Research Report

Institutional Design Principlesfor Accountabilityin Large Irrigation Systems

Douglas J. Merrey

INTERNATIONAL IRRIGATION MANAGEMENT INSTITUTEP O Box 2075 Colombo, Sri Lanka

Tel (94-1) 867404 • Fax (94-1) 866854 • E-mail IIMI@cgnet.comInternet Home Page http:/ /www.cgiar.org

ISSN: 1026-0862ISBN: 92-9090-335-X

Research Reports

IIMI’s mission is to create sustainable increases in the productivity of irrigated agricul-ture within the overall context of water basins and the analysis of water resource sys-tems as a whole. In serving this mission, IIMI concentrates on the integration of policies,technologies and management systems to achieve workable solutions to real problems—practical, relevant results in the field of irrigation and water resources.

The publications in this series cover a wide range of subjects—from computer mod-eling to experience with water users associations—and vary in content from directly ap-plicable research to more basic studies, on which applied work ultimately depends.Some research reports are narrowly focused, analytical, and detailed empirical studies;others are wide-ranging and synthetic overviews of generic problems.

Although most of the reports are published by IIMI staff and their collaborators, wewelcome contributions from others. Each report is reviewed internally, by IIMI’s ownstaff, by IIMI’s senior research associates and by other external reviewers. The reportsare published and distributed both in hard copy and electronically. They may be cop-ied freely and cited with due acknowledgment.

3iii

Research Report 8

Institutional Design Principlesfor Accountabilityin Large Irrigation Systems

Douglas J. Merrey

International Irrigation Management InstituteP O Box 2075, Colombo, Sri Lanka

The author: D. J. Merrey was formerly a senior staff member and is now a consultant atthe International Irrigation Management Institute.

The debt the author owes to Elinor Ostrom and her colleagues, Robert Hunt, and DavidFreeman for their pioneering work should be clear. He is grateful to IIMI for support-ing the preparation of this report; to David Seckler, Director General of IIMI, for criticalcomments on an earlier version of this report; to the participants at the InternationalConference on Irrigation Management Transfer in Wuhan, China, September 1994, andto IIMI staff members at the November 1994 Internal Program Review for their criticalcomments. A substantially revised version was presented at a seminar organized by theDepartment of Irrigation and Soil and Water Conservation, Wageningen University, theNetherlands. The author is grateful to the participants for the lively discussion and use-ful comments. The report was further revised based on that discussion and commentsfrom Chris Perry and Doug Vermillion. The author is also grateful to three external re-viewers, Romana de los Reyes, Mark Svendsen, and Harald Frederiksen whose criticalcomments were very valuable and constructive. Steve Breth’s editing helped immenselyto clarify fuzzy sentences. The author was not able to accommodate all the commentshe received, as valuable as they were and, therefore, he is entirely and solely respon-sible for any remaining weaknesses in this report.

Merrey, D. J. 1996. Institutional design principles for accountability in large irrigation systems.Research Report 8. Colombo, Sri Lanka: International Irrigation Management Institute(IIMI).

/ irrigation management / government-managed irrigation systems / large-scale systems / organi-zational design / water users associations / water users / farmers’ associations / farmer participa-tion / sustainability / water rights / performance evaluation / performance indexes / participatorymanagement / privatization / policy / research methods / case studies /

ISBN: 92-9090-335-XISSN: 1026-0862

© IIMI, 1996. All rights reserved.

Responsibility for the contents of this paper rests with the author.

Editor: Kingsley Kurukulasuriya; Consultant Editor: Steven Breth; Artist: D. C. Karunaratne;Typesetter: Kithsiri Jayakody; Editorial/Production Manager: Nimal A. Fernando.

5iiiiii

Contents

Summary v

Introduction 1

Institutional Principles for Irrigation Management 2

Self-Governing and Government-Managed Irrigation Systems 2

Design Principles for Local Irrigation Organizations 3

Shared Management of Large Irrigation Systems 5

Institutional Arrangements for Large Irrigation Schemes 6

Institutional Design and Performance: Five Hypotheses 7

Are the Hypotheses Plausible? Analysis of Selected Cases 10

Autonomous Agencies Managing Single Systems (Hypothesis 1) 10

A Dependent Agency Managing Single Systems (Hypothesis 2) 12

Autonomous Agencies Managing Multiple Systems(Hypotheses 3 and 4) 13

Dependent Agencies Managing Multiple Systems (Hypothesis 5) 15

Recommendations 18

Methodologies for Future Research 18

What Policy Makers Can Do Now 20

Literature Cited 23

v

Summary

This report draws on the substantial literature on suc-cessful self-governing irrigation systems and appliestheir principles to the design of institutions on gov-ernment-managed systems. Its basic hypothesis is thatsingle irrigation systems managed by autonomoussystem-specific organizations accountable to their cus-tomers perform better and are more sustainable thansystems managed by agencies dependent on the gov-ernment (organizationally and financially), or byagencies responsible for many different systems. A

matrix of autonomy-dependence and single-multiplesystems is used to classify selected cases. These arediscussed to establish the plausibility of the basic hy-pothesis. Further detailed research would be useful torefine the hypothesis. This will be difficult to do withexisting data, but the report makes several sugges-tions for further research. The findings are sufficientlypersuasive that policy makers can use them in de-signing reform programs.

1

This report addresses a question that is fun-damental to the future productivity andsustainability of irrigated agriculture inmany developing countries: what institu-tional conditions and principles are mostconducive to achieving and sustaining highperformance in those gravity irrigation sys-tems currently owned and managed bygovernment agencies? The report exploresthe hypothesis that irrigation systems man-aged by financially and organizationallyautonomous system-specific organizationsthat are accountable to their customers per-form better and are more sustainable thansystems managed by agencies dependent onthe government and agencies responsiblefor many systems.

There is a substantial literature on in-digenous farmer-managed irrigation sys-tems. Researchers have generally empha-sized the strengths of such systems, identi-fied underlying principles believed to be thefoundation of their success, and proposedways governments can support their con-tinuation and improvement. But a large pro-portion of the total irrigated area in devel-oping countries receives its water throughsystems constructed, owned, and adminis-tered by government agencies. Most ofthese systems are individually relativelylarge compared with farmer-managed sys-tems in a given country. Researchers havedocumented the experiences with water us-ers associations (WUAs) at the tertiary and

Institutional Design Principles for Accountability inLarge Irrigation Systems

Douglas J. Merrey

Introduction

secondary levels of government-managedsystems, focusing on methods for organiz-ing WUAs and the potential roles, benefits,and advantages of WUAs (and rarely thedisadvantages and costs of WUAs). Some ofthis work recommends that governmentagencies must themselves change to enablethem to support WUAs.

However, this literature makes a crucialassumption: that the government agencywill continue to retain primary responsibil-ity for overall system management in thelarger systems and will share some of theseresponsibilities with farmer organizations.Although there are variations in what tasksand responsibilities are shared, turned overto farmers, or retained by the government,the most frequent pattern in Asia is one inwhich the government retains overall own-ership and financial responsibility for thesystem and control over the water resource,reservoirs, and main canals. Maintenanceand (perhaps) operation of lower-level ca-nals are turned over to WUAs. Representa-tives of farmers may or may not be con-sulted on policy issues affecting the largersystem.

The fundamental problem is that whilepublic organizations, under various formsof pressure, have agreed to share many ofthe responsibilities—especially the expensesand hard work—of system managementwith farmer organizations, there is no sig-nificant change in the power relations be-

2

tween officials and farmers. Mutual ac-countability is absent. Officials have no in-centives to foster independent WUAs.Farmer organizations remain dependent onthe public organization legally, financially,and psychologically. “Joint management” ascurrently practiced is often business asusual with cosmetic changes.

This report selectively synthesizes re-cent research results and introduces a con-ceptual framework for classifying systemsthat links two basic institutional design

principles. It then proposes five interrelatedhypotheses on the institutional determi-nants of performance. Although further re-search would be useful to refine the hypoth-eses, they are sufficiently plausible thatpolicy makers can act on their basis in de-signing policy and institutional reforms. Thereport therefore concludes with a brief dis-cussion of the methodologies that could beused for further research and makes recom-mendations for policy makers.

Institutional Principles for Irrigation Management

Self-Governing and Government-Managed Irrigation Systems

We can distinguish between systems thatare controlled and managed (and in mostindigenous systems, owned) by local userorganizations and systems that are ownedand, to varying degrees, controlled by gov-ernment agencies. The first category in-cludes both indigenous systems, which of-ten have long histories, and systems thathave been turned over by the governmentto user groups for management. Both typesare widespread throughout both developingand industrialized countries. Terms for thiscategory include “irrigation communities,”“communal irrigation,” and “farmer-man-aged irrigation systems.” In a few countries(e.g., USA, Mexico, Colombia), governmentstransfer management control to publiclychartered user-based district or irrigationassociations. In this report all these systemsare referred to as “self-governing” afterOstrom (1992). There is considerable thoughnot unlimited variety in the size, technol-ogy, and organization of self-governing sys-tems.

The other class of systems—those con-structed, controlled, and almost invariablyowned by government agencies—also in-cludes both ancient and modern systems.Various types of government agencies areinvolved including centralized departmentsusually dominated by civil engineers, areaor river-basin development authorities, andagricultural agencies. The dependence ofthese agencies on the control of political au-thorities, their financial dependence or au-tonomy, and the extent to which govern-ment control extends to the farm gate or isshared with local user groups at lower lev-els also vary.

There is no perfect or deterministic re-lationship between governance arrange-ments and size or scale (Hunt 1988). Thereare very large self-governing systems (inNepal, USA, and Argentina, for example)and quite small government-managed sys-tems (in Indonesia and Sri Lanka, for ex-ample). The general tendency in countrieshaving both types of systems, however, isfor self-governing systems to be relativelysmall and for government-controlled sys-tems to be relatively large.

3

There is a consistent though not perfectrelationship between governance arrange-ments and performance. Tang’s (1992) com-parative analysis of 47 systems showsclearly that the performance of government-controlled (or “bureaucratic”) systems isconsistently lower on several dimensionsthan that of local self-governing systems.Tang (1992, 50-57) used three outcomes asmeasures of the relative performance of irri-gation systems on which data were avail-able: adequacy of water supply, level of sys-tem maintenance, and “degree of rule con-formance among cultivators.” Arrangingthese measures on a Guttman scale, hefound they are closely related (coefficient ofreproducibility = 0.99). A chi-square analysisof rule conformance and maintenance byadequacy of water supply found that all 21of the systems with a good water supplyhad a positive rating in both rule conform-ance and maintenance, but of the 26 caseswith an inadequate water supply, only 8were rated positive on both rule conform-ance and maintenance (P <0.0001) (Tang1992, 56-57, tables 4.1 and 4.2). In compar-ing self-governing and bureaucratic sys-tems, he found that among the 36 havingcollective-choice arrangements, self-govern-ing systems were more likely to be ratedpositive on both rule conformance andmaintenance (P <0.05). For bureaucratic sys-tems, the result was the same—those withcollective-choice arrangements (n = 14)tended to be rated positive on both dimen-sions, while none of those without such ar-rangements were rated positive on both (P<0.05) (Tang 1992, 109, 116, tables 5.12 and5.13).

A comparative analysis of a sample of127 self-governing and government-man-aged systems in Nepal came to the sameconclusion (Benjamin et al. 1994). Perfor-mance was measured in terms of cropping

intensity, technical efficacy of infrastructure,and water availability. On all three dimen-sions, the performance of self-governingsystems was higher than that of govern-ment-managed systems (P = 0.00 in allcases) (Benjamin et al. 1994, 84-85, tables 6.1to 6.5).

These findings, based on statisticalcomparative analyses using specific perfor-mance data, support the general perceptionthat self-governing systems, by and large,perform better than government systems.

Other studies have also documentedimprovements in irrigation performance(defined in various ways in different stud-ies) of government systems resulting fromorganizing WUAs (Uphoff 1986, 1992;Aluwihare and Kikuchi 1991; Jopillo and delos Reyes 1988; Wijayaratna and Vermillion1994; Plusquellec 1989; Pradhan 1993; Paul19941). However, as Ostrom (1990, 167-173,180-181) has noted in reference to one ofthese cases, it is not clear how robust andenduring these newly established WUAswill be. Ostrom, Schroeder, and Wynne(1993, 219) further point out that temporarydecentralization and local organizations es-tablished as part of construction or rehabili-tation projects often do not have a signifi-cant long-term impact because the basic in-stitutional framework is not changed. Paki-stan provides a classic example of this ob-servation (Byrnes 1992).

Design Principles for LocalIrrigation Organizations

Scholars have been fascinated for decadesby the numerous successful examples of lo-cally constructed, owned, and managed irri-gation systems found throughout the world.This interest has resulted in many excellentcase studies and comparative analyses—arecent example is a study of the Chhattis

1See also Meinzen-Dicket al. (1994, app. I) for areview of this evidencebased on four broad cat-egories of performanceindicators used in theliterature: technical im-pact, productivity im-pact, financial impact,and environmental im-pact.

4

Mauja system in Nepal by Yoder (1994a).An important objective has been to identifyprinciples and lessons that can be used inpromoting or improving farmer organiza-tions in large government systems (Coward1980). Here I synthesize the major principlesemerging from the most recent work aimedat identifying the institutional principlescharacterizing successful self-governing sys-tems.

1. A supportive policy, regulatory and le-gal environment, that recognizes the ir-rigation community’s water rights(Ostrom 1992; Uphoff 1986; Vermillion1991; Yoder 1994b; Svendsen and Ver-million 1994; Vermillion and Johnson1995).

2. Capacity to mobilize resources ad-equate to meet the costs of operationsand maintenance including emergencyrepairs (Yoder 1994b; Freeman 1989;Svendsen and Vermillion 1994) andmodernization (Harald Frederiksen,personal communication).

3. Benefits exceed costs of participation,with proportional equivalence betweenbenefits and costs for each irrigator—that is, those with larger benefits pay alarger share of the costs (Ostrom 1992;Ostrom, Schroeder, and Wynne 1993;Freeman 1989; Hunt 1989; Vermillion1991; Vermillion and Johnson 1995).

4. Effective collective choice arrangementsor “organizational control of water” byusers (Hunt 1990), which will normallyhave the following characteristics:

• Organizational autonomy, withclearly defined boundaries (area andmembership), in which the users con-trol the allocation of water,2 and officialsderive their legitimacy and authorityfrom users and are accountable to users

(“internal charter of authority” in Rob-ert Hunt’s terms) (Hunt 1989, 1990;Freeman 1989; Ostrom 1992; Uphoff1986; Svendsen and Vermillion 1994;Garcés-Restrepo and Vermillion 1995).

• Financial autonomy, i.e., an institu-tional arrangement in which the irriga-tion management organization relies ondirect methods to raise most or all ofthe resources needed for operation andmaintenance (Small and Carruthers1991, 48-49; Svendsen and Vermillion1994) as well as capital investments(Harald Frederiksen, personal commu-nication).

• Single organizational entity man-ages a single infrastructural system(Ostrom, Schroeder, and Wynne 1993).

• Maintenance and conflict resolutionare tightly connected to the allocationand distribution of water and the orga-nization can enforce rules among itsmembers (Ostrom 1992; Freeman 1989;Hunt 1989; Svendsen and Vermillion1994).

• Transparent arrangements formonitoring performance (including fi-nancial) (Ostrom 1992; Hunt 1989; Ver-million 1991; Vermillion and Johnson1995).

• Nested (or federated) organiza-tional structure (Ostrom 1992; Freeman1989; Hunt 1989; Uphoff 1986; Vermil-lion 1991).

Other proposed principles whose uni-versality is less certain are the use of gradu-ated sanctions (Ostrom 1990, 1992), mainte-nance of written accounts and records(Yoder 1994b), water sufficient to meet cropwater demands (Freeman 1989),3 general as-sembly of members that chooses a commit-

2The literature based onindigenous self-govern-ing systems emphasizesthat capture of water atits source should also beunder the users’ control;in modern systemswhere self-governing or-ganizations obtain waterfrom a wholesaler, en-forceable contracts arethe equivalent.3Uphoff, Wickremasinghe,and Wijayaratna (1990)argue, based on datafrom Sri Lanka, thatfarmers are more willingto manage and maintainsystems when water isneither absolutely scarcenor abundant, but isrelatively scarce.

5

tee of officials (Uphoff 1986), use of special-ized paid staff for regular maintenance(Freeman 1989), and local recruitment ofstaff (Freeman 1989). “Leadership” is oftenproposed as a necessary factor, but it is notan institutional variable. Although leader-ship can substitute for good organization inthe short run, dependence upon individu-als’ leadership qualities alone in the absenceof strong institutions threatens thesustainability of an organization.

Shared Management of LargeIrrigation Systems

Several pilot projects and national programshave demonstrated the benefits of using so-cial organizers as “catalysts” or manage-ment consultants to help farmers to orga-nize effective WUAs (e.g., Uphoff 1992).These programs involve establishing WUAsat the tertiary level to improve cooperationamong farmers for local operation and man-agement. They differ primarily in how far“upstream” the role of WUAs is extendedand the extent to which WUA representa-tives are involved in overall system man-agement policy decisions. WUAs are oftenlimited to tertiary-level rehabilitation andoperation and maintenance (O&M) only(Pakistan, India, Egypt); sometimes, tertiaryWUAs are federated at the secondary canallevel and given responsibility for O&M atthis level (Sri Lanka, Philippines). In somecases, representatives of these secondary-level WUAs are involved in overall system-level decision making, for example througha joint project management committee (SriLanka).

However, two related issues that re-main problematic are (1) the relationshipbetween the WUAs and the governmentagency and (2) the nature of the agency it-self. The relationship between the agencyand WUAs may be formally contractual,

such as an agreement to deliver a givenquantity of water in return for an agreed feeor an agreement that the WUA will carryout maintenance and operational responsi-bilities in return for a payment. The lattermay be accompanied by a rebate on the ir-rigation service fee the association collectson behalf of the government, as in the Phil-ippines.

The key to success is accountability. Bothevidence and theory suggest that financiallyautonomous public irrigation agencies,which depend for a substantial portion oftheir funding on farmers paying fees, pro-vide better services (Small and Carruthers1991; Svendsen 1992). But it is rare to find acountry where the farmers can obtain com-pensation for losses resulting from bad irri-gation service. That is, even where contractsexist, their enforcement is difficult.

Accountability is especially problematicin large public service agencies with manydifferent stakeholders. Such agencies areusually structured hierarchically, with offi-cials’ primary accountability upward totheir supervisors and to the political level.Stakeholders of a large government irriga-tion department include politicians, the civilservice hierarchy, the users of the services(who are themselves not a homogenousgroup), other government departments andministries, private firms, and donors. Evenwhere corruption and other such problemsare minimal, it is difficult for politicians orsenior officials at the central level to effec-tively understand and represent the specificinterests of local users. Decentralizationwithin an agency may be a partial solution,but that does not solve the problem of ac-countability upwards. An accountabilitysystem based on hierarchical control alonewill not serve the interests of the most rel-evant stakeholders (Ostrom, Schroeder, andWynne 1993; Paul 1991) because the charterof authority is external, not internal. In

6

other words, officials are not accountable tousers.

Many researchers and practitioners rec-ognize that sharing control with WUAs aswell as providing effective support to self-governing systems have important implica-tions for the roles, functions, attitudes, andorganizational structures of public irrigationagencies. The most frequently recom-mended solution is that the agency undergoa process of “bureaucratic reorientation.”The National Irrigation Agency (NIA) of thePhilippines is an oft-cited example of a pub-lic irrigation agency that has been undergo-ing bureaucratic reorientation with somesuccess (Korten and Siy 1989).

To recapitulate the key point, mostscholars and practitioners advocatingWUAs for government-managed systemsassume that such organizations are compat-ible with the continuation of traditionalpublic agencies managing the higher levelsof the system. The special difficulties of de-signing effective institutional relationshipsin such systems are acknowledged, but notexplicitly or adequately addressed. Hunt(1989) is one of the few scholars to questionthe assumption that WUAs in large systemsare analogous to irrigation communitiesmanaging self-governing systems.4

Institutional Arrangements forLarge Irrigation Schemes

We have noted so far that the few specificinstitutional design principles that underliesuccessful self-governing systems support ahigh level of accountability between leadersand irrigators, that ensuring a high level ofaccountability to water users by officials incharge of government-managed systems isparticularly problematic, and that interven-tions based on an assumed analogy be-tween irrigation communities and water us-

ers associations in large government sys-tems have been only partially successful atbest. A more holistic approach is required todesign robust performance-oriented institu-tions for managing larger irrigation systemsthat are currently owned and managed bygovernment agencies.

Three of the four institutional principlesdiscussed above are broadly applicable toall irrigation systems: supportive policy,regulatory, and legal environment; capacityto mobilize adequate resources; and userbenefits being greater than costs and pro-portionally equivalent. This section focuseson effective collective choice arrange-ments—the fourth principle—and on theinteractions between two of its characteris-tics: (1) financial and organizational au-tonomy and (2) management of a single in-frastructure by a single organization. Thesetwo characteristics may be particularlypowerful predictors of performance of largeirrigation systems.

Relationship of the managing agency tothe government

The relationship between the managementagency and the government ranges from“autonomy” to “dependence” in both thefinancial dimension and the organizationaldimension. To simplify the presentation, Ihave combined these dimensions of the con-tinuum between autonomy and depen-dence.

Financial dimension. An irrigation manage-ment agency can be either financially au-tonomous, that is, self-financing, or finan-cially dependent on outside sources (gener-ally the state treasury) for most or all of itsfunds. In the latter case, it is the outsidesource—the government—that controls andallocates the level of resources available tothe managing agency. An irrigation manage-

4Hunt (1989, 1990) notesthe “weak analogy” inthe literature between ir-rigation communities—self-governing sys-tems—and WUAs.WUAs often have onlyduties, but none of therights, local control, andintegration with higherlevels characteristic ofself-governing systems.

7

ment agency may be considered only par-tially independent if it receives some fund-ing from outside sources (such as capitalfunds for construction) that are not repay-able, but is dependent on generating opera-tional funds internally.

Organizational dimension. The concept of “or-ganizational autonomy” is based primarilyon work by Hunt (1988, 1989, 1990), Ostrom(1990, 1992), Ostrom, Schroeder, and Wynne(1993), and Freeman (1989). Whether au-thority over decisions and activities is inter-nal or external depends on the source of the“charter of authority”—the state or theuser-members. Hunt (1990) extends thisidea of a charter of authority to the hypoth-esis that “organizational control of water” isthe key variable affecting farmers’ willing-ness to participate in irrigation managementtasks.

Therefore, to define the variablesclearly, autonomous refers to a charter of au-thority that is internal, i.e., by the consentand agreement of the members or share-holders, though the state may define therules governing such charters. It is thus au-tonomous from the state and accountable tothe shareholders—usually its customers.Dependent refers to a charter of authoritythat is based on an external source, usuallythe state. It is thus closely articulated withthe state and is a mechanism for state con-trol. Government-managed irrigation sys-tems are by definition dependent in termsof their charter of authority.

Relationship of the managing agency tothe irrigation system

Ostrom, Schroeder, and Wynne (1993) havesuggested that infrastructure managementby local entities is more effective when theentity manages a single system. I haveadapted this idea to large irrigation systems.

In many countries, a government agency isresponsible for hundreds, even thousands,of small systems scattered over a wide area.Monitoring performance and ensuring ac-countability for achieving defined perfor-mance objectives is a daunting task.

This variable has two extremes:

• a single agency or organizational entitywhose primary business is managementof one irrigation system (which in-cludes related drainage and flood con-trol as well as water supply infrastruc-ture)

• an organization or agency managing amultiplicity of systems

Institutional Design andPerformance: Five Hypotheses

In comparing institutional design principles,it is critical to examine their effects on theperformance, sustainability, and adaptabilityof irrigation systems.

Performance, sustainability, andadaptability

Many irrigation performance indicators aredescribed in the literature (Rao 1993 pro-vides the most comprehensive review). Sev-eral useful conceptual frameworks for as-sessing irrigation performance have alsobeen proposed (e.g., Small and Svendsen1992; Murray-Rust and Snellen 1993; Bos etal. 1994). More recently, IIMI has proposedand begun testing a “minimum set of per-formance indicators,” of which the mostcrucial describes the additional economicvalue of irrigated agriculture per unit ofwater (Perry 1996). However, at present twobasic constraints make it impossible to ap-ply a universal performance indicator forcomparative analysis of irrigation systems:

8

• There is no agreement among special-ists on what such a universal indicator(or set of indicators) might be, becauseof the variation in types of irrigationsystems, in their physical, social, andeconomic conditions, and in their objec-tives. This complexity is growing as ir-rigation systems increasingly functionas water supply systems with multipleuses.

• There are little data on the performanceof irrigation systems that can be usedfor comparative analysis. Tang (1992),in one of the few attempts, is depen-dent on making rather heroic assump-tions about the quality of performancedata, and because of the constraints im-posed by available data, he uses asample of systems that is highlyskewed toward South Asia.

The present report is not definitive be-cause of these problems. It seeks only toshow the plausibility of its hypotheses, notprove them in a scientific sense. Therefore,performance, as used here, refers to whetheran irrigation system is achieving its objec-tives (where these are clearly defined) ormeeting the users’ expectations. This judg-ment is based on data presented by the re-searchers cited below where possible,complemented by the researchers’ ownjudgments. In the cases discussed below, Ihave tried to state what criteria are beingused, but clearly the usage is subjective andimprecise.

The long-term sustainability of irriga-tion system performance is an important is-sue, but there are even less data availablethan for short-term performance. I usesustainability to refer to whether an irrigationsystem continues to meet its users’ expecta-tions over time. It is conceivable that a sys-tem generating a high economic value per

unit of water in a given period collapsesbecause of mining of a resource—watertables decline or soils become saline. Fur-ther, in many systems, rising water demand(including expansion of nonagriculturaluses of water) and increasing pollution areforcing changes in irrigation. Sustainabilitythus represents the time dimension of per-formance.

Because both conditions and expecta-tions change over time, I assume thatadaptability is a key factor underlyingsustainability. By adaptability I mean thecapacity to make changes—in organization,objectives, or infrastructure for example—to solve problems, enhance users’ welfare,and continue to meet users’ expectationsover time. Again, lack of data is a limitingfactor.

Hypotheses

I propose five hypotheses about the likelyrelationship between institutional principlesand performance. Figure 1 summarizes thehypotheses. It is a matrix that relates au-tonomy-dependence (both dimensions com-bined) and whether an agency manages asingle system or multiple systems. The firsthypothesis is the most basic; the other hy-potheses are derived from it.

Hypothesis 1. Fully autonomous organiza-tions accountable to their customers andmanaging single irrigation systems (those incell 1 of figure 1) will exhibit the highestperformance, will prove most adaptive tochanging conditions, and therefore willprove to be most sustainable.

Hypothesis 2. Dependent agencies managingsingle systems (cell 2) will exhibit mixedbut generally low performance; adaptabilityto changing environments will be poor, andtherefore sustainability will be problematic.

9

Hypothesis 3. Autonomous agencies manag-ing many systems (cell 3) may try to be in-novative and customer-oriented, but dis-persal of attention among a multiplicity ofsystems will limit accountability for particu-lar systems (except those politically fa-vored), and therefore will limit their perfor-mance; sustainability and adaptability willvary among systems but will generally belower than those in cell 1.

Hypothesis 4. Agencies that are only partiallyautonomous (i.e., organizationally or finan-cially but not both) (cell 3) will have onlylimited accountability to the users, whichwill limit their performance, adaptability,and sustainability; their performance andsustainability will be closer to those in cell4 than to those in cell 1.

Hypothesis 5. Dependent agencies manag-ing many systems (cell 4) will exhibit awide range of performance levels, but witha preponderance of low levels, and low(though variable) adaptability andsustainability.

In a recent stimulating paper, Perry(1995) suggested there are three determi-nants of “proper” irrigation management:

• defined water rights

• infrastructure capable of providing theservice embodied in the water rights

• assigned responsibilities for all aspectsof system operation

Perry argues that all other common is-sues (e.g., good maintenance, sound institu-tions) are really either a subset or a combi-nation of these three basic elements. Waterrights, infrastructure, and responsibilities in-teract and are interdependent—change inone will lead to changes in the other ele-ments. The quality and interaction of thesethree elements can be used to distinguishbetween functional and dysfunctional sys-tems, a distinction that Perry argues is offundamental importance in formulating in-terventions. The question is, are these threedeterminants alternatives to the hypothesesproposed in this report?

FIGURE 1.Matrix of irrigation system governance arrangements: Performance hypotheses.

RELATIONSHIP OF AGENCY TO GOVERNMENT

Autonomous Dependent

Agency manages 1. Achieve highest performance 2. Mixed but generally low performancea single irrigation system Most adaptive to changing conditions Low adaptability

Most sustainable Sustainability threatened[Hypothesis 1] [Hypothesis 2]

Agency manages 3. Performance will vary among systems 2. Wide range of, but generally low,multiple irrigation system but overall will be lower than cell 1, performance

higher than cell 4 Low adaptability and sustainability, withAdaptability and sustainability will vary variation among systems based onamong systems but overall will be local factors lower than cell 1, higher than cell 4

[Hypotheses 3, 4] [Hypothesis 5]

10

I argue they are not. While grantingtheir importance, they beg the issue of whatinstitutional framework is most conduciveto defined water rights, appropriate infra-structure, and organizational arrangementsfor implementation. How are water rightsto be defined and then enforced? How willinfrastructure be designed and subsequently

maintained to provide a service that fulfillsusers’ rights? And how will implementationbe organized to ensure its effectivenessthrough accountability to the users? Thisreport addresses these issues and thereforecomplements Perry 1995 in that, throughhypothesis 1, it proposes such an institu-tional framework.

Are the Hypotheses Plausible? Analysis of Selected Cases

A proper test of the five hypotheses wouldrequire precisely operationalizing terms likeperformance, sustainability, and adaptability.A sufficient number of cases would need tobe selected randomly to enable statisticalanalysis. Hunt (1988), Uphoff (1986, 1991),and others have discussed the difficulties inmeeting such rigorous standards in irriga-tion research. As noted above, there is nosingle, universally acceptable measure of ir-rigation performance that can be used tocompare systems. The data are inadequate,especially for government-managed sys-tems. Nevertheless, an analysis of casesfound in an exploratory review of literaturesuggests that the proposed hypotheses areplausible. Figure 2 groups selected systemsaccording to the cells depicted in figure 1.

Autonomous Agencies ManagingSingle Systems (Hypothesis 1)

Single systems managed by autonomousagencies (cell 1, figure 1) ought to exhibitthe best performance, be most adaptable asconditions change, and be the most sustain-able. This group includes all indigenousself-governing systems, many of which havecertainly demonstrated their sustainabilityand adaptability by surviving and continu-ously adapting for hundreds of years. It

also includes cases of very large irrigationsystems that have substantially autonomousmanagement organizations; the largest re-ported case irrigates 458,000 hectares (Hunt1988, table 1). Maass and Anderson (1978)describe contemporary cases that are largescale, have high performance in terms ofwater productivity, and have demonstratedtheir adaptability and sustainability by sur-viving for long periods of time.

The irrigation associations of Taiwanare famous for their high level of perfor-mance in terms of high crop yields and theadequacy, equity, and reliability of waterdistribution (Levine 1991; Bottrall 19815).The governance arrangements have gonethrough several stages since the occupationof the island by Japan in the early twentiethcentury and continue to evolve today. Irri-gation associations are financially autono-mous, as well as largely autonomous fromdirect government control, though closelysupervised by the government. A represen-tative assembly of water users elects a chair-man who hires staff and has overall execu-tive responsibility. In these systems, some ofwhich are quite large (up to 87,000 ha), sys-tem managers are highly accountable to theusers.6

The province of Mendoza in Argentinais another interesting case. Five river basinsystems totaling 360,000 hectares are under

5Bottrall (1981) com-pares four systems oneof which is an irrigationassociation in Taiwan.6For the past 2 years, theirrigation service feeshave apparently beenpaid by the govern-ment, not farmers, forpolitical reasons; and, in1994, the governmentbegan appointing presi-dents of irrigation asso-ciations because electedrepresentatives “wereserving their own pur-poses” (Chung-HuangHuang, personal com-munication, January1996).

11

the overall supervision of the General Irriga-tion Department (DGI), described as an“autonomous and autarchic” agency(Chambouleyron 1989). The DGI is managedby a superintendent appointed by the Pro-vincial Senate, and he is assisted by an ad-ministrative council consisting of farmer rep-resentatives. The DGI is primarily a regula-tory agency, with its own budget and rule-making capacity. Irrigation management iscarried out by the users through water usersassociations (WUAs), which are autonomousfrom the DGI and self-financing. TheMendoza system is adaptable—for example,a process of consolidating WUAs is currentlyunderway to further reduce managementcosts (Chambouleyron 1995). The system’sperformance is good in terms of water deliv-ery equity and the ratio of water deliveriesto targets (Bos et al. 1991). Again, account-

ability of the DGI and the WUAs to the us-ers is clearly very important.

Irrigation districts in the USA and Co-lombia are organizationally and financiallyautonomous and manage single water sup-ply systems. Recent case studies in bothcountries show that returns to agriculturehave been rising, costs have been contained,financial viability achieved, and the level ofservice has at least not deteriorated after thesystems were turned over to the districts(Svendsen and Vermillion 1994; Garcés-Restrepo and Vermillion 1995). Mexico iswell along in making similar reforms(Gorriz, Subramanian, and Simas 1995).

It is likely that a comparative studywould confirm the relatively high perfor-mance and capacity for sustainability ofthese types of systems, and thus confirm thefirst hypothesis.

a ORMVAs (Office Regionaux de Mise en Valeur Agricole) are a partial case: they are partially autonomous financially, but theircharter of authority is derived from the government.

b NIA is a partial or mixed case: it is partially, but not fully, autonomous in financial terms, and its charter of authority is derived fromthe government.

c In transition through a turnover program and therefore moving into cell 1 (autonomous, single system).

FIGURE 2.Matrix of irrigation system governance arrangements: Examples.

RELATIONSHIP OF AGENCY TO GOVERNMENT

Autonomous Dependent

Agency manages 1. Mendoza, Argentina 2. Egypta single irrigation Irrigation districts, USA Haryana, Indiasystem Irrigation districts, Colombia Punjab & Sind, Pakistan

Taiwan systems ORMVAs, Moroccoa

Communals, PhilippinesFarmer-Managed Irrigation Systems,

NepalIrrigation districts, Mexico (post-reform)Self-governing systems, by definition

Agency manages 3. National systems under NIA, 4. Irrigation districts, Mexico (pre-reformc)multiple irrigation systems Philippinesb Sri Lanka

NepalIndonesia technical systems (>500 ha)Most West Frontier Province &

Baluchistan, Pakistan

12

A Dependent Agency ManagingSingle Systems (Hypothesis 2)

Single systems managed by a dependentagency (cell 2 in figures 1 and 2) are usuallylarge systems. Some are sufficiently inde-pendent physically that programs to changethem to an autonomous single system (cell1) are feasible and could lead to improvedperformance in the long run (e.g., ORMVAsin Morocco). Other cases involve extremelylarge highly integrated systems that are cen-tral to the prosperity of entire countries orprovinces and that have multiple (includingnonirrigation) uses (e.g., Egypt, Haryana inIndia, Punjab and Sind in Pakistan). Thereis considerable variation in the performanceof these systems, and no single indicator isadequate for comparing them.

Egypt by many measures has a highlevel of performance: its crop yields andcropping intensity are high compared withlarge systems in other developing countries.If irrigation efficiency is measured on a ba-sin-wide basis, it is very high because ofsubstantial water reuse (Keller, Keller, andEl-Kady 1995). But this system is not, as yet,water-short; the managers attempt to, andlargely succeed in, meeting water demands.High macro-level performance is a productof having sufficient water to meet demandsand excellent climatic conditions for cropgrowth. But users have no clear waterrights, and there is no administrative capa-bility or other institutional means to man-age water scarcity. Although macro-levelperformance is high, there is wide variabil-ity in crop yields and cropping intensity, inpart reflecting differential access to water,and yields are far below the potential, giventhe growing conditions (John Mellor Associ-ates 1995). There are serious concerns aboutthe system’s future performance andsustainability as demand continues to in-crease for a fixed water supply, com-

pounded by increasing water pollution(IIMI 1995a).

Haryana State in India is often cited asa system with high performance in terms ofequity, reliability, and productivity per unitof water (Malhotra, Raja, and Seckler 1984a,1984b), but it is also facing serious waterlog-ging and salinity problems. Pakistan’s Indussystem (especially the provinces of Punjaband Sind), which is based on design prin-ciples similar to Haryana’s, suffers from lowand stagnating productivity and salinity(e.g., Kijne and Vander Velde 1992; Murray-Rust and Vander Velde 1994; Kijne andKuper 1995). Unlike Haryana, its institu-tions seem no longer able to manage thesystem effectively (Bandaragoda andFirdousi 1992). It has been argued thatovercentralized management has led to aninability to solve local problems and pre-vent them from threatening the integrity ofthe system (Merrey 1987). Like Egypt, bothHaryana and Pakistan face increasing scar-city and competition for a limited watersupply as demand increases.

These three cases are characterized byhierarchical management agencies in whichdecision making is centralized, and account-ability is upward to senior civil servantsand politicians, and not directly to their cus-tomers.7 They are organizationally and fi-nancially dependent. The governments ofHaryana in India and of Punjab and Sind inPakistan collect water rates, but these gointo the general treasury—there is no linkbetween the agencies’ budgets and collec-tions. Egypt charges no fee for irrigationservices. In all three, there are either propos-als or ongoing efforts for policy and institu-tional reform aimed at decentralizing deci-sion making and making the managementorganizations more autonomous and self-reliant in both organizational and financialterms (for Egypt, see IIMI 1995b, 1995c;Merrey forthcoming).

7During a presentationof the previous versionof this paper in theNetherlands, Jan Ubelspointed out that allmodern cases of highlycentralized manage-ment organizations forlarge systems were colo-nial creations and aretherefore a legacy leftby the colonial powers.Consistent with the“Wittfogel hypothesis”(Wittfogel 1957), theseorganizations were animportant basis for co-lonial governments’control of populations.

13

But given their size and complexity,making these management organizationscompletely autonomous is risky; reformsneed to be carried out step-by-step. Some ofthese systems are characterized by highrates of rent-seeking and are controlled bypowerful stakeholders. These stakeholderswill resist reforms, and may also capturethe benefits of well-intended reforms. How-ever, governments should persevere withreforms to improve incentives and account-ability, while exercising overall controlthrough strong regulatory and consultativemechanisms. Subsystems may be madecompletely autonomous and user-based,with enforceable contracts as the basis forreceiving and paying for water servicesfrom the larger entity. Policy making andregulation can be separated from provisionof service as suggested by Frederiksen,Berkoff, and Barber (1994).

Morocco is an interesting intermediatecase. Irrigation administration is decentral-ized to nine ORMVAs (Office Regionaux deMise en Valeur Agricole), which are semiau-tonomous regional offices. They are super-vised by the Ministry of Agriculture andAgrarian Reform and the Ministry of Fi-nance. Water charges are collected and usedto pay for O&M, but they are not adequateto cover all costs; the government fixes theprice of water and makes up differenceswith subsidies (Hofwegen 1994). ORMVAsare created and controlled by the govern-ment, which is the source of their authority.They are therefore not accountable to users.

Autonomous Agencies ManagingMultiple Systems (Hypotheses 3and 4)

Cell 3 in figure 2—autonomous agenciesoperating many systems—is nearly empty.One well-known case, the National Irriga-

tion Administration (NIA) in the Philip-pines, does not fully meet the criteria. First,NIA is only partially autonomous finan-cially. It continues to receive various subsi-dies, and it cannot set the amount of fees oreffectively enforce fee payment. Second, itscharter of authority is from the government,with no user control over NIA itself; NIA isdominated by its parent ministry. Therefore,NIA illustrates both hypothesis 3 (even au-tonomous agencies will show limited per-formance if they manage many systems)and hypothesis 4 (partial autonomy limitsperformance).

NIA is an interesting case because itsrelative autonomy has made its organiza-tional culture more innovative and perfor-mance-oriented than most systems in cells 2(single systems, dependent) and 4 (multiplesystems, dependent). NIA has been a pio-neer in developing methodologies and poli-cies to promote strong irrigation associa-tions in both “communal” systems (self-governing systems) and “national” systems,those owned and controlled by the govern-ment through NIA (Korten and Siy 1989).

NIA has created cost centers for mostnational systems, and it evaluates staff andsystem performance using a viability in-dex—the ratio of costs to expenditures. Thisprocedure is intended to motivate NIA staffto reduce costs and improve services to en-courage farmers to pay irrigation servicefees. Central to this program is the promo-tion of irrigation associations to take respon-sibility for O&M as well as fee collection inspecific subsystems (for example, a lateral).NIA has three models of joint managementin the form of contracts:

• Type I contracts involve maintenancecontracts for certain canal lengths.

• In Type II contracts, the irrigation asso-ciation also undertakes operations and

14

fee collection, keeping a certain percent-age of the amount collected over aspecified minimum.

• In Type III contracts, the association as-sumes full O&M responsibility and alsoamortizes a portion of the constructioncosts over a period of time.

Table 1 shows the status of the develop-ment of irrigation associations and the threemodels of joint management in the Philip-pines as of December 1992 (NIA 1992). Atthat time, 75 percent of the “organized” ir-rigation associations had taken a contract; ofthese, two-thirds held Type II contracts(maintenance and collection). Only 4 per-cent had taken Type III contracts (“completeturnover”).

Type III systems are not equivalent tofarmer-managed or communal systems andthe use of the term “complete turnover” by

NIA is inaccurate today.8 In the provisionson the obligations of NIA and the irrigationassociations under these contracts, NIA re-tains a stronger role than it does in commu-nal systems. More important, NIA retainsthe water rights, whereas communals havetheir own legally recognized water rights.The contracts contain no explicit referenceto ownership of the system being turnedover to the irrigation associations, and NIAretains a great deal of de facto control(Oorthuizen and Sloot 1993, 26-27).Wijayaratna and Vermillion (1994) discussthe disincentives for NIA to turn over moresystems completely to the irrigation associa-tions. Most Type III (“turned over”) systemsto date are the least viable in terms of cov-ering costs from irrigation service fees; thesystems that are most viable financially tendnot to be turned over as they are a source ofincome (“profit”) to NIA.9

8Jopillo and de los Reyes(1988, 3) refer to Type IIIas “transforming a gov-ernment-operated or na-tional system into afarmer-managed orcommunal system.”This was the original in-tention (Romana de losReyes, personal commu-nication) but was laterchanged.9In a case study ofa Type III system,Oorthuizen and Sloot(1993, 36–40) found thatthe annual amortizationpayments exceeded theirrigation services feesthe system farmers usedto pay; and both opera-tional performance andmaintenance quality re-mained at a very lowlevel. The irrigation as-sociation was successfulonly in recovery of fees.

TABLE 1.Status of irrigation association development in national irrigation systems in the Philippines, 1992.

Associations Area Members

no. % ha % no. %

NIA

Total — 645,789 100 496,760 100

Irrigation associations

Organizeda 1,745 — 531,635 82 409,215 82Registeredb 1,522 87c

Irrigation associations with contracts

Totald 1,315 75 405,814 76 278,560 68Type Ie (maintenance) 369 28 118,610 29 74,810 27Type IIe (maintenance &

collection) 900 68 273,143 67 190,525 68Type IIIe (complete

turnover) 46 4 14,061 4 13,225 4

Source: NIA 1992, 22.a Percentages relative to total NIA.b With Security and Exchange Commission.c Relative to organized irrigation associations.d Percentages relative to organized irrigation associations.e Percentages relative to irrigation associations with contracts.

15

Data on performance of the Philippinesystems are mixed. Jopillo and de los Reyes(1988, 210-212, table 77) show that irrigatedarea and cropping intensity increased afterfarmers became involved in system man-agement in the first set of national systemswhere irrigation associations were orga-nized, but they note it is difficult to separatethe impacts of physical improvements andfarmer organizations. Svendsen (1992)shows that based on official data on asample of irrigation systems, NIA’s reformshave apparently reduced system operationalcosts, while improving equity of water de-livery, and yields per hectare have remainedconstant. Merrey, Valera, and Dassenaike(1994) compare performance data for threesystems, one each in the Philippines, Nepal,and Sri Lanka, and show that the Philippinesystem performs substantially better thanthe other two as measured by water deliv-ery and agricultural and economic vari-ables. They attribute this difference to theinstitutional strengths of the Philippine sys-tem (specifically, a performance orientationand accountability for service to customers)that are lacking in the other two.

On the other hand, Oorthuizen andKloezen (1995), in a case study of the im-pact of NIA’s financial policies in a singlesystem, found that although fee collectionrose, the quality of the service declined (seealso Oorthuizen and Sloot 1993). Laurayaand Sala (1995), based on a study of foursystems in Bicol Province, Philippines, showthat performance is not very good. Averageyield per hectare is half the highest yields inall four systems. The authors suggest thisdemonstrates a significant yield gap. Theyalso suggest that the use of the viability in-dex as a measure of performance is leadingto underinvestment in maintenance andpossible deterioration of the systems. Re-searchers carrying out a restudy of theeconomy of a village in central Luzon have

found that the irrigation system managedby NIA has “collapsed,” and only thosefarmers who can afford pumps for ground-water are able to cultivate (M. Kikuchi, per-sonal communication, 1995). These observa-tions raise questions about systemsustainability.

Given the hypothesized relationship ofautonomous organizations managing singlesystems to performance, the obvious direc-tion for future reform would be to createfully autonomous user-based managemententities in each scheme (i.e., carry the cur-rent policy of turnover to its logical conclu-sion). These organizations should acquireclear water rights, thus converting such sys-tems into true “communals.” It would benecessary to devise incentives for NIA topursue this path, and it would require fur-ther reorientation of NIA to provide techni-cal and management support to irrigationassociations.10 Otherwise, the present lim-ited and dispersed accountability may in-hibit improving and sustaining performanceover the long term.

Dependent Agencies ManagingMultiple Systems (Hypothesis 5)

The cases in cell 4 of figure 2 are more dif-ficult to analyze because of the lack of sys-tematic data on the range of variation in,and overall performance of, irrigation sys-tems . A large number of the systems in thiscategory are performing poorly, and the re-sponsible agencies have been only partiallysuccessful in implementing reforms. Mexicohas embarked on a program to shift its gov-ernment-managed systems to managementby autonomous organizations accountableto users (Gorriz, Subramanian, and Simas1995). This entails transforming these sys-tems into single systems managed by au-tonomous agencies (cell 1), which we pre-

10Oorthuizen and Sloot(1993) emphasize thefailure of NIA to achieveits “democratization”objectives in the systemthey studied—the irri-gation association wascontrolled by a few in-fluential people.

16

dict will in the long run lead to improvedperformance. At present little data are avail-able on the performance impacts of this pro-gram.11

Sri Lanka and Nepal, like the Philip-pines, have been experimenting with infor-mal joint management arrangements forabout a decade. Sri Lanka and Nepal havesimilar organizational models for joint man-agement of government-owned irrigationsystems. Both include informal groups atthe tertiary levels that choose representa-tives to more formal organizations at thedistributary level (distributary canal organi-zations in Sri Lanka). A joint project man-agement committee including representa-tives of farmers and government depart-ments is intended to make overall water al-location and seasonal scheduling decisionsin Sri Lanka. In Nepal, this joint decision-making process seems less formal. Somesystems in Nepal have formal water usersassociations at the system level; these areless common in Sri Lankan systems—untilrecently they were not officially encour-aged—and their role is not clear. In Indone-sia, reforms have clearer objectives but re-main weak in terms of accountability ofmanagers to users.

Nepal

In a recent review of Nepal’s experienceswith joint management Pant, Valera, andPradhan (1992) suggest that while the short-term results are clearly impressive, thesustainability of the WUAs and the jointWUA-Department of Irrigation relationshipis doubtful. Nepal has not yet developed amanagement system that would ensure suf-ficient accountability of officials to farmers,farmer leaders to members of WUAs, orWUAs to the government.12 The large num-ber of relatively small systems spreadthroughout a large country with poor com-

munications and transport, compounded byshortage of funds and institutional weak-nesses, make it difficult to envision how thedepartment can ever manage these systemseffectively and improve their performance.A comparative study of three systems inNepal, the Philippines, and Sri Lankashowed that the Nepal system ranked low-est on all performance parameters for whichthere were data (Merrey, Valera, andDassenaike 1994); and a data-based com-parative study of self-governing and gov-ernment-managed systems in Nepal (Ben-jamin et al. 1994) also found the perfor-mance of the latter to be poor.

Sri Lanka

IIMI has been closely associated with theparticipatory management programs in SriLanka since 1986 (and I have been associ-ated with them since 1980). There are cur-rently three sets of government agenciesimplementing the participatory manage-ment policy. First, the Irrigation Departmentis trying to implement it in the small andmedium systems for which it is responsible,in the Management of Irrigation Systems(MANIS) program, usually through its ownstaff and with no external support until re-cently. Second, in 38 major schemes underthe Irrigation Department, a parallel divi-sion of the ministry in charge of irrigation ispromoting the development of the partici-patory management model as part of theIntegrated Management of Major IrrigationSchemes (INMAS) program. In some ofthese schemes, institutional organizers, con-sultants, and training programs are sup-ported by external assistance. Finally, since1992, the Mahaweli Authority has had aprogram similar to INMAS for the largeschemes under its control.

Reports from a recently completed com-parative study monitoring the performance

11Preliminary resultsfrom a survey of farmersin three turnover dis-tricts show they have aperception of improvedservice (Palacios-Velez1995).12Wilkins-Wells et al.(1995) describe a watershare system that has re-cently been pilot-tested;if this continues to besuccessful, it may be-come the government’spolicy, which is stated asbeing “the transforma-tion of agency-managedsystems to farmer-con-trolled systems”(Wilkins-Wells et al.1995, 230).

17

of 199 jointly managed systems provide amixed picture (IIMI and ARTI 1995). AllINMAS schemes in the sample were re-ported to have project management com-mittees, about 88 percent of the distribu-taries had organizations, and considerableprogress had been made in the “turnover”of distributaries to farmers. However, mostfarmer organizations remain dependent onthe department that gives them funds formaintenance. The concept of turnover is notprecisely defined. It is not ownership anddoes not include enforceable water rights.Distributary canal organizations take con-tracts for O&M from the Irrigation Depart-ment, like the Type I contracts in the Philip-pines, but unlike in the Philippines, thegovernment pays the farmers without col-lecting irrigation fees.

Overall, the comparative study by IIMIand ARTI (1995) shows that participatorymanagement has improved water distribu-tion and that maintenance quality in dis-tributaries and crop yields have notchanged. A significant minority of farmerorganizations have expanded into other rev-enue-generating activities, but they dependon government assistance for their success.The government’s O&M costs have notfallen, however expenditure has beenshifted from distributary canals (maintainedby farmer organizations, usually under acontract with the government) to main sys-tem maintenance. As in the Philippines andNepal, the program is stalled because thereis no commitment to a long-term objectiveand no willingness to consider more radicalreforms of the government agency.

In Mahaweli systems, some turnoverhas occurred and a joint management struc-ture has been put in place quickly. The sys-tems under the Irrigation Department’sMANIS program show the least progressbecause of the weak support. Overall, theinterviews carried out by IIMI revealed a

marked reluctance among both farmers andofficials of all departments to turn over fullO&M responsibility to farmer organizations.Studies of other irrigation systems provide amixed picture, with examples of both strongand weak farmer organizations (see TEAMS1992; Vimaladharma 1994). Two systems inwhich a local nongovernmental organiza-tion implemented experimental institutionalstrengthening programs have also had dis-appointing results (Athukorale, Athukorale,and Merrey 1994). In their comparativeanalysis of systems in the Philippines,Nepal, and Sri Lanka, Merrey, Valera, andDassenaike (1994) showed the performanceof the Sri Lanka system was very low on allparameters—though it was regarded by theIrrigation Department as a show-case sys-tem.

A recent review of the experience of theMahaweli Authority prior to 1992 arguesthat the impediments to strengtheningfarmer organizations and their system man-agement role lay primarily in the organiza-tional structure, incentives, and philosophyof the Mahaweli Authority itself (Merrey1995). But the problem is deeper. It is theresult of the reluctance of Sri Lanka to carrythe participatory management policy to itslogical conclusion. Neither the IrrigationDepartment nor the Mahaweli Authority(nor some future amalgamation of the two)will ever have the combination of resources,incentives, and accountability to system us-ers needed to be able to achieve a high levelof performance in the hundreds of systemsscattered around the island. Nor has thegovernment been willing to address the is-sue of financing irrigation services. Irrigationremains free, which greatly limits farmers’interest in taking over O&M responsibilities.

Radical reform to shift Sri Lankan (andindeed Nepali) systems to single systemsmanaged by autonomous agencies (cell 1),as is underway in Mexico, may be the most

18

promising long-term institutional means toachieve sustainable high performance.

Indonesia

In Indonesia, the government is turningover all systems under 500 hectares to wa-ter users organizations (i.e., shifting them tocell 1). On systems over 500 hectares, thegovernment is introducing irrigation servicefees. These fees are linked to giving farmers,through their water users associations andfederations, a clear voice in defining the ser-vice for which they are paying (Gerards,Tambunan, and Harun 1991; Gerards 1992).A recent carefully documented case study ofthis Indonesian program has confirmed the

positive association between the use of“voice mechanisms” (public pressure by us-ers on service providers) by farmersthrough their WUAs and the accountabilityof the service provider, and it has demon-strated that improved outcomes such ashigher cropping intensity can be attributedat least partially to improved accountability(Paul 1994). However, the provincial irriga-tion services, which continue to managethese larger systems, remain hierarchicaland dependent on the government both or-ganizationally and financially and continueto be responsible for multiple systems. Thenext step in Indonesia’s program should bereconsideration of the structure and role ofthe provincial irrigation services.

Recommendations

The findings stated in this report are sug-gestive and certainly not prescriptive. Theplausibility of the basic hypothesis has beenestablished. This basic hypothesis is thatsingle irrigation systems managed by sys-tem-specific organizations that are finan-cially and organizationally autonomous andaccountable to their customers, generallyperform better and are more sustainableover the long term. If this report is on theright track, the findings should be of greatinterest to countries struggling to improvethe performance and sustainability of gov-ernment-managed gravity irrigation sys-tems. Certainly there is an urgent need formore research, but policy makers can ini-tiate changes now.

Methodologies for Future Research

Only a few comparative studies of irrigationhave attempted to test hypotheses about ir-rigation performance using quantitative

data from samples of irrigation systems.Examples are the studies by Tang (1992),Murray-Rust and Snellen (1993), and Ben-jamin et al. (1994). All these studies strugglewith defining variables in a conceptuallymeaningful way and with finding sufficientdata. Data on irrigation system performanceand sustainability are particularly problem-atic. Many studies are carried out by re-searchers lacking the resources and skills tomeasure performance adequately (thoughtheir data on other dimensions are often ex-cellent); and even researchers specialized inperformance assessment do not agree on auniversally valid set of performance indica-tors. IIMI recently proposed such a mini-mum set for use in its work (Perry 1996),but at present there are few systems forwhich data are available to test these indica-tors. Establishing agreed performance indi-cators and collecting data on a sample ofsystems reflecting the diversity of irrigationconditions in the world constitute one of

19

the most important areas for research on ir-rigation in the next few years.

Anticipating that this may take sometime, another option is to continue usingimprecise proxies as, for example, Tang(1992) does. It is not necessary to wait tofind a single performance indicator (or setof indicators) applicable to all systems anddata based on this indicator, to continueseeking to understand the determinants ofperformance and how it can be improved.

A related problem is sampling. Nostudy has yet used a sample that can claimto be representative of the universe of irri-gation. This too may be impossible as theuniverse itself is not well known, but cer-tain kinds of systems favored by researchersare well-represented, while others are notrepresented at all. For example, two of thefour cases from Pakistan used by Tang(1992) are on the same irrigation system; allthe bureaucratic systems are from six Asiancountries and from one Middle Easterncountry. All but one self-governed system inhis sample are from Asia, with the Philip-pines heavily over-represented (12 of 29cases) (Tang 1992, 43-44, tables 3.1 and 3.2).A proper study using statistical measures ofassociation among variables must be basedon a more representative sample of systems.To create such a sample, it would be neces-sary to delineate the major irrigation areasof the world, identify what studies areavailable for each of these areas and howadequate they are, and attempt to develop asample that is reasonably representative ofworldwide variation.

Further refinement of the variablesused here will also be necessary, such as amore precise definition of financial and or-ganizational autonomy. But this is only thebeginning of the measurement problems in-volved. For example, how does one mea-sure the performance of an organizationmanaging hundreds of systems? The earlier

discussion of the Philippines implies thereis considerable intersystem variation. It maybe necessary to look at the data availablefrom different studies and attempt to arriveat a median and standard deviation for per-formance indicators.

No single researcher could carry outthis kind of rigorous, statistically valid,comparative study on a global basis. Onepossibility is for IIMI to establish a world-wide irrigation system database, modeledon the Human Relations Area Files13, whichcontains coded data on thousands of vari-ables from hundreds of societies around theworld and is used regularly by anthropolo-gists to test hypotheses. In principle, as aninternational institution, IIMI could estab-lish a worldwide sampling frame, identifythe cases to be used, develop guidelines forcollecting data and coding variables, codethe sample cases, and then both carry outanalyses for testing hypotheses and inviteother scholars to make use of the database.Such a database could be used for answer-ing many questions about the determinantsof irrigation performance.

The creation of this database would beexpensive and take years; unfortunately, it isunlikely to be implementable and its cost-effectiveness is uncertain. There are severalmore feasible research designs, though theresults may be less-conclusive. These in-clude case studies, before-and-after impactstudies, controlled comparisons, and actionresearch. Carefully done studies of casesrepresenting different organizational ar-rangements can be useful, particularly if anexplicit common analytical framework andperformance indicators are employed to en-able comparative analyses. Some of the re-cent case studies supported by IIMI wouldhave been more useful if there had beensuch a common framework (e.g., Yoder1994a; Svendsen and Vermillion 1994;Garcés-Restrepo and Vermillion 1995).

13The Human RelationsArea Files (HRAF) is anonprofit cooperativeorganization of a largenumber of Americanuniversities whichmaintains a coded data-base of primary ethno-graphic and other dataon a global sample ofsocieties (Moore 1970).

20

Ex post facto studies of performance beforeand after organizational changes are intro-duced constitute another promising ap-proach currently being explored by IIMI todocument the impact of management turn-over (Irrigation Management Reform Group1996). A third approach involves compara-tive analyses of similar systems with vary-ing organizational arrangements. Benjaminet al. (1994) use data from a large sample ofsystems in Nepal to compare the perfor-mance of self-governed and government-managed systems. Merrey, Valera, andDassenaike (1994) compare three rice-basedsystems in three Asian countries to examinehypotheses about the determinants of per-formance. A related approach is “controlledcomparison,” in which two (or several) sys-tems that are similar except for their organi-zational arrangements are carefully selectedand studied in depth to test carefully con-structed hypotheses. Finally, “action re-search” is an option that may be useful forboth researchers and policy makers: organi-zational innovations are implemented in afew systems through a learning process ap-proach facilitated by the researchers, com-bined with monitoring both these and othersystems where no changes have been intro-duced.

What Policy Makers Can Do Now

One reviewer of an earlier draft of this re-port suggested that further research is un-necessary, that we already know enoughabout what kind of institutional frameworkworks and does not work, and that coun-tries should get on with reforms based onthis knowledge. At the level of general orga-nizational and management principles, thereis much truth in this position. However,transferring arrangements deemed success-ful in one context to another entirely differ-

ent context and translating abstract prin-ciples into practical reality are highly prob-lematic. Would the USA model of irrigationdistricts be equally successful in, say, Indiaor China, with their very different social,cultural, economic, and political conditions?Creating the necessary conditions for suc-cessful organizational change is usually con-tentious and complex. The implementationstrategy and organizational design must fitthe specific context, and there must be suf-ficient time to adapt and institutionalize.Nevertheless, policy makers who wish toimprove the performance and sustainabilityof their water services need not wait for fur-ther research. Enough is known to designand implement successful programs.

The term autonomous, as used here, isnot to be confused with private. As Ostrom,Schroeder, and Wynne (1993) note, a stronggovernment role is necessary in most coun-tries, not only to avoid capture of the irriga-tion agency by an influential minority, butalso because the authority inherent in thegovernment is necessary for enforcing rulesto manage and conserve a common re-source. Local authorities with specific gov-ernment mandates for the integrated man-agement of particular irrigation systems,whose charter of authority derives from theusers, are an effective arrangement for man-aging irrigation systems.

In crafting institutions, special attentionneeds to be paid to incentives, both positiveand negative, through accountability of us-ers as well as managers. An integrated ap-proach based on the whole system, and notfragmented between user groups and gov-ernment departments, is essential. There-fore, countries currently promoting partialturnover programs face a serious problem:water users associations in parts of systems,partial turnover of O&M for entire systems,or shared management through jointfarmer-government committees are not

21

likely to become stable and effective institu-tional mechanisms for managing irrigationsystems unless there is clear provision formutual accountability between users andmanagers.

The recommendations for policy mak-ers are therefore clear, though difficult toimplement; some have been suggested ear-lier. Reform programs must be carefully de-signed, strongly supported at high levels,and based on pilot testing at the initialstages. Countries whose irrigation manage-ment agencies are dependent on govern-ments financially, organizationally, or bothshould design strategies to make the agen-cies more autonomous. Countries withhighly centralized agencies that managesingle large systems should proceed with

shifting to more autonomous decentralizedstructures with high levels of accountabilityto the users but in a regulatory frameworkthat ensures equity and cost-effectiveness aswell as enforcement of contracts withouthigh transaction costs. Countries in which asingle provincial or national irrigationagency manages a large number of systemsshould explore ways to encourage the emer-gence of autonomous user-based organiza-tions in each system or perhaps in specificwatersheds. Such reforms will be successfulonly if governments are willing to provide asupportive and enabling institutional andpolicy framework, positive incentives forlocal users to take full responsibility andauthority for their systems, and sufficienttraining and technical support.

22

23

Literature Cited

Aluwihare, P. B., and Masao Kikuchi. 1991. Irrigation investment trends in Sri Lanka: New construction and beyond.IIMI Research Paper. Colombo: International Irrigation Management Institute.

Athukorale, Karunatissa, Kusum Athukorale, and Douglas J. Merrey. 1994. Effectiveness of nongovernment orga-nizations in developing local irrigation organizations: A case study from Sri Lanka. IIMI Country Paper, Sri Lanka,No. 12. Colombo: International Irrigation Management Institute.

Bandaragoda, D. J., and G. R. Firdousi. 1992. Institutional factors affecting irrigation performance in Pakistan: Re-search and policy priorities. IIMI Country Paper, Pakistan, No. 4. Colombo: International Irrigation Manage-ment Institute.

Benjamin, Paul, Wai Fung Lam, Elinor Ostrom, and Ganesh Shivakoti. 1994. Institutions, incentives and irriga-tion in Nepal. Decentralization: Finance and Management Project. [Burlington, Vermont, USA]: Associatesin Rural Development

Bos, M. G., D. H. Murray-Rust, D. J. Merrey, H. G. Johnson, and W. B. Snellen. 1994. Methodologies for assess-ing performance of irrigation and drainage management. Irrigation and Drainage Systems 7:231–261.

Bos, M. G., W. Wolters, A. Drovandi, and J. A. Morabito. 1991. The Viejo Retamo secondary canal: Performanceevaluation case study: Mendoza, Argentina. Irrigation and Drainage Systems 5:77–88.

Bottrall, Anthony F. 1981. Comparative study of the management and organization of irrigation projects. World BankStaff Working Paper No. 458. Washington, D.C.: World Bank.

Byrnes, Kerry J. 1992. Water users associations in World Bank-assisted irrigation projects in Pakistan. World BankTechnical Paper No. 173. Washington, D.C.: World Bank.

Chambouleyron, Jorge. 1989. The reorganization of water users’ associations in Mendoza, Argentina. Irrigationand Drainage Systems 3:81–94.

Chambouleyron, Jorge. 1995. Determining the optimum size of water users’ associations. Irrigation and Drain-age Systems 8:189–199.

Coward, E. Walter, Jr. 1980. Management themes in community irrigation systems. In: Irrigation and agricul-tural development in Asia: Perspectives from the social sciences, ed. E. Walter Coward, Jr. Ithaca, New York, USA:Cornell University Press.

Frederiksen, Harald D., Jeremy Berkoff, and William Barber. 1994. Principles and practices for dealing with waterresources issues. World Bank Technical Paper No. 233. Washington, D.C.: World Bank.

Freeman, David. 1989. Local organizations for social development: Concepts and cases of irrigation organization. Boulder,Colorado, USA: Westview Press.

Garcés-Restrepo, C., and D. Vermillion. 1995. Irrigation management transfer in Colombia: Assessment ofseven districts. In Irrigation management transfer: Selected papers from the International Conference on IrrigationManagement Transfer, Wuhan, China, 20–24 September 1994, ed. S. H. Johnson, D. L. Vermillion, and J. A.Sagardoy. Rome: FAO.

Gerards, Jan L. M. H. 1992. Introduction of Irrigation Service Fee (ISF) in Indonesia: Institutional developmentin action for resource management. Irrigation and Drainage Systems 6:223–247.

Gerards, Jan L., Biron S. Tambunan, and Bachtiar Harun. 1991. Payment for irrigation services in Indonesia:Creating mutual accountability through participation and voice: Experience with pilot project introduction(1989-1991). Proceedings of the International Commission on Irrigation and Drainage Eighth Afro-AsianRegional Conference, Bangkok, 1991. Vol. B, supplemental paper. Duplicated

Gorriz, Cecilia M., Ashok Subramanian, and Jose Simas. 1995. Irrigation management transfer in Mexico: Processand progress. World Bank Technical Paper 292. Washington, D.C.: World Bank.

Hofwegen, Paul van. 1994. Use and utility of performance indicators, propositions and interventions. ResearchProgramme on Irrigation Performance Progress Report No. 1 (August 1994 draft). Delft, Netherlands: In-ternational Institute for Infrastructural, Hydraulic, and Environmental Engineering. Duplicated.

Hunt, Robert C. 1988. Size and the structure of authority in canal irrigation systems. Journal of AnthropologicalResearch 44:335–355.

Hunt, Robert C. 1989. Appropriate social organization? Water users associations in bureaucratic canal irriga-tion systems. Human Organization 48:79–90.

24

Hunt, Robert C. 1990. Organizational control over water: The positive identification of a social constraint onfarmer participation. In Social, economic, and institutional issues in third world irrigation management, ed. R. K.Sampath and Robert A. Young. Boulder, Colorado, USA: Westview Press.

IIMI (International Irrigation Management Institute). 1995a. Water resources, irrigation operations and institu-tional issues: An analysis of the Ministry of Public Works and Water Resources, Government of Egypt. Strengthen-ing Irrigation Management in Egypt Report No. 2. Colombo.

IIMI. 1995b. An action plan for strengthening water resource management in Egypt. Strengthening Irrigation Man-agement in Egypt Report No. 3. Colombo.

IIMI. 1995c. Cost recovery for water services to agriculture. Strengthening Irrigation Management in Egypt. Cairo.IIMI and ARTI (Agrarian Research and Training Institute). 1995. Monitoring and evaluation of the participa-

tory irrigation system management policy. 3 vols. (Draft, February 1995). Colombo. Duplicated.Irrigation Management Reform Group. 1996. A standard methodology to assess the impacts of irrigation man-

agement turnover (Draft, May 1996). Colombo: International Irrigation Management Institute. Duplicated.John Mellor Associates, Inc. 1995. Strategic priorities for sustainable agriculture in Egypt—Impact on growth,

employment, trade, and privatization. Report of a reconnaissance mission to generate a strategy for USAID.Washington, D.C. Duplicated.

Jopillo, Sylvia Ma G., and Romana de los Reyes. 1988. Partnership in irrigation: Farmers and government in agency-managed systems. Quezon City, Philippines: Institute of Philippine Culture, Ateneo de Manila University.

Keller, Andrew A., Jack Keller, and Mona El-Kady. 1995. Effective irrigation efficiency applied to Egypt’s Nile Sys-tem. Strategic Research Program Working Paper Series No. 5-1. Cairo: Ministry of Public Works and WaterResources.

Kijne, Jacob W., and Marcel Kuper. 1995. Salinity and sodicity in Pakistan’s Punjab: A threat to sustainability ofirrigated agriculture? Water Resources Development 11 (1): 73–86.

Kijne, Jacob W., and Edward J. Vander Velde. 1992. Salinity in Punjab watercourse commands and irrigationsystem operations. In Advancements in IIMI’s research 1989–91. Colombo: International Irrigation Manage-ment Institute.

Korten, Frances F., and Robert Y. Siy, Jr., eds. 1989. Transforming a bureaucracy: The experience of the PhilippineNational Irrigation Administration. Manila: Ateneo de Manila University Press.

Lauraya, Fay M., and Antonia Lea R. Sala. 1995. Performance determinants of irrigation associations in nationalirrigation systems in Bicol, the Philippines: An analysis. IIMI Country Paper, The Philippines, No. 4. Colombo:International Irrigation Management Institute.

Levine, Gilbert. 1991. The Taiwan irrigation associations: Observations from the outside. In Farmers in themanagement of irrigation systems, ed. K. K. Singh. New Delhi: Sterling Publishers.

Maass, Arthur, and Raymond L. Anderson. 1978. . . . and the desert shall rejoice: Conflict, growth and justice in aridenvironments. Cambridge, Massachusetts, USA: MIT Press. (Reprinted 1986 by Robert E. Krieger Publish-ing, Malabar, Florida, USA).

Malhotra, S. P., S. K. Raja, and D. Seckler. 1984a. A methodology for monitoring the performance of large-scaleirrigation systems: A case study of the warabandi system of northwest India. Agricultural Administration17:231-259.

Malhotra, S. P., S. K. Raja, and D. Seckler. 1984b. Performance monitoring in the warabandi system of irrigationmanagement. In Productivity and equity in irrigation systems, ed. Niranjan Pant. New Delhi: Ashish Publish-ing House.

Meinzen-Dick, Ruth, Meyra Mendoza, Loic Sadoulet, Ghada Abiad-Shields, and Ashok Subramanian. 1994.Sustainable water user associations: Lessons from a literature review. Paper presented at World Bank WaterResources Seminar, Landsdowne, Virginia, USA, December 13–15, 1994. Duplicated.

Merrey, Douglas J. 1987. The local impact of centralized irrigation control in Pakistan: A sociocentric perspec-tive. In Lands at risk in the Third World: Local-level perspectives, ed. Peter D. Little, Michael M. Horowitz, andA. Endre Nyerges. Boulder, Colorado, USA: Westview Press.

Merrey, Douglas J. 1995. Potential for devolution of management to farmers’ organizations in an hierarchicalirrigation management agency: The case of the Mahaweli Authority of Sri Lanka. In The blurring of a vi-sion—The Mahaweli: Its social, economic and political considerations, ed. H. P. Muller and S. T. Hettige. Ratmalana,Sri Lanka: Sarvodhaya Book Publishing Services.

25

Merrey, Douglas J. Forthcoming. Governance and institutional arrangements for managing water resources in Egypt.Liquid Gold Series, Wageningen, Netherlands: International Institute for Land Reclamation and Improve-ment.

Merrey, Douglas J., Alfredo Valera, and Lalith Dassenaike. 1994. Does assessing irrigation performance make adifference? Results from a comparative study of three irrigation systems. Quarterly Journal of InternationalAgriculture 33 (3):276–293.

Moore, Frank W. 1970. The Human Relations Area Files. In A handbook of method in cultural anthropology, ed.Raoul Naroll and Ronald Cohen. New York: Natural History Press. (Reprinted 1973 by Colombia Univer-sity Press, New York).

Murray-Rust, D. Hammond, and Bart W. Snellen. 1993. Irrigation system performance assessment and diagnosis.Colombo: International Irrigation Management Institute.

Murray-Rust, D. Hammond, and Edward J. Vander Velde. 1994. Conjunctive use of canal and groundwater inPunjab, Pakistan: Management and policy options. Irrigation and Drainage Systems 8 (4):201-232.

NIA (National Irrigation Administration). 1992. Annual report of the National Irrigation Administration of thePhilippines. Manila.

Oorthuizen, Joost, and Wim H. Kloezen. 1995. The other side of the coin: A case study on the impact of finan-cial autonomy on irrigation management performance in the Philippines. Irrigation and Drainage Systems 9(1):15–37.

Oorthuizen, Joost, and Gerven Sloot. 1993. Privatization in irrigation: A case study of the Osgong Irrigation SystemIrrigation Association. Rural Development Studies, vol. 9, no. 3. Quezon City, Philippines: Philippine Peas-ant Institute.

Ostrom, Elinor. 1990. Governing the commons: The evolution of institutions for collective action. Cambridge: Cam-bridge University Press.

Ostrom, Elinor. 1992. Crafting institutions for self-governing irrigation systems. San Francisco, California, USA:Institute for Contemporary Studies Press.

Ostrom, Elinor, Larry Schroeder, and Susan Wynne. 1993. Institutional incentives and sustainable development:Infrastructural policies in perspective. Boulder, Colorado, USA: Westview Press.

Palacios-Velez, Enrique. 1995. Performance of water users associations in the operation and maintenance ofirrigation districts in Mexico. In Irrigation management transfer: Selected papers from the International Confer-ence on Irrigation Management Transfer, Wuhan, China, 20–24 September 1994, ed. S. H. Johnson, D. L. Vermil-lion, and J. A. Sagardoy. Rome: FAO.

Pant, Shiva Raj, Alfredo Valera, and Ujjwal Pradhan, eds. 1992. Proceedings of the National Workshop on Participa-tory Management in Agency-Managed Irrigation Systems in Nepal. Kathmandu: Department of Irrigation andInternational Irrigation Management Institute, Nepal Field Operations.

Paul, Samuel. 1991. Accountability in public services: Exit, voice, and capture. World Bank Working Paper No. 614.Washington, D.C.: World Bank.

Paul, Samuel. 1994. Does voice matter? For public accountability, yes. Policy Research Working Papers 1388. Wash-ington, D.C.: World Bank.

Perry, C. J. 1995. Determinants of function and dysfunction in irrigation performance, and implications forperformance improvement. Water Resources Development 11 (1):11–24.

Perry, C. J. 1996. Quantification and measurement of a minimum set of indicators of the performance of irriga-tion systems. Colombo: International Irrigation Management Institute. Duplicated.

Plusquellec, Herve. 1989. Two irrigation systems in Colombia: Their performance and transfer of management to users’associations. World Bank Working Paper No. 264. Washington, D.C.: World Bank.

Pradhan, Prachandra. 1993. Participatory irrigation management through water users’ associations: An exer-cise in Kano River Irrigation Project (KIRP), Nigeria. Paper presented at Ninth IIMI Internal Program Re-view, International Irrigation Management Institute. Duplicated.

Rao, P. S. 1993. Review of selected literature on indicators of irrigation performance. IIMI Research Paper. Colombo:International Irrigation Management Institute.

Small, Leslie A., and Ian Carruthers. 1991. Farmer-financed irrigation: The economics of reform. Cambridge: Cam-bridge University Press.

26

Small, L. A., and M. Svendsen. 1992. A framework for assessing irrigation performance. IFPRI Working Papers onIrrigation Performance No. 1. Washington, D.C.: International Food Policy Research Institute.

Svendsen, Mark. 1992. Assessing effects of policy change on Philippine irrigation performance. IFPRI Working Pa-pers on Irrigation Performance No. 2. Washington, D.C.: International Food Policy Research Institute.

Svendsen, M. and D. Vermillion. 1994. Irrigation management transfer in the Colombia Basin: Lessons and interna-tional implications. IIMI Research Paper No. 12. Colombo: International Irrigation Management Institute.

Tang, Shui Yan. 1992. Institutions and collective action: Self-governance in irrigation. San Francisco, California,USA: Institute for Contemporary Studies Press.

TEAMS. 1992. Turnover of O&M of distributaries to farmers’ organizations, Polonnaruwa District. Final re-port submitted to International Irrigation Management Institute. Colombo. Duplicated.

Uphoff, Norman. 1986. Improving international irrigation management with farmer participation: Getting the processright. Boulder, Colorado, USA: Westview Press.

Uphoff, Norman. 1991. Managing irrigation: Analyzing and improving the performance of bureaucracies. New Delhi:Sage Publications.

Uphoff, Norman. 1992. Learning from Gal Oya: Possibilities for participatory development and post-Newtonian socialscience. Ithaca, New York, USA: Cornell University Press.

Uphoff, Norman, M. L. Wickremasinghe, and C. M. Wijayaratna. 1990. “Optimum” participation in irrigationmanagement: Issues and evidence from Sri Lanka. Human Organization 49 (1):26–40.

Vermillion, Douglas L. 1991. The turnover and self management of irrigation institutions in developing coun-tries. A discussion paper for a new program of IIMI. Colombo: International Irrigation Management Insti-tute. Duplicated.

Vermillion, D. L, and S. H. Johnson, III. 1995. Globalization of irrigation management transfer: A summary ofideas and experiences from the Wuhan Conference. In Irrigation management transfer: Selected papers from theInternational Conference on Irrigation Management Transfer, Wuhan, China, 20–24 September 1994, ed. S. H.Johnson, D. L. Vermillion, and J. A. Sagardoy. Rome: FAO.

Vimaladharma, Kapila Pathirana. 1994. Agriswiss experiences in participatory management of irrigation projectsin Sri Lanka. Paper presented at Seminar on Agriswiss M&E Project Experiences, 31 January 1994, at Agrar-ian Research and Training Institute, Colombo. Duplicated

Wijayaratna, C. M., and Douglas L. Vermillion. 1994. Irrigation management turnover in the Philippines: Strategyof the National Irrigation Administration. Short Report Series on Locally Managed Irrigation Report No. 4.Colombo: International Irrigation Management Institute.

Wilkins-Wells, John, David J. Molden, Prayag Pradhan, and Shyam P. Rajbhandari. 1995. Developing sharesystems for sustainable water users associations in Nepal. In Irrigation management transfer: Selected papersfrom the International Conference on Irrigation Management Transfer, Wuhan, China, 20-24 September 1994, ed. S.H. Johnson, D. L. Vermillion, and J. A. Sagardoy. Rome: FAO.

Wittfogel, Karl. 1957. Oriental despotism. New Haven, Connecticut, USA: Yale University Press.Yoder, Robert. 1994a. Organization and management by farmers in the Chhattis Mauja Irrigation System, Nepal. IIMI

Research Paper No. 11. Colombo: International Irrigation Management Institute.Yoder, Robert. 1994b. Locally managed irrigation systems: Essential tasks and implications for assistance, management

transfer, and turnover programs. IIMI Monograph No. 3. Colombo: International Irrigation Management In-stitute.

Research Reports

1. The New Era of Water Resources Management: From “Dry” to “Wet” Water Savings.David Seckler, 1996.

2. Alternative Approaches to Cost Sharing for Water Service to Agriculture in Egypt.C. J. Perry, 1996.

3. Integrated Water Resource Systems: Theory and Policy Implications. Andrew Keller,Jack Keller, and David Seckler, 1996.

4. Results of Management Turnover in Two Irrigation Districts in Colombia: Douglas L.Vermillion, and Carlos Gracés-Restrepo, 1996.

5. The IIMI Water Balance Framework: A Model for Project Level Analysis. C. J. Perry, 1996.

6. Water and Salinity Balances for Irrigated Agriculture in Pakistan. Jacob W. Kijne, 1996.

7. Free-Riders or Victims: Women’s Nonparticipation in Irrigation Management in Nepal’sChhattis Mauja Irrigation Scheme. Margreet Zwarteveen, and Nita Neupane, 1996.

8. Institutional Design Principles for Accountability in Large Irrigation Systems. Douglas J.Merrey. 1996.

Research Report

Institutional Design Principlesfor Accountabilityin Large Irrigation Systems

Douglas J. Merrey

INTERNATIONAL IRRIGATION MANAGEMENT INSTITUTEP O Box 2075 Colombo, Sri Lanka

Tel (94-1) 867404 • Fax (94-1) 866854 • E-mail IIMI@cgnet.comInternet Home Page http:/ /www.cgiar.org

ISSN: 1026-0862ISBN: 92-9090-335-X