+ All Categories
Home > Documents > Methodological Issues in the Analysis of Decentralization

Methodological Issues in the Analysis of Decentralization

Date post: 20-Feb-2022
Category:
Upload: others
View: 2 times
Download: 0 times
Share this document with a friend
26
Methodological Issues in the Analysis of Decentralization John M. Cohen and Stephen B. Peterson Abstract* Since the 1950s, a large body of literature has emerged that reviews various dimensions of decentralization. It is comprised of theoretical exercises, comparative studies of selected cases, individual country studies, focused inquiries into particular aspects of the intervention, teach- ing materials, and government or aid agency design and implementation manuals. This body of writing is now so wide-ranging, diverse, and substantial that it merits consolidation in a state-of-the-art paper. Such an exercise was carried out by the authors for the United Nations and issued as “Administrative Decentralization Strategies for the 1990s and Beyond” (Research Study Prepared for the Governance and Public Administration Branch, Division for Public Administration and Management Development, Department for Development Support and Management Services, United Nations Secretariat, November 1995). A version of this report will be published by Kumarian Press in 1997 as Administrative Decentralization in Late Developing Countries. This Discussion Paper consolidates some of the background research undertaken by the authors while preparing the United Nations study. The methodological issues reviewed will not be published in the Kumarian Press book. Hence, this paper seeks to preserve issues identi- fied during the research process for academics and aid agency professionals. The paper begins by describing the range and scope of the numerous books, monographs, journal articles, governmental studies, and consulting reports that describe and analyze various forms and types of decentralization. Then it reviews several methodological problems marking this literature that limit its utility to governments and aid agencies seeking to use decentraliza- tion strategies to promote development processes. These include careless use of conceptual definitions and terms, misconceptions and unrealistic expectations, unsystematic presentations, an overemphasis on cases of failure, lack of comparability among diverse case studies, neglect of historical patterns that generate complexity, inappropriate linear assumptions, and naive arguments that bureaucracies should be dramatically reduced and power and responsibility for public sector tasks be transferred to local communities, private sector firms, and organizations. John M. Cohen is an Institute Fellow at HIID. Stephen B. Peterson is a Development Associate at HIID.
Transcript
Page 1: Methodological Issues in the Analysis of Decentralization

Methodological Issues in the Analysis of Decentralization

John M. Cohen and Stephen B. Peterson

Abstract*

Since the 1950s, a large body of literature has emerged that reviews various dimensions ofdecentralization. It is comprised of theoretical exercises, comparative studies of selected cases,individual country studies, focused inquiries into particular aspects of the intervention, teach-ing materials, and government or aid agency design and implementation manuals. This bodyof writing is now so wide-ranging, diverse, and substantial that it merits consolidation in astate-of-the-art paper. Such an exercise was carried out by the authors for the United Nationsand issued as “Administrative Decentralization Strategies for the 1990s and Beyond” (ResearchStudy Prepared for the Governance and Public Administration Branch, Division for PublicAdministration and Management Development, Department for Development Support andManagement Services, United Nations Secretariat, November 1995). A version of this reportwill be published by Kumarian Press in 1997 as Administrative Decentralization in Late DevelopingCountries.

This Discussion Paper consolidates some of the background research undertaken by theauthors while preparing the United Nations study. The methodological issues reviewed willnot be published in the Kumarian Press book. Hence, this paper seeks to preserve issues identi-fied during the research process for academics and aid agency professionals.

The paper begins by describing the range and scope of the numerous books, monographs,journal articles, governmental studies, and consulting reports that describe and analyze variousforms and types of decentralization. Then it reviews several methodological problems markingthis literature that limit its utility to governments and aid agencies seeking to use decentraliza-tion strategies to promote development processes. These include careless use of conceptualdefinitions and terms, misconceptions and unrealistic expectations, unsystematic presentations,an overemphasis on cases of failure, lack of comparability among diverse case studies, neglectof historical patterns that generate complexity, inappropriate linear assumptions, and naivearguments that bureaucracies should be dramatically reduced and power and responsibility forpublic sector tasks be transferred to local communities, private sector firms, and organizations.

John M. Cohen is an Institute Fellow at HIID.

Stephen B. Peterson is a Development Associate at HIID.

Page 2: Methodological Issues in the Analysis of Decentralization

1

Methodological Issues in Analysis of Decentralization

Over the past four decades a large body ofliterature has emerged that reviews variousaspects of decentralization interventionsand reforms. It is comprised of theoreticalexercises, comparative studies of selectedcases, individual country studies, focusedinquiries into particular aspects of theintervention, international and local train-ing materials, and government or aidagency design and implementation manu-als. Most of these studies are produced byacademics, either through their ownresearch activities or through grants andconsultancies provided by aid agencies. Thebody of writing on decentralization is nowso wide-ranging, diverse, and substantialthat it is tempting to conclude that a carefulreview of its content could generate usefulguidelines for designing and implementingdecentralization strategies in the 1990s.However, this is not the case. To a largeextent this is because of the methodologicalproblems reviewed in this paper.

This Discussion Paper consolidatesbackground analysis undertaken by theauthors during the process of preparing astate-of-the-art paper for the UnitedNations Secretariat, which was issued as“Administrative Decentralization Strategiesfor the 1990s and Beyond” (Research StudyPrepared for the Governance and PublicAdministration Branch, Division for PublicAdministration and Management Devel-opment, Department for DevelopmentSupport and Management Services, UnitedNations Secretariat, November 1995). Theanalysis in this paper is not included in thecommercial publication based on theUnited Nations study, which will be pub-lished for the authors by Kumarian Press in1997 as Administrative Decentralization inLate Developing Countries.

The paper begins by describing therange and focus of the numerous books,monographs, journal articles, governmentalstudies, and consulting reports thatdescribe and analyze decentralizationexperiences in late-developing countries.After demonstrating the richness ofinformation on this topic, attention is givento the methodological difficulties involvedin comparing such studies and using themto generate reliable propositions that canguide government decision-makers and aidagency professionals. Towards this end thepaper reviews analytical problemsgenerated by careless use of conceptualdefinitions and terms, misconceptions andunrealistic expectations, unsystematicpresentations, an overemphasis on cases offailure, lack of comparability among diversecase studies, neglect of historical patternsthat generate complexity, inappropriatelinear assumptions, and naive assertions bysome that bureaucracies should bedramatically reduced and power andresponsibility for public sector tasks betransferred to local communities and pri-vate sector organizations.

Range and Focus of DecentralizationLiterature

To identify, review, and summarize the lit-erature on decentralization one must becommitted to the research enterprise. Thisis because undertaking such research ishampered by a number of methodologicalproblems. These begin with the size anddiversity of the literature on the topic,which is comprised of an extremely largenumber of academic publications, officialgovernment documents, commissioned aidagency studies, local and internationaltraining materials, and internal governmen-

Page 3: Methodological Issues in the Analysis of Decentralization

2

tal and aid agency documents related tospecific decentralization innovations andexperiences. To locate this literature onemust be both knowledgeable about thetopic and well trained in the use of univer-sity, aid agency, and United Nations-basedresearch engines.

Beyond this, the literature is dividedamong three major languages (English,French, and Spanish), as well as written inthe national languages of late-developingcountries. Yet few researchers have theskills to go far beyond the language theywork in. Rarely, do bibliographies identifysource material written in a Third Worldlanguage, even though such sources canoften be quite valuable in the insights theyoffer.1

There are additional reasons why thislarge, diverse, and steadily growing bodyof literature is difficult to fully access anduse. First, as noted shortly, it is generallyagreed that there are four forms ofdecentralization: political, administrative,spatial, and market. Because there is a closecorrelation between these forms ofdecentralization and standard social sciencefields, decentralization studies are oftencoded in the terminology of disciplines. Forexample, economists and planners tend tofocus on “market” and “spatial” forms ofdecentralization and their publications arecompartmentalized into their discipline’ssub-fields in ways that make them difficultto find through standard “key word”search methodologies. Second, whileacademic and professional writing has beenextensive in three languages, it is notparticularly easy for those writing inEnglish to access Spanish or Frenchpublications, reports, and documents. Thisproblem is compounded by a tendency ofthose writing in French and Spanish to fail 1 For example, books, articles, and papers ondecentralization are written in such languagesas: Hindi, Chinese, Arabic, Indonesian, andSwahili.

or review or drawn upon the analytical andempirical work of those writing in English.As a result of these two problems crossreferencing of materials is infrequent andbibliographic searches in one languageoften fail to access materials in otherlanguages. Third, a number ofdecentralization studies that are commis-sioned by aid agencies, United Nationsfamily organizations, and Non-Governmen-tal Organizations (NGOs) never rise suffi-ciently to the surface to be capable of beinglocated in a standard bibliographic search.This is unfortunate, for some of these stud-ies are more analytically or professionallyuseful than published academic studies.2

However, they can be found, largelythough paying close attention to footnotesin published studies. Unfortunately mostbibliographic searches in the area of decen-tralization fail to go this far.

Perhaps because of the size and linguis-tic diversity of this literature, it is notuncommon to hear aid agency consultantsand academics dismiss it as “not solid oruseful.” This simply is not the case. The lit-erature on decentralization contains someof the best analytical arguments and casestudies found in the social sciences, some ofwhich are identified in the footnotes to thispaper. There is much in the literature tobuild on. Dismissing it and starting afresh,as seems common in the mid-1990s, is bothlazy and wrong-headed. Rather, the analy-sis, findings, and lessons of the past fourdecades must be sought out and masteredby those wishing to understand and pro-

2 For example, one of most coherentrestatements of the conceptual definitions of theforms and types of decentralization is found ina consulting report prepared for USAID’s Officeof Housing and Urban Programs: DennisRondinelli, Decentralizing Urban DevelopmentPrograms: A Framework for Analyzing Policy(Washington, D.C., USAID, Office of Housingand Urban Programs, 1990).

Page 4: Methodological Issues in the Analysis of Decentralization

3

mote decentralization strategies into thenext century.

Sifting Areas of Research Focus

While the number of publications anddocuments related to decentralization hasgrown at a steady annual rate, since the late1940s, there have been three periods duringwhich output on the topic has accelerated.Each period has had a different focus.

The first period occurred in the early1960s, when the focus was primarily on de-centralization as an administrative ap-proach for local-level governance in thepost-colonial era.3 During this period thefocus was largely on administrative aspectsof decentralization, with particular concernwith the legal organization of center-fieldoffice relationships and the role of localauthorities or municipalities within a cen-trally managed government.

The second period took place in theearly 1980s when aid agencies were urginggovernments to consider decentralizationstrategies both to better reach the rural andurban poor and to increase their participa-tion in the development process. Examplesof the work carried out during this periodare found in publications stimulated theUnited Nations Nagoya Centre for Regional

3 A. A. Maass, ed. Area and Power: A Theory ofLocal Government (Glencoe: Free Press, 1959); U.K. Hicks, Development from Below: Local Govern-ment and Finance in Less Developed Countries(Oxford: Oxford University Press, 1961);Samuel Humes and Eileen M. Martin, LocalGovernment in Developing Countries (Hague:Martinus Nijhoff, 1961); H. Maddock,Democracy, Decentralization, and Development(Asia Publishing House, 1963); United Nations,Decentralization for National and LocalDevelopment (New York: Working Group on theAdministrative Aspects of Decentralization forNational Development, 1962); H. H. Alderfer,Local Government in Developing Countries (NewYork: McGraw-Hill, 1964).

Development,4 United Nations organiza-tions in general,5 or in university-basedresearch projects, most notably thosefunded by USAID under its rural-directed“New Directions” initiative.6

4 G. Shabbir Cheema and Dennis A. Rondinelli,eds., Decentralization and Development: PolicyImplementation in Developing Countries (BeverlyHills: Sage Publications, 1983); Dennis A.Rondinelli, John R. Nellis, and G. ShabbirCheema, Decentralization in Developing Countries:A Review of Recent Experience (Washington, D.C.:World Bank, Staff Working Paper No. 581,1984); E. P. Wolfers, et al., Decentralization:Options and Issues: A Manual for Policy Makers(London: Commonwealth Secretariat, 1982).5 Their most visible work products of thisperiod include: United Nations, LocalGovernment Reform: Analysis of Experience inSelected Countries (New York: Department ofEconomic and Social Affairs, ST/ESA/SER.E/2,1975); United Nations, ImplementingDecentralization Policies and Programmes(Nagoya: Centre for Regional Development,1981); and its special studies, such as ShabbirCheema’s 1983 study ImplementingDecentralization Programmes in Asia; Food andAgriculture Organization, Administering Agricul-tural Development for Small Farmers: Issues inDecentralization and People’s Participation (Rome:FAO, Economic and Social Development PaperNo. 20, 1981); International Labour Organiza-tion, “World Employment Programme ResearchWorking Papers on Decentralization” (SpecialPapers Prepared on Decentralized Planning andParticipation, Geneva, 1981).6 Examples include the Berkeley-based Decen-tralization Project: David K. Leonard and DaleR. Marshall, eds., Institutions of Rural Develop-ment for the Poor: Decentralization and Organiza-tional Linkages (Berkeley: Institute of Interna-tional Studies, University of California, 1982);and the Cornell-based Decentralization and Par-ticipation Project: Norman T. Uphoff andMilton J. Esman, Local Organization for RuralDevelopment: Analysis of Asian Experience (Ithaca:Rural Development Committee, CornellUniversity, Rural Local GovernmentPublication No. 19, 1974) and their LocalOrganizations: Intermediaries in Rural

Page 5: Methodological Issues in the Analysis of Decentralization

4

Interest in decentralization strategiesdeclined during the 1980s because aidagency concerns shifted from urban andrural development projects and programsto non-sectoral, specific reforms generallyincluded in structural adjustment pro-grams.7 What writing did appear on thetopic during the 1980s and early 1990s waslargely confined to workshop reportsfocused on administrative aspects ofregional and sectoral planning, notably inregard to infrastructure, health, and educa-tion. Among the exceptions to tendencytoward neglect of the topic are a few WorldBank studies focused on legal, financial,and administrative aspects of decentraliza-tion as a potential development strategy8

and United Nations studies of the early1990s that sought to capture the experience

Development (Ithaca: Cornell University Press,1984); John M. Cohen and Norman T. Uphoff,Rural Development Participation: Concepts andMeasures for Project Design, Implementation andEvaluation (Ithaca: Center for InternationalStudies, Rural Development Committee, Mono-graph Series No. 2, 1977); and their Feasibilityand Application of Rural DevelopmentParticipation: A State of the Art Paper (Ithaca:Center for International Studies, RuralDevelopment Committee, Monograph SeriesNo. 3, 1979).7 An exception to this assertion is found in thecooperative agreement between USAID andAssociates in Rural Development, Inc., which istitled “Decentralization: Finance and Manage-ment Project.” It has, however, tended to focuson privatization of services and communitycontrol of environmental resources,contributing little direct value to governmentdecision-makers or aid agency professionalsseeking to implement strategies ofadministrative decentralization.8 Silverman, Public Sector Decentralization; RémyPrud’homme, On the Dangers of Decentralization(Washington, D.C.: World Bank,Transportation, Water, and Urban DevelopmentDepartment, Policy Research Working PaperNo. 1252, 1994).

of late-developing countries involved inimplementing decentralization strategies.9

A third and currently on-going periodemerged in the early 1990s. It focuses pri-marily on political aspects of decentraliza-tion, seeking to understand whether decen-tralization of any form or type canstimulate the emergence of goodgovernance, constrain sub-national ethnicconflict, promote democratic practices,facilitate the growth of civil societies, andincrease the privatization of public sectortasks. Examples of recent work calling forattention to political dimensions ofdecentralization are generally found in:academic studies focused on sub-nationalethnicity and democratization,10 aid agency

9 United Nations, Seminar on Decentralization inAfrican Countries: Banjul, Gambia 27-31 July 1992(New York: United Nations Department of Eco-nomic and Social Development and AfricanAssociation for Public Administration and Man-agement, 1993); Administrative Modernization inCentral and Eastern European Countries: A CaseStudy on Decentralization and Public Administra-tion, Report on an Expert Working Group Meetingin Budapest, Hungary, 1991 (New York: UnitedNations, Department of Economic and SocialDevelopment, 1992); Séminaire sur laDécentralisation en Afrique: Tunis, 10-14 Février1992 (New York: Nations Unies, Départment duDéveloppement Économique et Social, 1992);Descentralización en America Latina: Santa Cruz dela Sierra, Bolivia, 3-5 Noviembre 1993 (NuevaYork: Naciones Unidas, Departamento deApoyo al Desarrollo y de Servicios de Gestion,1994); United Nations DevelopmentProgramme, Workshop on the DecentralizationProcess: Bern, Switzerland, 20-23 April 1993(Report on Workshop, New York, ManagementDevelopment Programme, 1993). Related tothese workshop findings is: UNDP, Workshop onthe Decentralization Process.10 For example: I. William Zartman, ed.,Collapsed States: the Disintegration and Restorationof Legitimate Authority (Boulder: Lynne RiennerPublishers, 1994); L. Picard and R. Zariski, eds.,Subnational Politics in the 1980s: Organization, Re-

Page 6: Methodological Issues in the Analysis of Decentralization

5

inquiries into the promotion ofdemocratization and improved gov-ernance,11 and United Nations studies onsuch topics as the capacity of decentraliza-tion to promote equitable human develop-ment.12

organization and Economic Development (NewYork: Praeger Publishers, 1987).11 For example: Richard C. Crook and AlfMorten Jerve, eds., Government and Participation:Institutional Development, Decentralisation andDemocracy in the Third World (Bergen: Chr.Michelsen Institute, 1991); Richard Crook andJames Manor, “Enhancing Participation andInstitutional Performance: Democratic Decen-tralization in South Asia and West Africa’(Paper Prepared for Overseas DevelopmentAdministration, London, 1994); Harry Blair,“Assessing Democratic Decentralization”(Paper Prepared for USAID by the Center forDevelopment Information and Evaluation,Washington, D.C., 1995); James Manor, “ThePolitical Economy of Decentralization”(Preliminary Draft Paper Prepared for theAgriculture and National ResourcesDepartment, Sector Policy and Water ResourcesDivision, World Bank, 1996); Joel Barkan, et al.,Decentralization and Democratization in Africa(Ford Foundation Report on Decentralizationand Democratization, Grant No. 920-0572,forthcoming 1997); and several paperspresented at the Annual Meeting of theAmerican Political Science Association, August,1996): Joel Barkan, “Decentralization,Democratization, Déja Vu?: ContemporaryLessons From Anglophone Africa”; Harry Blair,“Supporting Democratic Local Governance:Initial Concepts and Some PreliminaryFindings”; Jennifer Widner, “State-Building andCommunity: Understanding LocalGovernmental Performance in Africa.”12 For example: Jeni Klugman, Decentralization: ASurvey of Literature from a Human DevelopmentPerspective (New York: United Nations, HumanDevelopment Report Office, Occasional PaperNo. 13, July 1994).

Limited Coverage of Bibliographies

There is no published bibliography thatcomes close to capturing the range andfocus of the literature on decentralization.The most useful bibliography covering thegeneral literature on decentralization wasprepared in 1983 by Britain’s Common-wealth Secretariat.13 While updated andgiven a new preface in 1986, it is obviouslyout of date.14 To the authors’ knowledge nocomparable bibliography has beenattempted since the Secretariat’s effort,though useful up-dating bibliographies canbe found in many of the materials footnotedin this paper. For example: a useful, thoughbrief bibliography is found in a 1992 WorldBank analysis of administrative and marketdecentralization strategies.15

The Secretariat’s bibliography is a selectone that primarily cites readily available

13 Commonwealth Secretariat, Decentralizationfor Development: A Select Annotated Bibliography(London: Commonwealth Secretariat, 1983).Related bibliographies of the time include: L.Beyna, et al., Managing Decentralization: AnAnnotated Bibliography (Syracuse: MaxwellSchool of Syracuse University, DevelopmentManagement Research Series No. 1, 1977); D. R.de Burlo, Local Government and RuralDevelopment in the Third World: An ExploratoryBibliography (Syracuse: Maxwell School ofSyracuse University, Local RevenueAdministration Project Bibliography Series I,1980). A review of the publications of the 1980sis found in: Malcolm Wallis, “LocalGovernment and Development: A Guide to theLiterature,” Environment and Urbanization , III, 1(1991), pp. 121-9.14 Commonwealth Secretariat, rev. ed., Decen-tralization for Development: A Select AnnotatedBibliography (London: CommonwealthSecretariat, 1986), with preface by D. Conyersthat pinpoints major issues.15 Jerry M. Silverman, Public Sector Decentraliza-tion: Economic Policy and Sector Investment Pro-grams (Washington, D.C.: World Bank, AfricanTechnical Department, Technical Paper No. 188,1992).

Page 7: Methodological Issues in the Analysis of Decentralization

6

materials in English directly related todecentralization concerns and considered tohave analytical, comparative, or empiricalvalue. As such, its list of over 500 studiesreflects only a small part of the larger bodyof literature on decentralization.

The annotated Commonwealth Secretar-iat bibliography divides the reviewed litera-ture into several topics: (1) the conceptualdefinition of decentralization, focusing onforms and types and their relationship toprinciples of centralization; (2) the role ofdecentralization, with special reference toparticipation, improved management ofprograms and projects, and secessioniststrategies; (3) the management of decen-tralization and its relationship to suchissues as legal framework, political struc-ture, division of powers, planning, adminis-trative organization, financial management,and implementation; and (4) the monitoringand evaluation of decentralization reformsand programs. The Secretariat then dividedspecific studies into two types: (1) regionalprofiles and (2) country case studies.

The Secretariat’s bibliography was thesubject of a content analysis by DianaConyers, which gives a useful picture of thebasic patterns marking studies of decen-tralization.16 Her tabulation of the 537 itemsin the bibliography revealed some startingbiases in the literature. Briefly, only some80-90 items were considered by Common-wealth researchers to have a general, com-parative, or over-arching character. Theremaining studies were all region or coun-try focused. Of these, a disproportionatenumber focused on a few countries, mostnotably Nigeria, India, Papua New Guinea,The Philippines, Tanzania, Ghana, Kenya,and Sudan. The major reason for this is thatmost research on administrative decentrali-zation is funded by aid agencies and in 16 Diana Conyers, “Decentralization and Devel-opment: a Review of the Literature,” PublicAdministration and Development, IV, 2 (1984), pp.187-97.

these countries there has been substantialaid agency and NGO investment. Thisprobably explains why there are far fewerstudies of decentralization in Latin Ameri-can countries. As for Nigeria and India,they have been studied extensively becausethey are large countries based on principlesof federalism and have a number ofnationals who are students of comparativepublic administration.

With little effort, the authors of thispaper identified an additional 130 studies,all published after the completion of thebibliography. As with the original work, theadditional items were selected for theirquality and availability to academics andaid agency professionals. This underlinesthe fact that the number of studies pub-lished since the early 1980s is much greaterthan 130. Indeed, the authors found the lit-erature grew steadily during the two yearperiod they prepared their state-of-the-art-paper (1994-96). Nearly every development-related journal publishes several articleseach year with the word “decentralization”in the title. So too, every year edited bookson a variety of topics appear that includechapters that focus on some dimension ofthe topic. Finally, United Nations organiza-tions, aid agencies, and NGOs prepareinternal documents, some of which aremade public in documentation series, thatfocus on decentralization topics.

Further, a review of the footnotes ofmany of most emerging studies shows theyare based on a growing number of localdocuments unlikely to be available outsidethe country of focus. This is well illustratedby the citations in recent articles on Zim-babwe and Ethiopia.17 Ways need to found

17 Emery M. Roe, “More than the Politics ofDecentralization: Local Government Reform,District Development and PublicAdministration in Zimbabwe,” WorldDevelopment, XXIII, 5 (1995), pp. 833-43; John M.Cohen, “Ethnic Federalism in Ethiopia,”

Page 8: Methodological Issues in the Analysis of Decentralization

7

to assist researchers to locate and givegreater attention to such internal govern-mental or donor studies and documenta-tion. At present the only way to find thesefugitive materials is through personal visitsto government or aid agency offices or dis-cussions with other researchers working inthe local task environment.

Unlike the Secretariat’s study, whichincluded a few cases related to EasternEurope and the Antipodes, the authors’review of the post-1983 items excludedcountries previously classified as membersof the First or Second Worlds. So too, theyexcluded all statistical and planning reportsby governments on their local governmen-tal units, materials developed for trainingsessions at national institutes of publicadministration, and consulting reports car-ried out for particular aid agency projects orprograms. Needless to say, such documentsare numerous and rapidly expanding asgreater attention is given by governmentsand aid agencies to strategies of decentrali-zation.

An updated Secretariat bibliographicprofile suggests that the trends identifiedby Conners have continued. In total, only59 of the World Bank’s 90 low and middle-income countries are the subject of at leastone decentralization study available inlibrary or aid agency archives.18 However,since the 1983 Commonwealth Secretariat’sbibliography was researched, most studiesundertaken over the past decade havefocused on India, Nigeria, China, Bangla-desh, South Africa, and Pakistan, indescending order of emphasis. The authors’review reveals that a number of developingcountries have no widely available, focused

Northeast African Studies, II, 2 (New Series)(1995), pp. 157-188.18 Numbers exclude middle-income countries inthe Newly Independent States category. WorldBank, World Development Report 1990: Poverty(Oxford: Oxford University Press for the WorldBank, 1991).

studies on their patterns of decentralizationand that a number of countries have notbeen studied for some time. Some countryfocused studies do include chapters ondecentralization patterns in a few selectedcountries.19 Decentralization patterns arealso covered in the U. S. Department ofState’s Area Handbook Series, which coversmost late developing countries. Many of thestudies in this series, however, are dated.Finally, there are specialized referencebooks that profile or describe governmen-tal-administrative patterns in all countries.20

However, neither this paper’s review northe Secretariat’s bibliography counts suchsub-sections, edited chapters, or country-by-country reference profiles. Data on thecountries included and the number ofstudies of them are set forth in Figure 1(page 7).

19 Good examples of edited books on decentrali-zation, ones that combine analytical andcountry case study chapters, are: PhilipMawhood, ed., Local Government in the ThirdWorld: the Experience of Tropical Africa(Chichester: John Wiley & Sons, 1983); Cheemaand Rondinelli, Decentralization andDevelopment.20 For example: Arthur S. Banks, et al., eds.,Political Handbook of the World: 1994-1995(Binghamton: State University of New York,1995); Donald C. Rowat, ed., International Hand-book on Local Government Reorganization: Contem-porary Developments (Westport: GreenwoodPress, 1980).

Page 9: Methodological Issues in the Analysis of Decentralization

8

NUMBER OF REFERENCES NAME OF COUNTRY

1-4 Algeria, Barbados, Benin, Burma, Cameroon, Chile,Costa Rica, Cuba, Ecuador, Ethiopia, Fiji, Gambia,Guatemala, Guyana, Honduras, Iran, Ivory Coast,Jamaica, Malawi, Morocco, Nepal, Nicaragua, Niger,Senegal, Sierra Leone, Taiwan, Trinidad, Tunisia,Turkey, Uganda

5-9 Brazil, Colombia, Egypt, Jamaica, Korea, Mexico, Peru,Solomon Islands, South Africa

10-14 Botswana, China, Indonesia, Zimbabwe

15-19 Sri Lanka, Thailand, Zambia

20-29 Bangladesh, Ghana, Kenya, Malaysia, Pakistan, Sudan

30-39 Papua New Guinea, Philippines, Tanzania

40+ India, Nigeria

Figure 1: Distribution of Country Specific Studies

This tentative and unsystematic updat-ing of the Secretariat’s bibliography sug-gests that only a handful of regional andover-arching comparative studies havebeen published for Africa, Latin America,and Asia. Examples of some of these are:James S. Wunsch and Dele Olowu’s analy-sis of the failure of the centralized state inAfrica,21 Arthur Morris and Stella Lowder’scollection of studies on decentralization inLatin America,22 and Hasnat Abdul Hye’scollection of articles on local government inAsia.23

A number of reports and studies relatedto the various forms and types of decen-tralization are published every year. These

21 James S. Wunsch and Dele Olowu, The Failureof the Centralized State: Institutions and Self-Gov-ernance in Africa (Boulder: Westview Press,1990).22 Arthur Morris and Stella Lowder, eds., Decen-tralization in Latin America: An Evaluation (NewYork: Praeger, 1992).23 Hasnat Abdul Hye, ed., Decentralisation, LocalGovernment Institutions, and Resource Mobilization(Comilla: Bangladesh Academy for RuralDevelopment, 1985).

are frequently difficult to access. As aresult, it is not easy for specialists to stayabreast of the growing knowledge base.Hence, it is important to repeat therecommendation of the 1983Commonwealth Secretariat’s study that a“clearing-house” or “resource center” beestablished to keep professionals, aca-demics, and practitioners up-to-date on thistopic.24 Clearly, given the currentemergence of interest in political andadministrative forms of decentralization,some aid agency needs to fund these kindsof resource services.

Methodological DifficultiesHampering Comparative Analysis

Serious methodological issues confrontefforts to carry out studies of decentraliza-tion or to summarize the lessons learnedfrom four decades of studies and reports.Specification of these can help alert analyststo the difficulties they face in efforts to

24 Conyers, “Decentralization and Develop-ment,” p. 194.

Page 10: Methodological Issues in the Analysis of Decentralization

9

compare or consolidate the state of knowl-edge about the objectives, strategies, inter-vention guidelines, and predictable out-comes of decentralization strategies. This isthe objective of the following sections.

Confusion Over and Careless Use ofConceptual Terms

Progress has been made over the past fewdecades in defining the concept of“decentralization” and identifying its formsand types. Nevertheless, it is still difficultfor academics and professionals to discussdecentralization strategies because 40 yearsof reports and studies have freighted theconcepts and terms with inconsistent defi-nitions and usages. In particular, the con-cepts related to the term are used differ-ently by academics and aid agency profes-sionals writing is English, French, Spanishand other languages. Until these conceptsare better agreed upon, made comparativeacross languages, and used consistentlyand carefully, it will be difficult to makeprogress in improving the analytical frame-works and design handbooks that facilitatecommunication among those designing andimplementing administrative decentraliza-tion strategies.

The importance of and rationale forpromoting conceptual clarity and compara-tive agreement is based on establishedmethodological cannons that are central toachieving effective description, explanation,and prediction. Specialists on the conductof inquiry in the social sciences have longbeen concerned with carefully defining themeaning of concepts.25 Only when this isdone will it be possible to make progress in

25 Foremost among these specialists are: R. M.Hare, The Language of Morals (Oxford:Clarendon Press, 1952); A. Kaplan, The Conductof Inquiry (San Francisco: Chandler Press, 1964);Bernard S. Phillips, Social Research: Strategy andTactics, 2nd ed. (New York: The MacmillianCompany, 1971).

building propositions that can contribute totheory building and have comparative util-ity to those seeking to use theory to guideaction.

The role of language in developmentstudies has recently become a major topic ofdebate. Drawing on her earlier describedreview of the literature on decentralization,Conyers argues:

...the language used in developmentstudies...is plagued by ambiguitiesand inconsistencies, which lead toconfusion, misunderstanding, andconflict in discourse.26

Conyers then goes on to address amajor methodological problem: the failureof development specialists to define andagree upon the meaning of“decentralization.” In this regard, sheargues that it will not be possible to carryout meaningful, comparative, and empiricalstudies on decentralization until there iswidespread agreement among internationaldevelopment professionals and academicson the definition of the concept. Hence, shecalls for studies to:

...reduce the terminological confusionand misconceptions caused by the of-ten ambiguous and emotive languageused to discuss decentralization.27

Considerable conceptual elaborationoccurred during the early 1980s, largelybased on the work of G. Shabbir Cheema,

26 Conyers, “Future Directions in DevelopmentStudies, p. 594. This paper builds on: R.Apthorpe and B. Schaffer, “Development PolicyDiscourse” (Paper Presented at DevelopmentStudies Association Annual Conference, Brigh-ton, 1983).27 Ibid. This also the conclusion of the USAID-funded Berkeley Decentralisation Project:Martin Landau and Eva Eagle, “On the Conceptof Decentralization” (Paper Prepared for Decen-tralization Project, University of California atBerkeley, Institute of International Studies,1981), p. 7.

Page 11: Methodological Issues in the Analysis of Decentralization

10

John R. Nellis, and Dennis A. Rondinelli.28

Today there is growing agreement in theEnglish-based literature on the conceptualdefinitions that emerged from their efforts.Yet, there are still academics and practitio-ners who are either unaware of or indiffer-ent to the progress that has been made indefining decentralization terms. As a result,even the current literature is marked bycareless use of terms. So to, debates aboutdecentralization reforms within govern-ments and at development-related confer-ences and workshops are frequentlyclouded by terminological confusion, withparties often talking past each other. This isparticularly the case at international work-shops or among members of internationalresearch and design teams.

The Cheema, Nellis, and Rondinelliapproach is based on the classification ofdecentralization by form and type. Briefly,forms of decentralization are classified onthe basis of objectives: political, spatial,market, and administrative. Each form isthen divided into types. The most elabo-rated approach to types is found in regardto “administrative decentralization,”namely: deconcentration, devolution, anddelegation.29

Since the focus of this paper is on meth-odological issues in the analysis of decen-tralization, it is useful to briefly summarizethis increasingly dominant, English-basedconceptual approach. Those interested inthe complexities of this approach, as well ascompeting conceptual approaches, particu-larly those of French and Latin Americanspecialists, should review Chapters I and IIof the earlier cited forthcoming book by theauthors.

28 Their basic work on conceptual terminology isdrawn from the studies cited in Footnote 4.29 Distinctions between these forms and types ofdecentralization are well made in: Rondinelli,Decentralizing Urban Development Programs, pp.9-15.

Briefly, “political” forms of decentrali-zation are typically used by political scien-tists interested in democratization and civilsocieties to identify the transfer of decision-making power to lower-level governmentalunits or to citizens or their elected represen-tatives. “Spatial” decentralization is a termused by regional planners and geographersinvolved in formulating policies and pro-grams that aim at reducing excessive urbanconcentration in a few large cities by pro-moting regional growth poles that havepotential to become centers of manufactur-ing and agricultural marketing. “Market”forms of decentralization are generally usedby economists to analyze and promoteaction that facilitates the creation of condi-tions allowing goods and services to beproduced and provided by market mecha-nisms sensitive to the revealed preferencesof individuals. This form of decentralizationhas become more prevalent due to recenttrends toward economic liberalization, pri-vatization, and the demise of commandeconomies. Under it, public goods andservices are produced and provided bysmall and large firms, community groups,cooperatives, private voluntary associa-tions, and NGOs. Finally, “administrative”decentralization is the focus of lawyers andpublic administration professionals seekingto describe or reform hierarchical and func-tional distribution of powers and functionsbetween central and non-central govern-mental units.

Much of the literature on decentraliza-tion is focused on only one of these fourforms of decentralization: administrative.Again, briefly, there is growing agreementthat this form is characterized by threetypes. “Deconcentration” is the transfer ofauthority over specified decision-making,financial, and management functions byadministrative means to different levelsunder the jurisdictional authority of thecentral government. This is the least exten-sive type of administrative decentralizationand the most common found in late devel-

Page 12: Methodological Issues in the Analysis of Decentralization

11

oping countries. “Devolution” occurs whenauthority is transferred by central govern-ments to local-level governmental unitsholding corporate status granted understate legislation. Federal states are by defi-nition devolved, though the extent oflegally defined and shared powersdevolved by the federal government tolower-level governmental units can be quitelimited. Devolution, such as the establish-ment of chartered municipal authorities isnot common in unitary states, largelybecause many late developing countries arecharacterized by weak central governmentsweary of losing political or administrativecontrol to local governmental units.“Delegation” refers to the transfer of gov-ernment decision-making and administra-tive authority and/or responsibility forcarefully spelled out tasks to institutionsand organizations that are either under itsindirect control or independent. Most typi-cally, delegation is by the central govern-ment to semi-autonomous organizationsnot wholly controlled by the governmentbut legally accountable to it, such as stateowned enterprises and urban or regionaldevelopment corporations. Increasingly,central and local-level governmental unitsare delegating tasks, such as refuse collec-tion and road repair, by contract to privatefirms.

Recently, additional types of adminis-trative decentralization have been sug-gested: for example, top-down and bottom-up principle agency,30 privatization-deregu-lation,31 and hybrid.32 Descriptions of these

30 Silverman, Public Sector Decentralization, pp. 2-3.31 Privatization is described as a type in: DennisA. Rondinelli, James S. McCullough and RonaldW. Johnson, “Analysing Decentralization Poli-cies in Developing Countries: A Political-Econ-omy Framework,” Development and Change, XX4/3 (1989),pp. 72-4.

additional types will not be presented herebecause most specialists have concludedthat principal agency and privatization canbe incorporated effectively within the ac-cepted definition of “delegation” and thathybrid is not a type but a description of his-torical reality in countries that haveexperimented with different administrativedecentralization interventions since inde-pendence, with a resulting mixture of typesin the delivery of particular public sectortasks, a methodological issue that will beaddressed shortly.

As noted earlier, a problem with theconceptual definitions for the types ofadministrative decentralization just sum-marized as the growing majority view isthat their supporters are largely confined tothe those working and writing in English. Itis not uncommon to find those working inother languages to be unfamiliar with thisprogress. For example, the proceedings of aLatin American workshop sponsored bythe United Nations illustrate that theconcept of “decentralization” was usedthroughout with no distinction by type,leaving the discussants, as well as thereaders of the proceedings, to decide whattype of administrative decentralization wasbeing discussed.33 More grievously, atanother United Nations sponsoredworkshop on the topic in FrancophoneAfrica the following French-based andconflicting definition was given:

Decentralization is defined as thetransfer of responsibilities and finan-cial resources from the central gov-ernment and its bodies to: (1) admin-istrative units located in the regions(delegation); (2) regional or localauthorities (devolution); (3) publicsemi-autonomous public organisms atthe national, regional or even local-

32 Silverman, Public Sector Decentralization, pp.15-17.33 United Nations, Descentralización en AmericaLatina.

Page 13: Methodological Issues in the Analysis of Decentralization

12

level – (uni- or multi-functional de-concentration); and (4) organisms thatare not part of the state or public localauthorities, that is to say private forprofit or not for profit volunteer or-ganizations (privatization).34

This definition and the terms used areat variance with the just described majorityview. For example, there is no indicationthat the definition is focused on adminis-trative decentralization as opposed to otherforms of decentralization; what is describedas “delegation” appears to be“deconcentration; and the terms “uni- ormulti-functional deconcentration” are notfound in the English-based literature ontypes of administrative decentralization.This problem results from the fact that thedocument is in French, and French special-ists have shown themselves not to be par-ticularly aware of the English-based litera-ture. Since they typically do not cite thisliterature it is difficult to know if they disa-gree with or reject it. Given the pressures tobetter design and implement administrativedecentralization reforms and programs,how can professionals and academics workon a comparative international basis if suchinconsistency is prevails within workshops,seminars, and publications?

In this regard, a number of French spe-cialists use the word “decentralization” tomean “devolution.” This was the case inwith a discussed paper written in Englishby a French expert for the World Bank.35

Clearly, careless translation or acceptance ofthe French notion that “decentralizationequals devolution” undermines the concep-

34 United Nations, Séminaire sur la décentralisationen Afrique, p. 4.35 Prud’homme, On the Dangers of Decentraliza-tion, ignores the English literature and arguesthat: “Decentralization is an ambiguousconcept, its borders not welldefined...decentralization is commonly used torefer to what is described above as devolution.This paper follows that usage.”

tual clarity required to formulate usefulanalytical frameworks and guidelines.

Finally, a good example of problemscaused by terminology is found in the useof the term “municipio” in regard to local-level governance in Latin America. In manyLatin countries a municipio can be an urbanarea, a zone within an urban area, a mixedurban-rural area, or an entirely rural area.Yet, careless specialists writing in Englishhave translated the term as “municipality”and treated this governmental unit from theperspective of British local government. It ispossible to attend international workshopswhere those from Latin America are talkingabout their experiments in promotingdecentralization to municipios (Bolivia, forexample) while their counterparts fromAmerica or Britain think “urban municipal-ity” and fail to realize that these experi-ments are taking place on a country-widebasis.

Beyond these problems related to lin-guistic biases, there is a further problem. Anumber of academics continue to inventnew concepts and definitions and govern-ments or use the terms according to theirown predilections. This leads to confusionin the literature and among academics andaid agency professionals.

For example, academics contracted byUSAID invented new terms in publicationsunder the “Decentralization: Finance andManagement Project.” In this regard theyidentified “polycentric” or “non-centralgovernance” as types of administrativedecentralization that should be promotedby countries willing to pursue democraticpolitical decentralization.36 Close inspectionof these newly proposed types reveals that

36 Elinor Ostrom, Larry Schroeder, and SusanWynne, “Institutional Incentives and RuralInfrastructures” (Draft State-of-the-Art PaperProduced for USAID Funded Decentralization:Finance and Management Project, April 1989),pp. 165-68.

Page 14: Methodological Issues in the Analysis of Decentralization

13

they are little different from devolution.37

Governments, too, have a tendency toinvest new terms. For example, in the 1930sBolivian officials formulated the term“nuclearization” to ensure that certain cul-tural responsibilities remained under thecontrol of indigenous Andean populationswhile literacy and education programswere being implemented by the center.38

Careless use of decentralization con-cepts also results from a literature that overthe past 40 years has used a wide range ofterms to describe lower-level hierarchies ofgovernance. For example, it is possible tofind within a single study a wide variety ofterms describing these units, such as“regional government,” “district govern-ment,” “field administration,” “rural gov-ernment,” “urban government,” “municipalgovernment,” and “local authority govern-ment.” In the past, the terms useddepended on the preferences and objectivesof authors. For example, in their 1974ground-breaking monograph on the effectof decentralization and local organizationson Asian development, Milton J. Esmanand Norman Uphoff used the term “localgovernment,”39 even though the term wasviewed by many specialists as being con-fined to classical British local governmentauthorities or chartered municipalities, asworked out by Lord Rippon in the 19th cen-

37 This is also the conclusion reached by Silver-man, Public Sector Decentralization, p. 49, fn 1.38 Peru was the first country to try the nucleimodel for education: Andres Cardo, “Peru: theEducational Reform and the Nuclear System,”Prospects, XI, 2 (1981), pp. 180-191. For anotherexample, see: Carlos E. Olivera, “The Nucleari-zation Programme in Costa Rica,” in EducationalAdministration and Multilevel PlanImplementation: Experiences from DevelopingCountries, edited by Carlos Malpica andShapour Rassekh (Paris: IIEP, 1983), pp. 100-18.39 Uphoff and Esman, Local Organization forRural Development.

tury.40 Alternatively, in a comparative studyof a wide range of late developing countriesother specialists used the term“decentralized government.”41 Both sets ofauthors were talking about the same kindsof lower-level governmental units.

In sum, linguistic preferences, inventionof new terms, and inconsistent use of estab-lished terms creates methodological confu-sion in any comparative review of decen-tralization. Perhaps more importantly, italso complicates the presentation of com-parative, international analytical frame-works for guiding the design and imple-mentation of administrative decentraliza-tion strategies. For this reason, it is impor-tant to be conscious and explicit aboutterms used. Obviously, these problemswith conceptual clarity and terminologycannot be solved by fiat. However, it isessential that academics and aid agencyprofessionals involved withdecentralization strategies be aware of themethodological and analytical problemsgenerated by conceptual carelessness orinventiveness and seek to use the commondefinitions of forms and types.

Diversity in the Focus of Studies

One of the major tasks faced during thedesign of an inquiry into forms of decen-tralization is what the study will focus on.The literature exhibits great diversity in theresolution of this task. Among the kinds ofquestions that can be addressed arewhether to: (1) compare several countries orexamine only a single country; (2) reviewall or selected local-level governmentalunits within a given country; (3) selectcountries or local-level governmental unitsaccording to theoretical principles or on the

40 Philip Mawhood, “Decentralisation and theThird World in the 1980s,” Planning and Admini-stration, XIV, 1 (1987), pp. 12-13.41 Rondinelli, Nellis, and Cheema, Decentraliza-tion in Developing Countries.

Page 15: Methodological Issues in the Analysis of Decentralization

14

basis of research access and funding agencydemands; (4) use the same researcher, if thestudy is comparative across a number ofcountries, for all studies or a singleresearcher to synthesize studies carried outby a larger group of researchers assigned todifferent countries; (5) use a rigorous or aflexible research framework for compara-tive country studies; (6) look at the fullrange or a single type of administrativedecentralization; (7) generalize about orgive specific attention to inter-local gov-ernmental unit variability; (8) focus on thehistory of decentralization patterns or onlyon the current patterns of decentralization;(9) focus on expected outputs of local-levelunits or broader development outcomesaccompanying deconcentrated, devolved,or delegated interventions; (10) emphasizequalitative or quantitative methodologies;(11) seek to establish causal explanations,and, if so, what the key variables will beand how they will be controlled over time;(12) try to insulate analysis from interna-tional factors and forces; (13) draw on thestandard stock of quantitative data, legisla-tion, regulations, and published studies andreports or carry out original fieldwork; (14)seek to obtain local-level information to testfor the extent and variability of the patternsbeing studied, particularly because datacollection on such patterns can take consid-erable time and have substantial cost impli-cations; and (15) look at counter-factualquestions relative to decentralization, par-ticularly in regard to “what might havehappened had there been no intervention orhad a different type of intervention beenattempted.”

Each of the comparative and country-specific studies reviewed by the authors aspart of their larger study has answered thisset of questions in a different way. Signifi-cantly, this makes it difficult to use the lit-erature reliably to generate a set of inter-subjective propositions or guidelines rela-tive to administrative decentralizationstrategies and interventions.

Comparison of both comparative andtargeted studies is further complicated bythe fact that administrative, political, spa-tial, and market forms of decentralizationinitiatives or reforms are affected by anumber of external variables. Differentspecialists present varying lists of suchvariables. For example, one recent WorldBank study specified: (1) vibrancy of civicassociations; (2) exposure of local-level gov-ernment officials and the citizens theyadminister or serve to information on pub-lic affairs; (3) levels of ethnic, religious, andinterregional conflict or compatibility; (4)education of local-level government officialsand the inhabitants under their jurisdiction;(5) origin, training, and stability of localgovernment public servants; (6) patterns ofdemographic and migratory change; (7)levels of development, market integration,and urbanization; (8) patterns of economiccontrol and their relationship to theinteraction between local officials, elites,and citizens; (9) extent and pattern oflinkages between central and local-levelgovernments; (10) patterns of central-localgovernment control over revenue andbudgets; (11) patterns of allocation relativeto capital and recurrent functional respon-sibilities; and (12) logistical and communi-cation linkages and patterns.42 This exhaus-tive and explicit list is an exception to therule. Further, the variables listed are quitedifferent from those listed in other studies.Indeed, many studies simply do not offerspecification of what the key variables are,even in comparative exercises.

Whichever explicit or implicit variablesan analyst chooses, gathering data on themacross a range of countries and local gov-

42 Jennifer A. Widner, “Evaluating the Influenceof Socio-Cultural Variables on the Performanceof Local Governments” (Paper Prepared for theWorld Bank on Research Design Issues forMeasuring Governmental Effectiveness ofDecentralized Subnational Governments, Wash-ington, D.C., May 1994) pp. 11-19.

Page 16: Methodological Issues in the Analysis of Decentralization

15

ernmental units is likely to be time consum-ing and costly, particularly if empiricalmethodologies aimed at generating propo-sitions are used. Again, studies of the formsand types of decentralization vary greatlyin regard to the extent to which they gatherthe information and data required to ad-dress these types of comparative variables.

Authenticity and Variability Problems

Substantial obstacles to comparison ofstudies are also generated by severalauthenticity factors. It is often difficult forthose using published case studies todetermine the amount of time and effortresearchers have allocated to verifying, inthe field, the findings they present. That is,it is difficult to know if the administrativefunctions decentralized to public, quasi-public, or private units by legislation andsupporting regulations correspond with theactual authority and responsibility held bythese units. This is particularly the casewith many studies that are compiled incapital city offices from legal documentsand interviews with central governmentofficials. Clearly, urban-tarmac biases, rainyseason problems, and other constraintsmake it difficult to get to more remote areasto measure the extent and variability oflocal governance in a given country.43 Forthis reasons, findings in some studies areunreliable because a nation-wide conclusionis deduced from findings in readilyaccessible local governmental unit.

Authenticity problems also arisebecause it is often difficult for researchers togain the data necessary to evaluate thedegree to which financial functions havebeen transferred under an administrative

43 This problem is well described as spatial,urban and tarmac biases in: Robert Chambers,Rural Development: Putting the Last First(London: Longman, 1983), pp. 13-18.

decentralization intervention.44 Most gov-ernmental units, parastatals, firms, andNGOs are secretive about revenue andbudgetary matters. So too, many studentsof administrative decentralization are nottrained or interested in local revenue andfinance questions. As a result, in moststudies one finds that this major indicatorof decentralization patterns is eithermissing or glossed.

Further, many studies do not determineor establish if viable local communities andtheir citizens are sufficiently empowered tocarry out devolved or delegated public sec-tor tasks.45 Robert Putnam’s recent work inItaly demonstrates the extent of inquiryrequired to evaluate local governmentauthenticity.46 In that study he reviews thevibrancy of civic and other associations,voter turnout in referenda, attitudes of localofficials to political equality and compro-mise, and local citizen feelings aboutpower, justice, and government perform-ance. Few studies of devolvement in thelate developing world go this far to verifyor test the authenticity of administrative

44 This problem is discussed at length in: NaomiCaiden and Aaron Wildavsky, Planning andBudgeting in Poor Countries (New York: Wiley,1974); A. Premchand, ed. Government FinancialManagement: Issues and Country Studies(Washington, D.C.: International MonetaryFund, 1990).45 Civil society is the sphere of social interactionbetween the household and the state which ismanifest in norms of community cooperation,structures of voluntary association, and net-works of public communication. See, for exam-ple: Michael Bratton, “Civil Society and PoliticalTransition in Africa,” Institute for DevelopmentResearch, XI, 6 (1994). For a review of the rela-tionship between the state and civil society see:John W. Harbeson, Donald Rothchild, andNaomi Chazan, eds., Civil Society and the State inAfrica in Africa (Boulder: Lynne ReinnerPublishers, 1994).46 Robert Putnam, Making Democracy Work(Princeton: Princeton University Press, 1993).

Page 17: Methodological Issues in the Analysis of Decentralization

16

decentralization reforms and programs.Nor would they have the historical evi-dence or background data to carry out aninquiry comparable to Putnam’s. Yet, thepropositions on civic society generated byPutnam has led a number aid agency pro-fessionals to use decentralization strategiesas a way to promote the “civil associations”Putnam found to be so central to explainingpolitics and development in Northern Italy.

The difficulties these and other meth-odological problems raise for comparativeanalysis were persuasively demonstrated ina 1993 exchange over Dele Olowu and PaulSmoke’s efforts to use case studies to com-paratively determine what makes for suc-cessful, devolved local-level governance inAfrica. Olowu and Smoke’s methodologyidentified several factors critical to success:(1) location in an area with an adequateeconomic base; (2) well-defined responsi-bilities in a satisfactory legal framework; (3)capacity to mobilize sufficient resources; (4)supportive central government activities;and (5) appropriate management practices,including development of productive inter-nal and external relations and satisfactoryresponsiveness to constituents.47

Subsequently, Ole Therkildsen chargedthat: (1) these indicators should be supple-mented by attention to equity, political plu-ralism, and resource mobilization; (2) thenumber of cases were limited and not sub-ject to comparable standards; and (3) inade-quate attention was given to methodologi-cal questions of validity and causality.48

Olowu and Smoke responded that whilethey recognized these points, it was notpossible to carry out scientifically designed

47 Dele Olowu and Paul Smoke, “Determinantsof Success in African Local Governments: AnOverview,” Public Administration and Develop-ment, XII, 1 (1992), pp. 1-17.48 Ole Therkildsen, “Successful African LocalGovernment: Some Methodological andConceptual Issues,” Public Administration andDevelopment, XIII, 5 (1993), pp. 51-5.

comparative studies of devolved authoritybecause: (1) it is difficult to select casestudies along scientific lines; (2) it is toocostly in terms of data and time to carry outa methodologically satisfactory study of thequality needed to deal with the lack ofcomparability between and within coun-tries; and (3) it is far more helpful to thoseaddressing administrative decentralizationissues to formulate general sets of insightsthan to develop empirically generated uni-versal laws.49 Clearly, as this exchange sug-gests, it will be a long time before studies ofdeconcentration, devolution, or delegationsolve the range of methodological problemsjust outlined.

In sum, because of thesemethodological problems, action-orientedcomparative propositions and guidelinesbased on case studies are difficult to verify.Hence, developing analytical principles anddesign or implementation guidelines isfraught with risk, for they are twiceremoved from empirical reality.

Negative Social Science

Many case studies focused on experienceswith deconcentration, devolution, or dele-gation are marked by negative social sci-ence. As a result, far more attention is paidto the problems and failures of strategies,reforms, and programs than to potentiallypositive impacts of such interventions.Negative analysis, the search for what wentwrong, affects efforts to learn from casestudies and generate useful analytical prin-ciples for government and aid agency deci-sion-makers, professionals, and practitio-ners.50

49 Paul Smoke and Dele Olowu, “SuccessfulAfrican Local Government: Methodological andConceptual Issues Reconsidered,” PublicAdministration and Development, XIII, 5 (1993),pp. 507-14.50 The need for and the scarcity of successfuldescriptions of developmental interventions is

Page 18: Methodological Issues in the Analysis of Decentralization

17

On the other hand, the dominant pat-tern among those writing comparativeanalytical reviews of administrative decen-tralization is to accept failure while seekingto offer action-oriented advice for betterdesign and implementation of such reformsand programs. Yet rarely do these over-arching manuals acknowledge the effects ofthe negative character of the case studiesfrom which they are drawing their propo-sitions from.

The problem of negative social sciencein development studies is well described byRobert Chambers. He argues there are twocultures in Western universities and aidagencies:

...a negative academic culture mainlyof social scientists, engaged in unhur-ried analysis and criticism; and amore positive culture of practitioners,engaged in time-bounded action.Each culture takes a poor view of theother and the gap between them isoften wide...Academics are trained tocriticize and are rewarded for it.Social scientists in particular aretaught to argue and to find fault...Asupposedly successful project is a redflag to some academics, a challenge tosee whether it can be turned into afailure by finding hidden harmfulefforts or errors which officials try toconceal.51

This observation is significant, becausestudies based upon the case studies ofnegative social science done by academicsstand in Chamber’s “wide gap” as theyseek to learn the lessons of comparativeexperience. Both the methodologicalproblems reviewed above and the negativecharacter of most studies make it verydifficult for those seeking to specifyanalytical frameworks, propositions, and

persuasively made in: Samuel Paul, ManagementDevelopment Programs: The Lessons of Success(Boulder: Westview Press, 1982).51 Chambers, Rural Development, p. 30.

guidelines that can be useful to governmentofficials or aid agency professionals.

Those making such translations are, inmost cases, academics funded by aid agen-cies. To be sure, practitioners recognize thatthe critical attitudes of negative social sci-entists have made a substantial contributionto the understanding of administrativedecentralization. But the academics arecommitted, indeed employed, to assist theprofessionals and practitioners in actingpromptly and effectively. As a result, theymine the negative experience to generateaction-oriented insights. If the assumptionsand arguments that administrative, politi-cal, spatial or market decentralization canplay an important role in assisting centralgovernments to address problems of devel-opment and basic human needs are correct,then their act of translation is probably jus-tified. Nevertheless, it is imperative thatpractitioners drawing upon their positivefindings and suggestions recognize thatthey are distilled from a largely negativeliterature and may be flawed as a result.

Decentralization Naïveté

John D. Montgomery is arguably the majorproponent of the view that central policiesand programs can be carried out effectivelyand sustainably through strategies ofadministrative or political forms of decen-tralization. For a number of years he stud-ied how administrative and political formsof decentralization could facilitate theimplementation of national programs ofreforms. In his classic study of the imple-mentation of land reforms, he demon-strated, perhaps for the first time empiri-cally, that when programs were carried outby centralized processes only they tendedto strengthen local bureaucratic power butdid little for peasant incomes or politicalpower, that deconcentrated processesshowed more local benefits, and thatdevolved processes had the best prospectsfor introducing distributive or democratic

Page 19: Methodological Issues in the Analysis of Decentralization

18

results, though not perfectly.52 But in pro-viding this evidence in support of adminis-trative or political decentralization, he didnot argue for severely limiting the centerand transferring government power to localgovernments, communities, and people.

A major problem plaguing those writ-ing about the rationale and strategy foradministrative or political decentralizationis summed up in the title of Montgomery’swell known article: “The Populist Front inRural Development: or Shall We Eliminatethe Bureaucrats and Get On With theJob?”53 There, he points out that throughoutthe 1970s a growing number ofdevelopment specialists challenged thestatist view and lost confidence in thecapacity of central ministries to effectivelyand efficiently promote developmentobjectives. At the same time, they becameincreasingly attracted to strategies thatwould increase the involvement of localcommunities and private sectororganizations in the performance ofgovernment functions at the local level,particularly in regard to the design andimplementation of projects.

This trend was clearly recognized by theBerkeley Decentralization Project, whichpointed out that a growing number of pro-ponents of devolution and delegation

52 John D. Montgomery, “Allocation Authorityin Land Reform Programs,” AdministrativeScience Quarterly, XVII, 1 (1972), pp. 62-75;“Land Reform and Popular Participation: SomePossibilities for the Philippines,” Solidarity, X(Jan./Feb. 1976). Among other applications, heextended this evidence to integrated ruraldevelopment projects in: “Decentralizing Inte-grated Rural Development Activities,” Cheemaand Rondinelli, Decentralization andDevelopment.53 John D. Montgomery, “The Populist Front inRural Development: Shall We Eliminate thebureaucrats and Get on With the Job?” PublicAdministration Review, XXXIX, 1 (1979), pp. 1-8.

tended to base their advocacy on implicitand explicit assumptions that:

...individuals and small groups knowbest their own self interests, that com-petition between them leads to effi-ciencies, and that there is someroughly equal initial endowment ofcapacity.54

Worried about this trend, the Berkeleygroup noted in the early 1980s that suchassumptions may be inappropriate if animportant goal of government is to estab-lish control over often diverse and deeplyconflictual political, economic, and societalinterests, and that such competition canappear to central government politiciansand decision-makers as wasteful of energyand resources. Indeed, a central finding ofthe Berkeley group was that effectivedevelopment and decentralization strate-gies requires an strong public sector center.“Strong” is not defined in terms of defenseforce, police, or any other similar notions.Rather, it is defined in terms of the compe-tency of the public sector to carry out allo-cative functions long identified by RichardA. Musgrave, namely: (1) stabilization andmaintenance of high levels of employmentand output; (2) achievement of a desireddistribution of wealth and income; and (3)efficient allocation of resources.55

Nevertheless, by the mid-1980s argu-ments for devolution and participation hadbeen consolidated in the concept of“people-centered development.”56 Since

54 Stephen Cohen, John Dykman, Erica Schoen-berger, and Charles Downs, “A Framework forPolicy Analysis” (Paper Prepared for Decen-tralization Project, University of California atBerkeley, Institute of International Studies,1981), p. 149.55 Richard A. Musgrave, The Theory of PublicFinance (New York: McGraw-Hill, 1959).56 Basic thinking on this development strategy isfound is such studies as: David C. Korten andRudi Klauss, eds., People Centered Development:Contributions Toward Theory and Planning Frame-

Page 20: Methodological Issues in the Analysis of Decentralization

19

then there has been a growing number ofarguments for severely reducing the role ofthe centralized state and transferring powerthrough devolution to local governmentalunits or delegation to local civic associa-tions, private sector firms, and NGOs.

Related to this movement was a shift inattention of some specialists from concernsabout better governance to ways topromote local organizations that areprotected from centers held to be predatoryand unable eliminate rural and urbanpoverty. That is, at the same time thatspecialists were elaborating upon principlesof “people-centered development,” theyshifted their primary focus from addressingtraditional public administration concernsabout public sector services to formulatingstrategies for dealing with problems ofinequality in rural and urban areas. Herethe emphasis was more on developingstrategies to address poverty by gainingaccess to local knowledge andorganizations than on formulating guide-lines for designing and implementing moreeffective administrative, political, spatial, ormarket decentralization reforms and pro-grams.57

works (West Hartford: Kumarian Press, 1984);Guy Gran, Development By People: Citizen Con-struction of a Just World (New York: Praeger,1983).57 This trend is well illustrated in the titles offinal reports for two important USAID-fundeddecentralization studies: Cornell’s 1974 LocalOrganization for Rural Development(subsequently published as: Uphoff and Esman,Local Organization; and Berkeley’s 1982Institutions of Rural Development for the Poor.Leonard and Marshall, Institutions of RuralDevelopment for the Poor. The Berkeley project isparticularly instructive. Because of a strongemphasis by its director, David K. Leonard,toward assisting the poor, and his view thatcentral and local governments dominate andexploit such people, the resulting study focusedprimarily on how to insulate the poor fromgovernment and to build or strengthen local

One of the best known arguments fortransferring development responsibility tothe people is found in the writing of DavidC. Korten.58 He asserts that centralizedbureaucracies and top-down aid agencyefforts have proved largely ineffective atreaching the poor rural and urban majority.In their place, he makes a case for dramati-cally increasing the role of local people inthe development process, primarilythrough voluntary associations and NGOs.Others go on to argue that only suchcommunity-based organizations can dealwith difficult resource-based problems.Most notable here is the work of ElinorOstrom, who asserts local groups in civilsociety are better at protecting thecommons than either the state or themarket.59

Clearly, there is merit in arguments forlocal responsibility, provided one does not,as has often been the case with advocatesfor new development strategies, throw the organizations that could assist them to achieveeconomic progress and a better way of life. As aresult, USAID critics felt that inadequateattention was given to formulating managementtechniques for administrative decentralizationor central-local government linkages preciselybecause the objective of the study was toprotect the poor from government. In responseto criticism that the resulting work product ofthe project was not going to assist governmentsand aid agencies seeking to promoteadministrative decentralization reforms andprograms, the director of the project wrote:“We find it impossible to disaggregate the taskof designing decentralized organizations fromthe task of devising strategies for reaching therural poor.”58 David C. Korten, Getting to the 21st Century:Voluntary Action and the Global Agenda (WestHartford: Kumarian Press, 1990).59 Elinor Ostrom, Governing the Commons: TheEvolution of Institutions for Collective Action (NewYork: Cambridge University Press, 1990); andher Crafting Institutions for Self-Governing Irriga-tion Systems (San Francisco: Institute for Con-temporary Studies, 1992).

Page 21: Methodological Issues in the Analysis of Decentralization

20

baby out with the bath water. Those whogo beyond Korten and argue thatdevelopment can only occur through localcontrol do more than make unsubstantiatedstatements, they do a disservice to poorrural and urban people.60 This is because, asnoted above, a major lesson of the past fourdecades is that without central leadership,adequate resources, and effective centraland local government administrative coor-dination and linkages, sustainable devel-opment is unlikely. Joel Samoff makes thisabundantly clear when he states:

...there is no absolute value in eithercentral direction or local autonomy.Both are important...at differentmoments. Both must coexist. That un-derstanding provides useful leaven-ing to the populist tone...61

This view has been supported by a longline of decentralization specialists, rangingfrom Brian Smith who argues that decen-tralization is not desirable in itself butdepends on the politics, economics, andsociology of particular systems and locali-ties,62 and Herbert Werlin, who provides aconvincing analytical approach to thenecessity of shifting central-local power andauthority relationships,63 to David Slatter, aMarxist who argues that decentralization

60 Naive analysis is revealed in a statement byan Aid Action staff member, who said: “Unlesslocal people have ownership or control of proj-ects they will not take responsibility...peoplehave local knowledge. Their ideas are alwayssuperior to those of experts.” John Vidal, “LocalKnow-How Works Wonders,” Guardian Weekly(July 17, 1994), p. 25.61 Joel Samoff, “Decentralization: The Politics ofIntervention,” Development and Change, XXI, 3(1990), p. 527.62 Brian C. Smith, Decentralization: the TerritorialDimension of the State (London: George Allen &Unwin, 1985).63 Herbert H. Werlin, “Elasticity of Control: AnAnalysis of Decentralization,” Journal of Com-parative Administration, II, 2 (1970), pp. 185-209.

only makes sense when it leads to greatersocial or territorial equality, which is oftennot the case.64

A recent example of “naive decentrali-zation” is found in the writing of Wunschand Olowu, academics deeply frustrated bythe failure of highly centralized Africanstates to promote effective, sustainabledevelopment and political integration anddriven to find a solution to central neglectand under-performance.65 In their stronglyworded argument for shifting power fromthe central government to local units of self-governance, they assert:

...the unleashing and encouragementof African peoples’ ‘self-organizing’capabilities through what we call‘self-governance’ is a critical pre-req-uisite for human development to oc-cur. We will argue that the excessivecentralization of institutions of gov-ernment in Africa since independencehas seriously impeded Africans’ abil-ity to do this and to engage in the artof self-governance.66

Wunsch and Olowu have concludedthat the highly centralized, unitary state isnot capable of promoting equitable, demo-cratically-based development processesbecause it is dominated by politicians andpublic servants who are concerned withruling rather than serving and with captur-ing monopoly rents and developmentbenefits. They are insensitive to the diver-sity of tasks, problems, and opportunitiesthroughout the country, prone to error inpolicy; incapable of country-wide manage-ment, slow to recognize and adjust to mis-takes, exploitative of the majority of its citi-zens; unaccountable to the people, and in-

64 D. Slater, “Territorial Power and the Periph-eral State,” Development and Change, XX, 3(1989), pp. 501-31.65 Wunsch and Olowu, The Failure of the Central-ized State.66 Ibid., p. 7.

Page 22: Methodological Issues in the Analysis of Decentralization

21

hospitable to local initiative, empowerment,and civic pluralism, particularly whenlinked to ethnicity. Further, Wunsch andOlowu argue that conventional deconcen-tration and devolution are simply replacingone model of “tutelage” and elite controlfor another, albeit at a lower-level. Forthem, effective administrative and politicaldecentralization is “self-governance,” whichthey describe as:

...institutionalized empowerment ofthe people, and the expansion of theirability to engage in collective choiceand action at a variety of scales ofhuman organization... (self-govern-ance permits) citizens to join with oneanother to take collective action: totax, contract, hire, fire, borrow, man-date, and face the consequences ofthese choices. Similarly, it can facili-tate many rule-based organizationswhich allow citizens to act in amediating position between indi-viduals and the sovereign actions ofgovernment.67

To be sure, one of the major lessons ofthe past four decades is that centralizedpolitical and administrative structures aloneare not sufficient to bring about democratic,equitable, and sustainable developmentprocesses. On the other hand, the sameholds true for grassroots organizations,which on their own have tended towardfailure. To their credit, Wunsch and Olowurecognize that central leadership andresources are important to effective devolu-tion or delegation of public sector respon-sibility if material and social improvementare to be promoted and political integrationattained. But they offer few insights intohow to formulate a strategy for simultane-ously reforming and otherwise strengthen-ing the power of both the center and localunits to effectively make a transition

67 Ibid., pp. 294, 274.

toward the improved administrativestrategies of devolution and delegation.68

Unfortunately, people-centered devel-opment initiatives can have negative effectson local communities. This is becausedevolution or delegation to weak local-levelinstitutions or organizations frequentlyleads to increased central government pene-tration. After an extensive review of the lit-erature on using local organizations forrural development, Lenore Ralston and herBerkeley colleagues concluded:

Regardless of the form selected, de-centralization in systems with weaklyorganized local units usually leads tofurther penetration by the centralpower, which more often than not re-sults in the extraction of what few lo-cal resources remain, including themost able of the local leaders. Despitelegislation and administrative or-ders...decentralization usually favorsthe central government or the localelite.69

So too, care must be taken relative toclaims about the superiority of local partici-pation, which has been found to merelycreate new arrangements for bureaucracies

68 Indeed, a review of the limited advice they dogive suggests it is too overgeneralized to behelpful. Among the generalizations offered are:“Decentralization to a large variety of local gov-ernments regulated by law makes possible theexistence of a large number of flexible, diverseorganizations ...(that allow) persons, groupsand communities to act collectively to pursuetheir various conceptions of thegood...Redundancy, duplication and overlaphave been seen as important design strategies toimprove organizational performance.” Ibid., pp.286-89. How to achieve these and other benefitsand objectives cited in this section is notarticulated.69 L. Ralston, J. Anderson and E. Colson, Volun-tary Efforts in Decentralized Management: Oppor-tunities and Constraints in Rural Development(Berkeley: University of California, Institute ofInternational Studies, 1981), p. 113.

Page 23: Methodological Issues in the Analysis of Decentralization

22

to control local development.” For example,a recent village level study carried out inThe Gambia, where central efforts to intro-duce administrative decentralization havebeen bogged down for a decade and decon-centrated line ministry field agents and dis-trict commissioners hardly touch the livesof rural people. Noting that NGOs wererapidly expanding to fill the vacuum cre-ated by a weak center, the authors went onto caution:

NGOs can do many good things butthey cannot substitute for the state atthe local-level on a national scale.Their claims of a participatory ap-proach should not be regarded asequivalent to being democratic, trans-parent, and accountable. In conse-quence, donors need to place a muchhigher priority on how they can assistlocal government in the medium andlong term (there are no quick fixes inreal-world institutional development)in The Gambia, and more generally inAfrica.70

In sum, academics and professionalsdrawing on the decentralization literaturemust be aware of the methodological prob-lems created by people-centered decentrali-zation arguments that lead more from thehope than from evidence. That is, theymust attempt to identify the ideologicalbiases of studies they use and to questionthe empirical evidence in those studies ifthey are to use them to formulateframeworks and guidelines for promotingstrategies of administrative or politicaldecentralization. A failure to do so can leadto recommendations that are unrealistic orworse.

70 Daniel Davis, David Hulme and Philip Wood-house, “Decentralization by Default: Local Gov-ernance and the View from the Village in TheGambia,” Public Administration and Development,XIV, 3 (1994), p. 268.

Difficulty Recognizing HistoricalVariability

There has been a tendency in the literatureto describe countries as dominated by aspecific type of administrative or politicaldecentralization. For example, Indonesia isdeconcentrated and Papua New Guinea isdevolved. While deconcentration may wellbe the predominant pattern in a givencountry, the literature reveals the fact that:(1) many states have deconcentratedsystems, which ensure central presencethroughout the country, coexisting withdevolved lower-level governmental unitsover which the central government hassome authority and power; (2)administrative decentralization reformsneed not be government- or territory-wide;and (3) it is possible to use different typesof administrative decentralization reformsto address different functional problemswithin the same government. For example,while Kenya claims to be experimenting indevolution it is clearly a deconcentratedsystem, even though there are stilloperating but substantially weakened localauthorities that have survived the colonialera.71 This point about mix and variability isextremely important and cannot beoveremphasized. This is particularly thecase since administrative decentralization inthe 1990s will probably be dominated by aninstitutional pluralism that is characterizedby a mix of administrative strategies.

Silverman recognizes this analyticalpoint when he states:

Most system-wide institutional ar-rangements are characterized by thecoexistence of elements of at least fourof these (types of administrative) de-

71. For example: Joel D. Barkan and MichaelChege, “Decentralising the State: District Focusand Politics of Reallocation in Kenya,” Journal ofModern African Studies, XXVII, 3 (1989).

Page 24: Methodological Issues in the Analysis of Decentralization

23

centralization, together with otherhighly centralized functions.72

He identifies this situation as a type ofadministrative decentralization, which helabels “hybrid.” So too, Philip Mawhooddescribes it as “mixed.”73 However, asnoted earlier, this is not a type ofadministrative decentralization. Rather it isan empirical observation that asgovernments address the questions of whoshould carry out what public sector rolesand tasks and how they should do it, theyend up using a range of specific types ofadministrative decentralization. Over time,some tasks and the functions that supportthem remain centrally provided, some aredeconcentrated or devolved to lower levelsof government, some are delegated to awide range of semi-public and privateorganizations, and some are recentralized.

This observation suggests that design-ing and implementing administrativedecentralization reforms and programs isfar more complicated today than in the1960s. It is no longer possible to reach sim-ple solutions based on dichotomous orcompartmentalized thinking. This isbecause most countries’ center-peripheryrelationships are characterized by: (1) his-tories of colonial intervention and rule; (2)shifting post-independence regimes andapproaches to political, administrative,spatial, and market decentralization strate-gies; (3) economic crises and changingdevelopment strategies; (4) numerous aidagency experiments and innovations; (5)informal patterns of administrative behav-ior; (6) diversity in degree of regional isola-tion, infrastructural linkages, and economicdevelopment; and, most importantly, (7)uneven implementation and adoption ofprevious reforms and programs. Further

72 Silverman, Public Sector Decentralization, pp.15-16.73 Mawhood, “Decentralisation and the ThirdWorld,” pp. 13-14.

complicating this is the fact that manyregions (e.g. provinces or districts) in latedeveloping countries are more populousand administratively complex than theentire country was at independence.

The methodological significance ofthese points is underlined by the followingobservation:

The structure of government in allcountries changes constantly, some-times rapidly and dramaticallythrough administrative reforms andreorganization, sometimes slowly andimperceptibly through the interactionof social, administrative, and politicalforces.74

In sum, it is essential to recognize thatfour decades of experience with adminis-trative and political decentralization havecreated, in many countries, complex pat-terns of deconcentration, devolution, anddelegation. “Public administration archeol-ogy” must be carried out for history has leftvery complex relations and residues inmost late developing countries, many ofwhich are neglected or unrecognized instudies. Care must be taken, therefore, instudying a given country’s administrativeand decentralization patterns or indesigning and implementing furtherreforms that affect them.

Noting such variability and complexity,some critics argue that it makes no sense toconstruct a general framework for design-ing and implementing decentralizationreforms and programs.75 For them, history,politics, and institutional structures are tooidiosyncratic to be analyzed or addressed 74 Rondinelli and Cheema, Decentralization andDevelopment, p. 9.75 Conyers notes in “Future Directions in Devel-opment Studies,” p. 598: “...the decentralizationprograms introduced in the 1970s and 1980svary so much that differences between onedeveloping country and another may be asgreat as differences between developed anddeveloping countries.”

Page 25: Methodological Issues in the Analysis of Decentralization

24

by such a general model as that whichemerged from the work of Cheema, Nellis,Rondinelli, and Silverman.76 While it is truethat proposed decentralization reforms andprograms are implemented in constantlychanging task environments that are bur-dened with history, the fact is that generalframeworks can be useful in identifyingissues and suggesting strategies. Indeed,such frameworks can be very helpful insorting out the problems and potentials thatmark countries. The problem posed by theidiosyncratic nature of task environmentsmust be recognized. However, it is not suf-ficient to reject the recent efforts to formu-late general frameworks for analyzing,designing, and implementing decentraliza-tion reforms and programs.

Misleading Linear Assumptions

There is a tendency on the part of somespecialists to view administrative decen-tralization as a linear process, inevitablyleading to a reduced central governmentand substantial devolution and delegationto local communities and private firms andorganizations. Evidence found in the litera-ture appears to be contrary to this view.Because at independence many politicalleaders found themselves governing weakstates marked by low administrative capac-ity, internal ethnic and class conflict, andlimited financial resources, they tendedtoward centralization of authority andadministrative control over the provision ofgovernment goods and services. This hasparticularly been the case in Africa, thoughit is also true of a number of Asian andLatin American countries. This is why tosome observers it appears that the post-World War II period has been marked by apendulum moving between administrative

76 For example, Samoff, “Decentralization, p.523.

centralization and decentralization.77 Thepoint here is that despite assertions in somearticles on decentralization, the availableevidence suggests that there is no linearpattern or inevitable end-state in regard toadministrative, political, spatial, or marketforms of decentralization.

Assuming Democratization is aPrerequisite to Decentralization

There is also a tendency by some specialiststo simplistically argue that democratic elec-tions and responsible local officials andcouncils are essential for effective decen-tralization. For example, this argument ismade in the proceedings of a 1993 UNDPWorkshop on decentralization:

Decentralization is a political process,not an administrative option, andsimply delegating responsibilities outto out-posted central ministry officialswithout putting them under the con-trol of locally elected leaders will notresult in the desired improve-ments...The political process is per-haps the most important mechanismfor promoting decentralization, in thatbecause of it the demand for socio-economic development and forreforming and modernizing the statemachinery has been stimulated.78

From a methodological perspective thisstatement suffers from careless use of criti-cal terminology and idealistic assumptionsabout people-centered development, as dis-cussed earlier, and it confuses the relation-ship between political and administrativeforms of decentralization. Clearly, the

77 For example, Mawhood has describedAfrica’s experience with decentralization ascharacterized by such a pendulum effect.Mawhood, Local Government in the Third World.78 United Nations Development Programme,Workshop on the Decentralization Process, pp. 24,29.

Page 26: Methodological Issues in the Analysis of Decentralization

25

greater the amount of democratic decen-tralization that occurs, the wider the rangeof administrative or political forms ofdecentralization strategies available to gov-ernment decision-makers relative to theproblem at hand. In particular, the morelikely it will be that devolved or delegatedstrategies can be designed and effectivelyimplemented. On the other hand, in coun-tries that have experienced minimal politi-cal decentralization, the more promisingadministrative decentralization strategiesare probably confined to deconcentration.But just because a country is highly central-ized does not mean that it is unable toeffectively decentralize the provision ofcollective goods and services through thefield administration type of deconcentra-tion. In sum, democratization can facilitateadministrative and/or political decentrali-zation strategies, but its absence does notnecessarily mean that such strategiescannot be efficient or effective.


Recommended