Public institutions Institutions

Updated December 1997

Food and Agriculture Organization of the United Nations United Nations Capital Development FundInternational Fund for Agricultural DevelopmentGerman Agency for Technical CooperationSwiss Agency for Development and CooperationWorld Bank

16-18 December 1997
Technical Consultation on Decentralization

The Promise and Limitations of Decentralization, Introduction & Part 1

by James Manor
Institute of Development Studies
University of Sussex, UK


This study examines -- from a political economy perspective -- the implications of the current wave of decentralizations in less developed countries, especially its promise and limitations for rural development. It follows on from a companion study which assesses the origins of this wave, "Explaining the Popularity of Decentralization" (Manor, 1997). It is crucial that we understand its limitations, lest exaggerated expectations of decentralization - which are widely held - lead to unwarranted disillusionment that could cause it to be abandoned. That would be unfortunate, since it has very considerable promise.

This study is based mainly on empirical evidence drawn from experiments with decentralization in a large number of countries. This evidence is imperfect and incomplete, but still extensive enough to tell us a great deal. It should be stressed, however, that the findings here on the promise and limitations of decentralization may need to be revised after these experiments have more time to develop and make an impact - perhaps in areas where, at this early stage, they have achieved rather little. So the findings that appear here - even when they read like forceful assertions - are not intended as the final word on the subject.

The study is divided into four parts. Part 1 defines terms, to show that the word "decentralization" can mean many different things. This study focuses (for reasons set out in Part 1) on three types of decentralization: deconcentration or administrative decentralization, fiscal decentralization, and devolution or democratic decentralization. Each of these can occur in isolation, but any two (or all) of them can occur simultaneously. For reasons provided in Part 1, this paper concentrates upon experiments with decentralization which (in varied ways) possess democratic content.

Part 1 also notes the various levels to which power and resources are being decentralized - to the local level, or to one or more intermediate level(s), or to both. It is important to recognize that the creation of a federal system by empowering regions is quite different from the empowerment of authorities at or near the grass roots. Mix the two, and you get something different yet again.

This paper is therefore necessarily an assessment of varied decentralizations (plural), not decentralization (singular). The diversity of the phenomena under discussion here prevents this study from yielding the kind of clear, tidy findings some readers might desire.

Part 2 examines the encounter between decentralized institutions and the environments in which they must operate. The focus here is on how politics and state-society relations impinge on these institutions, and vice-versa.

Part 3 discusses the advantages and disadvantages (overwhelmingly the former) which attend decisions to decentralize to both the regional (or intermediate) level in a political system and to the local level, rather than to just one of these.

Part 4 assesses the promise of decentralization for rural development. It is better equipped to achieve some things than others. Only if we understand this can we protect it from the disillusionment that will surely follow when expectations about its utility in certain areas are shown to be unrealistically high [1].

Part 1. Defining terms

Let us begin by considering some definitions of types of decentralization. We will then focus more intensely on forms and democratic decentralization, and conclude with a brief comment on the problem of levels to which decentralization can occur.


Numerous definitions of "decentralization" emerge from the literature. To keep this analysis manageable, the term here refers to changes which occur within political systems. This rules out three of six things which are sometimes discussed under this label.

The first type of change to be excluded is sometimes called "decentralization by default." This happens when government institutions become so ineffective that they fail almost entirely to make the influence of central authorities penetrate down to lower-level arenas, and people at the grass roots become heartily cynical about government. When this occurs in countries with lively civil societies, voluntary associations or nongovernmental organizations at lower levels sometimes step in to generate development projects. Resources for such projects - which are either mobilized at the local level or obtained from nongovernmental sources higher up - accrue to these groups and a kind of "decentralization," unintended by government, takes place. We have only one well-documented case of this in the literature (Davis and others, 1994), but there are no doubt other examples [2]. These are worthy of study - although they are far from problem-free - but since the present paper already faces a hugely complex task in analyzing other variants of decentralization - which are intended by governments - we exclude them.

The second thing to be excluded is privatization - the handover of tasks formerly performed by state agencies to the private sector [3]. We omit it partly because it entails the transfer of tasks outside political systems and partly because the private sector firms which take them over (even from local authorities) are themselves often quite large (World Bank, 1995). Therefore, privatization often involves a shift of power and resources from one major, centralized power center to another.

We also set aside one further type of decentralization, namely, "delegation" - of some responsibilities for development programmes or projects to parastatal agencies. We exclude it, partly because it has only rarely been attempted and partly because when it has been tried, it has either failed to facilitate a genuine decentralization of decisionmaking or it has impeded project implementation, or both (Parker, 1995).

This leaves us with three key definitions. Following Parker's adaptation of Rondinelli's typology (Rondinelli, 1981; Parker, 1995) [4], we can describe them as follows:

  1. deconcentration or administrative decentralization,
  2. fiscal decentralization,
  3. devolution or democratic decentralization.
At times, two or all three of these types of decentralization occur simultaneously, but they can also occur in isolation.

The first of these, deconcentration, refers to the dispersal of agents of higher levels of government into lower-level arenas. Parker describes it as "administrative decentralization," and these two terms will be used interchangeably here.

One point needs emphasis. When deconcentration occurs in isolation, or when it occurs together with fiscal decentralization but without simultaneous democratization - that is, when agents of higher levels of government move into lower-level arenas but remain accountable only to persons higher up in the system - it enables central authority to penetrate more effectively into those arenas without increasing the influence of organized interests at those levels. The central government is not giving up any authority. It is simply relocating its officers at different levels or points in the national territory. In such circumstances, it tends in practice to constitute centralization, since it enhances the leverage of those at the apex of the system. This is especially true in less developed countries where ordinary people,

have small influence over any allocations in the modern sector, such as involve finance and the direction of skilled manpower. Their lack of knowledge excludes them from the affairs of government. This is particularly true of rural people, whose society and economy are still largely based on subsistence agriculture, and who are insulated from decisionmaking centers by poor communications. In this situation a "deconcentrated" field office takes most of its decisions - even major ones - without being subject to local pressures, though it may sometimes enter into voluntary consultations with local notables. Demands from central government are much stronger than those from the local population and the field officer (less secure than his counterpart in the West) is constantly concerned to satisfy his political masters" (Mawhood, 1993, pp. 2-3).
These themes are echoed in Olson (1971), Bates (1983) and Becker (1983 and 1985). A vivid example of this can be found in Moi's Kenya (Ng'ethe, 1993). (See also, Nellis, 1983; Rondinelli, 1983; and Oyugi, 1993.) When deconcentration produces, in effect, the opposite of decentralization, it hardly warrants consideration in this study. But it can also be linked to mechanisms which give people at lower levels some voice in the decisions made within state institutions, and in those cases it can produce a degree of genuine decentralization.

Second, the term decentralization sometimes refers to downward fiscal transfers, by which higher levels in a system cede influence over budgets and financial decisions to lower levels. This authority may pass to deconcentrated bureaucrats who are accountable only to superiors at higher levels, or to unelected appointees selected from higher up. That sort of arrangement is subject to the same concerns voiced by Mawhood in the quotation just above, although fiscal decentralization in isolation may be somewhat less prone to these tendencies than deconcentration. But since fiscal decentralization unattended by any steps towards democratization rarely increases the influence of organized interests at lower levels, this makes it difficult to regard it as an example of genuine decentralization. However, such fiscal transfers can also be linked to mechanisms which give people at lower levels some voice, and when that occurs, no one would describe it as anything other than decentralization.

Finally, there is devolution - the transfer of resources and power (and often, tasks) to lower-level authorities which are largely or wholly independent of higher levels of government, and which are democratic in some way and to some degree [5]. (The problem of what qualifies as democratic is tackled in section just below.)

This study concentrates on experiments with decentralization which entail some elements of devolution or democratization. It does so in part because this has been the main thrust of recent experiments with decentralization (Nickson, 1995) and because, as some World Bank analysts have recognized, "the decentralization of resources and responsibilities without...(democratizing) political reforms would have been incomplete and, probably, not conducive to socially effective results" (World Bank, 1995, pp. 2).

It does so because "There is no feasible substitute to an approach in which local governments, with the active participation of their communities, take the initiative and responsibility for the actions conducive to their institutional development" (World Bank, 1995, pp. 27). And because "Sustainable development of capacity at the local level is possible only when there is effective demand by local administrations and communities" (World Bank, 1995, pp. viii).

It is, at the least, exceedingly difficult for such demand to manifest itself unless some form of democratization has occurred along with deconcentration and/or fiscal decentralization. Anwar Shah has made similar comments in his work on fiscal decentralization (Shah, 1997).

To put the same point differently, we will see in this study that greater accountability of government institutions - the most crucial element in successful decentralizations (Crook and Manor, 1994) - can be hard to obtain even when substantial democratic elements are introduced into the decentralizing process. When they are absent, or when reforms entail only minimal steps towards democratization, the impediments to greater accountability tend to be well nigh insurmountable.

The insistence here upon the need for democratic content if decentralization is to possess much promise should not, however, blind us to one further, important reality: democratic decentralization on its own is likely to fail. Democratic authorities at lower levels in political systems will founder if they lack powers and resources - meaning both financial resources and the administrative resources to implement development projects (see Part 2). In other words, decentralization must be attended both by some fiscal decentralization (since that supplies financial resources) and by some deconcentration or administrative decentralization (since that supplies bureaucratic resources required for implementation). If it is to have significant promise, decentralization must entail a mixture of all three types: democratic, fiscal and administrative [6]. So throughout this study, when reference is made to decentralization which offers some promise, the presumption is that such a tripartite mixture exists. The assertion in the World Development Report 1997 that such tripartite mixtures rarely occur is erroneous - they are reasonably common [7] (World Bank, 1997, p. 121).

One last, modest comment is in order here. Decentralization is almost always the result of intentional decisions by policymakers. But there is such a thing as inadvertent decentralization. This is not the same as decentralization by default, mentioned above. It occurs when other policy innovations produce an unintended decentralization of power and resources as a by-product. Two main examples come to mind - a small number, but they have occurred in large, important countries. The first is Russia where authorities at lower levels have acquired greater powers than the central authorities intended as a result of oversights and unexpected developments (Gibson and Hanson, 1996). The second is China, where provincial governments have obtained more resources (and power over them) than central leaders wished. These examples of inadvertence are not entirely excluded from this discussion, but the emphasis here will be on intentional attempts at decentralization.

Forms of democratic decentralization

Democratic decentralization can take varied forms. It clearly includes cases in which the persons in authority within institutions at intermediate or local levels are elected by secret ballots [8]. Elections may be direct or indirect - in the latter case, elected representatives at lower levels select members or higher-level authorities. The persons who are elected may be the members of a council, or the executive head of an authority, or both. Elections are usually on a "first-past-the-post" basis, but they sometimes make use of some form of proportional representation.

All such cases qualify as examples of "democratic decentralization," but we need to avoid the narrow view that no set of arrangements other than these is admissible [9]. We should, for example, accept systems - often seen in Francophone countries - where competing political parties put forward lists of candidates for seats on an authority, with the party that gains the most votes becoming the only party represented therein. Francophone systems also often permit the heads of such authorities to hold positions in national-level parliaments and cabinets - in accordance with the principle of cumul des mandats (the accumulation of mandates) [10]. Such systems may be less healthy for the democratic process than those more familiar in the English speaking world. Losing parties that are utterly excluded from power in this winner-take-all game may feel deeply aggrieved, so that the legitimacy of such institutions suffers. And heads of decentralized bodies who are also active at the national level may be preoccupied with high politics, and thus less responsive to popular pressure from lower levels. But it would be unhelpful to exclude these systems here.

We also need to include certain other, unconventional arrangements. In the Philippines, nongovernmental organizations have officially been given voting powers on local councils (Brillantes, 1994). In Colombia, we encounter local level "construction projects...that involve community contributions in labor, materials or cash, and for which there is community supervision" (World Bank, 1995, p. 6). The italics (which I have added) identify the critical element here. Community contributions which provide people with no voice cannot be regarded as democratic, but when some form of supervision or influence is permitted, they have some democratic content.

Such supervision or influence over projects or decisions by local authorities is often informal - which is to say that it is not well-institutionalized (World Bank, 1995). This is a less than fully reliable means of rendering systems democratic, since it often depends on the good will of local office holders, some of whom may prefer to avoid it. This is worth stressing here, because some important World Bank documents (for example, World Bank, 1994) appear to reveal a preoccupation with this sort of involvement of local people in development project cycles, rather than with the promotion of well-established, elected institutions (Blair, 1995).

Nevertheless, when such participatory arrangements exist, it injects some democratic content into the system. We also need to accept other devices which provide people at the local level with some influence. These include efforts by local authorities to seek information on community needs and ways of addressing them, to establish local committees whose purpose is to foster active community participation, to organize and coordinate community involvement in projects (sometimes by hiring private firms to promote this), and to organize disenfranchised groups in order to assist them in voicing demands (which is sometimes done by central government agencies) (World Bank, 1995).

These devices are usually used by elected local authorities, but even in (the quite rare) cases where this occurs in the absence of elections, it is possible to say that a democratic element is - at least tenuously - present. Finally, we should recognize that occasions where authorities seek to draw community leaders and voluntary associations into consultations and decisions about development can - if they are not cynical exercises to give the appearance of openness before a government's initial intentions are implemented - bring at least a minimal degree of democracy into a system.

There are, of course, arrangements which appear democratic but usually amount to deceptions. For example, a system in which voters queue up to be counted behind symbols of their preferred candidates - used in the Nigerian local elections in March 1996 - is open to abuse and must be regarded as extremely dubious. And where unconventional devices amount only to half-measures, they need to be identified as such. But half-measures are better than none at all, and sometimes they permit people at the local level to begin to pry systems open in ways that those in authority did not anticipate.


To complicate matters further, decentralization can entail transfers of power to different levels within political systems. It can be bestowed on one or more intermediate levels (regions and/or subregions). When this involves the creation of fully blown governments at the regional level, it qualifies as federalism. But more limited powers may go to intermediate levels, in experiments which fall short of outright federal restructuring.

Other experiments entail transfers to the local level, or to arenas quite close to it. We need to recognize that there is a difference between experiments which empower intermediate levels and those that empower local levels. This that may sound obvious, but some analysts fail to grasp it [11]. And of course, some experiments entail simultaneous transfers to one or more intermediate levels and to local levels. (Such arrangements, about which questions often arise from policymakers, are discussed in detail in Part 3.)

This is not a short document, but even in a study of this length, it is impossible to provide a comprehensive analysis of all of the permutations that flow from the various types of decentralization listed above (and mixtures thereof in variegated sequences - see Part 4), plus the various levels to which transfers occur. Nor can we begin to exhaust the complex implications which arise from variations in the size of the national political arenas within which decentralization takes place. Decentralization in Botswana is bound to mean something different from decentralization in India. We can go quite a long way in this discussion despite this limitation, but readers should still be aware of it.

Readers will also notice that some of the experiments with decentralization discussed here are wholly or substantially located in urban areas [12]. This may seem odd, since this study deals with the utility of decentralization for rural development. But some urban initiatives carry useful lessons that have general application, and where that is true, they are considered here.

One last comment is in order here. The World Development Report 1997, in its excessively economistic discussion of decentralization, stresses the benefits that allegedly follow from "competition between levels of government" (World Bank, 1997, p. 122). It will become apparent - particularly from Part 2 of this study - that the main gains arise not from competition, but from cooperation between levels in decentralized systems.

  • To: The promise and limitations of decentralization, Part 2

    SD Homepage Back to Top FAO Homepage