What is corruption? How do we define it? Is it a bad thing? More precisely what are its welfare implications? What are the causes, direct and indirect, of corruption and how, using this knowledge, may we design policy to ensure that corruption is at its optimal level for welfare?
These are the questions that interest economists in corruption. This essay will examine some of these issues with an eye less to comprehensiveness than to attempting to provide a coherent organizational framework for analysis.
We may want to add in another agent to represent the monitoring and enforcement arm (E) of the principal. This is a two agent model with one agent charged with monitoring the other. For example, an internal affairs division to 'police' the police. This may introduce interesting questions of corruption not only in the monitored, but in the monitoring agent and of how the two sets of agents interact.
We could also introduce more detailed modelling of, and the sanctions available to, the principal. Consider the situation where the citizens are the principal with the government as the agent. Here we encounter many issues arising from the literature on 'free riding' and related problems.
For example while all citizens may agree that they wish to do something about corruption it may be very difficult to coordinate together to effectively sanction the agent (e.g. by voting the current government out of office). In many of the worse cases of corruption this appears to be the central problem. Governments, even those which in theory are democratically elected, are sometimes more like parasites than agents. Moreover the government in these situations possesses most of the effective tools which could be used to monitor it. This is, of course, the reason for the famous separation of powers - that is the division of government up into executive, legislative and judicial branches.
[acemoglu_2000] has two types of bureaucrat with different probabilities of being detected. [bannerjee_1997] has a continuum of bureaucrats whose cost/benefits for corruption vary. We might also allow bureaucrats to interact, or, for the customer to interact with several bureaucrats. The first possibility allows us to investigate the effect of policies, such as that used in Singapore to reduce corruption in the Customs department, where bureaucrats are paired together when interacting with a customer. The second allows us to investigate models of centralized vs. decentralized corruption [shleifer_1993].
Alternatively we might allow for a 'corruption externality': the (net) payoff to corruption is an increasing function (see [bardhan_1997:1331ff.]) of the percentage of corrupt bureaucrats (and honest bureaucrats' net payoff is an increasing function of the percentage of honest bureaucrats). For example: the probability of detection of corruption declines - or the cost of monitoring to increase - with the number of corrupt bureaucrats (safety in numbers).
Finally corruption can be related to the repeated interaction between bureaucrats and customers. For example suppose there are two sides to the corrupt transaction the bribed, the bureaucrat (B), and the briber, the 'customer' (C). We have a large group of B and of C who repeatedly interact in transactions (so a given B interacts with a different C each time and vice versa). A party will act corruptly/honestly in a given transaction if they believe the other party to be corrupt/honest (respectively). After each transaction the parties perform a Bayesian updating of their belief that the next agent they meet will be corrupt. This setup will permit the existence of multiple equilibria, in particular a totally corrupt and a totally honest equilibrium in which all agents of both types always act corruptly or honestly (respectively).
Provision of micro-foundations is essential for two reasons. First, so that we can understand the existence of the principal-agent relationship in the first place (e.g. why do we need that bureaucracy?). Second to allow us to perform a welfare analysis, that is, crudely, to trade-off the costs of the agent's corruption against the benefits for the principal.
For example, consider the usual case where the government is the principal. Much corruption results from the creation of rules and regulations. Why are these rules and regulations created? Why is government intervention necessary at all? First, and most fundamental: for the market to function at all requires a set of basic institutions to exist which the government must provide. For example, ensuring some security of property - necessitating the creation of a military, primarily for external threats, and also of civilian police forces. Laws to govern contract must be created and enforced thus necessitating a judiciary as well as other auxiliary bureaucracies.
Second, the existence of market failure may necessitate government intervention. Modern economics provides a whole set of examples of such failure not just confined to traditional staples such as the regulation of externalities and the provision of public goods but those stemming from imperfect information and other market imperfections.
This is also an important consideration in evaluating the 'efficiency approach' which presents corruption, counter-intuitively, as beneficial. It 'oils the wheels' of government or provides a substitute price mechanism (for example in the allocation of priority in a queue). However these results implicitly assume that either i) the principal has an incorrect utility function ii) that the principal is incorrectly maximizing its utility function (and therefore using an incorrectly designed mechanism).
Take the queue example. Allowing customers to bribe the queue manager minimizes the total discounted queue waiting time (see [bardhan_1997:1323]). However this result only holds if queue members are not credit constrained (it also ignores problems that might exist in the bribe relationship - e.g. how do we know the bureaucrat will allocate positions 'honestly' in line with the bribes). But surely the reason governments implement resource allocation by queue rather than by price is precisely because of equity issues stemming from credit constraints on poor members of society. The equity/efficiency trade-off derives precisely from the existence of constraints or from the use of utility functions in which 'fairness' is an explicit argument. For example the principal may possess a utility function that contains not only the utility of all citizens but also a measure of societal inequality such as the gini coefficient1.
An example of a model that incorporates both of the two points is [acemoglu_ea_2000]. By incorporating a production externality that bureaucratic regulation helps solve they explicitly model 'The Choice Between Market Failures And Corruption'2. At the same a general equilibrium approach allows them to endogenize government revenue and hence the ability to sanction bureaucrats through their wages.
The 'macro' approach covers studies or models that look more at the global structure of corruption in a state (and may not provide micro 'foundations' for the macro regularities at all). A particular example is [shleifer_ea_1993]. They offer an explanation of why, somewhat paradoxically, countries can have the same overall dollar amount of corruption but yet suffer very differently economic and social consequences.
The answer they present is based on a typology, borrowed from the industrial organization literature, of single monopolist vs. independent/complementary monopolist corruption. Some societies while being corrupt at least possess (centrally) regulated and ordered corruption. Classic examples of this are Communist Russia, South Korea and various autocratic states. In these cases while bribes are necessary the transaction takes place in a world of almost complete contracts and the costs associated with working out and managing the bribing process are minimized. In contrast in situation of many independent corrupt administrators an agent wishing to carry out business must come to an agreement with each of them in turn as has far less security that he will not have to pay more bribes to as yet unknown power-holders or that those already bribed will not come back for more.
This raises a point neglected in the literature: that corruption, at least in its more extreme and debilitating form, is more a symptom than a cause. A symptom of the underlying failure of the state. This has two implications. First, that other issues, perhaps even more serious for the general welfare, are likely to be the result of this situation: lack of secure property rights, a poor or non-existent legal system, poor or non-existent mechanisms for non-violent (political) settlement of disputes between social and ethnic groups. Second, that attention should be focused not on corruption and the precise mechanisms by which it comes about but on the solving the underlying failure; the failure of the state. At the most basic level states and governments are solving a coordination problem in the provision of a public good (whether by compulsion [olson_1993] or by consent - the social contract). The greatest disaster to a polity is not corruption but anarchy and disorder.
Countries appear to have very different cultural norms regarding monetary payments to government officials (even to the extent of whether payments should be considered corruption or simply a normal part of a business transaction).
Moreover important factors relating to the level of corruption, for example the independence of the judiciary, seem tohave less to do with formal rules and much more to do with informal (or uncodified) sets of norms and traditions. Consider the example of England. In theory the judiciary is controlled by the executive branch of government in the form of the Lord Chancellor. Nevertheless for the last several hundred years, and continuing right up to the present, the judiciary in England have exhibited a degree of independence both from the monarch and the government unparalleled by that in most other countries and, in particular, often far exceeding that of states where judicial independence is provided with much greater formal guarantees.