
Corruption in this kind of sense is widely recognized as a ‘problem’ in many
countries today. Indices measuring the degree of corruption in the countries of
the world are widely published: Business International and Political Risk
Services publishes an annual index, International Country Risk Guide; see also
Business International Corporation (1984).

In an academic context corruption is a term used in many disciplines
(including anthropology, sociology, politics, public administration and
development studies) and can refer to a very wide diversity of activities and
states of affairs. Shleifer and Vishny (1993), for example, ‘define government
corruption as the sale by government officials of government property for
personal gain’. But this is a much narrower definition of corruption than the
classical authors of political theory such as Plato and Aristotle used (Bouckaert,
1996). For them, and for some more recent authors such as Andreski (1978),
corruption was not so much a class of illegal practices as a kind of complete
subversion to which states and civilisations were vulnerable. In this more
classical sense corruption conveys the idea of wrecking or destroying
institutions. The more recent use of the term does not have this same
connotation, although authors differ widely in the moral content they give the
term. For purposes of this review we take a ‘middle of the road’ view, by
defining corruption more widely than Shleifer and Vishny but less broadly than
Plato and Aristotle. We include private sector institutions but make rather little
reference to the corruption by rulers of the basic machinery of the state.

2. The Literature on Corruption

The literature on corruption is both vast and diverse. There is a journal
(Corruption and Reform) which has been running since 1987 for which it
represents a focus of study, whilst a number of journals have devoted special
issues to the subject, some recent examples being: Stato e Mercato (April
1992); Crime, Law and Social Change (1994-95); Informacion Comercial
Espanola (May 1995); Indian Journal of Public Administration
(July-September, 1995); European Journal on Criminal Policy and Research
(vol 3(2), 1995); and Journal of Law and Society (March 1996).

Over the past decade corruption has attracted attention also within the
mainstream economics literature as theorists have become increasingly
interested in the structure of organizations and agency problems. The impetus
for the rapidly growing interest in corruption on the part of economists in
recent years is usually traced back to the pioneering work of Rose-Ackerman
(1975, 1978a). Texts on microeconomics and on the economics of organisations
are increasingly likely to contain discussion of agency problems (Katz and
Rosen, 1994) and even corruption itself (Brickley, Smith and Zimmerman,

It is this strand of literature on which we focus here for the most part. In
contemporary economics the notion of corruption is often used in a fairly
narrow and technical way to refer to a particular type of opportunistic
behaviour which rational agents will follow. In particular corruption results as
a cooperative outcome in games involving three or more players, where two (or
more) players can collude at the expense of a third party. Bribes to officials are
thus readily explained as transactions in which both buyer and seller of a favour
benefit at the expense of the organisation or hierarchy within which the official
works. This same basic principal-agent model has many applications, such as
to the ‘agency problems’ resulting from the divorce between ownership and
control of firms. Lazy managers indulging their workers’ preferences for an
easy life make life more agreeable for both the manager and the workers at the
expense of the profits remaining for the firm’s shareholders (Katz and Rosen,
1994). Such behaviour is often described as collusion rather than as corruption,
even though it meets most definitions of corruption (Tirole, 1986, 1992).

A survey of the literature on corruption could be organised in many ways.
It could be based on discipline or on the type of corruption or on dimensions of
corruption such as causes and consequences. We opt here for organisation
around various strands in the economics literature on the topic. Sections 3 and
4 set out a basic, or generic, model of corruption giving a flavour of
contemporary work in the field. Section 5 explores the Public Choice approach.
Section 6 explores some of the applications of corruption models to private
sector firms. Section 7 reviews bureaucratic corruption and Section 8
concentrates on political corruption. Section 9 is concerned with ‘macro’
models of corruption which focus on the consequences of corruption for
investment levels, growth rates and so on. Section 10 reviews policy to combat
corruption. Section 11 reviews the analysis of corruption in LDCs and Section
12 concludes.

3. Basic Economic Analysis of Corruption

A natural starting point for the economic analysis of corruption is to treat it as
any other crime and to apply to it the standard economic model of crime
developed originally in Becker (1968) and extended subsequently by many
authors such as Polinsky and Shavell (1979, 1984). In this basic model the
persons contemplating corruption take account of the expected benefits in the
form of bribes, favours or payment in kind and compare the monetary
equivalent of these gains with the expected costs in the form of a probability
that they will be detected and the monetary sum (or equivalent) of the
punishment should they be convicted. Such a formulation has close parallels
with the application of the Becker model to the economics of tax evasion by
Allingham and Sandmo (1972). Corruption is predicted to occur if the net
expected gain is positive.

The important element in corruption which is missing in the earlier
applications of the Becker model is, of course, that both sides to a corrupt
transaction are ususally committing a crime and not just one side. In the tax
evasion model the taxpayer has private information not available to the tax
official and all that is entailed analytically is a calculation on the part of the
taxpayer as to whether, or how much of the information, to disclose. In a
corruption model, each side has to be persuaded that it is worthwhile for both
sides to collude in sharing information which will not be disclosed to third

A simple algebraic formulation helps crystalize ideas on this basic model
and forms a useful base on which more sophisticated models can be
constructed. We suppose that a citizen stands to enjoy a rent, R. The ‘rent’ can
take a great many forms: it might be an unpaid tax liability, or avoidance of a
term of imprisonment, or the supernormal profits from the receipt of monopoly
rights and so on: it could be as simple as the gain from having a document
processed. A bureaucrat (or regulator or politician or another citizen) is in a
position to decide whether or not the rent is to be enjoyed by the citizen. In
exchange for a bribe B the bureaucrat promises to bestow the monopoly right,
to process the document or to keep the incriminating information secret or
whatever. There is nevertheless a risk, in the form of a probability, p, known
to both the citizen and the bureaucrat, that payment of the bribe will be detected
by a third party in which case both parties to the corruption will be liable for
punishment. For the citizen, discovery will mean liability to pay a fine F whilst
for the bureaucrat it will mean liability to pay a fine G and to repay the bribe

To keep the model simple, assume both parties are risk neutral so that the
calculation of expected gains and losses need contain no reference to utility
levels. Assume also that both parties to the corruption are sincere, so that we
can ignore both the possibility that the citizen is trying to lure the bureaucrat
into an illegal act and also any fear the citizen might have that the bureaucrat
will renege on the corrupt agreement and spill the beans after taking the bribe.
For the citizen, payment of a bribe will be worthwhile iff:

The first term refers to the case where the bribe goes undetected so that the
citizen enjoys the rent net of the bribe paid. The second term refers to the case
where the bribe is detected and the citizen, who has already paid a
(non-recoverable) bribe, is fined for corruption. Condition (1) can be
rearranged in the form of an upper bound, on the size of bribe the citizen
will find it worthwhile paying:

For the bureaucrat, the calculation is that taking the bribe is worthwhile
provided that:

where, again, the first term refers to the case where the bribe is undetected and
the second term to the consequence of detection. We asume that the bureaucrat
assigns the same value, p, to the probability of detection as does the citizen. The
condition can then be rearranged in the form of a lower bound, on the bribe
which will induce the bureaucrat to conspire in the corruption:

Corruption is assumed to occur in this simple model if there exists a bribe
at which the citizen and the official both find themselves better off. This
condition is satisfied if and only if the upper bound for the citizen exceeds the
lower bound for the official, that is . This is equivalent to the
requirement that:

The ‘enforcement parameters’ p, F and G describing the probability of
detection and the size of the punishment, have to be viewed against the
prospective gain to the citizen from corruption. Corruption occurs so long as
the rent at stake is sufficiently large to offset the expected punishment.

4. Implications of the Basic Model

Many of the main findings from the corruption literature are at least implicit
in the formulation of the previous section, namely:
1. There is nothing to prevent a ‘rotten core’. Corruption will occur if the
lower bound from (4) is below the upper bound on willingness to pay
a bribe derived in (2).
2. As condition (5) indicates, this is equivalent to requiring that the rent at
issue is sufficiently high relative to the parameters p, F and G defining the
probability of detection and the punishment structure. In policy oriented
models it is common to specify either the fines as related in some way to
the size of the rent at issue and/or to consider a group of citizens across
whom the value of R varies. In this way it is possible to model the extent
of corruption rather than modelling a (binary) choice problem for an
isolated citizen who either indulges in corruption or avoids it;
3. Increasing the fine for citizens, F, will tend to reduce corruption since it
raises the cost of detection and thus reduces the net pay-off from
corruption: BU falls while BL remains constant;
4. Likewise for the penalty, G, imposed on officials: this time BL rises while
BU remains constant, with the result that corruption again becomes less
5. Increasing the probability of detection, p, simultaneously reduces BL and
raises BU and thus erodes the difference (BU ! BL) on which corruption
6. There is in the transaction at least some of the character of a bilateral
monopoly, so long as neither side can readily contact alternative buyers or
sellers. The size of the bribe may thus be indeterminate, requiring the
introduction of some assumption about the bargaining which will take
place between the citizen and the bureaucrat;
7. Introducing competition into the model, for example through allowing the
citizen to canvass a range of (corruptible) bureaucrats, may result in
Bertrand competition between officials driving bribes down to the
competitive level. Corrupt officials try to attract bribes by offering more
attractive or cheaper terms than their competitors with the result that the
bribe gets driven down close to zero (Shleifer and Vishny, 1993).
We consider some of these findings and further extensions in sections

below concerned with more sophisticated models of corruption. One obvious
extension is to consider the monetary cost to the bureaucrat of being detected
not just in the form of a court-imposed fine, but as putting at risk the
bureaucrat’s pension rights and any wage premium (relative to available
alternative job openings) she is earning as a bureaucrat (see Becker and
Stigler, 1974).

A second line of development is to introduce a layer of officials
(supervisors or auditors) whose responsibility is to detect corruption by
‘front-line’ officials. In these multi-tier hierarchies it is possible to specify
different kinds of ‘supervisory technology’ and to distinguish ‘systemic’
corruption from corruption specific to a single tier (Bac, 1998).

Another extension is to consider corruption in a dynamic model rather than
in a purely static way. Individuals who sense that corruption is potentially
profitable for them may seek to ‘regularize’ corrupt arrangements so as to
minimize the risk of being surprised by detection and to reduce transactions
costs to a minimum. The systematic corruption of police by organized
criminals can be interpreted in this way as can various political arrangements,
including bureaucratic corruption in the ‘old Soviet’ type of economy (Shleifer
and Vishny, 1992).

From a policy perspective the central question to which this basic model
can most readily be applied is the choice of an optimal level of resources to put
into the detection of corruption and the choice of an optimal fine. As in other
areas, the tendency of this optimal law enforcement model, as pioneered by
Becker (1968) and reviewed recently by Garoupa (1997), is to produce
solutions suggesting high, possibly maximal, fines and comparatively low
probabilities of detection. There are two reasons why these sorts of results tend
to emerge. First, fines represent transfer payments rather than real resource
costs so that, other things equal, it is usually judged better to rely on the
deterrent power of the fine rather than on a high probability of being caught
(Polinsky and Shavell, 1984). Secondly, in an area such as corruption where
the incentives to report the crime are small, if not non-existent, it is likely to
be extremely costly to raise the probability of detection significantly.

One drawback of this model, albeit one which is shared by most models of
corruption, is the difficulty of collecting empirical data which could be used
to test it. The probability of detection, for example, would require information
about the proportion of offences involving corruption in which an individual
is detected and convicted. But one of the central features of corruption is that
it is, in essence, a cooperative venture which both participants have every
incentive to keep secret. It is thus extremely difficult to identify either
particular instances of it or its extent in aggregate. From an analytical
perspective the question becomes one of how individuals might construct or
update (subjective) estimates of probabilities in such circumstances, a matter
which is discussed in some of the literature on information, and particularly
on the value of message services (see ch. 5 of Hirshleifer and Riley, 1992).
There are very obvious parallels with the analysis of the ‘underground’ or
 ‘black’ economy: ingenious indirect methods may be used to make estimates
of how widespread are the transactions at issue, but nobody really knows (Pyle,

5. Corruption and Public Choice

Whilst the analysis of behaviour under uncertainty has provided a
comparatively straightforward line of approach, as we have shown in the
previous two sections, there has been a parallel line of intellectual enquiry
which has also been influential in the development of our understanding of
corruption. The application of economic reasoning to public bodies and
political processes has resulted in the development of an analysis of
government and bureaucracy which stresses the critical role played by
individual interests and objectives in the shaping of these institutions. The
formulation of ideas which came to be known as the theory of public choice
showed a recognition that public policy could not be thought of simply as the
pursuit by pure-minded automata of ‘the public interest’ but rather represented
the outcome of decisions by individuals (be they politicians, bureaucrats or
interest groups) with their own interests and their own agenda. Early analyses
in this vein were the work of Downs (1957) on the motives of politicians and
of Niskanen (1974) on bureaucrats.

The ‘public choice’ approach emphasizes the role played by the private
interests of all the players involved in the formulation and execution of policy.
For surveys of the public choice approach see Mueller (1979), or Cullis and
Jones (1992). The capture of politicians by interest groups is studied by Becker
(1983). Corruption is something which fits very naturally into a picture of the
world in which politicians, bureaucrats and citizens are all busily searching for
strategies for pursuing their own private motives at the expense of others,
whether they be taxpayers, other citizens or whatever. It might be noted also
that principal-agent models of the kind reviewed in Section 2 above are quite
close in spirit to the Public Choice approach, although they tend to a higher
degree of abstraction.

The analysis of rent seeking provides a fertile ground for creating a
motivation for corruption. Standard references on rent seeking are Tullock
(1967), Kreuger (1974), Posner (1975), Bhagwati (1982) and Tollison (1982).
For applications of rent-seeking to corruption see Blomqvist and Mohammad
(1986) and Mbaku (1992). Rational individuals assumed to be maximizing
their own net benefits without further reference to ethical constraints may
come to regard corruption almost as a moral imperative. The individual
seeking monopoly rights will probably consider corruption as a possible
substitute for (or complement to) other tactics such as political lobbying or
capturing bureaucrats. Corruption comes then to be chosen if it represents part
of a cost-minimizing strategy. A natural extension of this rent-seeking
approach to the analysis of corruption is suggested in Bowles (1998) in which
a citizen seeking rents considers various possible ‘pathways’ to achieve her
goal. The more pathways there are, the more opportunity there is to acquire the
rent at a lower cost, such as corrupting an official or a politician or whatever
may be regarded as alternatives to seeking the rents through some ‘legitimate’
channel such as legislative change or the exercise of an official’s discretion.

Are there any limits to this pursuit of individual self interest other than
possibly rather ineffective external controls? Is corruption inevitable if it pays?
Just about the only device through which any such limits can be imposed is to
suppose that individuals may incur some subjective costs or ‘stigma’ when
acting dishonestly. This approach borrows the term coined by analysts seeking
to explain apparently irrational instances of honesty in other spheres, as in
Akerlof (1980) and Benjamini and Maital (1985). It may look contrived, but
it is often the only practicable way of filling the ‘moral vacuum’ created by the
economic approach and explaining why people may desist from corruption
even when it appears ‘profitable’.

6. Private Sector Corruption

Within what one might term the ‘purely private’ sector, corruption has been
studied rather less than its public sector counterpart, although that is not
necessarily a reason for supposing that it is less widespread than in the public
sector. A private sector engineering contractor, for example, may find it
expedient to pay bribes to a prospective private sector buyer for just the same
reasons they pay bribes to government officials or politicians. Data deficiencies
are particularly acute in this field, since it is only really from reports of court
cases that remotely reliable evidence is likely to emerge. Whilst a public body
might seek to publicize anti-corruption efforts and successes to demonstrate a
tough anti-corruption policy, the private firm might prefer to resolve matters
away from the public eye in order to avoid the bad publicity and loss of
reputation which disclosures of corruption might bring. There are many
different forms corruption in the private sector can take: we review a brief
selection here.

Insider trading scandals occur from time to time on most of the major
financial markets, and they sometimes involve corruption. The insider, say the
accountant of a major company or a senior executive of the firm who has seen
the firm’s provisional profit results, holds information which when published
will, or at least might, move the firm’s share price. Rather than trade in the
shares herself, the accountant or executive might contemplate selling the
relevant information to a third party who is in a position to make profitable,
but less obvious, use of it by buying or selling shares in the firm. Any such sale
of information would be regarded in most countries as corrupt, even though
there are some economists who would regard such transactions as being
conducive of market efficiency for the most part, and would argue that the
main beneficiaries of a prohibition of a ban on insider information are
financial market professionals rather than the firm’s shareholders (Manne,
1966; Baye, 1997).

Valuation of goods, services or assets is clearly an area where there is scope
for corruption. The valuer can ‘create a rent’ by overvaluing an item, since the
owner of the item might stand to gain in creditworthiness as a result or might
get a better price if selling the item. This creates an opportunity for collusion
between the owner and the valuer of the item at the expense of the buyer.
Examples of this kind of collusion can be found in markets for sports players,
where the manager of a club buying a player takes a bribe from the club selling
a player (or from the player’s agent) to overestimate the player’s value. They
can be found also in the financial markets where a stock analyst is entertained
lavishly by a firm hoping that the analyst will publish a positive report about
the prospects for the firm’s share price. In a somewhat similar vein was the
‘payola’ scandal in which record companies used to pay bribes to DJs for
playing records by the company’s artists on the radio. Pechlivanos (1995)
refers to the example of the American Honda Motor Company whose managers
received kickbacks from prospective franchisees.

Andvig’s study (1995) of the North Sea Oil industry makes use of reports
from six trials in the British courts of so-called ‘information brokers’. Ten of
the thirteen defendants in these trials were found guilty: punishments ranged
from six months to three years in prison plus fines and compensation orders
from about £40,000 to £500,000. Spending on the procurement of equipment
for the North Sea is very high, at around £13 billion per annum. Those
involved in the award of contracts for the supply of this equpiment have
information about bids and tenders which is potentially of substantial value to
the firms competing for supply contracts.

Monopoly power is likely to be a pre-condition for private sector
corruption. A well informed competitive market will leave little room for it
since, at least in the longer run, corrupt firms will have higher costs and thus
be unable to survive. But a privatized utility might, for example, enjoy a local
monopoly in the provision of some service. It is quite imaginable that the
employee of the utility will exploit this monopoly for private gain, say, by
arranging connection to a power or telephone network unusually quickly for
a premium. A number of commentators in the literature on privatization have,
indeed, argued that the elimination of waste and inefficiency on the part of
public utilities requires not only a transfer of assets from public to private
hands but also the introduction of elements of competition to break down the
local monopoly power which is what enables the utilitity’s employees to ‘hold
up’ their customers (Kay and Thompson, 1986).

7. Bureaucratic Corruption

Public sector bureaucracies often have considerable powers and areas of
discretion: they are often in a monopoly supply position and their employees
are often tempted to make strategic use of these advantages for their own ends.
Anecdotes are legion. There is the immigration official checking papers who
can be persuaded with a $20 tip to ignore some real or imaginary
inconsistency. There are ships which cannot enter certain ports until they
supply customs officials with amounts of whisky and cigarettes as spelled out
on a schedule sent in advance of their arrival. There are hospitals in some
countries where patients cannot get drugs without paying bribes to staff. And
so on.

A hard-bitten economist, or world traveller for that matter, would think it
naive to imagine anything else. Even the less cynical would agree that there
are dangers that the pursuit of the ideal of public interest might sometimes slip
in favour of the pursuit by bureaucrats of their private goals. It is perhaps
inevitable then that public administration should have a rather tainted
reputation when it comes to corruption. But as the examples above illustrate,
in many cases the corruption is not an occasional lapse by an official but a
deeply-embedded modus vivendi. We consider in this section two particular
areas of bureacracy where allegations are often made and a good deal has been

7.1 Corruption of Tax Officials
Tax evasion has been widely studied, and it has for long been known that tax
administrations are particularly vulnerable to collusion between taxpayers who
underdeclare their income and tax officials who may be bribed to overlook
these underdeclarations when reporting to the tax authority. As we remarked
in Section 1 above, corruption of this kind was the motivation for the
development of ‘tax farming’ from the days of the Greek and Roman empires
until comparitively recently.

To give a sense of how deep-seated tax corruption might be, Chu (1990)
reports the results of a 1981 survey by the city government of Taipei in which
94 percent of the taxpayers polled admitted to paying off tax officials in
exchange for collusion in tax evasion. Interviews with the group of accountants
involved in the preparation of tax returns appeared to confirm that taxpayer
dishonesty was widespread and that almost a half of the tax officials with
whom the accountants dealt would actively solicit payoff.

Contemporary models of corruption of tax administration are usually
constructed in two stages. At stage 1, the taxpayer makes a declaration of
income following the standard Allingam and Sandmo (1972) approach, that
is by comparing the expected costs and benefits of declaring different
proportions of true income. The tax administration, meanwhile, selects an
audit probability for the taxpayer (which may be declaration-dependent) to
maximize its revenue net of the costs of audit. In the Graetz-Reinganum-Wilde
(1986) model, a Nash equilibrium is established which makes the policies
followed by the taxpayer and the tax administration mutually compatible.

At stage 2 the possibility is introduced that the tax auditor can suppress the
result of audits and thereby shield the taxpayer from a penalty for
underdeclaring income (Cadot, 1987; Chander and Wilde, 1992). The auditor
who chooses to take a bribe herself runs the risk of detection. The tax
administration thus has two layers of auditing: audit of the taxpayer and audit
of the auditors. Auditing of either kind is costly, and the administration’s
problem is to figure out how best to split its resources between the two kinds
of audit work.

Models of this kind are inevitably rather complex, since there are many sets
of assumptions the analyst can make and there are many interactions which
can be explored. For example the split of the taxpayer’s gain from
underdeclaration between the taxpayer and the auditor can be treated as
exogenous, as in Chander and Wilde (1992) but is endogenized in other
models such as Virmani (1987), Goswami, Sanyal and Gang (1991) and
Besley and Maclaren (1993) (see also Flatters and MacLeod, 1995 and Toma
and Toma, 1992).

Despite their analytical sophistication, these models gloss over many
important practical problems. The ease with which tax auditors can be
investigated, for example, will depend on the quality of information and record
keeping. If taxpayer files can be (strategically) lost, if tax auditors have
discretion about which taxpayer returns to investigate or if the auditor is
subject to influence via the political system, then the model misses important
interactions. But this demonstrates simply that the tax official has to maintain
a delicate balance between various forces. The contemporary models have
made a good start on modelling this tension, but there remains plenty of scope
for ‘fine tuning’. Tirole (1996), for example, has developed a model in which
a reputation for corruption may be difficult to shake off, and this is a
development which might have an interesting application in the tax field.

7.2 Police Corruption
Police enjoy the same privilege as tax officials, namely access to public
resources with which to investigate illegal activities or earnings which
individuals wish to keep secret. The temptations to collude in keeping the
secret may be very great and there have undoubtedly been occasions when
police corruption has been widespread. An early economic study of corruption
was the work of Becker and Stigler (1974) who had been asked to find ways
of reducing corruption amongst Chicago police (for a recent summary of their
work see Brickley, Smith and Zimmerman, 1996).

Models of police corruption can be distinguished by whether they deal with
‘one-off’ or repeated incidents of corruption. Bowles and Garoupa (1997), for
example, treat ‘casual’ police corruption as a one-shot game, involving
isolated instances of a citizen and a police officer who are strangers to each
other. There are many instances, however, where it is alleged that corruption
is not of a ‘one off’ kind but is more deeply embedded in a continuing
relationship. Schelling (1967) gives as an example an illegal wire-service
gambling syndicate in Miami. Regular payments were made to police not only
in respect of protection of syndicate members from arrest but also in exchange
for systematic intimidation by police of bookmakers outside the syndicate.

Another example is the illegal drugs market where producers and suppliers
may establish a regular working relationship with enforcement agents (see
Klitgaart, 1988, and Lupsha, 1991).

These dynamic elements call for a repeat play treatment. There are
interesting parallels between continuing corrupt relationships and the
discussion of alternative forms of contracting between firms using the notion
of transactions costs (Williamson, 1979; Deakin and Michie, 1997).
Corruption cannot be supported by legally enforceable contracts and yet
corrupt agents are clearly able to sustain viable long-term relationships.

8. Political Corruption

The powers of government to create and/or to protect rents have been a rich
source of study for those interested in corruption. It was Lord Acton who
famously claimed that ‘Power tends to corrupt and absolute power corrupts

The power to enact legislation carries with it the possibility of creating
rents. The prospect for the amoral individual of being able to participate in the
processes by which rents are created and distributed may express itself in a
wide variety of ways, some at least of which can be regarded as corrupt. Abuse
of this power has occurred frequently, most notably with monarchs and
autocratic (often military) governments rather than with democratic
governments. Kings in mediaeval England, for example, would grant trading
and many other monopolies simply to raise cash for their own purposes,
whether noble or otherwise. Similar sorts of claims have sometimes been made
about socialist economies and about countries with military governments. For
review of these and other issues in the sphere of political corruption (see
Amuwo, 1986; Benson, 1978, Brooks, 1978 and other contributors to
Heidenheimer, 1978; and Coldham, 1995).

In a modern democracy politicians are not usually in a position to be able
to enact legislation purely for their own private benefit. But they may be able
to wield influence over the legislative process, and such influence makes them
vulnerable to corruption by those groups who will be affected.

The asking of leading questions in the legislature and the exploitation of
membership of influential committees or subcommittees are two of the more
obvious ways of exerting pressure on the shape of legislation. The corruption
element obviously arises when the politician ‘sells’ this influence to an
interested party. The grey area between explicit selling of influence and more
subtle routes through which influence can be exploited is sometimes referred
to as ‘sleaze’. It is in an effort to counter this kind of sleaze which has
prompted the Nolan Committee to press for greater disclosure of the outside
interests of British Members of Parliament. Consultancies and sponsorship can
significantly undermine the credibility of the claim by politicians that they are
pursuing the public interest.

The power to enforce legislation is likewise open to manipulation and
abuse. Governments make decisions all the time about the public resources to
be made available for the agencies responsible for enforcing legislation.
Influence over the way the resources are used and over agency priorities will
often have a market value since a greater intensity of enforcement is likely to
be of advantage to some party. Pressure for law enforcement is, of itself,
entirely legitimate. But if the pressure is wielded in return for payments of
some kind then we have corruption.

The power to influence public sector contracting with the private sector for
the provision of goods and services is an area where corruption is sometimes
alleged to occur. In large-scale civil engineering projects, ‘political factors’
sometimes play a role in deciding which contractor gets the job as well as more
traditional criteria such as tender price or track record. Giving a role to such
factors is a way of creating a space for political influence to carry weight and
thus a market value. In effect it is allowing the politician to collude with
outside interests against the bureaucrats responsible for procurement. It thus
creates the possibility of corruption since it gives the outsider the incentive to
bribe the politician to wield power over the bureaucrat (Becker, 1983).

This is a subtly different form of corruption from the one considered in the
‘regulatory capture’ tradition in which the regulator (the bureaucrat) and the
outside interest collude against the politician or the public interest. Collusion
between regulators and regulated industries, in the form of regulatory capture,
against the interests of consumers and/or taxpayers has been explored by
Stigler (1971), Peltzman (1976) and many others. Corruption in this context,
as elsewhere, could range from the payment of cash bribes through to
regulators expecting employment in the regulated industry on retirement from
the public service: the ‘revolving doors’ arrangement, on which see Adams
(1981) and Breyer and Steward (1979).

The expanding literature on public sector procurement has concerned itself
with the structure of contracts and incentives in these ‘hierarchical’ models,
where there are three or more parties involved and a number of forms collusion
could take. The standard theoretical work in microeconomics in this area is
Laffont and Tirole (1993), which draws on earlier work on multilevel
hierarchies by Tirole (1986, 1992), Laffont (1990) and Koffman and Lawarree

9. Consequences of Corruption

Many ‘macro’ studies of corruption have been concerned with the causes and
consequences of corruption at an aggregate level. These studies can be thought
of as complements to, rather than as alternatives to, the essentially micro
models of behaviour reviewed here thus far. The consequences of corruption
have been widely discussed and authors have reached widely differing

The conventional wisdom is that corruption is harmful, and for some
authors this seems almost self-evidently to be the case. A recent paper, for
example, Tanzi and Davoodi (1998) carries the title ‘Road to Nowhere: How
Corruption in Public Investment Hurts Growth’ (see further Alam, 1990; Ades
and Di Tella, 1995, 1998; Mauro, 1995, 1998, and Klitgaard, 1998). The
general thrust of the argument is that corruption harms growth by reducing the
incentives to invest. This distorts the allocation of resources and leads to
underinvestment and poor growth rates.

Some authors, however, argue that corruption may be as good a device for
screening projects as anything else. In this view the most profitable amongst
a group of competing projects will generate the greatest possible bribe and thus
have the best chance of succeeding. The only difference is that most of the
monopoly profits are appropriated by the corrupt persons choosing between the
projects rather than by the firms proposing them. More technically, this
‘neutrality’ view suggests that corruption is not inefficient but may redistribute

Yet others argue that corruption may be positively beneficial (Leff, 1964).
Corrupt bureaucrats who respond to payments for services rendered may work
harder and process matters more quickly than they would if they relied purely
on their public services salary. But it should be stressed that this is rather a
minority view.

Even if the outcomes of selection processes for projects will be the same
whether or not corruption occurs, it is more efficient to use an honest
procedure because of the resource costs of secrecy. Bribes themselves may be
only transfer payments and thus irrelevant from an allocative perspective, but
real resources are involved in arranging and protecting bribe offers, and these
represent real costs which could be avoided. Making corruption illegal raises
the costs of corruption and thus some of the rents potentially created by
projects get dissipated by rent-seeking behaviour (Shleifer and Vishny, 1993).
There are parallels with the continuing debate about the resource costs of rentseeking
more generally: see for example the debate about how transfers are
treated in cost-benefit analysis, about how the costs of crime are to be treated
and about the distortion of the results of cost-benefit analysis by vested
interests in the decision-making processes of government (see Weingast,
Shepsle and Johnsen, 1981; Lee, 1983; Lewin and Trumbell, 1993;
McChesney, 1993, and Jones and Cullis, 1996).

As well as theoretical arguments about corruption being harmful, there has
been some empirical literature pointing in the same direction. Mauro (1995),
using data for a wide cross-section of countries, argues that, holding income
constant, countries with higher rates of corruption (as measured by the
standard business indicators of corruption referred to above) have a lower ratio
of both total and private sector investment to income. Tanzi and Davoodi
(1998) take a rather different tack in arguing that countries with high levels
of corruption will engage in particularly high rates of public investment, and
will be particularly subject to low rates of return on these investments. The
argument is that corruption may encourage projects which are inefficient but
rewarding to corrupt politicians.

10. Control of Corruption

When it comes to measures to control corruption, results tend to be dictated by
the structure of the model of behaviour employed. Nevertheless there are a
number of recurring strands in discussion of the strategies which can be used
to combat corruption.

The basic principal-agent model of corruption rather naturally inclines to
the search for ‘top-down’ solutions. The victim in these models is the principal
(the government) and the corrupt parties are bureaucrats and citizens.
Principals who suspect themselves to be vulnerable to ‘agency problems’ can,
speaking in general, take various kinds of defensive action (Ross, 1973;
Grossman and Hart, 1983; and Hirshleifer and Riley, 1992). The ‘Mechanism
Design’ literature invites the principal to design a grand scheme of things
which anticipates the corrupt behaviour and makes it unprofitable. This
scheme is implemented through contracts for bureaucrats which are
constructed in a way which would make it irrational for the bureaucrat to take
bribes (Lui, 1986; Cadot, 1987; and Mookherjee and Png, 1995). Methods for
‘designing corruption out of the system’ have long been sought. The ‘selling
the store’ solution (see Rasmusen, 1994) is a rather extreme example, but has
relevant historical parallels in the use of ‘tax farming’ contracts by tax revenue
departments for many centuries. The literature also recognizes that there are
some circumstances in which it will either be infeasible to prevent corruption
(Koffman and Lawarree, 1993) or non-optimal (Olsen and Torsvik, undated).

A second common theme is the use of auditing methods to monitor the
quality of the reporting behaviour of agents in hierarchies. Many tax and
police authorities have a ‘vigilance’ or ‘special investigations’ branch with the
express task of identifying and pursuing corrupt officers. This approach has its
drawbacks because most of the factors which encourage ‘front line’ officers to
behave corruptly apply equally when it comes to an auditing tier. The audit
reveals potentially valuable information which someone is prepared to pay
bribes to suppress (Border and Sobell, 1987). Likewise, if corruption is
endemic within an institution then it is likely that any findings of corruption
will be liable to suppression (see Bac, 1996a, 1996b, 1998).

An early application of this approach is the argument of Becker and Stigler
(1974) that by paying police a salary higher than the prevailing market wage
it would be possible to create an incentive on the part of officers to retain their
job. In such an environment, the prospect of being found guilty of taking bribes
acts as a powerful threat since it would result in loss, over the rest of the
officer’s career, of the rents created by the wage premium. The excess supply
of prospective police officers this premium would cause is dealt with by
‘selling’ the posts. This payment can be interpreted as a bond posted by the
officer: the honest officer keeps the bond in the form of a premium wage whilst
the dishonest officer forfeits it.

A third theme which recurs quite commonly is the introduction of
competition as a means of attacking corruption. In this view, corruption is
fostered wherever there is a single source of supply. By opening up
alternatives, the capacity of the bureaucrat or tax official to hold up the citizen
is reduced and the willingness to pay bribes correspondingly lowered. This
might entail opening ‘multiple windows’, so that a citizen can contact a
different official if the first one tries to elicit a bribe (Shleifer and Vishny,
1993). It might equally entail opening up an economy to foreign trade in order
to prevent a domestic producer wielding monopoly power (Ades and Di Tella,

A fourth theme is the use of administrative devices within organisations to
discourage corruption. Many police and tax departments have a policy of
moving officers around quite frequently in order to prevent them from
becoming too familiar with an area. In terms of the relational contracting
literature this can be interpreted as an effort to raise the transactions costs of
dealing corruptly rather than honestly. By frequently disrupting the pattern it
may be possible, at the cost of having to give on-the-job training more
frequently, to impose higher ‘set-up’ costs on those wishing to corrupt officers.

11. Corruption in LDCs

In their folklore at least, many developing countries experience substantial
corruption. Many contributions have been made by sociologists and other
policy analysts (including some work by economists) to what has become a
substantial literature on the subject of corruption in LDCs: for general reviews
see Beenstock (1979), Winston (1979), Macrae (1982), Jagannathan (1986,
1988), Andvig (1989), Ward (1989), Alam (1989, 1990, 1995) and Charlick

An obvious question, of course, is whether there is anything about
corruption in developing countries to distinguish it from corruption elswehere.
The short answer to this is probably no: the institutions involved are the same
(business, bureaucracies and governments) and the incentives for the
individuals concerned are similar. The same kind of analysis can thus, at least
in principle, be applied.

But this is probably too glib. There are special reasons why corruption in
LDCs is of particular concern and there are also reasons why it may take a
rather different form and may impose greater costs than its counterpart in the
North. Development aid has in many cases been diverted for their own private
use by corrupt politicians and bureaucrats. Partly as a defensive measure
against criticisms from the taxpayers funding aid budgets, a quest for ‘good
governance’ is now high on the priority list of many international agencies,
and anti-corruption programmes have become an essential item for many of
the poorer countries (see, for example, United Nations, 1989; World Bank,
1992; Doig, 1995).

This quest can be problematic, however. There are some countries where
corruption has become deeply embedded, in part as a response to very low civil
service salaries. Just to survive and support their families, officials may need
to moonlight or to take bribes. Clearly any serious effort to reduce corruption
in such an environment can only work if part of the policy is to raise civil
service salary levels. But this tends to raise public expenditure and, other
things equal, may intensify what is already a serious debt problem. One could
argue, therefore, that in countries where it is difficult to raise money through
taxes to pay officials a little bribery is no bad thing because it is one way of
getting citizens to support public sector officials. This is not a very attractive
argument to economists looking for first best equilibria or to moral crusaders
but it probably helps explain why corruption is sometimes tolerated.

A further reason for treating corruption as a ‘special’ problem for LDCs is
that institutional differences resulting from different cultural and political
traditions can create some ambiguity about the definition of corruption. Critics
have argued that much of the literature of the 1950s and 1960s was judgmental
and morally critical, and that care is needed when studying corruption in
LDCs to avoid an ethnocentric view in which corruption is viewed as an
aberration of post-colonial societies which would eventually disappear as these
countries evolved towards a more Western form (Clarke, 1983; Werner, 1983;
and Lewis, 1996).

Many of the studies have been concerned with particular countries or
regions: Kaufer (1989), Kennedy (1989), Sands (1989, 1990), Kolenda (1990);
Rocca (1992), and Wong (1992)on China; Halayya (1985), Goswami, Sanyal
and Gang, 1991), Wade (1982) on India; Carino (1986) and Gleason (1995)
on Asia; Grayson (1984) on Mexico; Maingot (1994) on the Southern
Hemisphere; Werlin (1973) and Le Vine (1975) on Ghana; Amuwo (1986) on
Niger; Reno (1995) on Sierra Leone; Ouma (1991) on Uganda; Gould (1980)
and Szeftel (1983) on Zaire; Makumbe (1994) on Zimbabwe; Ekpo (1979) on
Sub-Saharan Africa and Kotecha and Adams (1981), Theobald (1993) and
Mbaku (1994) on Africa more generally.

12. Concluding Remarks

Corruption is a very widespread phenomenon and takes a great variety of
forms. It is difficult, for this reason, to define very exactly. It is also something
which is very difficult to prevent. It is illegal virtually everywhere, in some of
its forms, but its victims often are unaware that it is taking place. Both parties
to a corrupt transaction have an incentive to avoid disclosure since both will
usually have been acting illegally. Thus, even establishing that it has occurred
may be very costly. Even when corruption is suspected it is very often difficult
to adduce legally convincing evidence since parties to corrupt transactions can
comparatively easily cover their tracks by using tactics which offer ambiguity
and are susceptible to multiple explanations. It is not, however, in any respect
a ‘victimless crime’.

Analytically, corruption has been of great interest to economic theorists in
recent years because it is an archetypal problem of asymmetric information
involving collusion between agents and ‘outsiders’ against principals, or
between supervisors and agents against principals. Most of the models up to
now have been essentially static, but recent papers have shown that there is a
real prospect of developing dynamic models which can explain the persistence
of corruption. Much of the future for application of these models will probably
be in the fields of the economic analysis of the operation of government and
political institutions. But corruption remains an area which will continue to be
of interest to scholars of law and economics because it raises fascinating issues
about the enforcement of law in the broadest sense.



加入日期:2007/1/11 9:13:41浏览次数:2360
验证: 6811
联系站长: 柯华庆 lawgame@263.net 京ICP备09028584号
本网站由卡卡鱼网提供技术支持 网站总访问量:1996356