Click here for search results

Difficulties with Autonomous Agencies

Subtopics on this page:

There are diverse, competing, and frequently conflicting claims about the advantages of creating semi-autonomous entities within the public sector. Without a clearer understanding of what autonomization can achieve and  under what circumstances, the confusion of claims is allowing a debate over form to precede a debate over function. The consequence is that many autonomization adventures have been well packaged but unproven, while others have been successful but unsung.

Agencies – a wrench placed in the machinery of government?

Autonomization is not a new invention. Prior to the creation of Next Steps Agencies in the UK there were between 1000 and 5000 non-Departmental public bodies, state-owned enterprises, and health service providers. Similarly, there is a long Latin American tradition of "decentralized" entities (autonomous institutes, public enterprises, etc.). In general, machinery of government changes have taken place at an evolutionary rather than revolutionary pace – and given the uncertainties around design and the risk of errors this is probably appropriate. It appears, however, that the autonomization debate may have altered this evolutionary path, with autonomization emerging as a solution urgently in pursuit of a problem.

Types of agencies                                                   

It is possible to distinguish two primary types of autonomization: type 1 agencies (agencies that act within the executive – typically "executing agencies") and type 2 agencies (agencies that act to restrain the executive – typically "statutory commissions" or autonomous regulators). As arms-length instruments of line ministries for policy implementation, type 1 agencies are expected to improve operational efficiency in service delivery areas through greater managerial discretion and flexibility with relatively secure funding and organizational autonomy. Type 2 agencies, on the other hand, are designed to give statutory independence to agencies in charge of functions that require protection from politicians’ short-term orientations and capture by government in turn, including long-term policy-making (e.g., monetary policy) or quasi-judicial/regulatory functions. Within each type, agencies can be more or less institutionalized. Institutionalized agencies are subject to more deeply entrenched checks and balances, and tend to form a class of agency rather than one-off. By contrast, governments may create autonomous agencies as ad hoc responses to respond to particular political problems, to protect priority activities from opposing political influences, or to bypass the low technical capacity of public sector personnel.


Type 1
(Operational efficiency in service delivery)

Type 2
(Statutory independence for policymaking & quasi-judicial functions)


Ad hoc responses

Swedish ambetsverk
New Zealand Crown entities
Canadian Special Operating Agencies
UK Next Step agencies

Bolivia's Customs
Peru's Customs
Roads boards

Social funds

Independent central banks
Supreme audit institutions
Independent regulatory agencies and commissions



Relatively little is known about the success factors behind sustainable and effective agencies. Despite considerable managerialist rhetoric, it seems improbable that intra-public sector contracts are a particularly significant ingredient in type 1 agencies. Despite the celebration of their success in the new public management reforms of New Zealand and the UK, it is at least arguable that the result owes more to the tradition of discipline within the civil service than the particular nature of the contract with the agency. Experience from State Owned Enterprises is that performance contracts within the public sector are rarely effective in constraining enterprise behavior.

Externalities and collateral damage                               

Although the case for and against agencies is generally made in relation to the specific task of an agency, a broader examination may reveal areas in which agencies have potentially deleterious impacts on the public sector as a whole:

  1. Policy Lock-In. Are autonomous agencies undermining policy objectives by creating constituencies that will compel governments to maintain existing policies? Agencies are created but rarely closed or merged. Policy becomes what the agencies do, not what the government proposes. This can also create unhealthy institutional rivalry between the agency and the parent ministry which is legally responsible for a given policy domain. In dysfunctional bureaucracies where agencies are created to circumvent the ineffectiveness of the traditional ministries, it is often the case that agencies dominate de facto policy-making at the detriment of the parent ministry. Since agencies commonly lack formal mandate and authority for inter-agency policy coordination, the dominance of an agency in a particular policy domain could cause a serious problem of intra-government policy coordination. The result is often a bureaucracy that responds poorly to changing priorities.

  2. Policy Creep. Are autonomous agencies adopting quasi-fiscal activities that stretch beyond the original policy intention of government? Autonomous status can encourage agencies to engage in quasi-fiscal activities (e.g., fee-based services, special concessions to certain groups) that serve the same role as taxes and subsidies, and exceed the original policy intention of government.

  3. Budget Balkanization. Are autonomous agencies undermining the coherence of the budget? Unchecked agency creation can destabilize the budget in three ways. First, it can create an argument for earmarked funding, which undermines the strategic ability of government to shift funds to emerging priorities, thereby leading to budgetary rigidities and over-stretched funding. Second, it can create scenarios (a.k.a. "bleeding stump" arguments) in which government must provide additional resources or face the unthinkable — e.g., teachers on strike, nurses without jobs, etc. The arms-length nature of agencies makes "bleeding stump" arguments more likely since the imminent problem is less easily identified by the central agencies. Third, unchecked agency creation can create contingent liabilities for government by borrowing against assets or making other commitments.

  4. Patronage Den. Are autonomous agencies facilitating patronage? Unchecked agency creation can institutionalize patronage in appointments if the agency becomes the implicit bailiwick or property of a coalition party. This undermines the credibility of merit protection regulations for all agencies by offering a haven for patronage appointments. Particularly in governments backed by unstable coalitions, autonomous agencies – with the implicit promise of a distinct sphere of influence and public profile – can provide an easy route for appeasing fractious coalition members.

  5. Special privileges. Are autonomous agencies distorting public sector incentives? In order to attract qualified staff, autonomous agencies are often given exceptions from the government personnel regime to offer higher salary scales as well as other attractive benefits. The uneven incentive structures within the public sector make it difficult to lure competent individuals to traditional ministries, while those who remain in the non-autonomous parts of the government become unmotivated or even resentful of those receiving better incentives and benefits in autonomous agencies.


Autonomization is attractive to both reformers in countries and to donors because it allows quick circumvention of a number of institutional dysfunctions that plague public sectors in developing and transitional countries. However, there are significant costs associated with inappropriate autonomization. Therefore, benefits and costs should be weighed in each concrete circumstance. A model of autonomization that worked in one context may very well prove to be wholly inappropriate in another setting. A key challenge is to remove the magic bullet associations with autonomization, and restore it to its position as one dimension of institutional design among many within the public sector. The perennial issue when considering government provision of goods and services is not "How can we use autonomization?" The fundamental questions are "What is wrong with the present arrangement?" and "What would successfully solve that problem?"

The Public Agencies International Research Network (PAIR-NET) is a ‘closed’ secure forum for exchanging research and practice among scholars and practitioners active in ‘agency’ programs. It is part of a wider program of activity on public agencies, including research, conferences and workshops. Details are available at

Recommended readings                                                     

  • Axelrod, Donald. 1992. The Rise of Shadow Government. New York: Wiley.

  • Boston, J., J. Martin, J. Pallot, and P. Walsh. 1996. Public Management: the New Zealand Model.  Aukland: Oxford University Press.

  • Chancellor of the Duchy of Lancaster. 1997. Next Steps Agencies in Government - Review 1996. UK Cabinet Office, London.

  • Estache, Antonio, and David Martimort. 1999. "Politics, Transaction Costs, and the Design of  Regulatory Institutions." Policy Research Working Paper No. WPS 2073. Washington, D.C., World Bank.Fraser, A. 1991.Making the Most of Next Steps: The Management of Ministers’
    Departments and their Executive Agencies. London: HMSO.

  • Keefer, Philip, and David Stasavage. 2000. "Bureaucratic Delegation and Political Institutions: When Are Independent Central Banks Irrelevant?" Policy Research Working Paper No. 2356. World Bank, Washington, D.C.

  • Polidano, Charles. 1997. "The Bureaucrat Who Fell Under a Bus: Ministerial Responsibility, Executive Agencies and the Derek Lewis Affair in Britain." Working Paper No. 1. IDEM Public Policy and Management Series. [A revised version was published in Governance 12(1), 1999.]

  • Schick, Allen. 1998. "Why Most Developing Countries Should Not Try New Zealand Reforms." World Bank mimeo.

  • Scott, Graham, and Irene Taylor. 2000. "Autonomous Public Organisations in Thailand." Victoria Link mimeo, Wellington, New Zealand.

  • World Bank. 1995. "Contracting: What Works, What Doesn’t and Why." In Bureaucrats in Business: The Economics and Politics of Government Ownership. Oxford and New York: Oxford University Press for the World Bank, Washington, D.C.

This page was prepared by Nick Manning and Yasuhiko Matsuda of the World Bank. It was submitted on 7/27/00.

Permanent URL for this page: