Corruption: An Alternative Approach to its Definition and Measurement

Definitions of Corruption: A Review

By their very nature, definitions are incomplete descriptions of phenomena, consisting of a necessarily arbitrary selection of attributes chosen to define their character. What is a good definition? It ought to be unambiguous: the attributes themselves ought to be well defined and to constitute a necessary and sufficient condition for the existence of the phenomenon in question. A second criterion is more problematic. Strictly speaking, there is no need for a definition to adhere to common usage, and there are many examples where definitions are deliberately set at variance from popular understanding. However, the literature on definitions of corruption is full of criticisms, particularly of the public office definitions, for failing to accord with ‘public opinion’, or, what amounts to the same charge, for being ‘culturally relativistic’. Thus, most of the contributors to the definition debate have implicitly judged definitions by the ‘public opinion test’, perhaps for the sensible reason that the evolution of the term has indeed been closely connected to the breaking of public office norms and therefore to public disapproval. Public opinion has, for this reason, taken on a peculiar dual role in this definition debate – as a source for a definition (the public opinion definition) and as a criterion for evaluating other definitions. This practice is followed explicitly in what follows: conformity to public opinion constitutes a criterion that determines the quality of the definition of corruption.

The term ‘corruption’ can relate to a state of society as a whole or to individual acts. In what Johnston (1996) called the ‘classical’ view, corruption describes a society or a state that has lapsed from a standard of goodness. ¹ Alternatively, corruption can relate to qualities of individual actions. Whereas in the classical view social degradation is implicit in the term ‘corruption’, corrupt individual acts might well have beneficial social consequences — however much moral opprobrium and public condemnation attaches to them and however much rule-breaking (social or bureaucratic) they involve. The concern here is with individual actions, what Johnston called ‘behavioural corruption’, or more precisely the criteria that distinguish a corrupt from a non-corrupt act. ²

From this point of view, how are we to evaluate the development of discussion on the definition of corruption? I will argue in the following section – a view that is not universally shared – that little progress has been made since the groundbreaking discussions of the 1950s and 1960s, when the three types of definitions mentioned above emerged – the ‘subjective’ definition based on public opinion and two ‘objective’ definitions grounded in formal duties of the public office and the public interest, respectively. The undoubted advantages and obvious flaws of the different definitions will be contrasted, and I will show that the alternative types of definitions that have emerged invariably turn out to be variations of those basic types.

Because of its dual role as a way of both defining corruption and assessing other definitions, it is convenient to begin with a discussion of the public opinion definition – introduced by Scott only to be immediately discarded as unsatisfactory (1972, pp. 3–1:). The set of attributes defining corruption is assigned by ‘the public’, or some parts thereof, and herein lies, according to Scott, an insurmountable problem: there is unlikely to be agreement among the public at large as to what constitutes corruption, and, if there is disagreement, who is to decide which view to adopt? The definition, in other words, fails the operationability test. Despite the apparent cogency of the objection, the wholesale rejection was rash. Scott ought to have been surprised that the common international comparisons of levels of corruption derive from this type of definition, based as they are on such subjective perceptions of selected members of the public. Thus, at least the authors of the comparisons and their users must believe that there is sufficient common understanding of what constitutes corruption in the public sector to warrant the adoption of such a procedure. Moreover, what the public, however defined, thinks is corrupt is itself an important piece of information that is relevant, for example, in the design of anti-corruption programs. Thus, research programs into corruption in the public opinion sense certainly have their uses. More generally, analyses of public opinion – the attempt to isolate public office norms and values in different cultural contexts — led to major insights into the origins of public office corruption, since the conflict between such norms and values and the rule-bound action required by the bureaucratic Weberian state is thought to be one of the major causes of administrative corruption.

The most common ‘objective’ type of definitions today are public office definitions: corrupt actions violate rules of public office and are motivated by private gain. ‘Private’, in turn, is understood as a personal, close family, or private clique pecuniary or status gain or gain in influence (Nye, 1967, p. 419), a definition that will be adhered to in what follows. The advantage of this type of definition is that it scores high on the operationability count: what constitutes breaking formal rules and regulations is relatively easily established and, in principle at least, observable, as are the gains derived by the official in question. Unobservable in principle is the motivation, but this is of moderate practical concern, since someone who breaks rules and regulations and pockets a reward for doing so is not generally motivated by altruism.

At this stage, the public opinion definition of corruption comes into play in its second role – as the criterion of whether definitions conform to public opinion. Two criticisms of the public office definition are based on precisely this notion – that it violates public opinion.

One of the stock-in-trade criticisms runs as follows: ‘One does not condemn a Jew for bribing his way out of a concentration camp’ (Rose-Ackerman, 1978, p. 9). ³ The example is besides the point. Only the formal transgression by the official is corrupt: thus, only the guard is corrupt, not the prisoner. Moreover, the guard is only corrupt because he enriches himself in the process. Had he simply broken the rules and let the prisoner escape, the action would not be corrupt according to the definition. It is by no means obvious that the example proves that the public administration standard conflicts with public opinion.

The second widely touted criticism rejects the public office standard as culture-specific: upholding bureaucratic rules and regulations may be sanctioned by norms and conventions in Western societies but not in others (Bayley, 1966, p. 721; Scott, 1972, pp. 10–11). What the public condemns as a ‘corrupt act’ in the West today has not been so condemned in the past, nor is it generally condemned elsewhere. Two questions arise as a consequence: does this supposed chasm in the perception of corruption really exist (a question that will be dealt with in the second part of the paper); and what is the implication of this assertion even if it were true? The implication derives its relevance from the postulate of ‘cultural relativism’ not to impose standards on others, usually on non-Western societies, and thus questions the use of international comparisons based on ‘objective’ levels of corruption that rely on applying uniform standards independent of the ‘subjective’ perceptions of the inhabitants of the countries in the sample. However, even if standards diverge, a comparison of particular practices is still formally legitimate, even if the interpretation of those practices varies. Moreover, such comparisons do provide substantial information if, for example, the prevalence of certain practices such as administrative corruption has, or is suspected to have, uniform social, economic or political consequences.

The ‘legality’ attribute of this definition points to its most serious flaw: acts not illegal are not corrupt. This is true, for example, of influence peddling by political parties if not explicitly outlawed, and the legalisation of nepotism and bribery can largely free a country of corruption in this sense. From the legality rule, it follows that an identical act, depending on the state of legislation, can therefore be either corrupt or non-corrupt. This objection is of more than theoretical interest. The adoption of the OECD convention on combating bribery of foreign public officials in international business (OECD, 1997) took the non-corrupt act of bribing officials abroad into the realm of corruption in many countries. The public office definition is unable to deal with what may usefully be called ‘legislative corruption’.

One way of escaping the problem of the narrow scope of public office definitions is to extend the scope of corruption by defining it with regard to public interest: a

It was immediately pointed out that public interest definitions fail the operationability test: they ‘would find little acceptance’ because they presuppose an agreement on what constitutes ‘damage to the public and its interests’ (Scott, 1972, p. 3). Most importantly, however, as will become clear later on, the evolution of the term ‘corruption’ is more plausibly associated with violations of norms of distributional justice.

Johnston has recently claimed that the discussion on definitions has yielded major new insights – principal—agent—client (PAC) and ‘neoclassical definitions’ (1996, p. 525). He quotes Klitgaard approvingly: ‘This approach defines corruption in terms of the divergence between the principal's or public's interests and those of the agent or civil servant: corruption occurs when an agent betrays the principal's interest in pursuit of her own’ (1988, p. 21). However, if betrayal of trust becomes a sufficient condition for an act to be corrupt, then corruption becomes indistinguishable from cheating. Extending the range of corrupt acts in such a way is hardly a solution to the problem of defining corruption. In fact, the PAC approach is a method of analysis that deals with the interaction between a public official (the agent), the supervisor of the agent or ‘the public’ (the principal), and a private individual with whom the agent interacts (the client). It is not a definition.

Johnston's own attempt at forging a new definition – what he calls a ‘neoclassical’ approach – characterises corruption as ‘the abuse, according to the legal or social standards constituting a society's system of public order, of a public role or resource for private benefit’ (1996, p. 331). This amounts essentially to a combination of the public office and the public opinion standard. The public opinion concept enters because breaking social standards or social norms necessarily evokes public condemnation: ‘social standards’ regarding public order can only be ascertained by finding out what the public disapproves of. ⁴

We are thus left with a set of contradictory descriptions of the phenomenon of corruption, all of which have major disadvantages. Public office definitions lead to a serviceable heuristic that is adequate in many situations but unable to deal with legislative corruption. The public interest definition is neither operationable nor in line with ‘public opinion’. The public opinion definition suffers from the same lack of clarity, although it is the only concept that has underpinned actual empirical research involving comparisons of levels of corruption.

Is a state of affairs with multiple definitions to be deplored? In an ideal world this would certainly not constitute a problem, since, universally, all formal rules contribute to the public good and breaking them is condemned by the public. As it is, the three definitions do not generally agree and we are left with three approaches in uneasy competition. Subjective approaches to corruption in the form of the public opinion type that rely on ascertaining locally what is perceived to be corrupt will have the disadvantage of being relativistic, different in time and from society to society. Objective ones that define corruption a priori according to public interest or public office rules are criticised for being culturally insensitive by imposing a specific ‘cultural’ notion of corruption on the universe. On the other hand, only well-defined objective criteria allow international comparisons that go beyond measuring subjective impressions of different levels of corruption.

Ironically, much of the most fruitful research derives precisely from the tension between these definitions – the deviation of social norms from formal rules, of formal rules from public interest, or the asynchronous evolution of formal rules and social or moral norms. Questions arose as to why formal rules deviate from social norms, what the norms and values relating to corruption among members of the public precisely are, and regarding the connection of ‘public office corruption’ with the public interest. Far from hampering the research effort, the lack of a unified definition has positively stimulated it. So why reopen the discussion?

An Alternative Definition of Corruption

I suggest that an alternative approach based on the public opinion definition offers a better understanding of the phenomenon and additional scope for empirical research. Corrupt acts, I argue, are characterised by a holder of public office violating non-discrimination norms in order to gain a private advantage. Such a concept of corruption is thought to exist in any society of even moderate complexity. Public office norms based on such non-discrimination rules – and the impartiality principle that sustains them – are well-nigh universal.

The mainstream view of corruption denies the existence of strongly held specific public office norms – and implicitly, the distinction between public and private – in a wide range of societies:

What is the evidence for this supposed absence of distinction – in Rose-Ackerman's version ‘sharp distinction’ – between private and public?

In the case of Third World countries, the evidence invariably turns out to be equivocal. Linda Beck's illustration of the ‘patrimonial nature of accountability’ is a case in point. It deals with the LONASE scandal involving the skimming off of large sums of money from the Senegalese lottery:

What does this story prove? The villagers were clearly upset that their dealings were made public. But is this really an instance of the ‘patrimonial nature of accountability’? Or are the villagers perfectly aware of the public—private distinction and, since they presumably have benefited from the largesse of their well-connected patron, is their reaction caused by an awareness of their own moral dereliction that has just been publicised? Both interpretations are possible.

One may want to pause for a moment and look at exactly what the thesis of the ‘patrimonial nature of accountability’ implies. Assume two transactions involving private persons A and B and a government official C. In the first transaction, A receives (illegally) $100 worth of government resources from C, who happens to be his relative. B, in identical circumstances, receives nothing. Now take a second transaction: A has received a gift worth $100 from the private income of relative C (say, in the form of garden produce planted by C's wife). B, socially more distant, receives nothing. The cultural relativist position suggests that both A and B view these transactions as identical – as a simple gift exchange between A and C. The fact that government resources happen to be involved in one transaction is irrelevant for the assessment of the nature of the exchange. To the best of my knowledge, the literature of cultural relativism provides not the slightest piece of evidence that this is the case. That A – just like the villagers in Beck's example – is likely to engage in special pleading to justify the illicit extraction of government funds is to be expected and not necessarily evidence for the absence of the public—private distinction. The person who has just transgressed social norms is unlikely to be their unbiased interpreter. The people whose opinion would also have to be sought are those left out of the exchange. And they may well disagree with the thesis of the ‘patrimonial nature of accountability’.

On the contrary, far from being untainted by the public—private distinction, Nigeria's population, for example, does seem to support the re-establishment of public roles after a corrupt political elite has shown no inclination to fulfil these obligations. To popular acclaim, the Bakassi Militia, which, after a venal state elite had appropriated public offices as private sinecures and manifestly defaulted on their obligation to enforce public order and secure property rights, has taken over the role of police and judiciary, however rudimentary (NZZ, 2001a; Harnischfeger, 2003).

Evidence adduced to demonstrate the absence of the public—private distinction in pre-modern states is equally unconvincing. The missing distinction is generally illustrated by the failure to distinguish between the private and public expenditure of a ruler. Corruption in the form of the private misappropriation of public funds by the ruler is, in the absence of this distinction, impossible by definition. However, it does not follow either that a ruler could use government resources for any conceivable purpose without violating some public office norms. What is certainly true is that the forms of corruption have changed. The fundamental principle that throughout European history a sphere of public action existed where special norms applied is thereby not disproved.

Not only is there scant evidence for the alleged absence of the public—private distinction in pre-bureaucratic states in the West and in non-Western societies, there is little doubt that this distinction is ancient and fairly universal. In his admirable work, Noonan traces the concept to the Middle East, where in Mesopotamia and Egypt ‘from the fifteenth century B.C. on, there has been a concept that could be rendered in English as “bribe”, of a gift that perverts judgment’ (1984, pp. 13–14), although without the emergence of verbal counterparts to the purely negative terms like ‘bribe’ or ‘corrupt’ and without specific formal sanctions evolving. Bribery and corruption, he shows, are notions that have been with us since antiquity and have been debated in state (for example, Rome) and (Catholic) church ever since. Thus, the evidence is on the other foot: the public—private distinction is ancient and not specifically Western. ⁵

What do these specific public office duties imply? Noonan demonstrated the close link between the evolution of the concept of corruption and the role of the judge (the archetype of a ‘public official’) – a role that calls for some degree of ‘impartiality’, a behaviour constantly threatened by conventional gift-giving practices, concessions to personal proximity and personal advantage. Such ‘judge-like’ public roles from which some degree of impartiality is expected are likely to be found in most societies. Even in an acephalous society like that of the Nuer, Evans-Pritchard (1940) documented the existence of the leopard-skin chief who operated as a mediator in personal and tribal conflict, although without formal power of office and without sanctions at his disposal. The private—public distinction, however, was clearly implied by the role of mediator (p. 163), a function whose execution requires some degree of impartiality.

The principle of impartial action, inseparable from such functions as tribal mediator or judge, it seems, is understood everywhere as an element of the special duty of public office, even in relatively undifferentiated societies. Thus, upholding this principle and judging infringements of it by holders of public office must have been part of ordinary social life too. Philp's example of Samuel Pepys using his public position to extract sexual favours from the wife of an applicant in exchange for promises of employment is surely a case in point (1997, p. 27). Pepys, as his occasional qualms show, was perfectly aware that he was breaking public office norms when promising to discriminate in the allocation of jobs on the basis of sexual favours received, a practice as unacceptable in the seventeenth century as it is today.

What does it mean to talk of ‘impartiality’? The concept belongs to the realm of distributive justice and embodies the idea that those who are equally deserving ought to be treated equally. This still leaves empty the set of actions covered by it. Nevertheless, it advances the argument one step further: the close connection of the idea of corruption to the impartiality principle suggests that the two are inseparably bound together. At its most general, then, corruption involves a holder of public office violating the impartiality principle in order to achieve a private gain.

Such a view of corruption explains what, at first sight, might be surprising: the term is readily understood worldwide, but at the same time is interpreted differently. The apparent paradox is simply the result of the impartiality principle having no specific content: who deserves equally remains undefined. Public officials and state regulations discriminate all the time without causing offence – between those allowed to drive a car and those who are not, those who are in and out of prison, those who go to university and those who are not allowed to attend. Such types of discrimination are ‘allowed’, based on criteria judged to be ‘relevant’ in the particular case that ‘merits’ discrimination.

Thus, operationability requires that the impartiality principle be fleshed out by specific rules based on what might be called non-discrimination norms that prohibit certain types of discrimination. In other words, these are the rules that specify the categories of cases that deserve equally and ought to be treated equally, and whose unequal treatment constitutes a case of corruption. Thus, non-discrimination norms relate to the distribution of rights and obligations – who gets what, when and how – and provide bounds within which this distribution is to be confined if it is to pass as morally acceptable. Typical non-discrimination norms prohibit public officials from discriminating in favour of those who are socially close (nepotism) or who are able to pay the highest price (bribe). More generally, corrupt acts can now be defined as violations of non-discrimination norms governing the behaviour of holders of public office that are motivated by private gain.

This approach explains why, even without specification of particular non-discrimination norms, administrative corruption is widely resented: administrative rules are, however imperfectly, the embodiment of the impartiality principle and generally reflect non-discrimination norms. Thus, public office rules prevent dis-crimination between applicants for drivers' licences on the grounds of political adherence, religion, social distance or the ability to pay market clearing rates or bribes. Insofar as these rules do embody norms of non-discrimination, their breach is indeed condemned by the public. This is not to say that there are no other reasons unrelated to the impartiality principle for why breaking such rules might be unpopular. There are obvious public benefits to be derived from following public office rules. A stable system of property rights that a rule-bound administration provides reduces uncertainty and is a precondition for long-term planning, for functioning capital and credit markets and, ultimately, for economic development. Thus, economic well-being is closely related to the degree of predictability of the decisions of public administration and the courts. Indeed, this is where the public office and the public interest definitions meet: on the whole, and at least in the long run, endemic violations of public rules violate the public interest. Nevertheless, it is unlikely that people who condemn administrative corruption are motivated primarily by such long chains of utilitarian calculus. Notions of ‘distributive justice’ or ‘fairness’ or ‘unfair discrimination’ more plausibly actuate the feelings of those who condemn corruption. It would also explain the need for special pleadings of those violating the impartiality principle, as in the case of the villagers of the LONASE scandal, who probably knew very well that they had benefited from ‘unfair’ dealings, and, by benefiting from breaking administrative rules, violated some deeply held norms.

What non-discrimination norms are likely to emerge? It is convenient to follow the conventional distinction between the ‘input’ and the ‘output’ side of political processes, and hence between corruption relating to the nature of the political process itself (the ‘input’) and ‘corrupt’ legislation (the ‘output’).

Political corruption has caught the attention of the World Bank in the context of the transition economies of Eastern Europe. It announced the ‘introduction of the concept of state capture’, which is nothing but ‘political corruption’ of old:

Thus, a distinction is made between ‘legitimate’ and ‘illegitimate’ influence – but what distinguishes the two? That, of course, is precisely the question that definitions of corruption try to address. Although the World Bank authors do not deal with the issue explicitly, they operate with an implicit assumption – that there are some unspecified non-discrimination norms regulating the access to the political process, and violations of these norms are ‘illegitimate’ or corrupt. As any society of any complexity is likely to know some such rules, the problem of political access corruption of this kind will, again, be a fairly universal one. In a modern environment, violations might range from ‘donations’ to politically influential individuals or political parties in return for legislative favours to the corporatist structures of, for example, the Austrian Filz (‘felt’) – the personal networks of politicians, bureaucracy, state enterprises and rent-seeking industry that became widely resented by the public and contributed to the rise of its critic Jorg Haider – to lobbying that takes on shades of corruption in the eyes of the electorate.

Such political access corruption has been concerned with discrimination at the input stage of the political process: legislative corruption proper relates to the design of the rules and regulations themselves. Since corruption at the input stage generally aims at influencing the rules and regulations, it ought to manifest itself as legislative corruption. Political access corruption might therefore serve as a proxy for legislative corruption, because the latter is often unobservable or difficult to assess. One might debate the character, for example, of the recent decision by the US Justice Department not to continue to pursue its case against Microsoft, a major contributor to President Bush's election campaign. It is unlikely that the officials felt a sudden change of mind after the presidential election and found that Microsoft had not, after all, repeatedly abused its monopoly power, as even a conservative Washington DC appeals court had unanimously found (The Economist, 2001a). Taking political access corruption as a proxy for legislative corruption might clarify the issue: undue influence may have occurred, even if it is unclear that the decision itself was discriminatory and corrupt. ‘Input’ measures stand in the usual uneasy relationship to ‘output’ measures. It is perfectly conceivable that political access corruption might prevent legislative corruption: legislators are ‘bribed’, for example, not to engage in legislation that breaks non-discrimination norms. As in the much-discussed case of administrative corruption, such outcomes are primarily of academic interest.

Regulating the anti-corruption agencies opens a further field where norms regarding legislative corruption might apply: the deliberate design or the weakening of procedural rules in order to reduce access to information on norm infringement, or to reduce the punishment for norm transgressions. The problem of agreeing on norms, and actions regulated by them, is complicated by the existence of an optimal level of enforcement. Honourable reasons might generally be advanced for not waging wars on all forms of corruption (Anechiarico and Jacobs, 1996). Is, for example, the recent non-renewal of the commissioner for parliamentary standards in Britain, who apparently has ‘shown too much zeal in her pursuit of government ministers suspected of breaching parliamentary rules’ (The Economist, 2001b), a sign of corruption of the Blair government? Where does the Italian public locate Prime Minister Berlusconi's laws that increase the burden of proof required for a successful prosecution for corruption to such a degree that it amounts de facto to an amnesty for the past misdeeds that this very prime minister and many of his cronies are accused of (NZZ, 2001b)? Less debate will probably be aroused by legislative action that promotes administrative corruption for the benefit of the political elite.

The core of legislative corruption concerns violations of non-discrimination norms in the allocation of rights and duties, benefits and obligations. At least in today's Western societies, candidates for identification as corrupt legislation include discrimination according to social proximity (such as nepotism, including discriminating in favour of the public official himself), religion, social or political class, caste, tribe, race, sex, ethnic characteristics and, at least within broad bounds, political views. Legislation violating such non-discrimination norms will generally, after exceeding some boundary of tolerance, be considered corrupt. In some cases, wide concurrence prevails: few will defend discrimination in favour of the old nomenclature in Eastern Europe during the privatisation process. The assessment of where special privileges to legislators exceed the ‘acceptable’ – where they begin to draw resources beyond what is available to other citizens in equal circumstances – will be less clear cut. When do excessive salaries and perks begin to be considered corrupt? Take, for example, the case of the German party foundations, where public resources fund the employment of people with close party ties and dish out bursaries to students? Has the funding of these foundations exceeded the point where a perfectly informed ‘public’ would consider it a case of legislative corruption? Is the Austrian Filz corrupt? The answer will obviously vary. Its beneficiaries presumably defend it as an unavoidable accompaniment of successful and desirable corporatism – but few outside that narrow circle would view the practice as wholly untainted.

From the point of view of the impartiality principle, even pork-barrelling might exceed the bounds of corruption. As a method of give and take in parliament, of compromise and balancing different interests, it is an essential if not indispensable tool of practical politics. In the case of many spending programs, it may indeed be unavoidable if expenditures are ‘lumpy’: there are only a limited number of military bases to go round and they must be positioned somewhere. However, there are other cases closer to the corruption borderline. What about an education system of the Brazilian kind, where access to the free state elite universities is greatly helped by having attended a costly private high school, a process that restricts access to high quality tertiary education mainly to the rich? Is the impartiality principle violated if a large part of the health expenditures of a country like Tanzania goes towards the maintenance of one hospital in the main city where it happens to serve disproportionately the minor members of the governing elite (those not able to escape to the First World for treatment)? Or, to take another blatant African example, where the tribal supporters of the ruling coalitions have been systematically favoured in the allocation of government resources – health, education and infrastructure (Mueller, 1984; Sandbrook, 1985, p. 71).

How does administrative corruption of the public office definition type relate to the impartiality approach? Levels of corruption, according to the public office definition, will generally be judged by such indicators as the prevalence of bribery to avoid taxes and regulations and the irregular use of funds by public administrators. If the public office rules that forbid such practices embody public office norms, administrative corruption is simply a subset of the broader view of corruption suggested here. Despite the tension between public office rules and non-discrimination norms, there is little doubt that public office regulations do, in many cases, reflect such norms. Moreover, since there is a strong presumption that arbitrariness in the allocation of rights and obligations increases rapidly with the degree of violations of public office rules, the level of administrative corruption will be highly correlated with violations of non-discrimination norms. Thus, analysis of the public opinion definition based on the impartiality rule suggests, as has been pointed out already, that this definition is better than its reputation. Measuring corruption by violations of public office rules therefore seems to be a sensible heuristic.

The argument that has been developed has yielded a definition of the public opinion type: prevailing norms determine when a particular non-discrimination rule is broken and a corrupt act has occurred. As such, it suffers from all the problems of public opinion definitions – particularly the difficulty of agreeing on the boundaries of corruption. If, however, it is accepted that the definition of corruption ought to embody the meaning as it has evolved over the centuries, this problem is indeed unavoidable — corruption is breaking norms. Moreover, the definition does have some compensating advantages. Most importantly, it explains why the concept of corruption is universally understood (and thus not culture bound) and why administrative corruption is so widely resented.

What are the implications of this approach for measuring corruption? A definition has nothing to contribute to the debate on the reliability and precision of measures of corruption: it is exclusively concerned with the scope of the actions covered, and thus with the validity of the measurement — does it measure what has been nominally defined by corruption? Ideal measurements of the level of corruption generally capture the incidence of all practices that are corrupt according to the definition – all actions that violate non-discrimination norms. The validity criterion has implications for the discussion of private-to-private corruption, for the scope of legislative corruption and for the cultural relativity problem arising when corruption is measured.

Does corruption include actions within the private sector? That the term ‘corruption’ encompasses private-to-private transactions has been suggested by a number of participants in the debate (Transparency International, 2003). ⁶ Since a holder of public office does not necessarily have to be a politician or government employee, actions of employees or managers of a private company can, in principle at least, be corrupt if they violate particular public office norms. Corruption then becomes a phenomenon of the private sector, too. However, the set of corrupt acts cannot include any norm violation that occurs in the private sector: it must be one that goes beyond the obligations of private morality of keeping promises and contracts, and generally acting in good faith.

An extension of the concept of corruption to the private sector can be useful in cases where private companies have been delegated public office tasks involving non-discrimination norms. The case of privatised prisons may serve as an example. If an employee of such a prison were to accept payments in exchange for favourable treatment, that might indeed constitute a case of corruption. What transports the action into the realm of corruption is the violation of widely shared non-discrimination norms by an agency that has taken on public office functions. The concept of corruption, according to this view, ought to be limited to cases of this kind and not extended indiscriminately to the private sector.

If surveys try to assess the level of corruption generally, they need to cover corrupt practices comprehensively if they are to be valid. The impartiality rule suggests a range of additional practices that ought to be part of a comprehensive survey but are not generally included in current surveys. Corrupt acts have been categorised into two broad areas: administrative and legislative corruption. The latter category has been broken down into corruption influencing the policy-making process itself (‘input’) and that influencing the outcome (‘output’):

Although existing surveys that focus on practices (as opposed to the level of ‘corruption’ generally) do cover both violations of administrative rules ⁷ and corrupt influence on legislative processes, this is not the case with the ‘output’ side of legislative corruption. None of the surveys contain questions on the quality of legislative measures and programs (or the political will of the government) to contain corruption – the power and independence of the auditor general or the judiciary, or legal impediments to whistle-blowing (such as the nature of whistleblower legislation and libel laws). None of them looks at distributive justice in terms of privileges granted to narrow special interest groups or to the political elite itself. However, if it is indeed accepted that corruption is a question of violating the impartiality principle and the non-discriminations norms associated with it, then overstepping certain bounds in providing legal privileges to particular groups and in failing to constrain the spread of corruption will be perceived as corrupt. Thus, the new look on corruption does suggest a broader and richer concept and comparison.

The cultural relativity problem arises because non-discrimination norms are culture-specific. Who is equally deserving, or, putting it differently, where discrimination is forbidden, may vary from society to society. How seriously comparative studies are affected by the cultural relativity problem will depend on what precisely is measured and compared. It is useful to distinguish three types of comparisons: comparison of the different kinds of practices that are considered corrupt at different times and places by different sections of the population; comparison of beliefs about the causes, extent and consequences of corruption; and comparison about actual levels of corrupt practices.

The study of corruption is, in the first instance, a comparative study of different practices that public opinion considers corrupt in different contexts. Such a comparison makes sense if the concept of corruption is universally understood as a violation of public office norms, and, if this argument is correct, more specifically of the impartiality principle. Methodologically, such studies follow the lines that have been taken, for example, by the World Values Surveys (Inglehart, 1997, 2003). Such comparisons ought to be able to solve the issue of whether the suspected systemic differences in public opinion on what constitutes corruption do exist, and, if they do, whether they exhibit the pattern that is often postulated – that they are related to economic factors such as poverty or ‘cultural’ factors such as extensive family or kinship systems. ⁸

The second type of comparison compares beliefs about corruption. Of interest, therefore, is not primarily the frequency of the occurrence of certain practices, but beliefs among different strata of society about the pervasiveness and importance of corruption – however defined by the respondent – in political decision-making and the allocation of government resources. The interest of such studies derives largely from the influence these beliefs have on government and regime legitimacy. ⁹

The third type of measurement attempts to capture actual levels of corruption. Since these actual levels are unobservable, surveys are generally used as a proxy. ¹⁰ Such ‘expert’ polls will ideally encompass those who have cognisance of corrupt practices. Here, the cultural relativity problem comes into play: these comparisons involve standards of measurement according to a set of predetermined formal and informal standards that ignore local norms. This becomes a problem primarily where surveys do not specify what is meant by corruption and respondents are asked unspecified general questions – to assess the ‘level of corruption’ or whether ‘bribing and corruption prevail or do not prevail in the economy’. Evidently, this procedure elicits culture-specific responses. This is one of the weaknesses of Transparency International's corruption perception index, which is based on many questions of this kind (Lambsdorff, 2002). ¹¹ Other indices, such as BEEPS (the business environment and enterprise performance survey), avoid this problem by focusing on specific practices: the sale of parliamentary votes, laws and presidential decrees to private interests; the sale of court decisions; central bank mishandling of funds; contributions paid by private interests to political parties and election campaigns; patronage defined as public officials appointing their friends and relatives to official positions; and bribes paid to public officials to avoid taxes and regulations (Hellman et al., 2000, p. 21). ¹² As long as these practices are interpreted uniformly, the cultural relativity problem does not enter at this stage. Whether the practices are indeed condemned universally can now be treated as a conceptually separate question to be analysed with the help of studies about attitudes to corruption. Thus, even here the problem of cultural relativity is of limited importance as long as comparative studies focus on the incidence of reasonably well-understood practices for which there is a priori evidence that they break non-discrimination norms.

Conclusion

I have argued that corruption is a postulate of distributional justice and thus a relational concept. Public office rules will, even in moderately complex societies, impose certain restraints on the behaviour of public officials – restraints that are based on the impartiality principle. The impartiality principle, in turn, is embodied in non-discrimination norms that have a long history and are not confined to Western countries. As corruption has always been closely tied to the violation of such norms, corruption is a widely shared notion with an ancient past, too. Public officials act corruptly when they violate non-discrimination norms that regulate the allocation of a polity's rights and duties in order to derive a personal advantage. Whereas the concept of corruption is one that is well-nigh universal and thus hardly suffering from cultural specificity, its content – the specific non-discrimination norms — is not; what in practice will count as corrupt will depend on prevailing norms and conventions. This explains the paradox that although the term ‘corruption’ is readily understood and readily applied everywhere, there is hardly any agreement on where precisely the boundary between a corrupt and non-corrupt act should be drawn. The focus on the impartiality principle leads, moreover, to an upgrading of the public office definition precisely because public office rules are, to a large degree, the embodiment of the impartiality principle.

An approach to corruption that puts the principle of impartiality at the core of its meaning arrives at the conventional candidates for the measurement of corruption that are suggested by the corruption literature: violations of norms relating to political influence (political access corruption) and, with some reservations, administrative corruption. However, it goes beyond the conventional notion because of the generality of the non-discrimination principle, which includes in its scope any public action involving violations of non-discrimination norms, and thus, at least potentially, such dubious practices as pork-barrelling and log-rolling, or discriminative allocation of public benefits and obligations generally.