Acknowledgements: This paper could not have been produced without the guidance and support of Randi Ryterman and Jim Anderson of the World Bank; the assistance of Samjhana Thapa, Fang Rong, Aziz Ahmad, and Michael Radosh, all students at the University of Maryland; and the expert input of Melissa Thomas, Diana Rutherford, and Gabriela Mossi of IRIS. Deserving of special recognition are the authors of the three original case studies that served as source material for part 4 of the paper, namely Luigi Manzetti of Southern Methodist University, Teresa Benedict of Transparency International/ Malaysia, and Haji Semboja and James Kajuna of the Economic and Social Research Foundation, Tanzania. Any faults are the author’s own.
Table of Contents
1. Introduction 1
Structure of the paper 1
2. Researching and Assessing Anti-Corruption Agencies 3
Defining and Measuring Success 4
Towards a Workable Set of Performance Indicators 8
Explaining Success 10
3. Previously Documented Cases 14
The Single-Agency Paradigm: Hong Kong and Singapore 14
Responsibilities and Powers 16
Safeguards and Relationships 18
Performance Monitoring 21
Summary and Assessment 23
Variations on the Single-Agency Model 24
Establishment, Responsibilities and Powers 24
Safeguards, Relationships 26
Performance Monitoring 30
Summary and Assessment 32
The Alternative: Multiple-Agency Approaches 33
The U.S 33
The EU and Others 37
4. Original Case Studies 40
Responsibilities and Powers 45
Safeguards and Relationships 50
Performance Monitoring 62
5. Assessment and Conclusion 67
Explaining Success and Failure in the Mission 67
Performance Measures: 67
Explanatory Factors: 69
Annex 1: Tables 84 Annex 2: Agencies and Their Structures 100
Annex 3: Agency Enabling Acts and Enforcement 122
Annex 4: Research Protocol 127
Abbreviations ACA Anti-Corruption Agency (in general, Malaysia)
CCCC Comision de Control Civico de la Corrupcion (Ecuador)
CCM Chama Cha Mapinduzi (governing party in Tanzania)
CPIB Corrupt Practices Investigation Bureau (Singapore)
CVC Central Vigilance Commission (India)
DCEC Directorate on Corruption and Economic Crime (Botswana)
DG Director General
DOI City Department of Investigation (New York)
DOJ Department of Justice (US)
DPP Director of Public Prosecutions
EPA Environmental Protection Agency (US)
FTC Federal Trade Commission (US)
TI Transparency International
IAS All India Service (India)
ICAC Independent Commission Against corruption (Hong Kong, New South Wales Australia)
ICRG International Commercial Risk Guide
IG Inspector General (US)
IGG Inspector General of Government (Uganda)
KICAC Korea Independent Commission Against Corruption
NACSAP National Anti-Corruption Strategy and Action Plan (Tanzania)
NBI National Bureau of Investigation (Malaysia)
NCCC National counter Corruption Commission (Thailand)
OECD Organization for Economic Cooperation and Development
OEP Oficina de Etica Publica (Argentina)
OGE Office of Government Ethics (US)
PCAGC Presidential Commission Against Graft and Corruption (Philippines)
PCB Prevention of Corruption Bureau (Tanzania)
PO President’s Office (Tanzania)
SCPC Service Central de Prevention de la Corruption (France)
SEC Securities and Exchange Commission (US)
SFO Serious Fraud Office (UK, Ghana)
SIGEN Sindicatura General de la Nacion – executive branch oversight institution (Argentina)
TOR Terms of Reference
1. Introduction The problem of how best to control corruption has challenged policymakers from the dawn of civilization. Strategies and institutional responses have varied, but in recent decades the approach of choice has increasingly become: establish an anti-corruption agency. This ostensibly straightforward nostrum actually poses a lot of difficulties. How much authority, and which specific powers, to give it? How big should the agency – and its jurisdiction – be? What should we expect of such an agency, and how will we know whether it has been successful?
As a result of the prevalent interest in anti-corruption agencies, and the many questions they raise, the World Bank determined the need to assess the experiences of these institutions. The present paper responds to that need. In it, we review experiences with a wide range of agencies from around the world, in both industrialized and developing countries, and exhibiting a variety of designs. In recent decades, Singapore and Hong Kong have provided the impetus, and the paradigm, for the establishment of strong, centralized agencies in this field. Other countries, such as the U.S. and India, have chosen the alternative strategy of strengthening anti-corruption capabilities in government, but spreading these powers across multiple agencies. These differences of approach can have important implications, although background conditions such as effective legal institutions appear to be more critical. Given the level of interest in centralized or single-agency approaches, our focus lies there. We both analyze existing documentation on anti-corruption agencies and present three new, original case studies.
In assessing these experiences, determining success poses knotty problems. Many agencies’ missions are broadly defined in terms of reducing corruption or changing values. Measuring an agency’s impact on the level of corruption, let alone public values, appears impossibly problematic. Even where objectives are more concretely defined, the data on agency outputs, and especially on intermediate outcomes, are spotty. Keeping these difficulties in mind, we review the experiences of anti-corruption agencies using the tools at hand. In doing so, we attempt to provide practical guidance to the World Bank and member states who are considering the establishment of such an agency. We also attempt to identify promising strategies for developing a more rigorous assessment framework over the long term.
Thus, the paper addresses these questions: What lessons can policymakers learn from the record of experience with anti-corruption agencies? What appear to be the principal components and determinants of agency success? What is needed in order to make more meaningful comparisons possible?
Structure of the paper The following section addresses the main informational and methodological problems involved in assessing the performance of an anti-corruption agency (ACA). In that part, we review the extant literature on those agencies, which proposes numerous models, features, and lessons of success. This review points to some hypotheses about the determinants of ACA success, as well as the difficulty of clearly defining and measuring it. Rather than attempt to measure overall impact, which cannot be done with sufficient rigor, we look to the performance measures actually used by some ACAs and comparable agencies.
The bulk of the paper (parts 3 and 4) consists of “biographies” of illustrative anti-corruption agencies. The information used here comes from several sources. Based on a review of extant material on ACAs, the author, jointly with the World Bank staff overseeing this study, created a detailed research protocol (Annex 4). Researchers in three countries (Argentina, Malaysia, and Tanzania) used the protocol in collecting documents and interviewing relevant persons both in and outside the anti-corruption agencies in those countries (see part 4 below). In addition, IRIS researchers used the protocol as the basis for research and for queries by phone and e-mail in several other countries. We also used both studies of well-known ACAs and the official documents and websites of those agencies. In this way, we compiled information on the history and operations of these agencies, including technical, political, budgetary, personnel, and evaluation issues. We present a synthesis of this information, using the protocol to structure the discussion.
In part 5, we provide an analysis of the main findings from the cases, along with our conclusions and recommendations for the World Bank, member countries, and researchers in this field. Detailed case study information and tables appear in the annexes.
2. Researching and Assessing Anti-Corruption Agencies In taking up this subject, one first needs to develop a workable definition of an anti-corruption agency (ACA) for purposes of study. Every society has some institutions and procedures having as part of their mission the prevention, detection, or punishment of corruption. These include criminal laws, court systems and prosecutors, inspectors general, supreme audit agencies, civil service codes and hierarchies, frameworks of administrative law and freedom of information, and ombudsman institutions. What we focus on here, by contrast, are agencies (and their accompanying systems of rules) that have been established primarily as means of combating corruption – at least as advertised. This, too is quite a broad category, since it potentially embraces a host of institutions such as free-standing agencies; specialized sub-units, of which there are many in ministries of finance and justice; ad hoc panels and prosecutors; and existing institutions (e.g. courts and prosecutors) with a new mandate and enhanced anti-corruption capabilities.
In this study, our attention focuses on permanent agencies whose primary function is to provide centralized leadership in one or more of the core areas of anti-corruption activity – including policy, analysis and technical assistance in prevention, public outreach and information, monitoring, investigation, and prosecution. This leaves a range of other relevant entities outside the scope of consideration – or only briefly touched on. Even somewhat narrowly defined in this way, anti-corruption agencies are numerous and their ranks are growing. They respond to a variety of concerns and crises, and are modeled on a mix of organizational forms. The World Bank (1999) proposes a somewhat more restrictive definition, describing the ideal ACA as a body that: reviews and verifies official asset-declarations; carries out investigations of possible corruption; and pursues civil, administrative, and criminal sanctions in the appropriate forums. This definition – really a recommended model – identifies several of the core features of ACAs that distinguish them from other institutions, but leaves out some of them (e.g. information and outreach, analysis and technical assistance) and at the same time is too specific (e.g. including litigation and direct responsibility for asset declarations) to accommodate the range of agency forms now in use. Hence, we prefer a somewhat looser definition.
The history of ACAs appears not to be known in much detail in the international community of anti-corruption specialists, particularly those working in developing and transition countries. This history is frequently presented as starting with the establishment of the commission in Singapore in the 1930s, its restructuring in the early 1950s, or even the Hong Kong bureau founded in the 1970s. In fact, a quite similar model began operations in New York City in the 1870s. The relatively late arrival of such institutions derives in part from the wide recognition of corruption as an important dysfunction of public administrations only in the 19th century. A further influence was the general tendency, until well into the 20th century, to address corruption in the form of discrete scandals and by means of ad hoc commissions. (see Johnston 1999) Indeed, as the next sections of the paper will show, special anti-corruption mechanisms are still predominantly ad hoc or temporary in many countries, especially in Europe and North America. Most countries address corruption using multiple agencies, rather than concentrating powers and leadership responsibility in a single agency. However, the centralized or single-agency approach is sufficiently widespread as to merit careful study.
Defining and Measuring Success
What are anti-corruption agencies expected to do, and how does an observer know if such an agency has been successful? In general, ACAs are set up as a political response to scandal, and provide a mechanism for political leaders to reassure voters and reformists that action is being taken to bring corruption under control. The extent to which the objectives of a new agency reflect a desire for systemic change, as opposed to a drive to score political points, is rarely clear – and is rarely intended to be clear. This poses a dilemma for the observer in defining and measuring success. What yardstick to use? The measurement of political gain would be the easiest: how well does the reforming leader or party do in subsequent elections? However, in this paper we are more interested in the reformist objective, which is to constrain corruption. Political gain might factor in, for example, where poor performance by an ACA coincides with significant gains to the founding party – thus hinting at an empty political gesture.
If the agency’s objective, then, is to reduce or constrain corruption, how is one to identify and measure success in these terms? In the literature on ACAs, these issues are not often confronted, and appear in most cases to be assumed away. Kaufman (1998: p. 66) captures the problem as follows:
Often, mistaken conclusions are derived from analysis devoid of proper counterfactuals and controls. Ascribing success to anti-corruption watchdog bodies in Botswana, Singapore and other heralded cases by focusing on the details of the watchdogs themselves, without considering the impact of fundamental reforms in the broader environment, is on example of this bias.
Linked to this “anti-counterfactual bias” are several other fallacies, including the “tackling-the-symptom bias” (single-mindedly fighting corruption symptoms with targeted interventions while ignoring structural causes) and the “injection bias” (focusing on “greenfield stand-alone” agencies to the detriment of fundamental needs in the larger institutional environment). This is a damning indictment, one that implicates a wider array of institutional reform initiatives, though it applies with special force to anti-corruption programs. Kaufman helpfully points out the fallacy, and offers the alternative of focusing much greater effort on systemic issues of economic policy and structural incentives in the public sector. However, on the issue of assessing the value of extant ACA models, he offers no alternative.
Following Kaufman, if we were to attempt the measurement of an ACA’s value-added or net outcomes, we would confront a series of knotty methodological problems – as well as a need for significant time and resources. Even if these problems were overcome, the assessment effort would confront a highly complex set of phenomena. Box 1 presents the hypotheses one would need to pursue, and thus illustrates the difficulties involved.
Box 1: Hypotheses on the Impact of ACAs 1) There is a series of institutional design factors proposed in the literature on ACAs (see below) that describe necessary features of an effective anti-corruption agency. Although these conditions are consistently associated with success, they are not by themselves sufficient. The other necessary conditions for success of an ACA are exogenous, and include public order, political stability, the absence of macroeconomic crisis or crippling distortions, and some basic features of the rule of law.
2) Even if all these sufficient conditions are met, the dimensions of corruption could still overwhelm an ACA. Such agencies cannot cure thoroughly unsound governance environments. Rather, they can address corrupt areas (even deeply corrupt ones) within an environment that is otherwise sound in governance terms – i.e. they can add value in anti-corruption terms to a structure of working institutions that have proven unable to tackle some significant pockets of corruption. More specifically, an ACA requires effective support within most, if not all, of the following areas: the top governing circle in the executive, the top level of the judiciary, the supreme audit agency, the legislature, the top business figures and organizations, and the general public. In other words, there need to be some other effective institutions and networks that have not themselves been seriously compromised by corruption, or otherwise persuaded to look the other way.
3) Conversely, ACAs lacking the features cited above could still succeed in the short-term, but are unlikely to outlive the current political alignment or regime. Such a success would have to be based on mass political support for policies and measures against corruption (e.g. in the wake of a scandal), or alternatively a strong anti-corruption plank in the platform of the ruling party or formation. Absent other conditions, anti-corruption initiatives will depend on ad hoc bodies or on personalities (the chief of state and a few allies), but will not result in sustained and successful institutions.
4) A country’s (or jurisdiction’s) ability to create an effective ACA is a reflection of its ability to evolve effective institutions of governance (courts, civil service, competitive party system, etc.) more generally. The one is consistently associated with the other. This suggests that investing resources in an ACA within a context of high corruption will be both hopeless, wasteful in terms of higher priority needs going unmet, and indeed dangerous, since these are the situations where ACAs are most likely to be politicized and predatory. Short-term success (as in #3) may be feasible, but this depends on the interests and alignment of powerful individuals rather than on institutional design – and a strong ACA may become abusive in the wake of a political shift.1 We know what ACAs purport to achieve – but what do they actually do? As in Hong Kong and its many imitators, these agencies usually describe their anti-corruption programs in terms of prevention and deterrence, often with some element of public mobilization mixed in. Such a program involves a certain defined menu of functions and tasks that most ACAs share. Having collected information on some thirty ACAs, we are able to identify six functions that they commonly perform:
Preventive research, analysis, and technical assistance
Ethics policy guidance, compliance review, and scrutiny of asset declarations
Public information, education, and outreach.
In examining agencies’ performance of these tasks, we can certainly ask how well the agency carries them out. But this would not, by itself, enable us to determine the agencies’ impact on levels of corruption – which is what they are ostensibly designed to influence. There are two main difficulties here. One problem is that, even if we determine that an agency performs the above categories of tasks extremely well, this does not tell us whether it has the right mission.
The political science literature has traditionally been skeptical about the value of any robust anti-corruption mission. Huntington (1968: p. 386) famously stated:
In terms of economic growth, the only thing worse than a society with a rigid, over-centralized, dishonest bureaucracy is one with a rigid, over-centralized, honest bureaucracy.
With specific reference to anti-corruption programs and agencies, Williams (2000: p. 143) brings the following indictment:
An epidemic of corruption scandals often induces moral panic and over-reaction. In such contexts, corruption commissioners can almost seem to act as modern witchfinder-generals, playing on popular fears and asserting guilt where there was once a presumption of innocence. They deem customary political behaviour to be improper and they reject the legitimacy of established electoral systems….They encourage a climate of suspicion and distrust, thus undermining confidence and public trust in public figures and the wider political system.
These criticisms flow from the “corruption as grease” argument – both in the sense of easing transactional burdens in contexts of rigid bureaucracy, and in the sense of facilitating political action. A similar sentiment is reflected in Anechiarico and Jacobs (1996), but in that case, the concern is about the threat of an inquisitor standing over the shoulder of civil servants, thereby making it difficult to respond flexibly to needs in the public sphere – which often requires cutting a few corners.
The contemporary economics literature on corruption, and the apparent consensus of reformists and policymakers, suggests that measures to reduce corruption to no more than a minimal, efficient, level are justified by the distortions – economic, social, political – that would otherwise take hold. Corruption as “grease” intensifies distortions by bringing more rules and more bureaucrats into the game of attracting grease payments. (Kaufman 2000) Grease payments often form some part of the base of a graft pyramid that extends to the top of the administrative and political systems.
As distinct from anti-corruption measures generally, anti-corruption agencies might or might not have a useful mission. In one sense, they reflect the prevailing “tackling-the-symptom bias” (Kaufman 2000) or thermodynamic law that necessarily calls forth an anti-corruption institution when the stimulus of corruption is acknowledged. The consensus at least in the economics literature is that there are many more – and more important – causes of high corruption levels than the lack of an ACA or some other kind of repressive mechanism. (See, e.g. Bardhan 1997, Klitgaard 1988, Rose-Ackerman 1978) These other causes range from public sector institutional design to civil service pay, dependency on certain fiscal sources, poverty and lack of private economic opportunity, the emergence of bribery as a dominant social norm, the industrial organization of corruption networks, and others. If we were to line up the six task areas for which ACAs are designed against the full range of tasks involved in restraining corruption, we would get something like the following.
Anti-Corruption Task Area
Other agencies handle?
Receive and respond to complaints
Intelligence, monitoring, and investigation
Central investigations bureau, prosecutor
Prosecutions and administrative orders
Attorney general, special prosecutor
Preventive research, analysis, and technical assistance
Special policy units, performance audit agencies
Ethics policy guidance, compliance review, and scrutiny of asset declarations