François Molle and Mary Renwick

IWMI is a Future Harvest Centersupported by the CGIAR

Economics and Politics ofWater Resources DevelopmentUda Walawe Irrigation Project,Sri Lanka

87

RESEARCHR E P O R T

I n t e r n a t i o n a lWater ManagementI n s t i t u t e

SM

IWMI is a Future Harvest Centersupported by the CGIAR

Postal Address:P O Box 2075 ColomboSri Lanka

Location:127, Sunil Mawatha Pelawatta Battaramulla Sri Lanka

Tel:+94-11-2787404

Fax:+94-11-2786854

E-mail:iwmi@cgiar.org

Website:http://www.iwmi.org

I n t e r n a t i o n a lWater ManagementI n s t i t u t e

ISSN 1026-0862ISBN 92-9090-584-0

Research Reports

IWMI’s mission is to improve water and land resources management for food,livelihoods and nature. In serving this mission, IWMI concentrates on the integrationof policies, technologies and management systems to achieve workable solutions toreal problems—practical, relevant results in the field of irrigation and water and landresources.

The publications in this series cover a wide range of subjects—from computermodeling to experience with water user associations—and vary in content fromdirectly applicable research to more basic studies, on which applied work ultimatelydepends. Some research reports are narrowly focused, analytical and detailedempirical studies; others are wide-ranging and synthetic overviews of genericproblems.

Although most of the reports are published by IWMI staff and theircollaborators, we welcome contributions from others. Each report is reviewedinternally by IWMI’s own staff and Fellows, and by external reviewers. The reportsare published and distributed both in hard copy and electronically (www.iwmi.org)and where possible all data and analyses will be available as separatedownloadable files. Reports may be copied freely and cited with dueacknowledgment.

i

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

Research Report 87

Economics and Politics of Water ResourcesDevelopment: Uda Walawe Irrigation Project,Sri Lanka

François Molle and Mary Renwick

IWMI receives its principal funding from 58 governments, private foundations, andinternational and regional organizations known as the Consultative Group onInternational Agricultural Research (CGIAR). Support is also given by the Governmentsof Ghana, Pakistan, South Africa, Sri Lanka and Thailand.

The authors: François Molle is a Senior Researcher at the Institut de Recherche pour leDéveloppement (IRD), France, currently seconded to the International Water ManagementInstitute (IWMI), Colombo. Mary Renwick is a Senior Fellow of Economics and WaterPolicy at the Water Resources Center, University of Minnesota and an Adjunct AssistantProfessor in the Department of Applied Economics, University of Minnesota.

This research contributes to the Comprehensive Assessment of Water Management inAgriculture (www.iwmi.org\assessment). The research was supported by a grant fromthe Government of the Netherlands to the Comprehensive Assessment and by the“Collaborative Research on the Accounting of the Agricultural and Non-AgriculturalImpacts of Irrigation and Drainage Systems” funded by the Taiwan Council of Agriculture,the Taoyan Irrigation Research and Development Foundation, the Tsao-Jiin MemorialFoundation for R&D for Agriculture and Irrigation, the Environmental Greening Foundation,and the Chi-Seng Water Management and Development Foundation. The authors thankDeeptha Wijerathna, Ranjith Ariyaratne and Shyamalie de Silva for their collaboration ingathering data on crop production.

Molle, F.; Renwick, M. 2005. Economics and politics and of water resources development:Uda Walawe Irrigation Project, Sri Lanka. Research Report 87. Colombo, Sri Lanka:International Water Management Institute (IWMI).

/ water resources development / irrigation programs / river basins / poverty / cropproduction / political aspects / history / water use / cost recovery / cost-benefit analysis/ water requirements / decision making / domestic water / fisheries / rehabilitation /farmer associations / farmers / feasibility studies / development projects / Sri Lanka /

ISSN 1026-0862ISBN 92-9090-584-0

Copyright © 2005, by IWMI. All rights reserved.

Cover photograph is by François Molle.

Please send inquiries and comments to: iwmi@cgiar.org

iii

iii

Contents

Summary v

Introduction 1

History of Uda Walawe Irrigation and Resettlement Project 4

Project Performance 21

Costs and Benefits of UWIRP 26

Decision-Making Process 38

Lessons Learned and Conclusions 51

Appendices 55

Literature Cited 61

Abbreviations and Acronyms 67

Units and Currencies 68

Glossary 68

v

Summary

This report examines the history of water re-sources development and investment decisionsfor the Uda Walawe Irrigation and ResettlementProject (UWIRP), located in the southern dryzone of Sri Lanka. The project was initiated inthe early 1950s just after Sri Lanka gainedindependence. UWIRP was part of Sri Lanka’snew postcolonial vision for economic develop-ment and modernization; a vision that wassupported by multilateral and bilateral fundingagencies because irrigation was seen as anengine for growth in the 1950s and onward. Theoriginal plan for UWIRP is most aptly describedas a highly ambitious social, economic andphysical engineering project aimed at creating amodern and profitable agriculture sector. Thisplan envisioned bringing 32,780 hectares (81,000acres) of arid land into highly efficient agriculturalproduction by constructing a reservoir andirrigation facilities, and moving landless farmersto the newly developed lands.

The description of over 50 years of progres-sive development shows a wide range of difficul-ties that constantly undermined the efforts madeto implement the plan. They range from designfailure and shoddy construction to poor mainte-nance, from agricultural diversification constraintsto administrative inertia, and from massive landencroachment problems by squatters to politicalupheavals. What is striking is the contrastbetween the (sometimes blunt) assuredness ofthe hypotheses made in feasibility reports andthe reality on the ground, and between thesimplistic technological and social engineeringdrive of the consultants and the complexity ofregional development.

Rather than adaptively adjusting to theseunforeseen changes, planners continued toreturn time and again to the original plan andlaunch successive attempts at development. Foreach phase, investment was justified based on

specified outcomes. Typical ex post irrigationassessment performance measures (projectedversus actual capital costs, implementationschedules, and achievement of objectives suchas irrigated area and crop output) were used toidentify the extent of divergence betweenplanned and actual outcomes. These measuresunderscored the rather poor performance de-tailed in the historical analysis. However, acomparative analysis of performance measuresfrom UWIRP with other developing-countryirrigation projects reveals that UWIRP’s relativelypoor performance, as measured by typicalindicators, was on par with many other projects.These results support research from earliercomprehensive studies (e.g., WCD 2000) that exante irrigation plans tend to systematicallyoverstate proposed outcomes.

The rather grim history of UWIRP, coupledwith poor performance measures, is met withsome unanticipated results in the preliminaryanalysis of costs and benefits for UWIRP.Despite rapidly escalating cumulative projectcosts, the growth in cumulative direct benefitsfrom agricultural production has outstrippedcosts. The growth in benefits from agriculturalproduction stemmed primarily from two factors—a surge in international rice prices in the late1970s and early 1980s, which boosted the valueof rice produced, and the establishment andexpansion of highly profitable banana productionin the area. The successful dissemination ofbanana cultivation and associated marketingchannels were due to the vision and efforts ofboth farmers and local agricultural extensionagents. Since the UWIRP irrigation system isused for many other purposes than crop irriga-tion, the report provides preliminary estimates ofthese other noncrop benefits, such as inlandfisheries, home gardens, hydropower, drinkingwater supply and tourism.

vi

These benefits need to be compared toother costs imposed by the project beyondcapital investment costs, such as environmentaland social costs. Although only limited informa-tion exists on the environmental costs of UWIRP,the social costs of UWIRP have been relativelylow in comparison to other large-scale dam andirrigation projects, which frequently involvereallocation of local populations. The UWIRPproject area was sparsely populated beforedevelopment and those living in areas desig-nated for development (including later squatters)were made into project beneficiaries as settlers.Further research is needed on indirect projectcosts, such as human health effects due tounsafe drinking water and bioaccumulation ofagrochemical byproducts.

The report seeks to understand the decision-making processes of various interventions overthe years and the outcomes of these decisions.The analysis includes the identification of thevarious decisions that have influenced the

project’s evolution, the rationale behind thesedecisions, and how these decisions were imple-mented. It shows, in particular, the interplaybetween how projects are perceived, planned,implemented and managed by various actors(donors, government, implementing agencies andconsultants), all characterized by their respectivestrategic interests and accountability. Particularattention is given to the gap between planners’vision and reality on the ground, and to the abilityof implementers to effectively bridge this gap.

Overall, this research illustrates the difficul-ties of assessing not only project performancebut also project outcomes; the outcomes of aproject are governed by the evolution of thebehavior and choices of the different actorsconcerned, in which their interests, mindsets andstrategies are embedded. It uncovers underlyingprocesses that shaped the evolution of theproject and highlights the limitation of viewingdevelopment as a mere set of technical andsocial engineering endeavors.

1

Economics and Politics of Water ResourcesDevelopment: Uda Walawe Irrigation Project,Sri Lanka

François Molle and Mary Renwick

Introduction

Irrigation projects promise substantial benefits atthe outset. While they have made substantialcontributions to economic development, theprice paid to secure these benefits has oftenbeen higher than expected, not only in terms ofcapital outlay but also in terms of social andenvironmental costs. Questions are being raisedabout the distribution of benefits and costs fromirrigation projects and their real contribution tomeeting development needs.

This report examines the history of waterresources development in the lower Walaweriver basin, located in southern Sri Lanka. Itseeks to tell a comprehensive story of “whathappened” and “why,” by examining plannedversus actual outcomes of successive phases ofdevelopment since the 1950s. A variety ofqualitative and quantitative measures are used,including a detailed historical account ofactivities, quantitative performance measures,estimation of project costs and benefits, and ananalysis of decision-making processes. Theoverarching objective is to look at the history ofWalawe river basin development as an object ofresearch and to identify some of the features of“development” in the making, thus learning frompast experience and practice.

Structure of the Report

The report includes the introduction and six othersections. The second section is mainly

descriptive and recounts the successive phasesof water resources development in the lowerWalawe basin during the twentieth century,focusing on the Uda Walawe Irrigation andResettlement Project (UWIRP), which wasinitiated after Sri Lanka’s independence.Construction work under UWIRP began in the1960s and the final expansion is scheduled forcompletion by 2005. The original visions forUWIRP have not been realized as yet.

The third section assesses the performanceof successive phases of investment using typicalquantitative measures employed in ex postanalyses of development projects. Some of theperformance measures employed includeprojected versus actual capital costs,implementation schedules and achievement ofobjectives, such as irrigated area and cropoutput. The performance measures for UWIRPare compared with other irrigation projects toevaluate its relative performance. The fourthsection provides estimates of UWIRP benefitsand costs. Measures of both crop productionand noncrop production benefits (fisheries,domestic water, home gardens, tourism andeconomic multiplier effects) are presented, andthe evolution of cumulated costs and benefitsare compared.

The fifth section seeks to understand thedecision-making processes for variousinterventions over the years and the outcomesof these decisions. It identifies the variousdecisions that have influenced the project’s

2

evolution, the rationale behind these decisions,and how these decisions were implemented. Itdiscusses the interplay between how projectsare perceived, planned, implemented andmanaged by various actors (donors, government,implementing agencies and consultants) giventheir strategic interests and accountability.Particular attention is given to the gap betweenwhat the planners envision and reality on theground, and the ability of implementers toeffectively bridge this gap.

Overall, this research illustrates not only thedifficulties of assessing project planning,implementation and performance, both ex anteand ex post, but also that the outcome of theproject is governed by the evolution of thebehavior and choices of the different actorsconcerned, in which their interests, mindsets andstrategies are embedded. It highlights theserious shortcomings of viewing development asa set of technical and social engineeringendeavors, and uncovers underlying processesthat shape the evolution of the project.

Walawe River Basin and its PhysicalCharacteristics

The Walawe river basin1 covers an area ofapproximately 3,000 km2 and extends from theridge of the central highlands of Sri Lanka, at analtitude of over 2,000 meters, down to thesouthern coast (figure 1). The basin offers aclear contrast between, on the one hand, its

highlands and its intermediate mountainousassociation of ridges and valleys and, on theother, the lowland plain itself. Precipitation variessignificantly in the basin from over 3,000 mm inthe northwestern tip to around 1,000 mm alongthe seashore. Approximately half theprecipitation is transformed into runoff and theremainder is either used by vegetation orevaporates. Some of the runoff percolates intoshallow groundwater aquifers located in theplains but aquifer levels fall quickly after therainy season ends. The highlands are cut bymany valleys in which small streams, oftenperennials, can be found. They feed the Walaweriver, which has an average discharge to the seaof 1.1 billion cubic meters per year.

Agriculture in the basin is mainly rainfed andincludes plantations (tea and coconut),afforestation and chena (slash-and-burn)cultivation. Irrigation is practiced in themountainous areas through diversion of smallstreams (see Molle et al. 2003). In the plain,there are 600 small tanks (reservoirs) datingback a thousand years or more, that currentlyprovide only limited irrigation because many aresilted, breached or in disrepair.

The Uda Walawe Irrigation andResettlement Project (UWIRP) is located in thelower part of the basin. The Uda Walawereservoir is located in the middle of the basinand supplies water to downstream areasthrough two main canals known as the RightBank Main Canal (RBMC) and the Left BankMain Canal (LBMC) (figure 2).

1The Kachchigala and Karagan oya basins, which are small basins adjacent to the lower Walawe basin, have been hydrologically linkedwith the Walawe basin under the Uda Walawe Project.

3

FIG

UR

E 1

.La

yout

and

phy

siog

raph

y of

the

Wal

awe

river

bas

in.

4

FIGURE 2.Uda Walawe irrigation area.

History of Uda Walawe Irrigation and Resettlement Project

The natural, social and economic landscape ofthe Uda Walawe basin is a living legacy of acomplex myriad of interactions between humansand the natural environment. The story of UdaWalawe contains a rich ancient history ofirrigation that dates back about 2000 years tothe days of the Ruhuna kingdom, which isfollowed by a long period of abandonment. Inthe late nineteenth century, British colonialistsinitiated agricultural development activities in thebasin, including irrigation, but the area remainedrelatively sparsely populated. In the postcolonial

period, one of the first acts of the governmentwas to initiate plans for the development of theUda Walawe basin. A centerpiece of the planwas a large-scale irrigation and rural-development project that eventually becameknown as the Uda Walawe Irrigation andResettlement Project (UWIRP).

The postcolonial independence period of UdaWalawe was a period of rapid change. Tofacilitate the discussion, this period is subdividedinto the following four phases of development inthe basin.

5

• Phase I: From independence to theconstruction of headworks (1948-1967)

• Phase II: Downstream development of theRight Bank (RB) area (1969–1978)

• Phase III: Improvement and rehabilitation ofthe RB area (1984–1994)

• Phase IV: Rehabilitation and extension of theLeft Bank (LB) Project area (1995 onwards)

Before examining the successive phases ofUWIRP, a brief history of the area up to thecolonial period is presented.

Ancient History

Modern water resources developments in theUda Walawe basin are superimposed over a richhistory of water control and irrigation in theregion. Archaeological studies indicate that theWalawe river valley was populated severalthousand years ago and known archaeologicalsites date back to 6,500 B.C. (Deraniyagalan.d.). Magama, an important city of the oncepowerful ancient Ruhuna kingdom mentioned inthe Mahawamsa as early as 246 B.C., waslocated in the middle of the present UWIRPproject area (Fernando n.d.; Collins 1932).Because of its location in the dry zone of SriLanka, tanks and irrigation systems wereparamount to the survival of ancient civilizationsand many ancient kings devoted significantefforts to the creation and maintenance ofanicuts (small diversion weirs) and tanks(Ghanawimala 1942; Narada 1992). As a result,there are hundreds of small irrigation tanks andanicuts scattered throughout the Walawe basin.The density of ancient irrigation tanks in someareas is one of the highest in Sri Lanka, withapproximately 1 tank per 2 km2 area (Mendis1967). Some of these tanks are still in use,especially in the lower part of the basin.

The exact chronology of these developmentsis unknown but it is believed that the basin wasa prosperous area from the second century B.C.and began to decline around the thirteenth

century. Until the mid-twentieth century, thepopulation in the Walawe plain dwindled. Themountainous area retained denser populationsbut underwent significant depopulation after the1818 insurrection against the British. A numberof factors likely influenced the collapse ofcommunities in the plains, including repeatedattacks from enemies, malaria (Ghanawimala1942; Ghanawimala 1967; Narada 1992) or theattraction of better opportunities in the wet zone(Obeyesekere 1984).

British Rule: 1815-1948

By the time the British arrived, the Walawe basinwas sparsely populated. Those who lived in thearea relied on the ancient tanks for water fordomestic and agricultural purposes, althoughmost people in the area practiced slash-and-burnagriculture known locally as chena. In 1818, aGovernment Agent visited one of the mainvillages and described it as a thick jungle areainfested with malaria and populated with only 18persons (Abeyratna n.d.).

Until the extreme drought of 1866, whichcaused a high mortality and elicited strong callsfor the development of large-scale irrigationprojects, the British had only undertaken a fewminor settlement and rehabilitation works on theancient tanks and anicuts in the Walawe basin(Steele 1867). The first large-scale interventionby the British was in the southernmost part, inHambantota district, starting with therehabilitation of the Liyangastota anicut in 1889.Figure 2 provides a visualization of these andfuture works. The Liyangastota anicut, which ispart of the present UWIRP, diverted water to acanal on the RB of the river to irrigate about3,000 hectares a few miles further south. Thegovernment sold 75 percent of the irrigable land,with a theoretical maximum of 2.02 ha (5 acres)per person, and reserved the remaining 25percent for poor people (GOC 1889). Otherminor interventions included the development ofHingura and Embilipitiya village tanks around1890 and construction of the Ridiyagama tank in

6

1922. More irrigated land was developeddownstream of this tank in two phases (1925and 1941), with 3,000 hectares of crown landsold to local people. The investments in irrigationwere justified in dual terms—improving healthand livelihoods of the poor and earning incomefor the Crown (Bastiampillai 1967).

By the early 1940s, the British initiated moreambitious plans for the Walawe River BasinScheme. These development plans laid outschematic “blue prints” for a large-scale irrigationand rural-development project (eventually knownas UWIRP), a large upstream reservoir (wewa)for hydropower called Samanala, and numerousother dams. Planning efforts were discontinuedbecause of the Second World War and thecoming grant of independence to the colony.

Phase I: From Independence to theConstruction of Headworks—1948-1967

Shortly after independence, the Government ofCeylon continued the earlier efforts by the Britishin the Walawe river basin. The governmentrehabilitated some ancient tanks2 and continuedsurvey work and feasibility studies for theWalawe basin scheme. The settlements inWalawe were some of the first in the wave ofagrarian settlements of the post-World War IIperiod. The government sought to establishsettlements in the dry zone in response to aperceived population explosion and agrariancrisis. Settlements were construed as a type ofsocial engineering, whereby poor farmers from thewet zone were given equal access to land andthe threat of land concentration and division wascombated by limiting the transfer of ownership—allowing only one successor to inherit anallotment, despite the contradictions with

culturally sanctioned expectations of families.3

Efficiency concerns were supposed to beaddressed by setting a rationalistic procedure inthe selection of settlers that took intoconsideration data on farm size, land endowment,“farming skills” and experience.

Throughout the 1950s, numerous surveysand studies of the Walawe basin were carriedout. The Government of Canada supported aSurvey of the Land and Resources of Ceylonthat was published in 1960. This documentprovided the basic factual information on soils,land use and water resources for formulation ofplans. It also outlined a “plan for thedevelopment of a large part of the lowland plainof the Walawe Basin,” which included the UdaWalawe reservoir and the Samanala reservoirand eight other dam locations. Concurrently,other plans were developed for the basin,including one published by the U.S. firm,Engineering Consultants, Inc. (ECI). ECIenvisioned a scheme composed of five dams,including Samanala and Uda Walawe dams. Theplan focused on water requirements, economicbenefits and the potential for hydroelectricity(ECI 1960). The ECI Plan was later used as abasis for the Ceylon Irrigation Department Planof 1963.

While plans for the Uda Walawe progressed,in 1956 the government rather hastily initiatedand implemented a more moderate-sizedirrigation and resettlement project that wouldultimately become part of the UWIRP, known asthe Chandrikawewa irrigation scheme. TheChandrikawewa scheme was one of the firstsettlement schemes and included theconstruction of the Chandrikawewa reservoiracross a right bank tributary of the Walawe river(Hulanda river) and the settlement of a total of1,800 farmers on 2,023 ha (5,000 acres) ofjungle. Each family received a lot of 2.02 ha (5

2The Mahagama tank on the LB was among the tanks rehabilitated after the Second World War. This work included the irrigation of 580hectares of rice land and the resettlement of farmers around the tank.3This resulted in families keeping strong ties with their original villages, and they militated against the emergence of a new local solidarityand society.

7

acres; 3 acres of rice land and 2 acres ofhighland for homestead4). The rights to some ofthe land blocked out for the Chandrikawewascheme were held by a few purana (indigenous)villages (Amunugama 1965). Most villagers wereabsorbed into the scheme as settlers but somevillage elders had over 6 ha (15 acres) of landeach and refused to cede their land forredistribution. Their allotments were eventuallyexcised from the blocking-out plan. In addition, agroup of people who had settled 10 years earlierin 3 villages had been left out of the initialselection of settlers for the Chandrikawewascheme and they refused to move. Disputesbetween “official” de jure settlers and de factosettlers, including encroachers, were frequent.

While the Chandrikawewa scheme wasbeing implemented, planning efforts continued inthe basin. In 1963, the Irrigation Departmentunveiled a plan5 for the Uda Walawe reservoirbased on design work by ECI that eventuallybecame known as the UWIRP. The plan focusedon the engineering design details for the projectheadworks, with the goals of generatingelectricity and providing irrigation. The plan wasto irrigate 15,378 ha (38,000 acres) in yala and20,234 ha (50,000 acres)6 in maha7 along theright and left banks of the Uda Walawe river,and provide supplemental irrigation water forapproximately 6,070 ha (15,000 acres) in theLiyangastota anicut command area. The plan ofthe Irrigation Department focused only on thedesign of the infrastructure and made it clearthat the main concern of the moment was to go

ahead with the construction of the dam. Plans fordownstream development were left for a later datewhen “the system of land development to beadopted and the type of agricultural crops to begrown” would be determined. The plan includedtwo main components: (i) a 4-kilometer long and36-meter high earth-filled storage dam andreservoir of 240 million cubic meter live storage,with electrical plants for a combined powergeneration of 5.4 megawatts; and (ii) two irrigationsluices with channels to supply the right and leftbanks.

The project implementation scheduleenvisioned invitation for tenders in 1963,construction beginning in 1965 and completion ofthe entire project, including irrigation channelsfor 20,234 ha (50,000 acres), by 1974. In early1964, the government invited construction bidsand construction began later that year. The RiverValleys Development Board (RVDB) served asthe local implementing government agency.Shortly after construction began, the RVDBbegan allocating land for settlement in Tracts 2-7(see Appendix B), known as the Embilipitiyablock, in advance of the design of the irrigationsystem, a procedure referred to as “advancealienation” (ADB 1995).8 The RVDB broughtsettlers to the land without the preparation ofblocking-out plans; a process of land planningthat takes into consideration the topography andfuture location of canal networks.9 The RVDBengaged in advanced alienation based on the ideathat the settlers could clear and level their landso that they would be ready for production by the

4This area of plot was less than what the colonists were given, 5 acres of irrigated land and 3 acres of dryland, before 1952 (IEC 1954).The allotment of 3 acres of rice land was reduced to 2 acres in 1964 (1 acre = 0.405 ha).5Irrigation Department, 1963: “Development of the Uda Walawe Basin: Proposed Uda Walawe Reservoir Irrigation and Hydropower Project.”6There is inconsistency in the values given for the total area to be developed, partly because some plans include existing developmentand others do not. Later plans identify a goal of 32,780 to 34,398 hectares (81–85 thousand acres) of irrigable land. This estimate includes27,114 ha (67,000 acres) to be developed downstream plus existing development in the Chandrikewewa area (about 2,025 ha or 5,000acres) and Liyangastota anicut area (about 5,062 ha or 12,500 acres).7Maha is the “wet season” from October to March and yala is the “dry season” from April to September.8Tract 1, in the upstream area on the LB, was allocated for a sugar research station and factory, with a small section set aside for theestablishment of a hatchery for inland fisheries. Subsequently, the sugar research station was handed over to the Sevanagala SugarCorporation (ADB 1995).9This resulted in difficulties with future canal networks. Proposed canals intersected some lots, others had to be irrigated from two or moreofftake points, and others did not have frontage to field canals that necessitated construction of ditches through the land of other farmers.

8

FIG

UR

E 3

.E

volu

tion

of w

ater

res

ourc

es d

evel

opm

ent

in t

he lo

wer

Wal

awe

basi

n: 1

900-

2005

.

9

completion of the headworks and would also earnan income assisting with construction activities(ADB 1995). Between 1964 and 1967, about 800ha of land on the RB were cleared and varioussettlement infrastructures were constructed (ADB1979).

Initially settlers were selected based on theprevailing government policy that relied onpredominantly social criteria—landlessness,family size and unemployment. However, localmembers of Parliament eventually reviewed thelists of candidates giving way to allegedpatronage. The RVDB provided each settler withapproximately 1.0 ha of irrigable land and 0.4 haof highland for homestead. Along with authorizedcolonists (and even before them), a largenumber of encroachers began to move into thearea in anticipation of future land grants. HuntingTechnical Services later estimated that by 1968,nearly 60,000 people were living in the area. Inaddition to the 23,000 purana villagers, therewere 9,900 legal settlers (located primarily in theEmbilipitiya area of the RB), 18,000 illegalencroachers (scattered throughout the projectarea) and 1,100 RVDB employees. Thus, by thetime downstream planning began, almost one-thirdof the population in the area consisted of illegalsettlers whose presence ultimately put enormouspressure on the settlement process.

Advance alienation, coupled with otherissues, meant that farmers received little, if any,assistance in crop selection and productionmethods. So they engaged in what they knewbest—subsistence-based agriculture—andcultivated rice using traditional methods. Plentifuland continuous free-flowing water allowed themto cultivate rice even in highly permeable RBE(reddish brown earth) soils. Such high waterduties and infiltration losses created seriousproblems later on, when downstreamdevelopment occurred.

The reservoir was completed in 1968, “inrecord time” according to Mendis (1967) at a

cost of US$9.46 million (Rs 46 million).Contractors completed the canal and networksystems in Tracts 2–7 along the RB (up toChandrikawewa reservoir). Construction on theLB focused on connecting the new system tothe existing village tanks and the newlyconstructed Habaralu tank, and by 1968 the LBcanal was 17 km long (figure 3). Further detailsof the construction are in Appendix A.

Phase II: Downstream Development ofthe RB Area—1969-1978

Planning efforts

It was not until the headworks were completedand the irrigation system was well underway thatplans for a system of land development and atype of agricultural production were initiated.10

The first comprehensive feasibility study of theproject was done in 1967/68 by HuntingTechnical Services (referred to as “Hunting”hereafter), commissioned by The British Ministryof Overseas Development at the request of theGovernment of Ceylon. The Hunting Plan was ahighly ambitious social, economic and physicalengineering project aimed at creating a modern,highly efficient agriculture sector. It increased theproject area by over 10,117 ha (25,000 acres) to32,172 ha (79,500 acres) by including the areaupstream from the command that would berainfed.11 It included detailed plans for suchfactors as cropping patterns, productionpractices, water use and settler selection.

In formulating their plan, Hunting took intoaccount the prevailing macroeconomic policies ofimport substitution and food self-sufficiency thatwere to be achieved through modernization ofthe agriculture sector. The Hunting Plan alsorelied heavily on the earlier survey work carriedout by the Canadian Government in the 1950sand previous reports issued by government

10The 1969 ADB Appraisal Report on Walawe notes that “the investigation and planning of agricultural development in the area started muchlater than those of the engineering side” (ADB 1969).11Both the Hunting Plan and the earlier Irrigation Department Plan included 10,117 ha (25,000 acres) of existing development, the areasunder Liyangastota anicut and Chandrikawewa.

10

agencies and international engineeringconsultants. The soil composition of thecommand area favored the production of otherfield crops (OFCs). Soil surveys of the area hadrevealed two primary soil types—heavy low-humic gley soils (LHG) in the lowland areascovering about 30 percent of the land, which aresuitable for swamp rice, and relatively porousreddish brown earth soils (RBE) in the uplandareas covering the remaining 70 percent, whichare suitable for a variety of field crop such ascotton, sugar, chili and onion (HTS 1968b, 13).12

Soil and water survey data were used inconjunction with government macroeconomicpolicies to devise a cropping pattern. TheHunting Plan states that “within the limitsimposed by technical and social constraints,development plans for Uda Walawe should aimat the efficient production of rice, cotton, sugar,dairy products, chilies and onions.” Withcropping patterns determined, the whole designand planning process unfolded. The plan thatemerged assumed that all aspects of the projectcould be “engineered.” For example, it wasassumed that all irrigated land would be double-cropped annually, that the area under each cropwould be fixed, that “farmers will apply therecommended inputs,” and that 12,000 settlerfamilies would be selected based on theirpotential to become efficient modern farmers.The plan also assumed that the 18,000squatters, who had moved into the area inhopes of securing land, would be removed.

The very feasibility of the project itselfdepended, among other things, on efficientproduction of OFCs on the RBE soils. This, inturn, required the selection of the “right” settlers,adoption of the “right” practices (includingmodern chemical inputs and efficient water use),and eviction of large numbers of squatterscurrently on the land. Failure to achieve thisobjective would endanger the economic,environmental and social outcomes of the

proposed project. In order “to reduce growingforeign exchange costs,” targeted crops includedrice, cotton, sugar, chili and onion on 70 percentof the irrigated land, which amounted to all theland with RBE soils. Hunting stressed that theseRBE soils “should on no account be used forswamp irrigation because puddling is likely todestroy their structure,” and result in erosion andother problems (HTS 1968b, 18). The plan alsostressed the importance of evicting squatters “astheir continued occupation is bound to prejudicethe ordered settlement and could compromisethe whole concept of the project.”

Taking into consideration the Hunting Planand additional research conducted by a FAO/UNDP Mahaweli team, the governmentpublished a Plan of Development and EstimatedCosts for the Walawe Project (GOC 1969). Thisplan was based largely on the Hunting report butreduced the command area and croppingpatterns to accommodate higher water dutiessuggested by the FAO/UNDP Mahaweli team.On the basis of these revised water duties, theGovernment Plan estimated the total irrigablearea available for development as 27,357 ha(67,600 acres). This area included 3,432 ha(8,480 acres) of developed land (see AppendixC) to which 2,133 ha (5,270 acres) of land underirrigation from Chandrikawewa were laterincorporated.13

The Government Plan estimated the cost ofthe integrated development project, including allinfrastructure, at US$73.6 million (Rs 438 millionin 1969 prices). However, the cost-benefitanalysis was based on US$58.8 million (Rs 350million) to be distributed between 1964 and1972. This later estimate excludes theconstruction costs of main roads, hospitals,schools and police stations because, the planargued, many of these would have been providedregardless of the project. The reduced costestimate also excludes the cost of establishingindustrial units (sugar, brick, cottage, and sawmill

12The proportion of red soils was about 70 percent in both areas already developed and in the area proposed to be developed, based on theestimates provided by Hunting.13The Liyangastota anicut command area was excluded from this estimate.

11

industries) and anticipated benefits thereof. Of theUS$58.8 million allocated for 1964-1978,US$22.86 million had been spent by September1968.

Full development of the RB

In 1969, the Government of Ceylon approachedthe Asian Development Bank (ADB) to obtainfinancial assistance for the development of theUda Walawe Irrigation and Resettlement Project(ADB 1969). As it would be ADB’s first integratedrural-development project in Sri Lanka, ADB optedto finance development of the RB initially and toconsider the LB upon completion of the RB work.The ADB appraisal document largely reflected the1969 Government Plan with one criticalexception—it allowed for the production of rice ona portion of the area previously devoted to OFCsthat contained RBE soils. This decision and itsimplications are discussed below.

The area of the ADB project included theentire RB area (12,369 ha) articulated in theGovernment Plan. This area included the twoareas that were already settled—Chandrikawewaarea (2,134 ha) and Tracts 1-7 (1,520 ha)—andTracts 9-21 (8,715 ha) that were to be developed(see Appendix E for further details). The goals ofthe project were to develop and settle Tracts 9-21, improve agricultural productivity on existingland, rehabilitate the existing irrigationinfrastructure and construct new irrigation facilitieson Tracts 9-21, and construct village centers andother support infrastructure, including anagricultural experiment and extension center.

Following the Government Plan, the ADBproject sought to provide water supplies to ensuredouble cropping and enhance yields. It alsosought to accommodate about 3,440 new settlersfor intensive farming and to improve theagricultural productivity and living conditions forthe 3,100 earlier settlers. Due to high water usagein the existing RB command areas, improvingwater-use efficiency in already developed areaswas paramount to further development in the LBarea. The ADB deviated from the 1969

Government Plan (GOC 1969) in one criticalaspect; it permitted an expansion of riceproduction on to the RBE soils. According toNijman (1991), around the time ADB becameinvolved “the government decided that rice wouldbe grown on the well-drained soils of the areaupstream of Chandrikawewa reservoir. Thedecision was taken on the ‘insistence’ of a thenMinister.” Visvalingam (1986) cites this as a“disastrous example of the misapplication ofministerial omniscience,” although one maywonder whether farmers would have refrainedfrom growing rice, had this decision not beentaken. This expansion in rice area threatened analready fragile project on a number of fronts bytaxing available water supplies, limiting economicprofitability and diversification plans, damagingfragile soils and causing erosion.

At appraisal, the estimated cost of theproposed work was about US$26.45 million, witha government share of US$17.46 million andUS$8.99 million foreign financing from ADB. TheRVDB continued to serve as the implementingagency. The project commenced in 1970 with anestimated completion date of October 1973.However, because of a variety of problems theproject was extended to January 1977, when itwas about 90 percent complete.

The project suffered delays from the onset,beginning with a lag in hiring consultants tooversee development of project design andimplementation. Finally, it was launched in late1970 and construction began in early 1971.However, the rushed time frame between thehiring of consultants and engineers andimplementation resulted in “inherent deficienciesin the original engineering design” (ADB 1979).One major design limitation was a single-bankcanal system rather than the conventional double-bank canal system. Others included poorlydesigned control structures, lack of provision torecapture return flows, an inadequate number ofcross regulators and measuring devices, andlimited live storage of the Chandrikawewa (ADB1979). These design flaws were constraints toproper water management.

12

In 1971, shortly after construction began,three events—of the many in a long series ofevents to come—occurred that impededimplementation. First, political upheavals causedsevere disruptions in construction. Second, theGal Oya project was completed and a politicaldecision was made to transfer 5,000 unskilledworkers from that project to Uda Walawe. Themassive labor force, which could not beabsorbed, financially crippled the RVDB. Third,political tensions crept into the RVDB resulting inchanges in its top management, an occurrencethat would become common in the history of theRVDB. Over the 10-year period between 1969and 1979, the RVDB had nine differentchairmen.

Flawed design plans coupled with ongoingchanges in top management, a shortage ofexperienced technical staff and a surplus ofunskilled labor, resulted in poor-qualityconstruction of irrigation infrastructure andongoing delays resulted in a number ofunintended, but serious, consequences. Forexample, water use by settlers on previouslydeveloped lands in Tracts 2-7 andChandrikawewa (who were used to receiving andusing the supply of water intended for the entireproject area) was so high that a 1972 ADBreview of the project determined that theplanned irrigable area would have to be scaledback. This resulted in a decision to drop Tracts20 and 21 (676 ha) from the originally plannedproject area (see Appendix B).

Project delays and lack of administrativecontrol also created problems in land allocationand settlement. Altogether, 3,300 new settlerswere projected to settle in Tracts 9-21. The sizeof a holding was 2.02 ha (5 acres; 4.5 acres ofirrigable land and 0.5 acre for a homestead).Settlers were to be selected based on theirpotential to become efficient farmers using an“objective” point system. Land allocations initiallyfollowed the plan. Settlers were to be provided

with training to ensure they complied withrecommended cropping patterns and activelycooperated in system operation andmaintenance (O&M). The plans also called forconstruction of housing for new settlers.

Despite the plans to select settlers based oneconomic criteria, political and socialconsiderations crept into land allocationdecisions very early on, partly as a result of thetremendous demand for land by the largenumbers of encroachers in the project area. Asa result, “irrigable land allotments were reducedsuccessively from 4.5 acres (as originallyrecommended) to 3 acres and again to 2.5acres; homestead lots were reduced from 0.5acres to 0.25 acres” (ADB 1979). By 1979, thenumber of new settlers who were provided landunder the project was 4,143 (compared to theprojected number of 3,300), in addition to about2,800 settled earlier in the Chandrikawewa areaand Tracts 2-7.

Delays in placing legal settlers on the landfurther fueled encroachment and early efforts toevict the encroachers proved ineffective. Theproblems of encroachment were furtheraggravated by the lack of land surveys andmarkings for allotments and delays in issuingland permits to legal settlers. Encroachmentcreated disputes among settlers regarding rightsto land and water use but no legal action wastaken. The illegal status of the encroachersprevented their access to irrigation water,institutional credit and other agricultural supportservices. As a result, encroachers resorted toforce and water theft. These actions frequentlycaused damage to the irrigation structures andtended to disrupt attempts to introduce effectivewater management and agricultural developmentactivities in the project area. The RVDB largelyignored encroachment, which compounded theproblems over time.14

Due to mounting pressures for land and therather chaotic allocation process, settlers were

14Encroachment problems were not addressed until 1982 when the management of the project was transferred from RVDB to the MahaweliAuthority of Sri Lanka (MASL). By that time, some encroachers had been on the land for well over 15 years.

13

allocated land faster than the RVDB could supplyhousing. The original plan to provide frame-and-roof houses for all settlers was abandoned by1972 and in lieu of a house the RVDB provided asubsidy of US$50.25 (Rs 300). Only 933 of thetargeted 5,380 housing units were constructed.This decision had ramifications for futurecommunity development plans—it led to scattereddevelopment and defeated the objectives ofcluster-style development, including the plannedprovision of shared infrastructure, such as wateron tap and promotion of community cohesion.

The hasty and uncoordinated land-allocationprocess also resulted in a total lack of trainingand education to settlers, to ensure theircompliance with planned cropping patterns andsystem O&M. Proposed agricultural extensionefforts, subsidies for chemical inputs, anddevelopment of marketing channels for OFCsoccurred only on a limited basis, if at all. By theend of the project period, the averageconsumption of fertilizers was only about 20percent of the recommended quantities. The lackof marketing channels created severe problemsfor some settlers who produced OFCs,particularly cotton. There are many stories offarmers who followed the advice of extensionofficials and grew cotton only to find themselveswith no buyers and piles of rotting cotton.

The uncoordinated settlement process, incombination with other factors (including creditshortages, lack of marketing channels andinexperience with OFCs), created anenvironment where farmers opted for what theyperceived as the least risky approach. Theyfocused almost exclusively on rice productionusing traditional production practices withoutregard for soil types. This resulted in higher-than-anticipated water usage, which exacerbatedthe systemwide water-management problem asthe irrigable area expanded, particularly fordownstream water users.15 Thus, the decision to

expand (or tolerate the expansion of) upstreamrice production combined with poor watermanagement created substantial inequitiesamong project beneficiaries. Without water,downstream settlers began to abandon the land,which further fueled the encroachment problemas encroachers often moved onto these landsonce they were abandoned.

Due to all these problems, the project facedrepeated delays. The original project completiondate of October 1973 was repeatedly extendeduntil January 1977, when the project was closedwith about 90 percent of the anticipated workcomplete. Due to these delays and higher-than-anticipated costs for construction equipment andmaterial, the project faced substantial costoverruns. The project cost at appraisal ofUS$26.45 million (US$8.99 million foreign fundsand US$17.46 million local funds) escalated toUS$33.63 million (US$8.82 million foreign fundsand US$24.81 million local funds), resulting in acost overrun of 27 percent. The government borethe full brunt of the overrun, which increased itsshare of the cost by 42 percent. Escalatingproject costs, combined with the loss of potentialbenefits of Tracts 20 and 21, were offset by anunanticipated surge in international rice prices.Between 1968 and 1979, increased riceproduction combined with higher prices yieldedhigher-than-estimate returns per unit area (peracre) and resulted in foreign exchange savingsestimated at US$35 million. Thus, the short-runeconomic sustainability of the project dependedprecariously on continued high rice prices.

By 1979, structural deficiencies, poor watermanagement and inequitable distribution ofavailable water supplies continued to seriouslythreaten project objectives. The original plansanticipated that the Uda Walawe reservoir wouldprovide sufficient water resources to irrigate acommand area of 32,937 ha (comprising 12,662ha on the RB and 20,275 ha on the LB).

15By 1979, of the 6,680 hectares in Tracts 9-19, the irrigated area never exceeded 3,846 hectares in maha and 2,834 hectares in yala.Planned versus actual cropping patterns also diverged significantly. The plan envisioned that 59 percent of the land in Tracts 9-19 wouldproduce OFCs (cotton, chili and red onion in yala, and maize and groundnut in maha); however, the actual area devoted to OFCs rangedonly between 2 and 7 percent.

14

However, the entire supply of the reservoir wasdevoted to only 39 percent of the plannedproject area, on average16 (ADB 1979).

According to the ADB Project CompletionReport (ADB 1979):

“… irrigation water use in the projectarea far exceeds original expectationsand threatens to curtail furtherdevelopment on both banks. Only about70% of RB area envisioned for irrigationat the time of appraisal is actually servedand this area is consuming 3 times thewater proposed for the entire RB area.Only the lack of development on the LBhas permitted this excessive use.”

Phase III: Walawe Irrigation andImprovement Project—1984-1994

Planning of rehabilitation works

The rather abysmal outcome of ADB’s firstintegrated rural-development project in Sri Lankaushered in planning for a subsequent phase ofrehabilitation on the heels of the first project.According to the 1979 ADB report, the “irrigationsystem has deteriorated to such an extent as torequire immediate rehabilitation and improvementin order to attain full utilization of the availableland and water resources.” Over the next coupleof years, the ADB continued to stronglyadvocate a rehabilitation project aimed atimproving system performance and was ready tostart a rehabilitation project in 1980. Nijman(1991) argues that the Government of Sri Lanka

showed little interest in rehabilitation and onlyagreed to the rehabilitation in April 1982because ADB made rehabilitation of UdaWalawe a precondition to funding several otherhigh priority government projects.17

When the government agreed torehabilitation, there was no rehabilitation plan inplace. The UWIRP was far from reaching thegoals of the original plans to develop and settleabout 32,000 ha (79,000 acres) on both banksof the Walawe river. As of 1982, only 56 percentof the planned command area had beendeveloped, which included 12,000 ha in the RBcommand area and 6,000 ha in the LBcommand area.18

Due to poor performance, the managementof the project was transferred from the RVDBto the Mahaweli Authority of Sri Lanka (MASL)in 1982. MASL took a number of actions. Itincreased agricultural productivity byestablishing two distinct growing seasons—maha and yala—and instituting a mandatoryfallow season, which disrupted the reproductivecycle of pests and disease agents resulting inincreased yields and improved watermanagement.19 MASL introduced rotationalirrigation schedules, as originally planned, insome areas of the project to make wateravailable for downstream users who hadpreviously been denied irrigation supplies. Thiswas strongly opposed by some upstream users.However, MASL took measures to improve datacollection and operation and maintenance,strengthened agricultural support services andinitiated a crop diversification program.

MASL also began to address theencroachment problem, which had essentially

16An average of 8,502 ha on the RB and 4,412 ha on the LB.17According to Nijman, “First, the Appraisal Mission for supplementary funding for Kirindi Oya made such funding to Kirindi Oya conditionalto the willingness of the government to rehabilitate Uda Walawe. In addition, the donor would consider funding of a road construction projectin the Accelerated Mahaweli Program only if the government was willing to agree with rehabilitation of Uda Walawe. Another incentive forthe government to accept rehabilitation of Uda Walawe was the donor’s explicit willingness at that time to consider future loans for thedevelopment of the Walawe Left Bank. The latter was envisaged to boost the EIRR of the donor’s overall investment in Uda Walawe” (seeNijman 1991, 119 for further details).18The LB command area included about 4,000 ha of existing rice lands and about 3,000 ha being developed under a state sugar project.The command area of the Walawe project excludes an additional 5,000 ha developed under the Liyangastota diversion dam.19Prior to MASL’s involvement, farmers planted rice throughout the year, so at any point in time one could observe rice fields in variousstages of maturation.

15

been ignored by the RVDB.20 By 1982, anestimated 4,100 illegal encroachers had usurpedland and many had been there since the project’sinception (ADB 1995). Encroachment also causeddisputes among settlers regarding rights to landand water use. The illegal status of theencroachers prevented their access to irrigationwater, institutional credit and other agriculturalsupport services. As a result, water theftsbecame a common practice. Some encroachersresorted to use of force and often damagedirrigation structures in the process. They alsodisrupted attempts to introduce more effectivewater management and agricultural developmentactivities in the project area. Although MASLgave the issue of regularizing encroachers toppriority, work progressed slowly because ofnumerous legal and administrative problems.

In a relatively short period of time, the MASLefforts showed impressive results. Throughbetter water management, MASL increased theannual irrigated area from about 11,000 ha to16,000 ha by 1984. By creating two distinctgrowing seasons and other extension efforts,paddy yields increased from 3.6 tons per hectareto 4.0 tons per hectare during the same period.Through a culmination of these efforts, overallrice production increased from 40,000 tons to60,000 tons, resulting in a one-percentageincrease in the economic internal rate of return(EIRR) from 7 percent under the earlier ADBproject to 8 percent.

These improvements were well received byADB. However, they argued that the project wasyielding only about 60 percent of estimatedbenefits and these were too inequitablydistributed, with farmers in the upstream areasreceiving most of the benefits at the expense ofthose downstream. ADB (1984) argued that theseimprovements were “ad hoc and temporary, entailhigh recurring costs that could not be sustainedover time without a systematic and well-designed

improvement and rehabilitation program to rectifydesign and construction deficiencies of theexisting system.” From ADB’s perspective,“because of the project’s current unsatisfactoryperformance, it would not realize its long-termobjectives unless a carefully designedrehabilitation and improvement program wasundertaken to rectify the physical and institutionalproblems encountered with the project” (ADB1982; Project Performance Audit Report).

In 1983, ADB financed a mission to identifyrehabilitation priorities. The findings from themission revealed that “[the] major problemsfaced by the Walawe irrigation system can beattributed to a past lack of management ratherthan to inherent design or construction defects”(Wolf 1983, iv quoted in Nijman 1991, 120).While they considered major rehabilitation worksinfeasible, they did suggest minor improvementsand rehabilitation to the irrigation system andcited the need to consider potential supply-augmentation options. ADB disagreed with themission’s findings, according to Nijman, and in1983 hired SOGREAH, an internationalengineering consultancy firm, to prepare afeasibility study for rehabilitation.

By 1984, SOGREAH completed a feasibilitystudy for rehabilitation. It presented a detailedproposal that involved “a major rehabilitationwithin all distributary-channel subsystemscombined with a program of organizing the waterusers by means of institutional organizers”(Nijman 1991, 120). According to Nijman, “Thefeasibility study thus proposed an engineeringsolution for the excess water use in UdaWalawe, without giving further project objectiveswhich would tackle the managerial performanceof the managing agency apart from training andan operation and maintenance manual” (Nijman1991, 120).

Several months later, ADB submitted aProject Appraisal for the Walawe Irrigation and

20A number of factors fueled encroachment. The primary cause was delays in placing legal settlers in possession of allotments. Theabandonment of land by legal settlers, especially in the tail-end areas that received little or no water, resulted in more encroachment.Encroachment was further aggravated by the lack of allotment surveys and land marking and the failure by the RVDB to issue land permitsto legal settlers. Prevention of encroachment was difficult because no legal action was taken as soon as encroachment occurred.

16

Improvement Project (WIIP) based on theSOGREAH plan. According to this document, theprimary objectives of the rehabilitation projectwere to work towards realizing the goals of theoriginal 1969 plan for Uda Walawe in a manner“consistent with the Government’s sectoraldevelopment objectives of increasing agriculturalproduction, farm income and employmentopportunities” (ADB 1984). The overridingobjective was to increase agricultural productionand water-use efficiency in the RB commandarea and use resulting water savings for furtherdevelopment planned for the LB area. In short,rehabilitation was needed to take the project towhere the 1969-79 construction phase wasexpected to take it. The project included (ADB1995): the rehabilitation of the RB irrigationsystem, including the main irrigation system andon-farm distribution systems; construction ofwells for potable water supplies to scatteredsettlements and village centers; rehabilitation ofroads; provision of essential equipment andvehicles to sustain O&M of the system;consulting services and training; and plans tostrengthen ongoing adaptive research to promotecrop diversification.

While these were the explicit projectobjectives, the project also contained implicitobjectives for MASL, including implementation ofintensive training programs in irrigationmanagement at system level and farm level, andincreasing the involvement of farmers inirrigation scheduling decisions, regularizingencroachers, adjudicating water conflicts andcollecting irrigation fees. The ADB Appraisalreport states, “it is believed that under theproject, the improved irrigation system with anassured and adequate irrigation supply,regularization of encroachers, intensive trainingin irrigation management and involvement offarmers in decision making regarding irrigationscheduling, adjudication of water conflicts, and

collection of the irrigation service fee will providethe necessary impetus for strong and stablewater user groups” (ADB 1984, 16-17).

The estimated cost of the 5-year project atappraisal was US$13.7 million with US$11.0million in financing from ADB. The projectcommenced in October 1984 with an estimatedcompletion date of October 1989.

Implementation of the rehabilitation works

The project suffered delays from the onset. Morethan a year passed before the internationalconsultant, Sir M. MacDonald and Partners(MMP in what follows) was hired and had arrivedat the site. MMP was responsible for advisingand assisting the local engineering consultants—Central Engineering Consultancy Bureau orCECB—who would do the actual design work.CECB was also responsible for supervisingconstruction. The design work took over 2 yearsto complete due to a number of factors, such aspoor topographical information, problems withearlier plans, and CECB’s lack of experience indesign works for rehabilitation of irrigationinfrastructure. Thus, detailed design plans werenot completed until 1988, one year before theplanned completion date of the project.

Beyond technical difficulties with projectdesign, work was delayed due to civildisturbances associated with insurgent activitiesof the JVP (Janatha Vimukthi Peramuna orPeople’s Liberation Front) related to theupcoming presidential elections. Insurgentsharassed contractors by demanding financialassistance and killed employees of the engineerand contractors. Field staff became reluctant toundertake survey work and hold discussions withfarmers about design plans.21 This difficultsituation was compounded by the limitedavailability of local funds and regular shortages ofcement. The Government Treasury did not

21JVP insurgents engaged in widespread intimidation of project staff. Curfews were imposed in the project area and that disrupted work. InJuly 1987, insurgents used a contractor’s bulldozer to damage the Mahaweli Economic Agency (MEA) manager’s office. Shortly afterward,insurgents burned down the design consultant’s office, which resulted in the loss of field notes and survey and design work for Tracts 12-14.Under the guise of security operations, field staffs were stopped and their vehicles seized. A number of field-staff houses were also seized.This created a climate of poor morale and distrust (ABD 1994).

17

release funds as stipulated in budgetaryprovisions. The situation was aggravated bySOGREAH’s failure to properly estimate costsand incorrect expenditure forecasting that createda higher-than-anticipated demand for local funds.It soon became clear that the actual cost of theproject would vastly exceed initial expectations.

In March 1988, the government estimatedthat to complete the rehabilitation project itwould require an additional US$11.6 million inaddition to the originally planned US$13.7 million(Nijman 1991). Of this amount, ADB agreed toallocate an additional US$3.8 million to cover thedepreciation of the dollar and the governmentfinanced US$2.2 million (Nijman 1991). Thegovernment approached ADB for asupplementary loan of US$5.5 million but it wasdeclined because of the anticipated loweconomic return. ADB estimated that, with thesupplemental loan, the EIRR for the project,including sunk costs, would be about 7.5percent. Without the supplemental loan, thedownstream tracts of the RB command areawould not be rehabilitated. After muchdiscussion, the government applied for andreceived a loan through the Agricultural LoanProgram, which did not contain EIRRrequirements. At that time, ADB indicated itwould not be willing to fund future work in theLB area.22

Beyond rehabilitating the irrigation system,the ADB appraisal report planned for theconstruction of 100 shallow and 150 deeptubewells for scattered developments. Shallowwells were quickly abandoned because theywere recharged from return flows of surface-water irrigation and tended to dry out during thedry season. Little attention was paid to deep-welldevelopment early in the project and it was notuntil June 1990 that a contract was tendered fordrilling. Although a 1984 feasibility study bySOGREAH suggested water-quality problemsmight be an issue (especially salt and fluoride

problems), a careful analysis was not completeduntil 1992 when 186 wells had been drilled. Ofthese 186 wells, only 40 complied with WorldHealth Organization (WHO) drinking water qualitystandards; excessive levels of fluoride wereidentified as a particularly acute problem inaddition to high levels of iron. Signs were postedat wells that did not meet WHO water-qualitystandards stating that the water should not beused for drinking purposes. However, anecdotalevidence collected by International WaterManagement Institute (IWMI) field staffsuggested that many people continued to usethese wells for drinking water due to lack ofalternative sources.

While the engineering side of the projectsputtered forward haphazardly, the MahaweliEconomic Agency (MEA) worked towardscreating active water-user groups. In the projectarea, as elsewhere in Sri Lanka, farmerorganizations had been tried in the past withlimited success. Government policy required alllegally settled farmers to join water-user groups.So officially some 530 water-user or farmergroups existed in Uda Walawe, and they werelargely inactive.23 In late 1986, MEA establisheda subcommittee charged with the creation,organization and training of more active water-user groups. By 1988, 173 groups wereorganized and trained (ADB 1995). Despite theefforts, many of these groups quickly becameinactive. In 1987, IWMI was asked to providetechnical assistance to MEA on how to improveexisting irrigation institutions and to proposestructural and management innovations toimprove project performance (ADB 1995). Afteran extensive study of the ChandrikawewaBranch Canal, IWMI concluded that theshortcomings of MEA’s early initiative with water-user groups was due to: (i) lack of an integratedapproach to the water-user group program, (ii)lack of resources to achieve high expectations,(iii) delays in commencing rehabilitation, (iv)

22As a result, the government initiated discussions with the Bank of Japan about financing future rehabilitation and development in the LBarea of UWIRP.23Encroachers were forbidden from joining water-user groups.

18

political disturbances, and (v) lack of autonomyamong farmers in their relationship with MEAofficers (IIMI 1990a). IWMI made two principalrecommendations in its report. The first was toimprove integration and coordination amongvarious technical divisions operating in theproject area and, the second, to create strongerfarmer organizations with effectivecommunication channels to MEA staff.

IWMI also highlighted the lack of farmerinvolvement in the design process (IIMI 1990b).Up to this point, farmers were effectivelyexcluded from the design process for the projectbecause of the marked tendency towardscentralized decision making that characterizedproject planning and implementation in UdaWalawe from the onset. In an attempt to rectifythe situation, CECB consultants held discussionswith farmers once construction began to explainrehabilitation plans, obtain suggestions and makechanges if possible. These discussions continuedthrough the remainder of the implementationperiod and “substantially improved theestablishment of distributary and field canalfarmer groups by the MEA” (ADB 1995).

Farmers also became involved in theconstruction process when, in 1990, MASLmoved away from large-scale contracts towardmedium- and small-scale contracts forconstruction projects, and farmer groups wereallowed to bid for small-scale projects.24 Thechoice of involving farmer groups in small-scalecontracts was motivated by two additionalcompelling factors: “to foster a sense ofownership amongst the farmers and a recognitionof mutual responsibility for repairs andmaintenance, and to develop technical andorganization skills amongst the group” (ADB1995). This concept, which evolved from IWMI’s

research on farmer organizations, envisioned thatfarmer organizations along each canal wouldparticipate in the rehabilitation of their canal whileearning additional income. Initially, the conceptworked well. However, farmers becamedisillusioned because payments were made tofarmer organization leaders and many of themskimmed unusually high “overhead costs” beforedispersing payments to other farmers.25

Between 1990 and 1993, IWMI workedcollaboratively with MEA to develop anddemonstrate systems for strengthening farmerorganizations and for planning preseasonmaintenance and water deliveries in theMoraketiya branch canal (the main branch canalof the upper right-bank area). This processfostered mutual understanding on both sides.Based on this case study, a carefully structuredprogram for establishing and strengthening farmerorganizations was launched in 1993. A majorimpetus for this work was the strong politicalopposition to the 1984 government policy thatsignificantly increased per unit area (per acre)assessments for system O&M and stepped upcollection efforts. By 1988, the national irrigationO&M policy was repealed and replaced with onethat required farmers to share in system O&M.Under this program, the O&M of the distributaryand field canals were to be handed over fromMEA to farmer organizations (FOs).

The 1994 ADB report anticipated that FOswould take full responsibility for the O&M ofdistributary and field canals by the late 1990s.This was a very optimistic assessment sincemany of the FOs had still not become self-sustaining and autonomous. Like elsewhere, theeffectiveness of the participation process wasconstrained by the lack of real control by usersover water allocation and distribution, the latter

24During the same period, MASL terminated the services of CECB as a cost-saving measure and opted for in-house engineering servicesfrom the Mahaweli Engineering and Construction Agency (MECA).25However, there were exceptions in the distribution of profits among farmer organizations. In the Moraketiya branch canal area near Embilipitiya,farmers worked on a shramadana (voluntary labor) basis and used the profits to set up a fund to lend money to lower-income farmers forfertilizers (ADB 1995).

19

being partly constrained by structural limitationsresulting in supply uncertainties. One overridingobstacle to the turnover of O&M to farmers wasthe system of patronage developed at UdaWalawe.

In addition to organizing and engaging farmerand water-user organizations, MEA extensionstaff made a tremendous effort to increase cropdiversification. MEA instituted a mandatory fallowperiod and introduced faster maturing ricevarieties. In 1988, MEA began an aggressivediversification program that included training, fielddemonstrations, and provision of seeds andseedlings for OFCs (ADB 1995). This worked toincrease production of OFCs. In 1984, OFCswere cultivated on only 5-6 percent of irrigatedland. By 1994, this proportion had increased toroughly one third of the area, due to theexpansion of banana cultivation. It is hard,however, to separate the combined effects ofextension efforts, declining rice prices andoccasional water shortages.

Banana production was a great cropdiversification success. Banana was cultivated inthe lower part of the project area in the 1980s,but extension efforts mediated by influential pilotfarmers and technicians from MASL led to aspectacular spread of its cultivation. The resultswere remarkable. By 1994, bananas grew onapproximately 1,500 ha (14%) of the 10,500 haunder production in the RB area (ADB 1995) andreached 4,333 ha (out of 14,000 ha) in 2003. Theadoption of this high-value crop significantlyboosted farm income and the whole economy ofthe Embilipitiya town. While other diversificationefforts resulted in short-term conversion to OFCs,very little land converted to bananas reverted torice, although this has not been uncommon duringthe past two years.

Phase IV: LB Rehabilitation andExtension—1995-2005

Prior to completion of the RB rehabilitation andimprovement project, the Government of SriLanka approached the Government of Japan forassistance to conduct a feasibility study forupgrading and extending the LB command area.The Government of Sri Lanka initiated thesediscussions in 1987 after ADB indicated it wasno longer interested in financing further work inthe LB area. Despite the poor operational andfinancial performance of past investments inUWIRP, the Government decided to furtherdevelop the LB and complete the project asoriginally planned in 1969. This decision wasmade “in order to fully utilize and recover pastinvestment and ease the ever increasingpopulation pressure in the south of the country”(Nippon Koei and MASL 1995).

In 1991, the Japan InternationalCooperation Agency (JICA), working withMASL, commenced a feasibility study foragricultural development in the LB area, whichreport was issued in 1992. The plan involvedthree basic structural elements: (i) rehabilitationof existing irrigation facilities in the “Old” area,(ii) construction of new irrigation facilities in the“Extension” area, and (iii) improvements in ruralinfrastructure (Nippon Koei and MASL 1995). In1993, the Government of Sri Lanka requestedfinancial assistance from the Government ofJapan to implement the LB Irrigation andUpgrading and Extension project.26 TheOverseas Economic Cooperation Fund ofJapan conducted an appraisal in 1994 andagreed to provide financing for the engineeringaspects of the project (Nippon Koei and MASL1995).

26In 1993, Sri Lanka asked Japan for an aid grant to provide “urgently required development components” of the LB project, including roadimprovements, water-supply facilities at Suriyaweva, and procurement of water tankers to deliver drinking water to scattered settlers (NipponKoei and MASL 1995). The grant was approved in 1993. With design work by JICA, construction began in 1994 with completion expected in1995.

20

In 1995, Nippon Koei was hired to review andupdate the 1991/92 feasibility study conducted byJICA.27 From their perspective, the objectives ofthe rehabilitation and extension project were toincrease food self-sufficiency by increasing riceproduction, increase production of OFCs, increaseemployment opportunities, mitigate environmentaldegradation, boost regional economicdevelopment, and alleviate poverty (Nippon Koeiand MASL 1995). The scope of the projectconformed to the earlier feasibility study;however, the review suggested changes incropping patterns, rural infrastructure developmentand implementation schedule.28

Nippon Koei estimated the complete cost forthe project at US$144 million (1995 prices), whichincluded US$34 million for the rehabilitation andupgrading of the “Old” area and US$110 million forthe development of the “Extension” area. Theimplementation schedule included two phases.The first phase involved rehabilitation of the Oldarea with construction beginning in 1997 andcompletion in 1999. The second phase involveddevelopment of the Extension area withconstruction commencing in 1999 and completionby 2003.29

Nippon Koei concluded that the project was“technically feasible and economically sound,and the adverse environmental impacts couldbe minimized by mitigation measures” (NipponKoei and MASL 1995, p. S-6). The economicjustification for the project included anestimated EIRR of 17.8 percent and a cost-benefit ratio of 2.16, using a project period of50 years and a 10 percent discount rate. Theestimated EIRRs for the Old area andExtension area were 11 percent and 22percent, respectively. Estimates of benefitswere based on a projected tenfold increase infarm income from US$293 to US$2,927 (Rs15,000 to Rs 150,000) per hectare per year,which seems overly optimistic at best (NipponKoei and MASL 1995).

In 1996, rehabilitation and modernizationworks began in the Old area. In 2003, therehabilitation within the Old area was completedand construction of infrastructure for theExtension area was begun with hopes for a firstirrigated cropping season in yala 2004 (in at leasthalf of the area). It is too early to know what willbe the impact of the inclusion of this new area onthe management of the whole project.

27The terms of reference for Nippon Koei’s work included several objectives: (1) ensure that proposed cropping patterns are based on realisticassumptions; (2) identify the impact of the crop diversification policy on water requirements, water supply and canal/structure designs; (3)conduct a domestic and international marketing study for agricultural products; (4) study the feasibility of establishing an agricultural creditsystem; (5) estimate the price elasticity of demand for agricultural products; (6) examine the potential for constructing marketing infrastructurefor agricultural products; (7) conduct a detailed environmental study; (8) review and update the proposed settlement plan; (9) review andupdate the previous study on water and land resources, including hydrological analysis; and (10) design a concept for on-farm developmentaccounting for O&M of irrigation facilities (Nippon Koei and MASL 1995).28The proposed changes to cropping patterns involved the replacement of sugarcane with other OFCs because the government had failed tofind a private buyer for the Sevanagala sugar mill. The review also suggested changes in rural infrastructure development. Based on findingsby MASL, Nippon Koei recommended consolidating the number of village settlements, from 22 to 12, and increasing the number of familiesper settlement area. They also recommended increasing the role of international consultants in the implementation process due to MASL’sdecision to play a reduced role in construction management and supervision.29The implementation has been delayed by 2 years.

21

This section examines planned versus actualoutcomes for a number of key variables, suchas costs, implementation period, irrigated area,irrigation intensity and yields. To place theexperience of UWIRP in context, these findingsare compared to the performance of otherirrigation projects, where possible.

Capital Costs: Actual vs. Predicted

From its inception and through successivephases of investment, UWIRP experiencedsubstantial cost overruns.

Table 1 shows the estimated cost atappraisal and actual cost at completion forvarious phases of investment. In 1969, theestimated cost of the entire UWIRP wasUS$269.6 million (Rs 13,479 million in 1995prices) for the headworks, irrigation facilities anddownstream development of 32,937 hectares.Actual investment costs from 1964 to 1995 wereUS$322.92 million (Rs. 16,146 million in 1995prices) plus the proposed costs of the LBupgrading and extension project, which yieldedan estimated total cost of US$466.9 million (Rs

23,345 million), resulting in a cost overrun of 73percent. This estimate assumes that the entireproject is completed as envisioned. Ifinvestments are tallied through 1995, the costoverrun escalates rapidly.

While the costs of the project were 73percent higher than anticipated, the irrigablecommand area for the project fell significantlyshort of what was planned. By 1995, only 55percent of the planned command area wasirrigable. Thus, in 1995 the actual cost perirrigable unit area was roughly double what wasanticipated in 1969. These data suggest that notonly did costs substantially exceededexpectations, but actual benefits—measured bycommand area—fell far short of thoseanticipated.

Cost overruns characterize each phase ofdevelopment in UWIRP, with overruns of 27percent and 85 percent, respectively, for theADB-financed RB downstream developmentproject and the subsequent rehabilitation project.A number of factors contributed to cost overrunsover the years, including poor estimation ofcosts, design and construction problems,implementation delays, difficulties with

TABLE 1.Cost performance by phase of development (1995 prices).

Project Cost at appraisal Cost at completion Overrun (%)

Million

Entire project: 1969-2005 (estimated)a US$269.5 (Rs 13,479)b US$466.9 (Rs 23,345)c 73

ADB-RB downstream: 1969-1978 US$26.45d US$33.63d 27

WIIP: 1984-1994 US$13.7e US$25.3e 85

LB upgrade and extension Upgrade: US$33.5f Project ongoing Project ongoingExtension: US$110f

a “Entire project” as defined by the 1969 Government Plan, which estimated the total irrigable area available for development as 27,357ha (67,600 acres). This area included 3,432 ha of already developed land.

b Appraisal cost equals 1969 total plan of development costs of Rs 430 million expressed in 1995 prices (GOC 1969).c Completion cost equals total stream of actual investment costs from 1964 to 1995 plus expected costs of LB upgrade and extension

project expressed in 1995 prices (Source: Nippon Koei and MASL 1995).d Source: ADB 1979.e Source: ADB 1995.f Source: Nippon Koei and MASL 1995. June 1995 prices (US$1.00 = Rs 50.00).

Project Performance

22

contractors and unanticipated external factorssuch as the civil war disruptions.

A review of available literature on other damand irrigation projects suggests a markedtendency towards cost overruns. A meta-analysisof nine different projects conducted by the WorldCommission on Dams (WCD 2000) revealed anaverage cost overrun of 54 percent. To verify thevalidity of these findings, WCD conducted asurvey of 81 dam projects and found the averagecost overrun was 56 percent. However, costperformance varied substantially. Of the totalsample, 25 percent cost less than anticipated and75 percent experienced cost overruns.Multipurpose dams, like the Uda Walawereservoir, showed substantial variability rangingfrom a cost underrun of 22 percent to a costoverrun of 180 percent. Performance was worst inSouth Asia with average cost overruns of 138percent. A similar exercise by ADB revealed anaverage cost overrun of 16 percent for 23projects completed between 1968 and 1999(Lagman and Aylward 2000). Anothercomprehensive study conducted by the WorldBank’s Operations Evaluation Department foundthat large dams financed by the Bankexperienced an average cost overrun of 39percent (World Bank 1996).

Another measure of project performance isthe economic internal rate of return (EIRR). TheEIRR or cost-benefit analysis (CBA) or both aretypically used to provide economic justificationfor an investment. From the perspective of mostmultilateral development banks or agencies,

economic viability requires an EIRR of at least 10percent.

Table 2 shows the EIRR for various phasesof investment in UWIRP at appraisal andcompletion. The EIRR at appraisal, which rangedfrom 10 to 17.8 percent, plummeted to about 7percent upon completion. This suggests that,based on the EIRR threshold criterion of 10percent used by ADB, none of these investmentswere economically viable. The importance of theEIRR in decision making is discussed in thesection on the Decision-Making Process (page 38).

Implementation Schedules

The planned versus actual amount of time toimplement a project also provides an importantperformance measure. Delays in implementationlead to escalations in costs and reductions inbenefits. UWIRP experienced substantial delaysin all aspects of the project except theconstruction of headworks at the initiation of theproject (see table 3).

The implementation delays in UWIRPgenerally concur with experience in many otherirrigation and rural-development projects,suggesting a systematic bias in underestimatingthe project implementation period. A 1995 reviewof 57 ADB-funded projects in the rural-development and irrigation sector found that “allexcept two programs and one project,experienced delays averaging three years …projects in the irrigation subsector had the longest

TABLE 2.Economic internal rate of return (EIRR) and cost-benefit analysis (CBA) at project appraisal and completion.a

Project At appraisal At completion Overestimate

ADB-RB downstream EIRR = 12%b EIRR = 7%(b) 5%

WIIP: 1984-1994 EIRR = 10%c EIRR = <7.5% >2.5%

LB upgrade and extension EIRR = 17.8 (17.4)d Project ongoing Project ongoing

a An EIRR was not estimated at the inception of UWIRP.b ADB 1984, Appendix I, page 46.c This is the most pessimistic EIRR as it includes “sunk” costs of investment since 1964. However, it only includes 60 percent of the cost

of the dam.d The revised value appearing in Nippon Koei (1997) is 17.4. The EIRR for each area of the LB is 12.7 for the “Old” area and 19.9 for

the “Extension” area.

23

TABLE 3.Planned and actual implementation period by phase of investment.

Project At appraisal Actual Overrun

Period (years)

Entire project 11 To be completed 38 (245%)

Headworks 3 3 0

ADB-RB downstream 3 7 4 (133%)

WIIP: 1984-1994 5.5 10 4.5 (82%)

LB upgrade and extension Phase I: 2 6 4 (200%)

Phase II: 4 Project ongoing Project ongoing

30The most commonly reported causes of implementation delays include shortage of funds, institutional deficiencies including poorcoordination among executing and implementing agencies, optimistic assumptions about implementation schedule at appraisal, repairand remedial works, changes in project scope and design, lack of farmer participation during implementation, and other externalfactors such as civil disturbances, political instability, inadequate capacity of local contractors, and shortage of local supplies andmaterials (WCD 2000).31The improvement in water duty shown for the 1982-84 period, when MASL took over the management of the scheme is quite highand is said to result from stricter definition of calendars and scheduling. However, there is some doubt on the validity of the data(Nijman 1991). In particular, cultivated areas seem to have been overrated.

average time overrun of 3.7 years” (ADB 1995). Asimilar evaluation of 99 projects by the WorldCommission on Dams (WCD) found that only halfwere completed in the estimated period of time.Of the remaining 50 percent, about 30 percenthad delays of 1-2 years, 15 percent had delays of3-6 years and the remaining 5 percent had delaysof over 10 years (WCD 2000).30

Irrigated Area: Planned vs. ActualOutcomes

Table 4 shows the planned versus actualcommand area for UWIRP for successive phasesof investment. Only half of the originally plannedcommand area was developed by 2000.

The actual irrigated area provides a bettermeasure of performance than the projectcommand area, which is the potential irrigablearea under the project.

Table 5 shows the planned and actualchanges in irrigated area for various phases ofinvestment. The actual irrigated area hasconsistently fallen substantially short of project

plans. For the ADB-funded RB downstreamdevelopment project, actual irrigated area fell 47percent short of the planned area.31 For thesubsequent Walawe Irrigation and ImprovementProject (WIIP), actual irrigated area fell 17percent short of planned irrigated area. Althoughirrigated area has fallen short of expectations, ithas increased steadily since 1971.

Figure 4 shows this change by season,including three seasons with severe water shortagewhen cropping areas were drastically reduced (yala1977, yala 1992, and maha 2000/2001).

A WCD survey of 76 dam irrigation projectsfound that poor performance, as measured byirrigated area, was particularly pronounced in theearly years of a project. Over time, irrigated areagenerally increases, usually beginning around the5th year of the project continuing to the 30th year(WCD 2000). These findings are consistent withthose experienced in UWIRP and withobservations by Berkoff (2002) that irrigated areaand cropping intensity targets are oftenoverstated by significant amounts but that overtime the gap between planned and actual irrigatedarea tends to narrow.

24

TABLE 4.Planned and actual command areas of the phases of UWIRP.

Project At appraisal At completion Deviation

Command area (ha)

Entire project 32,793 17,615a –15,178 (–48%)

ADB-RB downstream 12,369 10,978 b –1,391 (–11%)

WIIP: 1984-1994 ~12,000 11,901 Small

LB upgrade and extension Phase 1: 2,900 Phase 1: Ongoing

Phase 2: 6,380 Phase 2: Ongoing

a Command area as of 2000.b RB Tracts 20-21 (1,391 ha) were dropped from the project after a 1972 review due to lack of sufficient water resources.

TABLE 5.Actual and planned irrigable areas at project appraisal and completion.

Project At appraisal At completionc Change at completion against:

Actuala Plannedb Actuald Plannede

Area (ha) %

Entire project M: 2,750 M: 32,793

Y: 2,750 Y: 32,793 N/A N/A N/A

T: 5,500f T: 65,586 h

ADB-RB downstream M: 3,017 M: 12,369 M: 7,024 M: +132 M: –43

Y: 2,309 Y: 12,369 Y: 6,036 Y: +161 Y: –51

T: 5,326 g T: 24,738 T: 13,060 T: +145 T: –47

WIIP: 1984-1994 M: 8,632 M: 12,000 M: 10,447 M: +121 M: –13

Y: 7,607 Y: 12,000 Y: 9,494 Y: +125 Y: –21

T: 16,239 T: 24,000 T: 19,941 T: +123 T: –17

a Actual irrigated area at appraisalb Planned irrigated command area at appraisal.c Actual irrigated area upon completion of project.d Percentage change in actual irrigated area before and after project.e Percentage of actual to planned irrigated area upon completion of project.f From ADB (1969).g 1970/71 average for RB command area only (ADB 1979, Appendix 10).h From Hunting (1968a).

Note: M = maha; Y = yala; T = total; N/A = not available.

25

32Cropping intensity (CI) is the ratio of the total crop area cultivated during one year to the command area (e.g., CI = 2 indicates fulldouble cropping).

FIGURE 4.Irrigated cropping area of UWIRP, 1971-2000.

TABLE 6.Cropping intensity, actual and planned.

Project At appraisal At completion Change (%)

Actual Planned Actual Planned

Entire project N/A 2 Project ongoing Project ongoing Project ongoing

ADB-RB downstream 1.69a 2 1.21b –28 –61

WIIP: 1984-1994 1.34c 1.85 1.68d +25 –11

a Actual estimated, based on command area in Tracts 1-7 and Chandrikawewa of 3,679 ha (9,091 acres) and actual irrigated area for1971, which was chosen because irrigated area in 1970 was unusually low with an irrigation intensity of 1.2.

b Average intensity for rehabilitated area (Tracts 1-7 and Chandrikawewa) and new area (Tracts 9-19). However, there was wide variationin irrigation intensity between upper and lower areas. The upper areas had irrigation intensities ranging from 1.46 to 1.78 and the lowerareas had intensities ranging from 0.43 to 0.90 (ADB 1979; PRC 1982, Appendix 8, p. 2).

c RB only. There was wide variation in irrigation intensities between the upper and lower command areas ranging from 200 percent inupper areas to 40 percent in lower areas.

d ADB 1995; PRC 1982.

Note: N/A = not available.

Cropping Intensity

Table 6 shows cropping intensity32 at appraisal—both planned and actual—and at completion fortwo phases of investment. In both phases, theactual cropping intensity fell short of the planned

cropping intensity. (Average irrigation intensitiesdeclined under the ADB-funded RB downstreamproject because per hectare water deliveries didnot keep pace with expansions in the commandarea.)

26

33Insufficient information is available on OFCs prior to 1984, but the area was very limited (e.g., less than 405 ha or 1,000 acres). OFCsinclude sugar cane, the current area of which has been taken at 3,000 ha based on remote sensing evidence. This is almost twice theofficial area but the latter number is admittedly understated in order to reduce the per area water charges to MASL.

Costs and Benefits of UWIRP

At the outset, irrigation projects promisesubstantial economic benefits, including increasedvalue of agricultural production and boosts to thelocal economy through demand for agriculturalinputs, marketing and transport of outputs, andincreased employment opportunities. Estimates ofthese benefits, particularly the value ofagricultural output, have been used in conjunctionwith capital costs to justify project investmentsthrough cost-benefit analysis (CBA). Generally,cost-benefit analyses, as practiced to justifyinvestments in UWIRP over the last 50 years,have been limited to an assessment of directbenefits generated by increases in cropping area,cropping intensity and yield and direct investmentcosts. This section examines some of thebenefits and costs of UWIRP. Beyond measuresof direct benefits, preliminary estimates of someindirect benefits are provided, including domesticuse of water, indirect irrigation of home gardens,fisheries, hydropower generation, the NationalPark, and economic multiplier effects. Estimatesof cumulative capital costs are also provided.Because estimates of indirect costs—negativeexternalities—are not captured in this analysis, afull social cost-benefit analysis is not provided.

Crop Production

The primary intended economic benefit of anirrigation scheme is, of course, the increasedvalue of agricultural production due to increasesin cropping area, cropping intensity and yield.Figure 5 shows the gross value of agriculturalcrop production over the period 1970-2002,expressed in 2002 US$ prices using theagricultural GDP as a deflator. It shows anupward trend in the gross value of rice until 1984.In the 1990s, the value of rice decreases both in

absolute and relative terms, due to the expansionof banana cultivation. The expansion of banana—a high-valued crop—substantially increased thetotal value of agricultural output. Other OFCs alsocontribute significantly to that change.33 In 2002,the gross value of banana was US$18.8 million(Rs 1,790 million) against US$11.2 million (Rs1,066 million) for rice and US$4.9 million (Rs 474million) for other crops. Figure 5 also shows thegross value of output per hectare. The 1990sshow a gain in the gross value added per hectareof approximately 50 percent from the 1970s, dueto both rehabilitation and expansion of bananacultivation.

The net benefit attributable to the project isthe gross value of agricultural output lessproduction costs. The question of whether familylabor and hired labor used must be assigned anappropriate opportunity cost is debatable. In thecase of an open economy with alternativeeconomic activities this is probably the case. Butwhen there exists high levels of unemployment(approximately 30% in Hambantota district,despite the project and other similar ones in theKirindi Oya basin) it cannot be assumed that inthe absence of the project the whole labor forcewould have readily shifted to alternativeactivities. In the case of a settlement such asUWIRP, which relieves local agrarian pressureby providing occupational opportunities that arelacking in the locality, the opportunity cost canbe hypothesized to be quite low. A review ofdifferent assessments of production costs in riceproduction since the beginning of the project didnot yield a clear picture of the evolution of therelative weight of the different categories ofcosts. On average, production costs amounted to45 percent of the gross added value, 40 percentof these costs being ascribed to labor, includingboth family (30-50%) and hired labor (50-70%). To

27

Notes: US$1.00 = Rs 95.40 in 2002.

Insufficient information is available on OFCs prior to 1984, but the area was very limited.

FIGURE 5.Gross value added of crop production of UWIRP in 2002 US$ values.

FIGURE 6.Net value added of crop production of UWIRP in 2002 US$ values.

Notes: US$1.00 = Rs 95.40 in 2002.

Insufficient information is available on OFCs prior to 1984, but the area was very limited.

28

strike a balance between these differentapproaches, we have tentatively chosen to valuelabor costs at half of the opportunity cost takenas the daily wage.

Cost-benefit analyses should not comparebefore- and after-project situations but with- andwithout-project situations. This means that evenif the project is not implemented, changes aregoing to take place. In the case of UWIRP, onemay assume that the sugar cane area wouldhave developed under rainfed conditions (just asit did outside the boundary of the project) andthat part of the area under UWIRP would havebeen cultivated under chena (or permanentrainfed crops). We must therefore deduct fromthe net benefit of the project the net benefit fromsuch rainfed agriculture, assuming tentatively thatit would have expanded gradually from pre-projectconditions up to 60 percent of the total UWIRPnet area in 2002. Figure 6 shows net economicvalue added for agricultural production from 1970to 2002, which equals the gross value ofagricultural output less the costs of production(including labor) and benefits that would haveoccurred in absence of the project (e.g., the“without-project” benefits).34

Domestic Water Use

The homesteads of settlers are dispersedthroughout the project area. An undulated reliefwith many intermittent small streams and quick-dropping shallow aquifers characterizes theWalawe plain. This means that in the absence ofsmall tanks no permanent source of water isavailable to humans. Infiltration of water from thecanal network and rice fields has replenishedshallow aquifers supporting scattereddevelopment in the project area (Meijer 2000).

Along main canals and along riverbeds, peopleobtain water from shallow wells that are usedprimarily for drinking purposes. Boelee and vander Hoek (2002) have found that in one locationalong the left bank project area, groundwaterlevels closely followed changes in canal waterreleases and that canal seepage accounted for 74percent of groundwater recharge.

Since the great majority of the settlerswithin UWIRP use wells for drinking water,understanding its value is important for futuredecision making. The value of drinking water isestimated using a replacement cost approach.An Irrigation Department study found that waterusers obtained about 20 liters per day percapita from shallow wells and other“nonprotected” sources. The minimum basicneed for water from a human rights perspectivehas been estimated at 50 liters per day percapita (Gleick 1996). The benefits derived fromthe wells can be valued by estimating the cost ofproviding an amount of water equivalent to thevolume abstracted. Recent communal watersupply projects in the region, mostly based onwells, showed a cost of Rs14/m3, assuming aproject lifetime of 20 years. To estimate the valueof drinking water in the project area an averagecost of Rs 15/m3 is used. This estimate seemsreasonable especially when one considers thatthe Lunugamwehera Water Supply Scheme, in theneighboring Kirindi Oya basin, provides drinkingwater at a cost of Rs 21/m3, includingamortization and operation and maintenancecosts.35

Based on an average replacement cost of Rs15/m3, an average current usage of 20 l/capita/dayand a population estimate of 120,000,36 theeconomic value of drinking water is conservativelyestimated at US$0.14 million (Rs 13 million in 2002values).37

34The value from rainfed cultivation has been deducted from all crop categories proportionate to the planted area.35For a project lifetime of 20 years, a daily production of 5.2 m3 provided to approximately 45,836 beneficiaries with running costs ofRs 0.74/m3.36Only one third of the population in the basin has access to pipe water. 120,000 corresponds to two thirds of a population ofapproximately 180,000 people in the lower Walawe basin.37This estimate can be considered very conservative as it uses a relatively modest per cubic-meter rate and 20 liters/capita rather than50 liters/capita as the basic necessity usage.

29

In addition, project canals are widely used forother domestic uses such as for laundry andpersonal hygiene that have economic value. Theyalso serve important social functions. A tour ofthe canals in the late afternoon demonstratestheir widespread use, particularly for socialinteraction and recreation for the children.Although these uses clearly have economicvalue, these values have not been measured.

Home Gardens

The homesteads of settlers are spread over allthe highlands bordering the irrigated areas.Recent remote sensing imagery shows that mostland in the project area is now developed andthat while settlements are, in general, not verydense they include significant areas of homegarden. The settlers highly value these homegardens for their produce, shade and aestheticvalue. Most home gardens have varied fruittrees (coconut, mango, jackfruit, breadfruit,areca nut, etc.), vegetables, bamboo groves,and shrubs and trees that provide buildingmaterials, firewood and material for handcraftsand medicinal use (Renault et al. 2000). Thesehome gardens also provide shade and additionalcommunal living space.

The development of these gardens aroundhomesteads would not be possible without projectwater. This is proven by observations made inareas where canals have been lined. Liningreduces the recharge of the aquifer, hencereducing water availability in the root zone oftrees, resulting in smaller fruits that often fallbefore reaching maturity (Meijer 2000). The linkbetween irrigation canals and trees is also evidentfrom the expansion of home gardens. Figure 7shows the extent of home gardens in the UWIRParea in 1999. These gardens, which are now aprominent feature of the landscape, constituting

28 percent of the land use (18,000 ha), werevirtually nonexistent prior to the commencementof the project, as indicated in the 1956 land-usemap.

This “unanticipated” development of homegardens and their fruit trees that feed on thewater table replenished by infiltration from canalsand rice fields, is generally overlooked. Researchcarried out in the neighboring Kirindi Oya basin(Renault et al. 2000) found that as much as 55percent of the total evapotranspiration in theKirindi Oya irrigation scheme came from noncropvegetation (against 28% from crops).

To assess the production of these homesteadgardens, a survey of their composition andproduction has been carried out on a sample of19 different plots averaging 0.24 ha (0.6 acre).The survey measured the composition andquantity of different home-garden productsincluding medicinal herbs, fuel wood, bamboo,fruit trees and timber. Home-garden products areboth consumed at home and sold in localmarkets. The value of total production, whetherconsumed or sold, was estimated at prevailinglocal-market prices.

The results of this analysis suggest that theproductivity of 0.405 ha (one acre) of homegarden is extremely high and the average grossproduct of fruits was recorded at US$248 (Rs23,611) per year, including US$185 (Rs 17,600)for marketed produce and US63 (Rs 6,011) forthe produce consumed at home.

In addition to fruit, timber is another valuablehome-garden product. The value of timber foundin the garden surveyed was estimated atUS$4,965 (Rs 473,700) on average, based on thesize and number of trees and market prices.Trees are cut at an average age of 25 years.Allocating the value of timber produce over 25years yields an additional gross value ofUS$198.6 (Rs 18,948) per year for home-gardenproduce.

30

FIGURE 7.Home gardens (in green) in the UWIRP area: 1999.

To estimate the value of home-gardenproduce for the entire UWIRP project area, weused 70 percent of the homesteaded area asappearing on satellite images to account forroads, houses, and other non planted areas,giving us a net area of 13,000 ha. The total fruitproduction is therefore worth about US$8 million(Rs 762 million) per year, to which can be addeda timber value of US$6.40 million (Rs 611million; considering a 25-year lifetime), giving atotal value of US$14.4 million (Rs 1,373 million)per year.38

It can be argued that in the absence of theproject some rainfed home gardens would havebeen developed and that even though their

composition would have been different, their valueshould be deducted from the full benefitsassessed above. Given the limited occurrence ofhome gardens in fully rainfed areas in the dryzone and the lack of data to assess their value,these have not been considered in this analysis.

Reservoir Fisheries39

Fisheries located in project reservoirs are often aneglected benefit of rural-development projects(Renwick 2001). Fisheries provide alternativeincome-generating opportunities and neededsources of protein for the rural poor without

38Production costs (seeds, possible treatments, etc.) are neglected here. This may compensate for the non-consideration of homegardens located along the main canal out of the command area.39This section is drawn from Perera 2003.

31

consuming any water. They also may provideindirect malaria control, as fish consumemosquito larvae, and support piscivorous birdspecies. This section examines the economiccontribution of fisheries in project reservoirs andcanals to UWIRP.

In 2003, IWMI carried out a survey ofreservoir fishery activities in UWIRP. This surveyrevealed the important role of fisheries in theproject area. Over 1,600 families are involved ininland fishing in UWIRP, of which about 660families are involved on a full-time basis and therest use fishing as a means to supplement familynutritional needs or incomes. In addition, about140 “middlemen” such as “bicycle fish sellers”and other retailers depend directly on fishing fortheir livelihood.

Survey work and related studies by Perera(2003) showed that in UWIRP an abundant fishcatch is obtained from a diversity of watersources, amounting to a yearly aggregate of1,600 metric tons (table 7). The largest catch isfrom the Uda Walawe reservoir, which accountsfor about 35 percent of total production. Anestimated 86 percent of the total catch is sold toretail markets and the remainder (14%) is usedfor home consumption. The estimated grossannual value of fish equals annual market salesof approximately US$691,000 (Rs 67 million)plus the value of fish consumed at home ofUS$103,000 (Rs 10 million; see table 7),amounting to roughly US$794,000 (Rs 77million) per year. Including only the catch that iscontained within the UWIRP project areas (e.g.,Walawe, Chandrika and Kiri Ibban reservoirs andproduction in small tanks and irrigation canalswithin the UWIRP project area40) the gross valueof the reservoir fishery catch is approximatelyUS$567,000 (Rs 55 million) a year.

The economic contribution of fisheries equalsthe gross value of the total catch less costsincurred in catching fish. In a detailed survey of

fishing costs in the nearby Kirindi Oya project,Renwick (2001) found that monetary costs offishing are relatively uniform among fishers androughly equals 20 percent of the value of thecatch. Deducting 20 percent of the value offisheries within UWIRP to cover production costsfrom above estimate, the net economic value ofinland fisheries is US$454,000 (Rs 44 million) peryear.41

This value is augmented if we consider theeconomic activities of well-organized middlemenin fish markets, notably in the Embilipitiya town.Like the fish vendors on bicycle who buy fishfrom fishers and sell it in towns, thesemiddlemen also make handsome profits. But, incontrast to the revenue accruing to fishermen,this revenue is appropriated by only a fewoperators who tend to agree and fix pricesamong themselves. This type of market failureseems to be quite common in the area and italso concerns other agricultural products. Sincethese middlemen usually buy fish at US$1.03(Rs 100) per kilogram and sell it at US$2.06 (Rs200) per kilogram, we may tentatively take halfof the profit as value added, allowing for bothcosts of marketing and the likely non-competitiveness of the market. Then the finalrough estimate for the yearly net added value isUS$743,000 (Rs 72 million), with benefits spreadacross households (home consumption),fishermen and middlemen.

Despite the importance of present production,there seems to be significant scope forimprovement of fisheries. Management andcontrol of fisheries are totally lacking in most ofthe tanks, with the exception of the Uda Walawereservoir and one small tank. Most fisheriessocieties are declining mainly due to lack ofnecessary guidance and assistance from theimplementing agency—the National AquaticDevelopment Authority. The lack of institutionalinfrastructure for fisheries stems from a history

40Of the small tanks 47% are within Uda Walawe. The most productive tanks are in fact those located within UWIRP and which benefitfrom the return flow of irrigation areas (or sometimes direct releases from main canals).41If noncash costs (including depreciation of owned boats, household labor used for maintenance and repair of boats and nets, etc.) areincluded, total costs amount to 33%, leaving a net benefit of Rs 36 million.

32

TAB

LE 7

.R

eser

voir

fishe

ries

in U

WIR

P: P

rodu

ctio

n an

d gr

oss

valu

e.

Wat

er s

ourc

esTo

tal

prod

uctio

nC

PU

EH

ome

cons

umpt

ion

Tota

l co

nsum

ptio

nQ

uant

ity s

old

Sel

ling

pric

eVa

lue

of s

ales

Mon

thly

inc

ome

(t)

(kg

)(k

g/da

y)(k

g)

(t)

(Rs)

(Rs)

(Rs)

Sm

all

tank

s31

1.1

81.

3f5

.525

5.6

52.0

013

,530

,421

6,0

65

Med

ium

tan

ks15

7.5

15

1.5

30.7

126.

859

.00

7,54

1,46

08,

174

Mau

ara

120a

13.3

1.5

13.5

106.

550

.00

5,32

5,00

0

Kiri

Ibb

an t

ank

12.9

7.5

1.0

3.4

9.5

50.0

047

5,20

03,

645

Rid

iyag

ama

tank

241.

92

31.

0118

.322

3.6

45.0

09,

980,

507

8,88

1

Rak

wan

a ri

ver

1.9

1.9

Wal

awe

river

21.3

5.2

16.1

50.0

083

5,20

0

Ber

agam

a12

.52.

51

055

.00

551,

650

Su

bto

tal

879

67A

vera

ge,

1.3

131

748

Ave

rag

e, 5

238

,239

,438

Cha

ndri

kaw

ewa

44.9

9.83

1.5

7.2

37.7

50.0

02,

109,

184

5,49

3

Uda

Wal

awe

tank

588.

229

.47

1.21

37.4

550.

845

.00

24,7

85,2

3512

,416

RB

MC

–151

.21

13

38.2

50.0

01,

911,

600

4,42

5

RB

MC

–20.

70.

7

LBM

C4.

90.

751.

13.

850

.00

190,

800

3,9

75

Rid

iyag

ama

MC

36.8

8.4

28.4

40.0

011

3,60

0

Gra

nd

to

tal

1,60

6A

vera

ge,

1.2

199

1,40

7A

vera

ge,

50

67,3

50,0

00b

a Est

imat

e. b

Rou

nded

val

ue.

Not

es:

t =

met

ric t

on (

= 1

,000

kg

); C

PU

E =

Cat

ch p

er u

nit

of e

ffort

(p

er b

oat

per

trip

); I

n 20

03,

US

$1.0

0 =

Rs

96.9

1.

Sou

rce:

Per

era

2003

.

33

of intermittent government support. During the1980s, the inland fishery industry achievedconsiderable growth with the support of thegovernment. The government provided extensionservices, fishing crafts and gear at subsidizedprices, free stocks of fingerlings, training andlegal support for fishers, made regulations forbetter control of fisheries, established fisheriessocieties and implemented communitydevelopment programs (De Silva 1988). As aresult of government support, the highest inlandfish production (39,800 tons) was recorded in1989 (Amarasinghe and De Silva 1999).

However, in 1990 the government withdrewits subsidies and support for inland fisheries, anddiscontinued all regulatory and stockingprograms, which seriously affected inlandfisheries and led to their sharp decline over thenext several years (Amarasinghe and De Silva1999). In the late 1990s, the governmentreinstated its financial and institutional supportfor inland fisheries and since then the fisherieshave gradually recovered (Amarasinghe 2003).When the government resumed its support withthe Fisheries and Aquatic Resources Act No.1996, it conferred control and management offish resources to Fisheries ManagementCommittees instead of local fisheries societies.Only the Uda Walawe Reservoir FisheriesSociety is effectively functioning at presentbecause of good catches in the reservoir andthe active supporting role played by the NationalAquatic Development Authority and theDepartment of Wildlife Conservation.

National Park

The Uda Walawe reservoir is a central feature ofthe Uda Walawe National Park (30,821 ha). Herdsof elephant and other wildlife congregate alongthe shores of the reservoir. While the value of thepark in terms of wildlife and biodiversitypreservation is beyond the scope of this

research, its value as a tourist spot is importantto the local and regional economy. A relativelysimple gauge of its importance is the number oftourists visiting the park each year. Although thearea was declared a park in 1971, it was only atthe end of the decade that villagers were movedout and resettled and poaching and logging werecurbed (Alwis n.d.). The problems, however,remain endemic and even include the cultivationof ganja (cannabis) within the park and in itsvicinities.

Available data on admissions to the parkand fees collected42 were obtained from parkofficials and DWLC-FAO 1997 (see Appendix F).In the 1994-2001 period, 452,000 persons, 20percent of whom were foreigners, visited thepark. The average revenue for the 1998-2001period was US$280,000 (2002 values). The parkemploys 60 persons full-time and there are 20guides who work in association with a 4x4vehicle company placed under the control of theProvincial Council. The revenue should includethe benefits generated by these two additionalactivities, while the O&M costs of the parkshould be subtracted. In the absence of suchdata, we may consider that these additionalcosts and benefits cancel each other and retainUS$280,000 as an order of magnitude of theyearly benefit generated by the park.

Energy Generation (at the Uda WalaweDam)

Both water towers delivering water to the rightand left bank main canals are equipped withturbines and electricity is generated as abyproduct of irrigation water release. Theseturbines produce an average of 15 GWh per year.To estimate the value of energy generated weuse the least-cost alternative to hydroelectricity inSri Lanka, which is thermal energy. Based onSomatilake (2002), the average cost ofpurchasing thermal energy from a private source

42In Sri Lanka, most tourist locations charge two-tier entrance fees—one for locals and the other for international tourists. Current fees are atUS$0.23 (Rs 23) for nationals and US$15.00 (Rs 1,500) for foreigners.

34

is Rs 7/kWh. Using this estimate the value ofenergy generated at UWIRP is US$1.1 million peryear (Rs 105 million per year; 2002 prices).

Backward/Forward Linkages and OtherMultiplier Effects

Apart from farmers themselves, many otheractors thrive on the development of irrigatedagriculture. Backward linkages include economicactivity generated as a result of agriculturalproduction through expenditures on such itemsas seeds, fertilizers and machinery. Forwardlinkages refer to post-harvest economic activitiessuch as storage, transportation, marketing,exports, etc. Economists measure economicactivity through backward and forward linkagesby “multipliers.” For example, a multiplier of 2.50implies that for each US dollar generated inagricultural output, another US$ 1.50 isgenerated through backward and forwardlinkages.

Although the concept of economic multipliersis well understood and accepted, calculatingmultipliers is an extremely involved process.Bhattarai et al. (2003) conducted a review of theliterature and found multipliers for irrigatedagriculture that ranged from 1.87 in Malaysia(Bell et al. 1982), through 1.87-2.18 in North-Arcot, India (Hazell et al. 1991) to 2.2-3.2 inSouth Africa (Hassan 2003).43 In their study ofirrigation in India, Bhattarai et al. (2003) estimatetwo different types of irrigation multipliers, percrop season and per hectare/year (includinglivestock). The first irrigation multiplier value is4.5, which implies that an increase of US$100 perhectare per crop season in the gross value ofcrop outputs in an irrigated area would generateanother US$350 of indirect benefits (or secondarybenefits) in the local economy. The per hectareirrigation multiplier was estimated at 3.15,

meaning that only 32 percent of the total benefitsof irrigation is actually obtained by Indian farmers,the rest of the benefits spilling over to theregional economy.

No attempt to estimate backward/forwardand consumption linkages has been made so farfor UWIRP. Although the above studies are notfully comparable in terms of methodology, theysuggest that a significant part of the benefits ofan irrigation project accrue to other sectors andactivities of the region, prompting questions onthe limitation of historical approaches used toestimate direct benefits.

Impact on Poverty

The impressive growth of the population in thearea from 2,000 in the 1950s to 200,000presently is in itself a telling illustration of theemployment and income-generating opportunitiesproduced by UWIRP. The impact of irrigationdevelopment on poverty has been recentlystudied in the Left Bank area of the project wherea detailed household survey instrument was usedto compare poverty measures for households withand without access to irrigation infrastructure(Hussein et al. 2002; Hussein et al. 2003;Hussein et al. 2004). Relying on survey data from858 households, study results revealed largediscrepancies in terms of chronic and transientpoverty between households in typical rainfed andirrigated areas. Table 8 exemplifies some of thedifferences found between irrigated, rainfed-typical, and rainfed-nontypical (sugarcane) areas.Household income in irrigated areas is 80 percenthigher than that in rainfed areas. However thisincome discrepancy is reduced in rainfedsugarcane areas, with income from irrigatedcropland 42 percent higher than that in rainfedsugarcane land. Irrigated agriculture is also muchmore labor absorbing than rainfed agriculture.

43Studies in developed countries have found higher values: 5 in New South Wales, Australia (Powell et al. 1985) and 6 in Canada (Hill andTollefson 1996).

35

Other Positive and NegativeExternalities

In addition to the benefits described above, otherpositive externalities can be added, such as floodcontrol (reduction of damage in Hambantota)44 andnonagricultural water use (factories). However, itis also necessary to consider negativeexternalities of the project. Since Uda Walawe isa settlement project based on the reclamation oflands hitherto little used (some lowland rice fieldsand some chena cultivation) the impact onexisting populations has been limited. On thecontrary, residents in the area were in generalselected as settlers and were given land. Twocategories of people have nevertheless seen theirlivelihoods threatened by the successivedevelopments of the project, i.e., those living onfisheries in the coastal lagoons and the cattlefarmers who need extensive areas for pasture(Birner 1996).

Approximately 500 families relied on shrimpproduction in the coastal lagoons before theybecame flooded with drainage water from thelower RB. This ecological change modifiedaquatic resources as well as flora, with aspectacular colonization of the lagoon by themangrove tree Sonneratia caseolaris (Jayatissaet al. 2002). Villagers in the area reported that

only 25 families are now deriving their mainrevenue from the lagoon. Most of the remainingfamilies are said to have shifted to marinefisheries (Senaratna and Clemett 2003).

Other possible externalities refer toenvironmental changes and their impact onwildlife and biodiversity, as well as healthimpacts of malaria or consumption of unsuitablegroundwater (fluoride). Perhaps one of the great,potential, hidden costs is human health risk fromunsafe drinking water and bioaccumulation ofagrochemical byproducts through regularconsumption of reservoir fish in the project area.

These different externalities have not beenassessed in detail for lack of means but theymust be cited here, even in qualitative terms, toillustrate the variety of impacts that should befactored in to achieve a more comprehensiveview and impact assessment.

Cumulated Costs and Benefits

The cumulative costs of all capital investments inUWIRP from its inception in 1964 through 2001are shown in figure 8. As of 2001, the totalcapital investment costs of the project reachedUS$251.6 million (Rs 24 billion), while O&M costsexceeded US$52.4 million (Rs 5 billion),

TABLE 8.Socioeconomic indicators in three settings, Walawe Left Bank area.

Indicator Irrigated area Rainfed-typical area Rainfed-nontypical area (sugarcane)

Average family size 5.05 4.9 4.75

Average land size (ha) 1 1.47 1.59

Annual net value of product (Rs/ha) 63,192 15,060 52,620

Labor use (person-day/ha/year) 67.2 28.0 58.3

Annual household income (Rs) 108,542 60,295 76,134

Crop income (% of total) 52 33 51

Ratio of food to nonfood expenditures 0.56 0.62 0.62

Source: Hussein et al. 2004.

Note: US$1.00 = Rs 97.50 (January 2004).

44For example, extreme flood events in 1912 and, particularly, in 1940 provoked extensive damage in the lower parts of the basin.

36

FIGURE 8.Cumulative capital and O&M costs for Uda Walawe: 1974-2001 (in 2002 US$ values).

Note: US$1.00 = Rs 95.40 in 2002.

45Detailed cost data were gathered from the Mahaweli Authority and supplemented with data from the Asian Development Bank (1969and 1994). Costs are expressed in 2002 prices using the agricultural GDP deflator series. Using the consumer price index yields verysimilar results.46Considering a lifetime of 25 years for the foregone investment.

expressed in 2002 prices.45 These costs shouldbe compared to cumulative direct benefits fromagricultural production. Depending on whether ornot labor costs are included, these benefits rangefrom US$392.0 million (Rs 37.4 billion) when alllabor costs are included, through US$448.6million (Rs 42.8 billion) when half of labor costsare included, to US$505.2 million (Rs 48.2 billion)without labor costs (see discussion above forfurther details). This gives a benefit-cost (BC)ratio of 1.63, 1.43 and 1.24, respectively. It mustbe noted, however, that the costs considered heredo not include replacement of infrastructure.

Figure 9 shows both cumulated costs andbenefits. It can be seen that the year in whichbenefits begin to exceed costs is 1987 whenlabor is excluded, 1990 with labor costsestimated at 50 percent of their opportunity costand 1993 with 100 percent labor costs.

Banana production plays a critical role inbolstering project benefits. Banana productioninvolved fairly low production costs relative to itsgross value. If we assume that all the area

converted to banana had remained under ricecultivation, then the BC ratio in 2000 would onlybe 1.15 (with 50% of labor considered).

Altogether, it may be more straightforward tocompare actual benefits for the current year,rather than for a long series, as many factors andconditions have changed. The current annual netbenefit of irrigated agriculture in the UWIRPproject area is approximately US$25 million (Rs2,380 million). Other water-related benefitsinclude: US$0.75 million (Rs 72 million) forfisheries, US$0.28 million (Rs 27 million) for UdaWalawe National Park, US$1.1 million (Rs 105million) for energy generation, US$14.4 (Rs 1,370million) for home-garden produce (45% of this intimber capital), and US$140,000 (Rs 13 million)for domestic water.46 Indirect benefits other thanthat from home gardens, therefore, amount toapproximately 9 percent of crop benefits. Figure10 summarizes all these primary and secondarybenefits and shows that home-garden productsother than timber generated more benefits thanrice production.

37

FIGURE 9.Cumulative crop benefits and costs for Uda Walawe: 1974-2002 (in 2002 US$ values).

FIGURE 10.Summary of crop and noncrop benefits derived from UWIRP (in 2002 US$ values).

Note: US$1.00 = Rs 95.40 in 2002.

38

The investments made in the twentieth century inthe Walawe river basin have been based ontechnical, economic, social and political grounds.This section reviews the most salientjustifications of decision making emerging fromthe descriptions of the preceding sections. It thenfocuses on how cost-benefit analysis andconcerns for cost recovery have been factored in,and on the role they play in decision making.Last, it attempts a wider reflection on theinterests and strategies of the different actorsinvolved in planning and decision making.

Chief Justifications and Ideologies

The goals and means of development during theBritish rule were clear. Colonial administratorssought to protect and uplift the poor people, andto involve them in productive capitalisticinvestments that would yield net revenues to theCrown. A good example of these twin goals canbe found in the following report (Steele 1867) onthe Hambantota province:

“Much might be done, and ought tobe done, in utilizing the streams whichrun through the district, pouring into theIndian Ocean their, not superfluous, butinappropriate waters which, rightlydirected, should replenish and invigoratethe thirsty land and people. By a well-devised system of anicuts, sluices andchannels, this would be feasible enough;and the day is, I trust, not far distantwhich we will see this simple, butmagnificent and really noble andphilanthropic, enterprise accomplished.Nor will it be a barren philanthropy, Imean, in point of pecuniary profit even.”

The postwar and postindependenceinvestments in the area were clearly part of thelarger policy of resettling populations in the dryzones of the country, both in the north-central andsouthern parts of the island. The policy wasmotivated by a growing population and inadequateaccess to land and water resources, generating ademand for land, which was already acute incolonial times. Sri Lanka was still an exclusivelyagricultural economy and was bent on achievingself-sufficiency in rice. Settlements weremodernist pieces of social engineering aimed atbuilding up a self-reliant and independent farmingpopulation provided with decent socioeconomicinfrastructures. Selection of settlers wassupposed to follow a mix of efficiency (e.g.,emphasis on modern agricultural production) andequity (e.g., providing land to landless poor,prohibiting land concentration and landfragmentation) objectives.47 The postcolonial statebuilt part of its legitimacy on taking over the“development mission.” The 1954 InternationalEngineering Company (IEC) report on Walawe(IEC 1954) provides a good description of thehost of aspirations that were supposed to befulfilled through settlements:

“It is demonstrated that the colonistson the project would be given anopportunity to enjoy a decent standard ofliving, have a nutritional diet, and havemoney available for luxuries, savings andrent. The project would provideappreciable social and economic benefitsto the people of Ceylon… including,electricity at reasonable costs, increasingfood production thus reducing imports,improvement of the external tradebalance, [and] providing additional jobs asa result of the labor required.”

Decision-Making Process

47See Farmer 1976 for more general considerations.

39

Likewise, the 1960 ECI report on the UdaWalawe scheme (ECI 1960) emphasizes that“there exists in the Walawe basin a large,unbroken area of lowland plain that appears tomerit agricultural development and yet, atpresent, is very sparsely inhabited. It is clear thatsuch a development would help alleviate Ceylon’seconomic problems.”48 Since the area is in thedry zone, “the irrigation system must be designedand constructed in such a way that the cultivatoris protected completely against famine resultingfrom drought.” The report also addresses cropdiversification, envisioning extensive areas ofcotton and sugarcane, since “the possibility ofdiversifying and balancing the agriculturaleconomy by the growing under irrigation ofcommercial crops and other food crops in additionto rice has not received enough attention inCeylon.”

The 1968 Hunting report epitomizes technicalhubris and the paternalistic or authoritarianplanning typical of the time. The three mainstated objectives of the project were “to attain ahighly productive cash agriculture generatingincome-earning opportunities and to promotemaximum alternative employment to farming,”49

and to reduce imports of agricultural products.The plan is presented as rigorous and technicallysound, the options taken as self-evident andindisputable, and cropping patterns as evident.The technical recommendations would becompulsory, squatters would be evicted andfarmers would benefit from “sympathetic advisoryservices.” Their social cohesion would be ensuredby adequate grouped settlements.

This flamboyant plan was not fullyendorsed by the ADB feasibility study (inparticular, irrigated areas to be developed weredisregarded in the plan and attention wasfocused on the RB only), but it retained muchof its modernist and paternalistic drive. TheUda Walawe project was promoted as an“Integrated Development Scheme,” a

fashionable concept at the time. It befits the“pseudo-comprehensive-programs” described byHirschman, with extravagant claims that givedecision makers the “illusion that the ‘experts’have already found all the answers to theproblems and that all that is needed is faithfulimplementation” (Hirschman 1967).

The settlement policy and need to create jobswere emphasized: “In view of the overcrowding inthe ‘wet zone’ and the prevailing high rate ofunemployment and underemployment, the projectwill alleviate considerably the acute man/arableland ratio and generate significant economic andsocial benefits to the country” (ADB 1969). Whileearly settlements were centered on subsistenceand poverty reduction, the new project was“production-oriented” and would erase traditionalfarming systems. We have seen earlier whatbecame of these grand plans.

Despite the disquieting facts thatnecessitated rehabilitation only 4 years after thecompletion of the project, the proposal forrehabilitation was packaged attractively; it was“low cost and quick yielding” (disregarding earliercosts). And it was “highly justifiable” because itwould remove “constraints which have inhibitedthe full realization of benefits under the originalprojects.” Since water management had failed todistribute water evenly, the project would“contribute to an increased and more equitableincome distribution among settlers in the projectarea.”

The objectives of the rehabilitation wereeasily shown to be in line with the government’ssectoral development objectives: increasingagricultural production, farm income andemployment opportunities in the project area. Inagreement with the then emphasis ofinternational agencies on turnover, cost recoveryand service provision, it was envisaged thatMASL would provide a “demand-driven advisoryservice to farmers where farmer organizationsrequest assistance in support of self-defined

48See Hirschman (1967): “A specific gift of nature [must] be fully developed and utilized. The argument that a ‘free’ natural resource isthere waiting to be harnessed exerts a continuing attraction on engineers, politicians, and the public.”49Employment was the most pressing issue in the implementation of the Mahaweli Project (Wickremaratne 1995).

40

objectives.” The vision of subsidized, indebtedand hand-to-mouth peasants transmuted intosmall entrepreneurs picking up and paying forservices, allowing them to achieve maximizationof production, is probably the most pervasive,resilient and appealing Holy Grail of developmentagencies. Unfortunately, if this vision helpedpresenting projects in orderly, efficient andattractive attires, it did not materialize in theWalawe basin.

After 25 years of persistent dominance ofrice cultivation, conflicts in the selection ofsettlers, and dismal efficiency of waterdistribution, everything pointed toward a lull inthe pace of investment. The development of theLB seemed unwarranted by the very highamount of water used in the RB. The ADBcontested the opportunity to embrace a newphase of development before management wasbrought in line with acceptable standards. Thecombined willingness of MASL and the Japanesedevelopment agencies to pursue development,however, prevailed. The new project did notdepart much, in its style, from earlier top-down,engineer-oriented projects with design options,settlement processes and economic feasibilitydependent upon assumptions of croppingpatterns, practices and productivity that hadbeen clearly proved ad hoc and unrealistic in thepast. The example of the intent to “change thepresent nomadic system of cattle managementto a privately managed permanent holdingsystem” illustrates this point (Nippon Koei andMASL 1995).

The feasibility study pointed to currentweaknesses of the Sri Lankan agriculturesector—“inactive production of staple food withdecrease in its self-sufficiency,” lowdiversification and decline in farm income—andargued that one “cannot take an optimistic viewof the agricultural sector in the country unless aneffective programme would be implemented toovercome the above constraints. The project hasvarious advantages over the agricultural

development in the dry zone and urgentimplementation is needed” (Nippon Koei andMASL 1995). The necessity of agriculturaldevelopment was also stressed, based on thesituation of unemployment in the country and onthe “government’s poverty alleviation drive.” TheSouthern Area Development Strategy was alsoseen to reflect the government’s special concernin the region.

In sum, the justifications given during fourdecades were based on modernist principles thatechoed very general objectives of thegovernment: raising national production and ruralincome, diversifying agriculture, alleviatingpoverty and increasing employment. All projectsdo this to a greater or lesser extent and, thus,have easily fallen in line with generalgovernment policies. As noted by Berkoff (2002),“irrigation is so obviously a good thing—who canbe against it?”

Cost-Benefit Analysis and EconomicRationale

Normative project planning assumes thatintervention decisions are taken based on aranking of their relative net benefits to society.Over the past 50 years, the societal benefits oflarge-scale public investments have beentypically measured using cost-benefit analysis(CBA). Despite the longstanding controversy onthe legitimacy of CBA’s claim for objectivity (seebelow), it is still widely used as a means toassess both the desirability and degree ofsuccess of projects. Decision-makers seekprojects where benefits exceed costs and theeconomic internal rate of return (EIRR) isadequate.50 The EIRR measures the return oninvestment and a value of 10 percent is generallyconsidered as a threshold for “feasibility” on an apriori basis and for “successfulness” on an expost basis. The necessity of projects to reachsuch a target EIRR for funding consideration by

50The EIRR is the discount rate such that the discounted stream of benefits equals the discounted stream of costs; it measures return oninvestment.

41

bilateral and multilateral donors heavily impactson the assumptions consultants make, not onlyabout prices, yields or cropping patterns, but alsoabout water requirements and canal design. Italso impacts on what type of work or action is tobe deemed necessary.

The 1960 report by the Photographic SurveyCorporation (PSC and SGC 1960) had a grandvision of the future of the basin. It recognizedthat “it is difficult to evaluate the ultimate benefitof this proposed irrigation and colonizationscheme to the national economy, but it wouldcertainly be very great.” The RVDB report of 1963estimated costs at US$14.41 million andcomputed expected optimistic revenues to comefrom annual land rents (US$10.38/ha) and waterfees (US$2.60/ha) during 50 years, as well asfrom the sale of electricity, to arrive at a cost-benefit ratio of 1.94. The Hunting report (1968)estimated the costs of the project at fulldevelopment at US$47.6 million and assessedthat “at maturity, the project, if successfullyimplemented, will provide 12,000 farmers withgenerous incomes” as well as “a substantialreturn to capital invested.” The report, however,fell short of calculating a cost-benefit ratio or anEIRR. It also captured only a portion of actualphysical infrastructural costs.

In the 1969 ADB project assessment report,a water charge of Rs 40/acre (US$16.6/ha) peryear was assumed (up from Rs 5/acre in theearlier feasibility study). The BC ratio of theproject, including past investments (dams asinitial irrigation works) and benefits fromagricultural production and water and land fees,was calculated at 1.1 (with a discount rate of10% per year) and the EIRR reached 11.1percent. The life of the project was taken as 40years. The ADB also estimated the BC ratio andEIRR ignoring earlier investment costs, whichyielded a rosier BC ratio of 1.7 and an EIRR of16 percent that helped cast the project into a

more favorable light. The report also noted thatthe project “will produce other direct and indirectsocial and economic benefits, which will beconsiderable but are difficult to quantify.”51

Ten years later, with a “newly” deliveredproject already in a worrying state of degradation,rehabilitation appeared inevitable. Cost-benefitanalyses were carried out for two assumptions—one incorporated a portion of past investment andthe other ignored earlier costs. Both CBAsincluded the full range of estimated benefits fromthe project. Due to the adverse impact of delays(10 years instead of 3), the elimination of twotracts in the southernmost area, unrealized“ambitious production targets,” and the increase incosts, the EIRR dropped to 7 and 8 percent, forthe two assumptions, respectively. In addition,the EIRRs would have been much lower withoutthe unexpected tripling of the price of rice sincedonor appraisal (Nijman 1991). “Numerousimportant non-quantifiable benefits” accruing to anow “flourishing region with considerable progressin various sectors” were also expected but noeffort was undertaken to estimate them.

The immediate rehabilitation of the RB projectarea held the promise of lifting the EIRR to thefatidic 10-percent level (or even to 35% if pastinvestment costs were not considered) if croppingintensity was to increase from 134 to 185percent. In the feasibility report, the CB ratio wasevaluated at 1.72 (with a 10% discount over 50years) and the EIRR at 17.3 percent. Theseestimates appear to ignore sunk costs and, as aresult, the project was necessarily “low cost andquick yielding.” Nijman (1991) reported thatseveral sound options, such as forming water-user groups, raising the dam of the Uda Walawereservoir by one meter and rehabilitating smalltanks, were discarded because of their negativeimpact on the EIRR.52

The ADB’s project performance audit of theRB rehabilitation in 1999 noted with satisfaction

51Detailed information contained in an appendix to the report also stated that in both cases costs for main roads and 50 percent of the costsfor townships and villages were excluded because they would be built anyway.52They raised the costs without raising the benefits, which were already derived from the most optimistic hypothesis.

42

the expansion of banana cultivation and its US$9million of yearly gross product “making the projectthe leading production area in the country,” whilerice yield (5 t/ha) also featured as one of thehighest in the country. These results raised theEIRR of rehabilitation to 11.3 percent (assumingthe first phase of the project was treated as sunkcosts) and the project was therefore classified as“successful.” On the other hand, while bananahad successfully restored the EIRR of theproject, it was noted that the sustainability of thisEIRR was threatened by a possible decline in theprice of banana resulting from increasedproduction associated with expansion of the areaunder cultivation. For example, a reduction of 20percent in banana prices would reduce the EIRRto 9.2 percent.

The review of the feasibility study of theextension of the LB area carried out in 1995 byNippon Koei included several changes withregard to the 1993 pre-feasibility report.Sugarcane, expected to grow on 3,000 ha, wassubstituted with banana, onion and vegetables,which yielded much better returns. Water supply,which the ADB considered as insufficient towarrant the development of the last phase underthe then management practices, was consideredto be secured by both the Samanala reservoirand the diversion from the Timbolketiya river.53

Although the 1995 revised feasibility study byNippon Koei increased the estimated cost of theproject from US$107 million (Rs 5.5 billion) inthe 1993 report to US$141 million (Rs 7.2billion), they estimated the EIRR at 17.8 percent.

EIRRs are routinely churned out to justifyinvestment projects that are often decided ontotally different grounds, although mosteconomists readily admit that the categories ofcost, benefit, project lifetime, etc. can bemanipulated to obtain very contrasting results

(Ingram 1971). While some acknowledge that theCBA is easily “corrupted,” others consider thevariability of the results as a product ofincompetence or bias, not a weakness of themethod (Williams 1972). Yet others present morephilosophical objections to the idea that valuesand preferences could be commensurate andcould be reduced to a single number.54 The EIRRundoubtedly suits bureaucratic processes ofdecision making, which are unable to deal withcomplexity and need to base decisions onassumedly neutral and objective criteria. Itprovides scientific support and legitimization, andmay serve as a powerful tool “to clothe politicallydesirable projects in the fig leaf of economicrespectability” (Marshall 1965). Despite thesealleged limitations and manipulations, CBAallows one to identify and weed out projects thatare absurd from an economic point of view(Ingram 1997).

Porter’s (1995) study of the history of CBAat the US Army Corps of Engineers (Corps)unearthed more subtle dimensions of thetechnique. Quantitative techniques provide asense of “mechanical objectivity” that seems tooverride the passions and interests that informpolitical debate. But they are also driven by therivalry between administrations (in Porter’s work,between the US Department of Agriculture, theBureau of Reclamation and the Corps), whichprovide incentives to seek a single standardizedmethod of the CBA.55 In regard to the Corps,Porter shows that “as the best harbors weredeveloped, levees erected and dam sites usedup, more and more of these so-called intangiblebenefits were made tangible, and quantified. Inconsequence, many projects that were turneddown, some decisively, in the 1940s or 1950swere eventually approved and built.” In otherwords, “the Corps was engaged in a perpetual

53While the MMP study of 1993 found that the Samanala reservoir would slightly decrease the frequency of deficit in the Walawe system, afeasibility study 2 years later arrives at the opposite conclusion.54For further discussion of the strengths and weaknesses of CBA, a subject of extensive literature, see Porter 1994; Kopp et al. 1997, andLohman 1997.55This attempt was largely unsuccessful, as each administration continued to incorporate its own interests and viewpoints in the techniquesit used.

43

effort to push back the frontiers of cost-benefitanalysis so that there would always be amanageable supply of economically approvedprojects.” As early as the 1940s, this led theCorps to consider five classes of “extendedbenefits radiating outward” categorized asmerchandizing, direct processing, other stages ofprocessing, wholesale trade and retail trade.

This stands in marked contrast to thecrudeness of the CBAs carried out by successiveconsultants or by the ADB for UWIRP. Wishfulhypotheses were made without much necessitybeing felt to justify them; the project lifetime wastaken as 50 years despite rehabilitation starting 4years after completion, and it had unrealisticcropping patterns aimed at boosting benefits, adhoc estimates of future irrigation efficiency andinfeasible cost-recovery scenarios. The EIRRs forthe rehabilitation phase were based on costs onlyfor the rehabilitation rather than the entireinvestment as a way to enhance the benefits ofthe proposed rehabilitation. In sum, while thebureaucratic necessity to raise the EIRR to 10percent was influencing the way the CBA wascarried out, the reports were quite liberal withparameters and did not show pressure to produce“harder” results. Significantly, derived benefits, inparticular the generation of job opportunities, werementioned in passing and there was noquantification. In other words, there was no signof the “accounting inventiveness” of the Bureau ofReclamation (Porter 1995). A tentative explanationfor this difference is given below.

Cost Recovery

Cost recovery is an important component of thepolicy of development banks. It is seen not onlyas a means to ensure sustainability, by providingfinancial resources for maintenance, but also as

a way to instill a sense of ownership, as well asresponsibility with regard to water-managementpractices. It is also vaguely assumed that makingfarmers pay for water might be conducive tolimiting water wastage. In practice, discussionson farmers paying for water are about recoveringpart of the costs incurred by the project for thegovernment coffers.

The ECI report in 1960 did not elaboratemuch on financial issues but endorsed theprinciple of water charges as a key element ofthe project: “experience in irrigated areas inother parts of the world has shown that thecontrol of water distribution and the eliminationof waste in irrigation depend as much on thelevying of a water charge or tax as on any otherfactor.” In a probable reference to colonial timesor to some western countries,56 the report saysthat charging for water “not only brings inrevenue, but in many countries has also provedto be an efficient means of bringing into beingand maintaining a desired land-use pattern.”

In the RVDB report of 1963, revenues fromland rent, water fees, and the sale of electricityduring 50 years were supposed to balance partof the investment costs. The concern for costrecovery was also present from the start in the1968 Hunting report, which emphasized that “animportant objective of the Walawe plan is thatfarmers should pay for certain services andfacilities provided to them.” Since fee recoverywas admittedly difficult to ensure, repaymentshould be “insisted on from the earliestopportunity as a pre-condition to the grant oftitle.”

The 1969 ADB appraisal report proposes torecover costs in the order of US$1,680 per 2-hectare farm (Rs 10,000 per 5-acre farm) over aperiod of 25 years, which could be consideredas “a considerable improvement compared to thepresent practice under which farmers pay only a

56The report shows a clear tendency—posing no problem at that time—to confuse contexts and to export issues and problems more familiarto northern America to developing countries. For example: “there is no water act in Ceylon defining the rights and limits applying to waterusers. It is assumed that existing developments would be given a prior right to the use of water. These rights should be evaluated by thelegal authorities, and an attempt should be made to determine how they may affect the economic evaluation of new projects, such as theone suggested for the lowland plain of the Walawe basin.” The issue was brought back by the ADB 1999 report, almost 40 years later,without a much clearer idea on how it could be put into practice.

44

nominal rent of Rs 20 per acre per year.” Thegovernment also “intends to collect a watercharge of Rs 40 per irrigated acre a year fromsettlers in the project area. This water charge willbe reviewed in the fifth year when farmers areexpected to realize more fully the benefits fromthe project. At that time, consideration will begiven not only to maintenance and operatingcosts, but also to recovery of the developmentcosts” (ADB 1969).

These intentions, however, would never berealized. Facing numerous and severe difficultiesin the settlement process, as well as in theimplementation of the construction work, theRVDB was obviously not keen to open a newfront on such a sensitive issue.

With the dismal evidence of the rapidity ofdegradation, the ADB understandingly placedemphasis on cost recovery in the negotiationsfor rehabilitation. The government explicitlyendorsed the principle of O&M cost recovery inthe National Agriculture, Food and NutritionStrategy and, in July 1983, issued the IrrigationOrdinance (amendment), which required the levyof Rs 100 per acre (US$10.50 per hectare)annually on gravity irrigation systems, effectiveApril 1984. “This rate represented about 50percent of average O&M costs on a country-wide basis. The government intention is toreview this rate periodically and increase it at 20percent per annum so as to attain full O&M costrecovery by 1989” (ADB 1984). With initialcollection efficiency in the project area at morethan 50 percent, the new policy seemed toreceive adequate support from the governmentand contented the Bank.

The ADB mission also obtained “suitableassurances” from the government that it would,in consultation with the ADB, introduce variableirrigation fees for rice and upland crops toencourage crop diversification, and reportregularly on the progress of the fee collectionand on improvements in the collectionmechanism. Because “subsidies for operation

and maintenance costs of the irrigation systemconstitute a heavy burden on the governmentbudget... the government has decided to phaseout irrigation subsidies gradually by recoveringthe O&M costs from the beneficiaries of theirrigation system” (ADB 1984).

It may be revealing that none of thesubsequent reports by MMP or the 1995Project Completion Report commented on whathappened to fee collection.57 The reportemphasized not only the need to strengthenfarmer organizations but also the “need toconsider financial incentives to encouragefarmers to make efficient use of water. Onemeans of doing so is to build on the proposal toissue bulk deliveries of water to distributary canalorganizations for onward sale to field canalgroups. This would raise the farmers’ awarenessof the cost of water, as well as generatingrevenue for maintenance and for the salary offarmer representatives and any gate operators.”

Nothing was said about how such policymight be implemented, why the earlier collectionfailed, and how the general rhetoric on thebenefits of water pricing might apply to theparticular context of Walawe where volumetricpricing was a far-fetched option. In addition, thereport also envisioned that “over the next fiveyears, farmer organizations [were] expected totake over the full costs of maintaining distributaryand field canals.”

The 1999 post-evaluation report lightlytouched upon the issue, and built upon the ideaof volumetric pricing, adding a reference to waterrights:

“While full economic pricing will notbe feasible for the foreseeable future, thesame objective could largely be achievedthrough the allocation of transferablewater rights, with areas with surplusencouraged to sell their excess rights todeficit areas. Transferability of water on avoluntary and compensated [sic] basis

57Fee collection was started but discontinued after 4 years because of widespread defaulting and vocal opposition from farmers. See Samadet al, Forthcoming.

45

will be a major tool to promote moreefficient allocation and use of waterresources” (ADB 1999a).

While such mechanisms are known to beuseful and powerful tools in certain contexts ofscarcity, notably in times of crisis (e.g., thedrought bank in California), they are also knownto demand that a set of drastic conditions be inplace for them to function (Molle 2004). Suchsuggestions in the Sri Lankan context areinappropriate for the time being and illustratehow outsiders’ mind-sets and blueprints aresometimes projected into utterly differentcontexts.

Water Requirements: Supply, Demandand Management

Because of its importance to projectmanagement and adequateness of supply tofarmers, it is instructive to insert here a commenton the malleability of supposedly fixed projectparameters, taking as an example the loss ofwater by percolation in irrigated fields. Water duty,or the amount of irrigation water planned for oneunit of land, as considered in project design, isoften believed to be an unambiguous parameterthat can be easily derived from sound technicalparameters on crop physiology and soil types, astextbooks indicate. In reality, crop waterrequirements are calculated based on data with asignificant level of uncertainty and, above all,high spatial and temporal heterogeneity (e.g., soilcharacteristics and management practices).

The design values of water duties are ofcrucial importance since they dictate thecapacity of the canals to be built and, thus, havea critical bearing on project costs and theresulting cost-benefit ratios. In what follows, welook at how assumptions on water requirementshave changed over time and the resultingimpacts on project design, management andeconomic profitability.

The study by IEC in 1954 considered a totalwater requirement of 8 feet (2.44 m) for two

crops of rice, based on past experiences insimilar areas, while the Irrigation Departmentestimate was 10.5 feet or 3.20 m (IEC 1954). TheECI report, in 1960, investigated the capacity ofthe future Uda Walawe dam and apparentlydiscovered the 10 feet (3.05 m) yearly waterrequirement mentioned in the Tender Notice data.Such a high figure was incompatible with theprojected irrigated area (110,000 acres or 44,515ha) and the storage capacity offered by the siteenvisaged. A study (“The Samanala WewaProject Report”) was opportunely reported topropose a value of 4 feet (1.22 m) per year thatwas eventually selected by ECI “as a soundrepresentation of the average overall conditionsfor the farm water needs.”

The 1968 Hunting report puts its faith inmodern management, which will allowrequirements for rice to be half the water used“under the traditional system of virtually unlimitedirrigation.” The farm-level irrigation waterrequirement is taken as 4.2 feet (1.28 m) peryear; “this is considerably less than the designrequirement used for the alienated rice areas inthe Uda Walawe area, but will be sufficient ifcorrect principles of water use are fullyunderstood by engineers and farmers alike” (HTS1968a). Consequently, water is found “to be insurplus above requirements and an additional12,000 acres of accessible highland is proposedto be brought under lift irrigation.” Design is notadapted to reality but, rather, reality is expectedto fall in line with “correct principles.”

The Government Plan (GOC 1969) revisedthese values and assumed a crop water need of5.5 feet (1.68 m), plus 40 percent conveyanceloss, giving a total water duty for rice of 7.7 feet(2.35 m). The 1969 ADB Appraisal Report is notmuch concerned about water issues, anunderstandable attitude since the project isabout the construction of the RB only. In 1979,the Project Completion Report finds out “thatonly about 70 per cent of the RB areaenvisioned for irrigation at the time of appraisalis actually served, and this area is consumingabout three times the water proposed for theentire RB area.” A poor design and operation,

46

added to widespread physical degradation,deprived many tail-end areas of water.

Indeed, the reports by PRC (1982), Wolf(1983) and SOGREAH (1984) give more attentionto water management. It is now recognized thatpercolation losses are very high in RBE soils andWolf takes them at 4 mm per day. SOGREAHassumes average infiltration loss in rice fields tobe 3 mm per day and states that “5 mm/day isprobably the maximum unavoidable loss onsloping land,” seemingly ignoring that many plotsdry up in one day. A dilemma occurs in that it isimpossible to plan an increase in cropping area inthe project without drastically reducing thisnumber. SOGREAH, therefore, envisions fourtypes of water-saving measures, which constituteoptimistic if not wishful ad hoc hypotheses.

Faced with the duty to implement therehabilitation works based on new designparameters, MMP’s (1986) inception reportidentifies effective rainfall and deep percolation ascritical factors, and stresses the absence of fieldmeasurement of actual losses. The percolationrate is taken as 5 mm/day and tests are advisedbut later reports do not indicate whether thesehave been carried out or not. Two years later,calculations of operational procedures appear tohave been made based on a general assumptionof 5 mm per day infiltration losses (MMP 1988).Oddly enough, the cropping patterns consideredare not the actual ones and include many moreOFCs than is the case in reality. The reportindicates that “it is not suggested that the[cropping] pattern… will in practice be adopted bythe farmers” but it is meant to “provide apracticable method for planning an equitabledistribution of water… the farmers will then adapttheir irrigation and cropping to match the availablewater within the specific constraints of the soilsand topography, their farming practices and theconditions of their individual farms” (MMP 1988).The rotations calculated and proposed aretherefore not meant to correspond to actualcropping patterns but, by incorporating some

constraint in supply, are expected to forceadjustments by farmers. A different possibleoutcome, of course, is that farmers will employillegal means and damage infrastructures toaccess water. Last, MMP’s (1992) handbookissued at the end of the project does notcomment on the issue. The tables given tocalculate water duties by canal, however,consider high values of deep percolation for tracts2, 3 and 4, that is, the very upper part of thescheme (see map in Appendix B), while keepinga value of 5 mm/day for the rest of the area.58

The feasibility study on the Walawe extensionproject (JICA 1993) addresses the issue ofpercolation loss more seriously. At last,systematic infiltration tests were carried out andyielded rates of 5, 10 and 20 mm per day forLHG, RBE (moderate drainage) and RBE (well-drained) soils, respectively. This gave waterduties that are between 2 and 8 times thoseconsidered by Hunting in 1968. By investigatingthe spatial variability of percolation losses, JICAopened the Pandora’s box and found outextremely high plot water requirements that havethe potential to jeopardize any attempt to addthe extension area to the existing scheme.Subsequent generous hypotheses on fieldapplication losses and water reuse were thenformulated to counterbalance this finding andreduce diversion requirements.

What this brief retrospective tells us is thatthe assessment of water requirements is both atouchy technical issue and one that governs themost crucial design parameters of the project aswell as the assessment of its economicperformance. The most astonishing observationis the range of variation of water duties, from 4.2to 15 feet (1.28-4.57 m) per year (let alone the36 feet or 11 m for well drained RBE soils foundby JICA). It is also apparent that for 50 yearsconsultants have avoided carrying out systematicpercolation tests, until those undertaken by JICAin 1993. On the contrary, it appears thatassumptions on water use have been largely

58It can be hypothesized that MMP’s earlier policy to consider losses of 5 mm/day and idealized cropping patterns was eventually judgedtoo hazardous for the upper block, where losses are highest. We don’t have the records needed to fully analyze how MMP eventually strucka balance between what was technically/financially possible and the implications of considering the real values of percolation losses.

47

taken as given, based on historical reports,inferred, or “retro-estimated,” based on thepotential defined by the dam capacity and thecorresponding irrigable area under its command.For example, ECI used comparisons with theUnited States. Hunting backed its water dutiesby arguing that modern management would halvethe water use of “traditional agriculture.”SOGREAH envisioned the wholesale adoption offour ad hoc water conservation measures. ADB(1969) and MMP (1986) assumed croppingpatterns that involved more area in OFCs thancould be realistically anticipated, and JICA(1993) minimized the implication of its ownfindings on percolation by adopting generoushypotheses on the rate of water reuse.

This “inventiveness” is no surprise for thosewho have been involved in project design.However, just as in the case of CBA,consultants cannot be fully blamed for their“adaptive” behavior. They are requested by thefunding agencies to arrive at results that are byand large predefined by general targets andspecific constraints. They must abide by thepolicy guidelines, avoid antagonizingmanagement agencies and other officials, andmust cast the project of their contracting agencyin a favorable light. Just as in the case of CBA,it appears that the range of hypotheses that onecan make on the numerous parameters of theproject is extremely large and malleable, andthat their impact on the EIRR can be surprisinglylarge (Williams 1972).

There is little scrutiny of and objections tothe choices made. “Who is to say theassumptions are wrong?” asks Berkoff (2002). Acollateral effect of such logic, however, is ofextreme importance. When structures aredesigned based on unrealistically optimistic waterduties, it creates an environment of future waterscarcity that ultimately become inscribed in thephysical landscape and on those who dependupon it. Hydraulic units and subunits will not getthe amount of water they need and users will beled to tamper with or destroy structures. Theresulting decay, ascribed to farmers’ indiscipline,tampering of structures and poor management, is

partly a consequence of the earlier “tampering” oftechnical parameters.

Actors’ Interests and Strategies

This section attempts to wrap up the differentobservations made earlier and to highlight howthe combined strategies of the various“stakeholders” may result in the sequence ofevents described earlier. These stakeholdersinclude multilateral funding agencies, the centralgovernment, local politicians (district andprovincial), the implementing agencies, theinternational or national consultants and the“affected populations” (see table 9). The actionsof stakeholders are determined, in part, by theirrole in the project, their particular interests,accountability and potential sanctions.

One of the most striking features of proposeddevelopment projects—especially one that doesnot involve land reclamation or resettlement—isthe convergence of interests around the project.Development banks need to lend money and theirrepresentatives are partly evaluated on theamount of funds they have lent. The centralgovernment, particularly the Ministry of Finance,is traditionally more cautious about spendingmoney, but the relatively advantageous conditionsof such loans plus the pressure felt by thegovernment to deliver public goods and tolegitimize its actions by taking up the“development mission” make the option attractiveor, at least, minimize opposition. This fosters anattitude of “getting the capital funds while youcan, worrying about the use of the resultingproject later, if at all” (Howe and Dixon 1993).Local politicians, especially those linked to theparty in charge, hope to benefit through patronageand obviously add their pressure to the demandfor investments. Implementing agencies, in turn,are eager to sustain the flow of resources thatstrengthens their power and soothes recurrentfinancial difficulties, particularly with regard toO&M. National and international consulting firmsseek project work opportunities, to respond to theobjectives of their clients, and to benefit from

48

TAB

LE 9

.M

ain

acto

rs in

the

deve

lopm

ent p

roce

ss a

nd th

eir

resp

ectiv

e st

rate

gies

.

Act

orR

ole

Par

ticul

ar i

nter

ests

Acc

ount

abili

tyP

ossi

ble

sanc

tion

Cen

tral

Cen

tral

pla

nner

: S

et p

roje

ctC

reat

e re

gion

al d

evel

opm

ent;

Acc

ount

able

to

soci

ety/

Loss

of

pow

er/lo

ss o

f

gove

rnm

ent

obje

ctiv

es,

plan

reg

iona

lst

reng

then

pol

itica

l leg

itim

acy

mul

tilat

eral

fun

ding

futu

re f

undi

ng

deve

lopm

ent

and

loca

l par

ty m

embe

rs;

agen

cies

desi

re f

or p

rest

ige

Loca

l po

litic

ians

Lobb

ying

: S

uppo

rt f

urth

erS

tren

gthe

n po

litic

al b

asis

In p

rinci

ple,

to

peop

le;

Non

-ree

lect

ion

deve

lopm

ent

proj

ects

and

loc

alth

roug

h pa

tron

age

in r

ealit

y, l

ittle

acc

ount

abili

ty

publ

ic i

nves

tmen

ts

Inte

rnat

iona

l/Te

chno

crat

: Id

entif

y pl

ans

to m

eet

Acc

epta

nce

of p

roje

ct;

furt

her

role

Acc

ount

able

to

cent

ral

Loss

of

futu

re w

ork

loca

l co

nsul

tant

sob

ject

ives

, en

sure

tec

hnic

al/e

cono

mic

in d

esig

n or

im

plem

enta

tion;

com

plyi

nggo

vern

men

t

feas

ibili

ty,

carr

y ou

t pl

ans

with

fun

ding

age

ncy

obje

ctiv

es

Loca

lTe

chno

crat

: Id

entif

y pl

ans

to m

eet

Sus

tain

the

gov

ernm

ent

budg

etA

ccou

ntab

le t

o ce

ntra

lLo

ss o

f po

wer

impl

emen

ting

obje

ctiv

es,

ensu

re t

echn

ical

/eco

nom

icat

trib

uted

to

it; j

ustif

y ex

iste

nce

and

gove

rnm

ent

agen

cies

feas

ibili

ty,

carr

y ou

t pl

ans

staf

fing;

str

engt

he b

urea

ucra

tic p

ower

Mul

tilat

eral

Pla

nner

/tech

nocr

at:

Hel

p id

entif

y de

velo

pmen

tLe

ndin

g m

oney

; se

lect

pro

ject

with

EIR

RA

ccou

ntab

le t

o se

lfLo

ss o

f cr

edib

ility

if

fund

ing

goal

s vi

a se

ctor

al p

lann

ing,

ide

ntify

pla

ns t

o m

eet

abov

e 10

per

cent

and

inc

reas

ing

cost

(and

par

tly t

o m

embe

rpr

ojec

t is

a s

ever

e

agen

cies

obje

ctiv

es,

prov

ide

fund

ing

and

othe

r te

chni

cal

reco

very

; de

velo

pmen

t of

a m

arke

t-ec

onom

y; c

ount

ries)

failu

re

assi

stan

cesu

stai

nabl

e de

velo

pmen

t

Aff

ecte

dLa

rgel

y pa

ssiv

e: c

ompl

y w

ith p

lann

ing

To b

e se

lect

ed a

s a

settl

er,

or f

orci

ngLo

osel

y ac

coun

tabl

e to

Loss

of

subs

idie

s;

popu

latio

nob

ject

ives

, lit

tle s

ay i

n pl

anni

ng a

ndde

cisi

on t

hrou

gh e

ncro

achm

ent;

loca

l im

plem

entin

g ag

ency

degr

adat

ion

of f

acili

ties;

impl

emen

tatio

nri

sk a

vers

ion

nons

elec

tion

49

further contracts. The concerned populations,which are by no means homogeneous in eithertheir a priori socioeconomic status or theirpotential to gain or lose as a result of the project,tend to welcome the project if they see directbenefit for their families.

Some minor groups, however, stand to lose,but they are hardly organized to a level that couldsignificantly influence the course of things. Insome cases, such as the purana villagers underthe Chandrikawewa reservoir who were successfulin keeping their larger plots of land, succeeded inpreserving their interests. In other instances, thisis not the case. As we have seen earlier, thepopulation making a living off the shrimp in thebrackish lagoon lost their livelihood when irrigationdrainage reduced the salinity of brackish waterand decimated indigenous shrimp populations.The expansion of irrigated areas or “stateenclosures” such as national parks orafforestation areas have also reduced the spaceavailable to indigenous chena cultivators andcattle farmers. The “nomadic” cattle owners of theLB are also likely to lose most of their grazinglands as development in the area continues.Chena cultivators are likely to be absorbed assettlers and since chena is deemed illegal mostof the time, they also have little scope foropposition.

Notwithstanding these third-party impacts,which are at best mentioned in passing in thedifferent project reports, UWIRP is characterizedby a convergence of interests. Basically, allactors in the process have a strong interest inmaking it happen. Sanction in case of failure alsoappears to be very low. The evidence that,irrespective of its outcome, the project willincrease the debt burden of the country is not aserious deterrent, as politicians struggle forreelection. Politics tends to be pervaded with ashort-term “after me the deluge” attitude.Consultants give priority to respond to the donors’objectives and are ready, for example, toincorporate cost-recovery mechanisms that suit aparticular bank’s general policies, ad hoc croppingpatterns or irrigation efficiencies that make theproject more profitable, irrespective of whether

they think these are realistic or not. If theassumptions made are not realized, failures areoften blamed on the shortcomings inimplementation, inadequate maintenance, poorperformance of water-user groups and non-adherence to proposed cropping patterns.Responsibility for unrealistic targets set atappraisal is avoided by deeming them as“ambitious” rather than unfeasible.

Despite significant improvements ininfrastructure after rehabilitation, national plannersand managers have failed to boost projectperformance. This poor performance was due tothe rapid turnover of Resident Project Manager,low accountability towards users, deferredmaintenance, incapacity to ensure adequatesupply to the tail-end areas despite very highsupply-demand ratios, lack of incentives tomanagers for improved performance, limitedconsultation with users and reluctance to co-manage the scheme. However, management in acontext of dilapidated infrastructure, poorcommunication and transportation means, andoccasional political meddling is not an easy task.Local politicians, indeed, have partly capturedproject benefits through patronistic practices inthe designation and choice of settlers, andsometimes through their intervention power andcapacity to influence spatial allocation of wateraccording to their needs.

In the 1960s and 1970s, multilateraldevelopment banks like the World Bank and theADB pressed countries to present “developmentplans replete with dirigiste policies and lists ofprojects to be financed” (Roberts 2004). They runlittle risk of being sanctioned by defaultingclients, since their peculiar status entails thatthey are assured of reimbursement. This fostersa supply-oriented logic of disbursementencouraged by incentives for country managers tomaximize loans (George and Sabelli 1993). Theseinstitutions are thus tempted to indulge in wishfulthinking and to idealize the context in which theproject is being developed. The RVDB, forexample, is conveniently considered to haveundergone “considerable changes in itsorganizational structure” and is fit to “promote

50

successful implementation of the multi-purposeWalawe scheme.” The ADB also tends to contentitself with the assurances given by thegovernment, knowing that political changes oreven their evident impracticability, often reducesuch commitments to dead letters, as theexample of cost recovery has clearly illustrated.59

The ADB has also demonstrated exaggeratedfaith in the power of the state to reorder socialsettings. The evidence of degradation ofinfrastructure prompted the injunction that “adecisive halt needs to be brought to thedeliberate destruction of project facilities. MASLneeds to prosecute culprits to stop suchpractices from further developing and solemnlywarn farmers that their area will be shut off fromwater delivery if further destruction takes place orif they do not repair the damaged structure.” Thecrucial role of the implementing agency seemedto come to the fore only in the ADB’s 1999 expost project assessment report, where it isrecognized that “for the participatory approach towork, it is essential that MASL forcefully backs itand be fully committed to implement it over thelong term. This requires attitudinal and behavioralchanges from the top to the bottom of theinstitution.” In 1999, all the recipes of the day arepotentially ready to put things back on the righttrack: “key issues that the policy could addressinclude the development of effective O&Msystems, water-right transferability and pricing.Urgent priority must be given to strengtheningFOs [the funding of which had been disregarded10 years earlier] if further decline of Walawe’sirrigation system is to be prevented” (ADB1999b). The report concluded with a pessimisticundertone, “it is not too late to redeem thesituation,” revealing some frustration with thestubborn resistance of projects to conform toidealized images.

The reputation of development banks,however, may suffer from outright project failures.To avoid such situations, banks sometimes getcaught up in avoidable successive phases ofrehabilitation or modernization. This is clear in the

present case study, where rehabilitation wasaimed at “realizing the benefits” of the precedingproject; that is, fixing up the failures of a projectby dubbing previous failures or shortcomings as“constraints.” Since the interest of the ADB forrehabilitation exceeded that of the centralgovernment, a telling shift in the relative sharesof funding occurred, with the ADB funding 80percent of the rehabilitation phase, against 26percent of the construction phase. In addition, theBank used the leverage of possible loans to thenearby Kirindi Oya Project to put pressure on thegovernment to accept the rehabilitation (Nijman1991). Getting the EIRR back to 10 percent wasa manifest concern of the Bank, even if thearithmetic involved might have been contested.

The point, indeed, is that except for theinternal review at the Bank itself, nobody seemsto have scrutinized or questioned the hypothesesand calculations made, or felt the need to test thesensitivity of parameters or to include a broaderdefinition of benefits and costs. While Porterattributed the development of the CBA in the USAto bureaucratic conflict in a context ofoverwhelming public distrust, the situation in theUSA is clearly at odds with that of the Walawebasin. First, loans are largely coming frommultilateral development banks and this type offunding is subject to much less internalcompetition or objection than the use ofgovernment coffers. Second, public scrutiny,participation or alternative viewpoints were almostnegligible in the process, diminishing the pressureto justify investments, particularly to presentalternative use of the funds, or to address third-party impacts. The absence of “accountinginventiveness” in the bank and consultant reportsis striking and reflects the lack of accountability,and the lack of scrutiny by outsiders. A moresophisticated CBA would be possible but wouldcome at a cost, and there is obviously noincentive to embark in such an exercise sincemost parties have converging interests and do notdispute the case. No costly models or surveysare required to make the case.

59“The government has given assurance that problems of land titling will be solved and that policies of free market, subsidy reductionon agricultural inputs, and price support will continue,” etc.

51

This consensus, however, was broken in the1990s, when ADB did not support furtherdevelopment of the irrigation area, based on theevidence that the lack of improvement in watermanagement and irrigation efficiency did not allowexpansion of the system.60 The government andMASL successfully resorted to alternativesources of funding, by approaching the JapaneseBank for International Cooperation and otherfunding agencies. As they were not involved inthe earlier frustrating phase of development, theJapanese institutions were receptive to therequest and conducted pre-feasibility andfeasibility studies, adjusting hypotheses toproduce an attractive project outlook and EIRRs.This shows that competition between lendingagencies can be adequately exploited to furtherdevelopment (Howe and Dixon 1993), even whenparticular circumstances have drasticallydampened the enthusiasm of a particular lender.

In sum, if we follow Ingram (1971), whostates that “the inclusion of a particular project ina package depends upon its effect on the

balance of support and opposition to the package,not the relationship of the project to any aims orgoals established for water management ingeneral or for the particular collection of projects,”then the “Walawe Package” appears as thenatural outcome of contributing and converginginterests faced with little or no opposition, andsubject to little scrutiny. This should not be takenas a criticism of poor or faulty planning, althoughsome elements were clearly unrealistic. Rather,we suggest that government interventions arealmost invariably based on overarching policies orpolitical priorities that come with strong self-justificatory undertones and that largely overrideformal cost-benefit analysis. This has been widelyobserved in both developing and developedcountries (del Moral et al. 2003). Irrespective ofwhether the priorities defined are sound or not,the convergence of interests leads tochoreographed cycles of project appraisal, design,implementation and assessments that unfoldunder the banner of rational planning andcomputed outcomes.

Lessons Learned and Conclusions

This comprehensive analysis of the Uda WalaweIrrigation and Resettlement project in Southern SriLanka has led us to look at its successivedevelopment phases through different lenses.

The main point coming out of thedescription of over 50 years of progressivedevelopment is the wide range of difficultiesthat have constantly undermined the effortsmade. They range from design failure, shoddyconstruction and poor maintenance toconstraints on agricultural diversification,administrative inertia and political upheavals.What is striking is the contrast between the

(sometimes blunt) assuredness of thehypotheses made in feasibility reports and thereality on the ground, between the simplistictechnological and social engineering drive ofthe consultants and the complexity of regionaldevelopment.

As a result, expectedly, the differentperformance indicators or, more simply, thecomparison between expected and observedresults of the different phases, reveal significantdiscrepancies. These discrepancies confirm themore general results found by various otherstudies on project performance.

60The 1999 ADB report states: “If supplies prove to be inadequate, decisions will be required by MASL on necessary changes to thedesign or phasing of expansion area development. It is suggested that MASL considers suspension of development of the LBexpansion area.”

52

The review of direct costs and benefitsshowed that crop benefits remained quitedepressed until the groundswell of bananacultivation, which came to occupy around 40percent of the command area. It also shows howthe price of agricultural products eventuallydictated the profitability of the project. While afull-fledged assessment of all direct and indirectbenefits and negative externalities was notattempted, the estimation of the value of otherwater-related benefits such as inland fisheries,the National Park, and home-garden produceshowed that many important components need tobe considered for a more comprehensive impactassessment.

The justifications of project developmentduring four decades were based on modernistprinciples that echoed very general objectives ofthe government—raising national production andrural income, diversifying agriculture, alleviatingpoverty and increasing employment. All projectsdo this to a greater or lesser extent and theywere easily shown to be consistent with generalgovernment policies. When presented asunquestionable necessities, national priorities canoverride benefit-cost ratios that are not sufficientto justify the project; they can also obviate theneed to investigate in more depth the diversity ofpossible positive and negative impacts.

Indeed, cost-benefit analyses, as appearing inconsultants’ reports, were found to be confined tothe estimation of crop-related direct costs andbenefits and showed little of the sophisticationthat has emerged in contexts where substantialscrutiny and competition for resources are therule. Assumptions on percolation losses, waterrequirements, cropping patterns, or managementefficiency were partly made under the constraintto obtain an EIRR of over 10 percent. The caseof the estimation of percolation losses wasillustrative of the “malleability” of projectparameters but it also showed the dramaticconsequences on water management andinfrastructure degradation that unrealisticassumptions made on paper may have on theground.

The crude nature of cost-benefit analysis isalso commonplace and might lead to the usualprescription that such analysis should be carriedout in a more thorough manner. However, if cost-benefit analysis does have a role in eliminating“dead duck” projects and contributing to betterchoices (Green 2003) it must be recognized thatsocial choice is eventually governed by a muchwider array of considerations than mere financialcalculations.

Finally, the history of UWIRP was examinedfrom the perspective of a set of actors with theirrespective interests, constraints and strategies.These actors include the central government,local politicians, international and localconsultants, local implementing agencies,multilateral funding agencies and affectedpopulations. The case of UWIRP is notable bythe convergence of interests to make the projecthappen; there was little apparent resistance toUWIRP. Accountability and sanction for failure ofthe different actors were found to be low ornonexistent. As a result, the unfolding of theproject was very much driven by the agenda ofthose in control of financial resources (the stateand lending agencies), with a bearing on statedpriorities, project options (e.g., decision torehabilitate), and the responses found byconsultants to fulfill their expectations.

The ongoing gap between the rationality ofthese actors and the events on the ground,between planning and the combined behaviors oflocal managers and users, cannot be understoodsolely on the basis of poor performance or non-adherence to rules and plans. The trajectory of aproject like UWIRP may be significantlyinfluenced by external and unpredictable factors(e.g., boom of banana) and the articulation andconflict between different rationalities at work,which are confronted in the common arena ofproject development.

In sum, this review of UWIRP in the lowerWalawe Basin underscores the importance ofrealistic planning, stakeholder participation, andadaptive management. Rather than adaptivelyadjusting to unanticipated changes, planners,

53

donors and managers continued to return timeand again to the original plan and launchsuccessive attempts at realizing the original plan.Planning and implementation proceeded through atypical top-down, blueprint approach thatconstituted a typical example of Hirschman’s“hiding hand” principle, whereby “ignorance ofignorance, uncertainties, and of difficulties” is theprelude of a “long voyage of discovery in themost varied domains, from technology to politics”

(Hirschman 1967). The obvious drawbacks ofsuch top-down approaches should warrant andencourage the adoption of process approaches ona wider scale (see Bond 1998). Participation ofconcerned stakeholders in planning is a widelyaccepted principle that brings about more equityand efficiency (Howe and Dixon 1993) but itentails a revolution of mindsets and practices andis eventually predicated upon wider societalchanges and democratization.

54

55

Appendices

APPENDIX A

Work Completed in UWIRP by 1968

1. Uda Walawe reservoir

2. Irrigation system:

(a) RB: Planned length 41.8 km (26 miles);to irrigate 10,522 ha (26,000 acres)• 37 km (23 miles) of main RB channel

completed• Irrigation facilities built for Tracts 1-7

covering 1,518 ha (3,750 acres)

(b) LB: Planned length 64.4 km (40 miles); toirrigate 19,425 ha (48,000 acres)• Work in progress: irrigation facilities for

Tracts 2-7, covering 1,518 ha (3,750acres) near completion

• Five tanks (Kiri Ibban, Mahagama,Gal, Vedigam and Habaralu) completedto irrigate 728 ha (1,800 acres)

3. Land Development and Colonization

(a) RB• 1,619 ha (4,000 acres) cleared in

Tracts 12 and 13• Sugar area: 809 ha (2,000 acres)

cleared• 2,023 ha (5,000 acres) cleared in

Tracts 1-7 by colonists on subsidy and1,800 settlers on land

(b) LB: 1,012 ha (2,500 acres) cleared bycolonists on subsidy and 600 settlers onland

(c) Youth schemes: 400 youth settled at KiriIbban tank on LB and on RB

(d) Public utilities: Roads; Civic centers

56

APPENDIX B

Tracts of the Uda Walawe Scheme

57

APPENDIX C

Planned Command Area for UWIRP

Command area

ha acres

To be developed

RB 8,711 21,525

LB 18,648 46,080

Total 27,359 67,605

Existing irrigable area included in the Government Plan

Embilipitiya Hingura Ara and other village tanks 1,151 2,844

Uda Walawe reservoir

RB Tracts 1-7 1,520 3,755

LB: Kiri Ibban Aru, Mahagama and Suriyawewa 761 1,881

Chandrikawewa (initially excluded but later added to Plan) 2,133 5,270

Total 5,565 13,750

TOTAL DEVELOPMENT INCLUDED IN PLAN 32,924 81,355

Existing irrigable area excluded from the Government Plan

Liyangastota anicut

LB 2,513 6,210

RB 2,546 6,290

Total 5,059 12,500

Source: GOC 1969 and HTS 1968b.

APPENDIX D

Proposed Cropping Pattern

Crop Areaa

RB LB Total

ha acres ha acres ha acres

Rice 2,883 7,125 5,573 13,770 8,456 20,895

Sugar 255 630 6,993 17,280 7,248 17,910

Cotton and other OFCs 5,573 13,770 6,082 15,030 11,655 28,800

Total area to be settled 8,711 21,525 18,648 46,080 27,359 67,605

Total area already settled within project area 3,946 9,750 1,619 4,000 5,564 13,750

Total irrigable area planned and settled 12,662 31,288 20,275 50,100 32,923 81,355

a Does not include 5,666 ha (14,000 acres) for homesteads, towns, etc., and existing settlements on 4,387 ha (10,840 acres) ofunirrigable land.

Source: HTS 1968b.

58

AP

PE

ND

IX E

Rev

ised

Pla

n o

f L

and

Use

un

der

th

e A

DB

Do

wn

stre

am D

evel

op

men

t P

roje

ct

Ric

eS

ugar

cane

Cot

ton

and

othe

r O

FC

sTo

tal

ha

acr

es

ha

acr

es

ha

acr

es

ha

acr

es

RB

Land

to

be d

evel

oped

Tra

ct 1

202

500

202

500

Tra

cts

2-7

1,54

63,

821

4,90

912

,130

6,45

515

,951

Tra

cts

9-21

3,48

08,

600

3,48

08,

600

Sub

tota

l5,

026

12,4

2120

250

04,

909

12,1

3010

,137

25,0

51

Exi

stin

g de

velo

pmen

t

Cha

ndrik

awew

a ar

ea2,

133

5,27

02,

133

5,27

0

Em

bilip

itiya

and

oth

er

villa

ge t

anks

1,15

12,

844

1,15

12,

844

Sub

tota

l3,

284

8,11

43,

284

8,11

4

TO

TAL

FO

R R

B8,

310

20,5

3520

250

04,

909

12,1

3013

,421

33,1

65

LB

Land

to

be d

evel

oped

Nor

ther

n zo

ne2,

651

6,55

06,

993

17,2

809,

644

23,8

30

Sou

ther

n zo

ne2,

922

7,22

06,

082

15,0

309,

004

22,2

50

Sub

tota

l5,

573

13,7

706,

993

17,2

806,

082

15,0

3018

,648

46,0

80

Exi

stin

g de

velo

pmen

t

Kiri

Ibb

an/o

ther

tan

ks76

11,

881

761

1,88

1

TO

TAL

FO

R L

B6,

334

15,6

516,

993

17,2

806,

082

15,0

3019

,409

47,9

61

Tota

l ne

w d

evel

opm

ent

10,5

9926

,191

7,19

517

,780

10,9

9127

,160

28,7

8671

,131

Tota

l ex

istin

g de

velo

pmen

t4,

045

9,99

54,

045

9,99

5

Tota

l pr

ojec

t ar

ea14

,644

36,1

867,

195

17,7

8010

,991

27,1

6032

,831

81,1

26

Sou

rce:

AD

B 1

969,

22.

59

APPENDIX F

Historical Records of Visitors and Income in Uda Walawe National Park

Year No. of No. of Total no. Annual Income

local visitors foreign visitors of visitors R s US$

1991 64,939 1,572

1992 296,150 6,685

1993 2,099,595 43,851

1994 18,849 6,554 25,403 4,209,124 85,274

1995 21,284 7,438 28,722 4,975,667 97,067

1996 33,602 6,180 39,782 7,974,000 144,169

1997 54,006 3,275 57,281 6,400,000 108,438

1998 48,101 15,733 63,834 18,800,000a 288,034a

1999 69,326 21,548 90,874 25,812,588a 365,669a

2000 71,144 17,641 88,785 22,477,489 287,878

2001 44,896 12,044 56,940 14,538,288a 163,646a

aEstimated values.

Source: Department of Wildlife.

60

61

Literature Cited

Abeyaratne M., ed. n.d. Ratnapura: As seen by government agents. Colombo: Department of Government Printing.

ADB (Asian Development Bank). 1969. Appraisal of Walawe Development Project in Ceylon. Manila: ADB.Unpublished report prepared for the exclusive use of the ADB. 116 pp.

ADB. 1979. Project completion report of the Walawe Development Project, (Loan Nos. 16- CEY(SF) and 17-CEY)in Sri Lanka. Manila: ADB. Unpublished (restricted) document. iii + 129pp.

ADB. 1982. Project performance audit report of the Uda Walawe Development Project. Manila: ADB.

ADB. 1984. Appraisal of the Walawe Irrigation Improvement Project in Sri Lanka. ADB Project Report. Manila:ADB.

ADB. 1995. Walawe Irrigation Improvement Project. Project completion report 1984-1996. Manila: ADB. Draft.

ADB. 1999a. Project performance audit report on the Walawe Irrigation Improvement Project (Loan 695-SRI[SF]).Manila: ADB. Unpublished report. 18 pp.

ADB. 1999b. Walawe Irrigation Improvement Project. Manila: ADB. Unpublished report, Loan No.695-SRI(SF). 13pp. + annexes.

Alwis, Lyn De. n. d. The birth of a national park: Redeeming Uda Walawe. www.lankalibrary.com/wlife/udawalawe.htm

Amarasinghe, U. S. 2003. New approaches to inland capture fisheries statistics in Sri Lanka. In New approachesfor the improvement of inland capture fishery statistics in the Mekong Basin. Proceedings of the ad-hoc expertconsultation, Udon Thani, Thailand 2-5 September 2003. RAP Publication 2003/01, FAO Regional Office forAsia and the Pacific, Bangkok, Thailand, pp. 116-121.

Amarasinghe, U. S.; De Silva, S. S. 1999. The Sri Lankan reservoir fishery: A case for introduction of a co-management strategy. Fisheries Management and Ecology. Asian Fisheries Science 6: 1-13.

Amunugama, Sarath. 1965. Chandrikawewa: A recent attempt at colonization on a peasant framework. The CeylonJournal of Historical and Social Studies 8(1&2): 130-162.

Bastiampillai, B. 1967. The revival of irrigation enterprise in Ceylon, 1870 to 1890. Ceylon Journal of Historicaland Social Studies 10: 1–26.

Bell, C.; Hazel, P.; Slade, R. 1982. Project evaluation in regional perspective: A study of an irrigation project innorthwest Malaysia. Baltimore, MD, USA: Johns Hopkins University Press, for the World Bank. x + 326 pp.

Berkoff, Jeremy. 2002. Economic valuation: Why is it often unsatisfactory? And does it matter? With reference tothe irrigation sector. Paper presented at the International Consulting Economists’ Association (ICEA) meeting,19 June 2002. http://www.icea.co.uk/archive/JBerkoffPaper190602.doc

Bhattarai, Madhusudan; Barker, Randolph; Narayanamoorthy, A. 2003. Who benefits from irrigation investmentsin India? Implication of irrigation multiplier estimates for cost recovery, and irrigation financing. Paper presentedat the ICID Asian Regional Workshop, Taiwan, 10-12 November 2003. International Commission on Irrigationand Drainage.

Birner, Regina. 1996. The role of livestock in economic development: Theory and empirical evidence. The caseof Sri Lanka. Ph.D. thesis. Gottingen, Germany: University of Göttingen.

Boelee, E.; van der Hoek, W. 2002. Impact of irrigation on drinking water availability in Sri Lanka. ICID-CIID 18th

Congress on Irrigation and Drainage, 21-28 July, Montreal, Canada. Paper No. Q51/R5.7 (http://www.cancid.org/tech_papers.htm#IWMI). International Commission on Irrigation and Drainage.

Bond, R. 1998. Lessons for the large-scale application of process approaches in Sri Lanka. International Institutefor Environment and Development. Gatekeeper Series No. 75. http://www.iied.org/docs/gatekeep/GK75.pdf

Collins, C. H. 1932. The Archeology of the Sabaragamuwa Binthenne. Journal of Ceylon Branch of the RoyalAsiatic Society 32 (85): 158-183.

62

De Silva, S. S. 1988. Reservoirs of Sri Lanka and their fisheries. FAO Fisheries Technical paper. Rome: Foodand Agriculture Organization of the United Nations (FAO).

del Moral, L.; Hill, G.; Paneque, P.; Pedregal, B.; Spash C.; Kevin, K. 2003. Evaluation practices in water projectdecision-making processes. Comparative analysis of five European cases: Alqueva dam (Portugal), Evinosreservoir (Greece), Ythan nitrate vulnerable zone (UK), the grensmaas (the Netherlands) and Ebro rivertransfer (Spain). Draft. www.us.es/ciberico/archivos_word/phnponendelmoral.doc

Deraniyagala, S. U. n.d. Early man and the rise of civilisation in Sri Lanka: The archaeological evidence. Mimeo.

DWLC-FAO (Department of Wildlife Conservation-Food and Agriculture Organisation of the United Nations). 1997.Management plan: Uda Walawe National Park. DWLC-FAO. 3 vols.

ECI (Engineering Consultants, Inc). 1960. Preliminary report. Uda Walawe Project: Development of the WalaweGanga Basin. Denver, Colorado: ECI. 41 pp.

Farmer, B. H. 1976. Pioneer peasant colonization in Ceylon: A study in Asian agrarian problems. Connecticut,USA: Greenwood Press Publishers. xv + 387 pp.

Fernando, A. D. N. n.d. Discovering the lost cities of Ruhuna. WWW Virtual Library Sri Lanka (accessed 26-01-2005). www.lankalibrary.com.

George, S.; Sabelli, F. 1993. Faith and credit. The World Bank’s secular empire. Boulder, Colorado: WestviewPress.

Ghanawimala, Rev. Kiriella. 1942. “Sabaragamuwe parani liyavili” (Ancient documents of Sabaragamuwa).Colombo: Jinadasa Press.

Ghanawimala, Rev. Kiriella. 1967. Sabaragamu darshanaya (The philosophy of Sabaragamuwa). Ratnapura, SriLanka: Shasthrodaya Printers.

Gleick, P. H. 1996. Basic water requirements for human activities: Meeting basic needs. Water International 21(1996) 83-92.

GOC (Government of Ceylon). 1989. Administration report, Hambantota district. Colombo: Ceylon GovernmentPress.

GOC (Government of Ceylon). 1969. Uda Walawe Project: Plan of development and cost estimates. Unpublishedreport. 40 pp.

Green, C. 2003. Handbook of water economics: Principles and practice. Hoboken, NJ: Wiley.

Hassan, R. M. 2003. Economy-wide benefits from water-intensive industries in South Africa: Quasi-input-outputanalysis of the contribution of irrigation agriculture and cultivated plantations in the Crocodile river catchment.Development Southern Africa 20(2): 171-195.

Hazell, P.; Ramasamy, C.; Rajagopalan, V. 1991. An analysis of the indirect effects of agricultural growth on theregional economy. In Green revolution reconsidered: The impact of high yielding rice varieties in South India,eds. P. Hazell and C. Ramasamy. Baltimore, Maryland: Johns Hopkins University Press. Ch. 8, pp. 153-180.

Hill, H.; Tollefson, L. 1996. Institutional questions and social challenge. In Sustainability of irrigated agriculture,eds. L. S. Pereira, R. A. Fedds, J. R. Gilley and B. Lesaffre. Dordrecht/London: Kulwer Academic Publisher.

Hirschman, A. 1967. Development projects observed. Washington, D.C.: The Brookings Institution.

Howe, J. A.; Dixon, J. A. 1993. Inefficiencies in water project design and operation in the third world: An economicperspective. Water Resources Research 29(7):1889-1894. (Also published in Water resources and economicdevelopment, ed. R. M. Saleth. 2002. Cheltenham, UK: Edward Elgar Publishing. Pp. 339-344.)

HTS (Hunting Technical Services). 1968a. The Uda Walawe report, Vol II, The Outline Development Plan. August1968. Hunting Technical Services Ltd.

HTS (Hunting Technical Services). 1968b. The Uda Walawe report: Volume I, Summary. Report for: Great Britain.Ministry of Overseas Development; Government of Ceylon. River Valleys Development Board. HuntingTechnical Services Ltd. Unpublished report. 64 pp.

63

Hussain, I.; Marikar, F.; Thrikawela, S.; Shinkai, N.; Sawada, Y.; Aoki, M. 2002. Impact assessment of irrigationinfrastructure development on poverty alleviation: A case study from Sri Lanka. Tokyo: Japan Bank forInternational Cooperation (JBIC); Colombo: International Water Management Institute (IWMI). JBIC InstituteResearch Paper No. 19. vi + 216 pp.

Hussain, I.; Hanjra, M.; Thrikawela, S.; Wijeratne, D. 2003. Impact of irrigation infrastructure development ondynamics of income and poverty: Econometric evidence using panel data from Sri Lanka. Tokyo: Japan Bankfor International Cooperation (JBIC); Colombo: International Water Management Institute (IWMI). vi + 269 pp.

Hussain, I.; Thrikawela, S.; Regassa, N.; Wijeratne, D.; Shinkai, N. 2004. Impact of irrigation on chronic andtransient poverty alleviation: The case of Walawe system in Ruhuna basin, Sri Lanka. Draft IWMI ResearchReport. Colombo: International Water Management Institute.

ID (Irrigation Department). 2002. Potential water resources for development of Ruhunupura: Pre-feasibility study.Draft report, 3 vols. Colombo: Irrigation Department (Planning Branch).

IEC (International Engineering Company, Inc.). 1954. Economic report on the Walawe Ganga Project. Report tothe Government of Ceylon. San Francisco: IEC.

IIMI (International Irrigation Management Institute). 1990a. Final report of the Technical Assistance Study (TA 846SRI), Irrigation Management and Crop Diversification (Sri Lanka). Colombo: IIMI.

IIMI. 1990b. Irrigation management and crop diversification (Sri Lanka). Volume 1- Synthesis of findings andrecommendations Kirindi Oya and Uda Walawe Projects. Volume 2 - Kirindi Oya Project. Volume 3 - UdaWalawe Project. Colombo: IIMI.

Ingram, Helen. 1971. Patterns of politics in water resources development. Natural Resources Journal 11:102-118.

Ingram, Helen. 1997. Water and equity: A direction for research. Common Property Resource Digest, No.43,September 1997.

Jayatissa, L. P.; Guero, M. C.; Hettiarachchi, S.; Koedam, N. 2002. Changes in vegetation cover and socio-economic transitions in a coastal lagoon (Kalametiya, Sri Lanka), as observed by teledetection and groundtruthing, can be attributed to an upstream irrigation scheme. Environment, Development and Sustainability 4:167–183.

JICA (Japan International Cooperation Agency). 1993. Feasibility study on Walawe Irrigation Upgrading andExtension Project. Prepared for the Democratic Socialist Republic of Sri Lanka, Ministry of Lands, Irrigationand Mahaweli Development. 3 vols.

Kopp, R. J.; Krupnick, A. J.; Toman, M. 1997. Cost-benefit analysis and regulatory reform: An assessment of thescience and the art. Washington, D.C.: Resources for the Future.

Kothare, V. D. 1961. Tentative development plan for the Walawe Ganga Basin. Preliminary report. 2 vols.

Lagman, Anneli; Aylward, Bruce. 2000. Survey of multilateral bank practice on financial and economic analysisof large dams. World Commission on Dams, Contributing Paper for Thematic Review III.1. http://www.dams.org/kbase/thematic/tr31.htm.

Lohmann, Larry. 1997. Cost-benefit analysis: Whose interest, whose rationality? Presentation to the Yale UniversityProgram in Agrarian Studies, 1997. New Haven, Connecticut, USA: Yale University.

Marshall, H. 1965. Politics and efficiency in water development. Water Research 294.

Meijer, K. 2000. Impacts of concrete lining of irrigation canals, Uda Walawe, Sri Lanka. M. Sc. thesis, Universityof Twente, The Netherlands. 65 pp. + annexes.

Mendis, D. L. O. 1967. Planning by network for construction of Uda Walawe headworks. Transactions of theInstitution of Engineers, Ceylon, pp. 23-94.

MMP (M. Macdonald and Partners). 1986. Walawe irrigation improvement project: Inception report. Cambridge,UK: Sir M. Macdonald & Partners Ltd. 97pp.

MMP (M. Macdonald and Partners). 1988. Cropping pattern for planning the operation of the right bank. WaterManagement Technical Note No 4. Cambridge, UK: Sir M. Macdonald & Partners Ltd.

64

MMP (M. Macdonald and Partners). 1992. Walawe Irrigation Improvement Project: Project Office Handbook.Cambridge, UK: Sir M. Macdonald & Partners Ltd.

Molle, François. 2004. Defining water rights: By prescription or negotiation? Water Policy 6(3): 207-227.

Molle, François; Jayakody, Priyantha; de Silva, Shyamalie. 2003. Anicut systems in Sri Lanka: The case of theUpper Walawe river basin. Working Paper No.61. Colombo: International Water Management Institute.

Narada, Rev. Vekiliye. 1992. History of Ratnapura. In Sabaragamu Udana 2, ed. Bandula Liyanage. Ratnapura,Sri Lanka: Sabaragamuwa Provincial Council.

Nijman, C. 1991. Irrigation management processes and conditions: A case study of Sri Lanka’s Walawe IrrigationImprovement Project. Colombo: International Irrigation Management Institute. 277 pp.

Nippon Koei; MASL (Mahaweli Authority of Sri Lanka). 1995. Review report on feasibility study for Walawe LeftBank Irrigation Upgrading and Extension Project: Volume I. Main report. Tokyo: Nippon Koei Co., Ltd.Unpublished report.

Obeyesekere, Gananath. 1984. The cult of the Goddess Pattini. Chicago and London: The University of ChicagoPress.

Perera, L. R. 2003. Assessment of multiple use of water resources: Inland fisheries in the Walawe basin. Colombo:International Water Management Institute. Unpublished draft.

PSC (Photographic Survey Corporation); SGC (Surveyor General of Ceylon). 1960. A report on a reconnaissancesurvey of the resources of the Walawe Ganga Basin, Ceylon. Colombo: Government Press. xiv + 359 pp.

Porter, Theodore. 1994. Objectivity as standardization: The rhetoric of impersonality in measurement, statistics,and cost-benefit analysis. Annals of Scholarship 9(1992): 19-59.

Porter, Theodore. 1995. Trust in numbers: The pursuit of objectivity in science and public life. Princeton: PrincetonUniversity Press.

Powell, R. A.; Jenson, R. C.; Gibson, A. L. 1985. The economic impact of irrigated agriculture in NSW. NSW,Australia: New South Wales Irrigators’ Council.

PRC (Engineering Consultants). 1982. Termination report: Water management, Uda Walawe irrigation project.

Renault, D.; Hemakumara, M.; Molden, D. 2000. Importance of water consumption by perennial vegetation inirrigated areas of the humid tropics: Evidence from Sri Lanka. Agricultural Water Management 46(3): 215-230.

Renwick, Mary E. 2001. Valuing water in irrigated agriculture and reservoirs fisheries: A multiple-use irrigationsystem in Sri Lanka. Research Report 51. Colombo: International Water Management Institute.

Roberts, J. 2004. Thirty-five years later in development assistance: Have we moved on, or just performed aminuet? Development Policy Review 22(5): 483-495.

Samad, M.; Molle, F.; de Silva, S. Forthcoming. Toing and froing: The politics of water pricing in Sri Lanka. InIrrigation water pricing policy in context: Exploring the gap between theory and practice, eds. François Molle,Jeremy Berkoff and Randy Barker. IWMI/CABI: Comprehensive Assessment of Water Management inAgriculture.

Senaratna Sellamuttu, S.; Clemett, A. 2003. Avoiding conflict between fishers and farmers: A study of potentialsocial conflict over scarce water resources in Kalametiya. Proceedings of the Fourth National Symposium onPoverty Research in Sri Lanka, organised by the Centre for Poverty Analysis (CEPA), the Programme toImprove Capacities for Poverty Research (IMCAP) and the Sri Lanka Association for the Advancement ofScience (SLAAS).

SOGREAH Consulting Engineers. 1984. Walawe Irrigation Rehabilitation & Improvement Project: Draft final report.Reports for ADB and Sri Lanka. Mahaweli Authority of Sri Lanka. Vols. 2-4; 279, 232 and 12 pages.

Somatilake, H. S. 2002. Water for energy. In World Water Assessment Programme. Sri Lanka case study. Ruhunabasins. Proceedings of a workshop held at Koggala Beach Hotel, Sri Lanka, 7-9 April 2002, eds. K. A. U. S.Imbulana, P. Droogers and I. W. Makin. Colombo, Sri Lanka: International Water Management Institute. Pp.129-141.

65

Steele, T. 1867. Report on the Hambantota district. Administration reports, 1867. Colombo: Ceylon GovernmentPress.

Visvalingam, A. C. 1986. Lessons for the Mahaweli from the Walawe Experience. Mimeo.

WCD (World Commission on Dams). 2000. Dams and development: A new framework for decision-making.London: Earthscan.

Wickremaratne, M. L. H. 1995. The rationale of the accelerated Mahaweli Programme. In The blurring of a vision—The Mahaweli: Its social, economic and political implications, eds. H. P. Muller and S. T. Hettige. Ratmalana,Sri Lanka: Sarvodaya. Pp. 24-70.

Williams, A. 1972. Cost Benefit Analysis: Bastard science? And/or insidious poison in the body politick? Journalof Public Economics 1, 2, 199-226.

Wolf, J. W. 1983. Walawe irrigation system.

World Bank. 1996. World Bank’s experience with large dams: A preliminary review of impacts. Washington, D.C.:Operations Evaluation Department, World Bank.

66

67

Abbreviations and Acronyms

ADB Asian Development BankCBA Cost-benefit analysisCEB Ceylon Electricity BoardCECB Central Engineering Consultancy BureauECI Engineering Consultants, Inc.EIRR Economic internal rate of returnFO farmer organizationGDP Gross Domestic ProductGWh gigawatt-hourha hectareIEC International Engineering Company, Inc.JICA Japan International Cooperation AgencykWh kilowatt-hourLB left bank of the Walawe riverLBMC Left Bank Main CanalLHG low-humic gley soilsMASL Mahaweli Authority of Sri LankaMC main canalMEA Mahaweli Economic AgencyMMP Sir M. MacDonald and PartnersOFC other field crops (field crops other than rice)O&M operation and maintenanceRB right bank of the Walawe riverRBE reddish brown earthRBMC Right Bank Main CanalRs Sri Lankan rupeeRVDB River Valleys Development Boardt metric tonUWIRP Uda Walawe Irrigation and Resettlement ProjectWCD World Commission on DamsWIIP Walawe Irrigation and Improvement Project

68

Units and Currencies

1.0 hectare (ha) = 2.471 acres1.0 acre = 0.405 hectare1.0 metric ton (t) = 1,000 kg

Value of US$1.00 in Sri Lankan Rupees (Rs)

1963 US$1.00 = Rs 4.761967 US$1.00 = Rs 4.861968, 1969 and 1970 US$1.00 = Rs 5.951971 US$1.00 = Rs 5.941972 US$1.00 = Rs 5.971983 US$1.00 = Rs 25.531993 US$1.00 = Rs 48.351995 US$1.00 = Rs 51.251999 US$1.00 = Rs 70.592000 US$1.00 = Rs 78.082002 US$1.00 = Rs 95.402003 US$1.00 = Rs 96.912004 (January) US$1.00 = Rs 97.50

Glossary

anicut diversion weirara non-permanent river or tributaryganga river (large)maha cultivation season from October to Marchoya river or tributarypurana indigenouswewa reservoiryala cultivation season from April to September

75. Prospects for Adopting System of Rice Intensification in Sri Lanka: ASocioeconomic Assessment. Regassa E. Namara, Parakrama Weligamage andRandolph Barker. 2004.

76. Small Dams and Social Capital in Yemen: How Assistance Strategies Affect LocalInvestments and Institutions. Douglas L. Vermillion and Said Al-Shaybani. 2004.

77. Simulating the Hydrology of Small Coastal Ecosystems in Conditions of LimitedData. V. U. Smakhtin, S. C. Piyankarage, P. Stanzel and E. Boelee. 2004.

78. Irrigation Kuznets Curve, Governance and Dynamics of Irrigation Development:A Global Cross-Country Analysis from 1972 to 1991. Madhusudan Bhattarai.2004.

79. Strategic Analysis of Water Institutions in India: Application of a New ResearchParadigm. R. Maria Saleth. 2004.

80. Robbing Yadullah’s Water to Irrigate Saeid’s Garden: Hydrology and Water Rightsin a Village of Central Iran. François Molle, Alireza Mamanpoush and MokhtarMiranzadeh. 2004.

81. Inadequacies in the Water Reforms in the Kyrgyz Republic: An InstitutionalAnalysis. Mehmood Ul Hassan, Ralf Starkloff and Nargiza Nizamedinkhodjaeva.2004.

82. Valuing Nutrients in Soil and Water: Concepts and Techniques with Examplesfrom IWMI Studies. Pay Drechsel, Mark Giordano and Lucy Gyiele. 2004.

83. Spatial Variation in Water Supply and Demand Across River Basins of India.Upali A. Amarasinghe, Bharat R. Sharma, Noel Aloysius, Christopher Scott,Vladimir Smakhtin and Charlotte de Fraiture. 2004.

84. An Assessment of Small-scale Users’ Inclusion in Large-scale Water UserAssociations of South Africa. Nicolas Faysse. 2004.

85. The Use of Remote-Sensing Data for Drought Assessment and Monitoring inSouthwest Asia. P. S. Thenkabail, M. S. D. N. Gamage and V. U. Smakhtin, 2004.

86. Strategies for the Management of Conjuctive use of Surface Water andGroundwater Resources in Semi-arid Areas: A Case Study from Pakistan. AsadSarwar Qureshi, Hugh Turral and Ilyas Masih. 2004.

87. Economics and Politics of Water Resources Development: Uda Walawe IrrigationProject, Sri Lanka. François Molle, and Mary Renwick. 2005.

Research Reports

François Molle and Mary Renwick

IWMI is a Future Harvest Centersupported by the CGIAR

Economics and Politics ofWater Resources DevelopmentUda Walawe Irrigation Project,Sri Lanka

87

RESEARCHR E P O R T

I n t e r n a t i o n a lWater ManagementI n s t i t u t e

SM

IWMI is a Future Harvest Centersupported by the CGIAR

Postal Address:P O Box 2075 ColomboSri Lanka

Location:127, Sunil Mawatha Pelawatta Battaramulla Sri Lanka

Tel:+94-11-2787404

Fax:+94-11-2786854

E-mail:iwmi@cgiar.org

Website:http://www.iwmi.org

I n t e r n a t i o n a lWater ManagementI n s t i t u t e

ISSN 1026-0862ISBN 92-9090-584-0