An Assessment of the Impact of Participatory Irrigation Management in Sri Lanka

This article was downloaded by: [The University of Manchester Library]On: 25 November 2014, At: 05:03Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number:1072954 Registered office: Mortimer House, 37-41 Mortimer Street,London W1T 3JH, UK

International Journalof Water ResourcesDevelopmentPublication details, including instructions forauthors and subscription information:http://www.tandfonline.com/loi/cijw20

An Assessment of the Impactof Participatory IrrigationManagement in Sri LankaM. Samad & Douglas VermillionPublished online: 05 Aug 2010.

To cite this article: M. Samad & Douglas Vermillion (1999) An Assessment of theImpact of Participatory Irrigation Management in Sri Lanka, International Journalof Water Resources Development, 15:1-2, 219-240, DOI: 10.1080/07900629949023

To link to this article: http://dx.doi.org/10.1080/07900629949023

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of allthe information (the “Content”) contained in the publications on ourplatform. However, Taylor & Francis, our agents, and our licensorsmake no representations or warranties whatsoever as to the accuracy,completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views ofthe authors, and are not the views of or endorsed by Taylor & Francis.The accuracy of the Content should not be relied upon and should beindependently verified with primary sources of information. Taylor andFrancis shall not be liable for any losses, actions, claims, proceedings,demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, inrelation to or arising out of the use of the Content.

This article may be used for research, teaching, and private studypurposes. Any substantial or systematic reproduction, redistribution,

http://www.tandfonline.com/loi/cijw20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/07900629949023

http://dx.doi.org/10.1080/07900629949023

reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of accessand use can be found at http://www.tandfonline.com/page/terms-and-conditions

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Water Resources Development, Vol. 15, Nos. 1/2, 219± 240, 1999

An Assessment of the Impact of Participatory

Irrigation Management in Sri Lanka

M. SAMAD & DOUGLAS VERMILLION

International Water Management Institute (IWMI), PO Box 2075, Colombo, Sri Lanka

ABSTRACT This paper presents the results of a study on the effects of participatoryirrigation management in Sri Lanka. The study is based on the application of a method-ology developed by the International Water Management Institute (IWMI) to assess theimpact of irrigation management transfer on the performance of irrigated agriculture.Performance is measured from several perspectives: ® nancial, irrigation management,and agricultural productivity levels. Piecewise linear regression models are ® tted tocompare trends in performance during the ® ve-year period before transfer and ® ve yearsafter. The results show that there has been a signi® cant drop in government’s recurrentexpenditure for irrigation over time. The decline is not con® ned to schemes whereparticipatorymanagementhad been introduced but to other schemes as well. The cost ofirrigation to farmers has remained the same before and after transfer. The analysisreveals that irrigation management transfer alone did not bring about signi® cantimprovements in the quality of irrigation or agricultural productivity levels. However,in schemes where both management transfer and physical rehabilitation had occurred,signi® cant improvements in agricultural productivity were noted.

Introduction

The worldwide interest in, and support for, transferring the management ofirrigation schemes from public agencies to water user groups and other non-governmental organizations, have prompted considerable research on variousaspects of irrigation management reforms and their impacts. This has resulted ina wide range of opinions on the subject. 1 The need for strong political supportfor the programme, clear policy direction, alternate strategies for irrigationmanagement, well-de® ned water rights, clarity about the process of creatingfarmer organizations, and conditions for successful management transfer aresome of the major issues discussed in the literature (Johnson III et al., 1995; Geijeret al., 1996; Meinzen-Dick et al., 1997; Vermillion, 1997). Yet, there is littlesystematic, comparative evidence to date on the impact of reforms on irrigationmanagement performance, government ® nances and the farming community(Vermillion, 1997). With some exceptions (e.g. Svendsen & Vermillion, 1994;Vermillion & Garces-Restrepo, 1996) most studies that deal with impacts ofirrigation management reform refer to short-term and immediate results.

To support systematic documentation of international experience withirrigation management reforms and their impact on the performance of irrigatedagriculture, the International Water Management Institute (IWMI) developed

0790-0627/99/010219± 22 $7.00 Ó1999 Taylor & Francis Ltd

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14

220 M. Samad & D. Vermillion

a standard methodology to assess and compare irrigation management trans-fers in a variety of settings.2 This paper reports the results of applicationof the methodology to assess the impact of irrigation management reforms in SriLanka. The study was designed and implemented with two objectives in mind:® rst, to ® eld-test the proposed methodology and, second, to determine whateffects management reforms have had on the performance of irrigation manage-ment and irrigated agriculture in Sri Lanka.

The paper begins with an overview of the irrigation management reformprogramme in Sri Lanka. We then outline the methodology. The section thatfollows presents the results of applying the methodology in Sri Lanka. The ® nalsection reviews the methodology and concludes with some general comments onthe Sri Lankan case study.

Irrigation Management Transfer in Sri Lanka

In 1988, following a decade of ® eld experiments, the Government of Sri Lankaformally adopted a policy of transferring full responsibility for the operation andmaintenance (O&M) of irrigation facilities below the distributary canal head ofmedium and major schemes to farmer organizations. The government retainedits control of the headworks and the main canal system. The programme,entitled `Participatory Irrigation System Management’, was implemented in alarge number of irrigation schemes in the country.3 Its main objectives were to:

· relieve the ® nancial burden on the government of funding recurrentexpenditures for irrigation;

· improve the maintenance of irrigation facilities and the irrigation service;

· enhance the productivity of irrigated land and water;

· promote a sprit of self-reliance among farmers in irrigation schemes(Abeywickrema, 1986; Brewer, 1994).

The main function of farmer organizations (FOs) is to deal with irrigationmatters, but statutory provisions permit FOs the right to implement agriculturalprogrammes for their area, market farm produce and distribute productioninputs (GOSL, 1991). Owner cultivators and occupiers of land in the designatedarea are eligible for membership in FOs. In most localities, cultivating a plot ofland irrigated by a particular distributary canal, regardless of the tenure pattern,is a suf® cient quali® cation for membership. Membership is conferred on onlyone person per plot of land. FOs can become legal if they register with theDepartment of Agrarian Services and the registration is approved by theCommissioner. Once they are registered, FOs receive authority under theIrrigation Ordinance to formulate rules on maintenance, conservation and man-agement of irrigation infrastructure under their jurisdiction, to deviseprocedures for distributing water within the area under their command andto impose and levy fees to recover the costs of operation and maintenance(IIMI/HKARTI, 1997).

Overview of Methodology

Objectives and Hypotheses

The general objective of the impact assessment methodology is to determine theeffects irrigation management transfer (IMT) has on the performance of irri-

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14

Impact of Participatory Irrigation Management 221

gation management and irrigated agriculture. In this study, performance ismeasured from several perspectives: cost to government and to farmers ofoperating and maintaining irrigation systems; quality of the irrigation service;and agricultural productivity levels. The principal hypotheses tested are:

· IMT leads to a reduction in government expenditure for operation andmaintenance.

· Where farmers did not have to pay for most of the cost for irrigation beforetransfer, IMT will lead to an increase in the cost of irrigation to farmers.

· IMT will lead to improvements in the quality of irrigation services to farmers.

· IMT will result in higher agricultural productivity per unit of land and water.

Research Design

The proposed methodology bases key evidence about impacts on differences inperformance before and after IMT, and with and without IMT. The rationale forthis approach is:

· Variability among irrigation schemes (which also causes differences in per-formance) is controlled through comparing performance before and afterturnover in the same irrigation system.

· Where governments implement IMT selectively, differences in performancemay be introduced between schemes in the selection process itself. This couldmake `with’ and `without’ comparisons misleading.

· Comparison of performance of systems that have or have not been transferredis to control against the possible effects of other time-related factors such aseconomic trends, which could cause generalized changes in performance overtime.

The assessment is based on two sets of data. The ® rst set is from an intensivestudy of two irrigation schemes (Nachchaduwa and Hakwatuna Oya) and thesecond from an extensive survey of 50 schemes. The intensive componentconsists of a rigorous and detailed analysis of changes in performance in the twoirrigation systems before and after management turnover, and validates a set ofperformance indicators that could be used in the extensive component.

The analysis is based on both time-series and cross-sectional data. Time-seriesdata covering a period of 10 years (® ve years before and ® ve years after transfer)were collected to measure changes in performance over time at the scheme level.Information was collected on a number of performance measures, including® nance, O&M, agricultural productivity and economic productivity. Recordsmaintained by various government agencies were the main source ofinformation.

Cross-sectional data were collected through a questionnaire survey of asample of farmers from the two schemes selected for the intensive study tomeasure impacts at the farm level. The objective of the survey was to obtaininformation from farmers about their perceptions of changes in selected per-formance attributes before and after turnover. Ninety farmers were selectedas a sample from each of the two schemes by strati® ed random sampling. Thesampling unit was a selected parcel of land in the irrigation scheme.The agricultural year immediately preceding the year of transfer was taken asthe `before’ period (pre -IMT reference year) and the latest complete agricultural

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Figure 1. Strati® cation of sample schemes: the extensive component.

year for the àfter’ situation (post-IMT reference year), provided it is at least fourto ® ve years after the year of transfer.

Key informant interviews were conducted primarily among the farmer organi-zation leaders to obtain information about post-IMT changes in the organization,operations, maintenance and ® nancing of irrigation management.

Two common intervening variables that could confound the assessment of theimpacts of management transfer are rehabilitation and rainfall. Where rehabili-tation occurs along with transfer, it is nearly impossible to distinguish betweenthe effects of transfer and rehabilitation. Similarly, abnormal rainfall in thechosen reference years could affect agricultural production and mask the effectsof IMT.

To control the effects of rehabilitation, schemes selected for the extensivesurvey were ® rst strati® ed into two groups: rehabilitated and unrehabilitated(groups 1 and 2).4 Each group is subdivided into IMT and non-IMT groups, asillustrated in Figure 1, and analysed separately. The confounding effects due torehabilitation would be the same in groups 1 and 2. It is assumed thatdifferences in performance between these two groups would be due to IMT andstochastic factors. Similarly, differences in performance in the schemes in theunrehabilitated schemes (groups 3 and 4) are assumed to be due to IMT andother factors. For the intensive component, two schemes were selected: one thathad been rehabilitated and turned over, and the other turned over but notrehabilitated.

As for rainfall, the reference year was selected only if annual rainfall did notdiffer by more than 25% (above or below) from the long-term average annualrainfall for the area. If this condition was not met, the `normal’ rainfall yearclosest to the reference year was selected.

Performance Indicators

In this study, performance is measured in both qualitative and quantitativeterms. Qualitative assessment is based on farmer perceptions of changes in

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Table 1. Performance indicators and the level at which they are measured.

Scheme Farm

Performance indicators level level Data source

Financial performance indicatorsAnnual operations and maintenance cost per hectare to X Secondary data

governmentIrrigation cash costs per hectare to farmers X Farmer survey

Value of family labour contributions for canal maintenance X Farmer surveyTotal irrigation costs per hectare to farmers X Farmer survey

Farmer perceptions of changes in irrigation costs X Farmer surveyOperational performance indicators

Relative irrigation water supply X Secondary dataRelative total water supply X Secondary data

Farmer perceptions about adequacy, timelines and equity X Farmer surveyof water supply

Agricultural performance indicatorsAnnual cropping intensity X Secondary data

Yield of major crop by season X Secondary dataFarmer perceptions of changes in rice yield X Farmer survey

Economic performance indicatorsStandardized gross value of output per hectare10 X Estimated

Standardized gross value of output per unit of water X Estimateddiverted

selected performance indicators before and after turnover. Quantitative analysisis based primarily on the standard set of indicators formulated by IIMI to assessthe performance of irrigation schemes (see Molden et al., 1998). The indicatorsrelate to ® nancial, agricultural, hydrological and economic performances acrossirrigation systems. The indicators require a limited amount of data, and theircomputation is straightforward. The set of indicators used to assess performancein this study and the level at which they were measured are shown in Table 1.

Analysing Trends in Performance

A major aim of the analysis was to determine the annual trends in selectedperformance indicators during the period 1985± 95, which covered ® ve yearsbefore turnover (1985± 90) and ® ve years after (1991± 95). A set of regressionequations was estimated based on data obtained from the 50 schemes selectedfor the extensive survey. The following performance indicators are used asdependent variables:

· government expenditure for O&M from 1985 to 1995;

· rice yields (yield/ha) 1985± 95;

· cropping intensity(CI), 1985± 95;

· standardized gross value of output per hectare (GVO/ha), 1985± 95;

· GVO per cubic metre of water diverted (GVO/m3), 1985± 95.

Cropping intensities, rice yields and GVOs per unit of land and water wereadjusted for seasonal and locational variations and analysed as annual values.

Piecewise linear regression models were ® tted to analyse trends in perform-ance in the two time periods: period before IMT (1985± 90) and the period after(1991± 95). The aim was to determine whether a performance indicator shows a

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


particular linear trend from 1985 up to 1990, the year of transfer, but follows adifferent trend thereafter. This involves testing whether there is a statisticallysigni® cant difference in the slopes of the regression lines for the two timeperiods.

A common set of explanatory variables was speci® ed in all equations. Theseinclude a time variable (T) to capture the effect of time (in years) on thedependent variable, and a dummy variable (D1) to indicate the periods beforeand after turnover.

The basic regression equation estimated was as follows:

Yt 5 b 0 1 b 1T 1 b 2(T 2 T*)D1 1 e (1)

Where: Yt 5 Performance measure (O&M costs, yield/ha, CI, GVO/ha,GVO/m3) in year t,

T 5 time in years (1985 ¼ ¼ ¼ ¼ 1995)T* 5 threshold period ( i.e., 1990, the year of transfer)D1 5 1 if T . 19900 if T , 5 1990e 5 random error

b 0 ¼ ¼ ¼ b 2 are parameters to be estimated

Assuming E(e) 5 0, parameter b 1 gives the slope of the regression line or thetrend during the pre-IMT period (1985± 90) and ( b 1 1 b 2 ) the trend in thepost-IMT period (1991± 95). A test of the hypothesis that there is a change in thetrend between the two periods is conducted by noting the statistical signi® canceof the estimated differential slope coef® cient b 2.

Results

Impact on Government Expenditures for O&M

The main interest of the government in transferring management of irrigation atthe subsystem level to farmer organizations was to reduce its own costs forirrigation. This section examines the trend in government expenditure for O&Mduring the period 1985± 95. The hypothesis advanced is that with the transfer ofO&M responsibilities to farmer organizations, the government’s recurrent costfor irrigation will be lower in the transferred schemes than in the non-transferred schemes.

The regression model (equation 1) was used to analyse trends in governmentinvestment in O&M (estimated in terms of 1995 constant US$/ha) during theperiod 1985± 95. The estimated regression coef® cients are given in Table 2.Figures 2a to 2d give the trend in government’s O&M expenditure in the fourgroups of schemes during the period 1985± 95. The results indicate that, in allfour groups, there is a statistically signi® cant declining trend ( 2 b 1) in govern-ment expenditure for O&M during the pre-IMT period. In the post-IMT period,there is a slight reversal in the trend ( 1 b 2) in all categories of schemes exceptthe no-IMT and rehabilitated group. However, the change trend is not statisti-cally signi® cant.

The results indicate that there has been a decline in recurrent governmentcosts for irrigation during the period 1985± 95 across all categories of schemesirrespective of whether IMT programmes have been introduced or not. The

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Table 2. Estimated regression coef® cients for trends in government expenditurefor O&M, 1985± 95.

Regression coef® cients

With IMT WithoutWith IMT Without and IMT and

and IMT and with without withoutVariable description rehabilitation rehabilitation rehabilitation rehabilitation

Constant ( b 0) 87.04 80.11 86.80 96.72

Trend in government’s O&M 2 0.879 2 0.794 2 0.885 2 0.983cost/ha in the pre-IMT period ( b 1) ( 2 5.684)* ( 2 4.269)* ( 2 8.271)* ( 2 5.023)*

The change in trend in 0.424 2 0.2867 0.346 0.428

government’s O&M costs in (1.373) ( 2 0.761) (1.603) (1.078)the post-IMT period ( b 2)

Adj. R2 0.534 0.4439 0.487 0.390

F. stat 43.42* 52.18* 102.47* 37.265*

Notes: *signi® cant at or less than 10% level. Figures in parentheses are t values.

results do not fully support the hypothesis that IMT leads to a reduction ingovernment expenditure for O&M.

Impact on Cost of Irrigation to Farmers

In Sri Lanka, irrigation water has traditionally been supplied free to farmers.Attempts made by the government in the past to levy a fee on farmers werelargely unsuccessful. The `costs’ of irrigation to farmers are primarily thecontribution of voluntary labour for canal maintenance and, in some instances,the payment made in kind to the person (yaya palaka) employed by the agencyto oversee the distribution of irrigation water. With the introduction of participa-tory management, the government expected the farmer organizations to recoverthe cost of O&M from farmers (Ratnayake, 1995). This section examines theimplications of participatory management for the cost of irrigation to farmers.The hypothesis advanced is that, as farmers did not pay for most of the cost ofirrigation before transfer, the adoption of participatory management will in-crease cash costs and labour contribution for irrigation.

The analysis is based on data obtained from a sample survey of farmers inNachchaduwa and Hakwatuna Oya schemes. Three kinds of irrigation costswere assessed: cash payments, payments made in kind, and the number ofperson-days of family labour contributed for canal maintenance. Farmers werealso asked about any ùnof® cial’ payments made to obtain irrigation water.Table 3 gives the annual irrigation costs reported by farmers in the post-transferreference year (1994± 95). The total cost of irrigation is about the same (approxi-mately US$15± 16/ha) for both schemes. Data show that after transfer farmersgenerally contributed more in the form of unpaid family labour (56% inNachchaduwa and 58% in Hakwatuna Oya) than in cash or kind for canalmaintenance.

In the survey, farmers were asked to compare irrigation costs in the post-transfer reference year with costs of irrigation before transfer. About 90% of

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Figure 2. (a) Trend in government O&M expenditure in schemes with IMT andrehabilitation; (b) trend in government O&M expenditure in schemes without IMTand with rehabilitation;

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Figure 2, continued. (c) trend in government O&M expenditure in schemes withIMT and without rehabilitation; (d) trend in government O&M expenditure in

schemes without IMT and without rehabilitation.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


farmers in both schemes claimed that there was no cash fee on irrigation beforeturnover. After the transfer of O&M functions to FOs, some organizationscharged a modest fee (Rs 50/acre/season or US$2.5/ha) for canal maintenance.The survey results showed that only a minority of farmers (23% in HakwatunaOya and 16% in Nachchaduwa) paid the maintenance fee.

Figure 3 gives farmers’ perception of changes in irrigation cost componentsbefore and after turnover. The majority of farmers in both schemes claimed thatpayments in kind and unpaid family labour contributions for canal maintenancehad remained almost the same before and after turnover. In both locationswell-de® ned procedures for cost recovery have not been established as yet. Datafrom the two schemes do not provide suf® cient evidence to suggest an increasein the cost of irrigation to farmers following the introduction of participatorymanagement.

Quality of Irrigation Service

It has been argued that, as farmers have a vested interest in the irrigationservice, involving them directly in irrigation management would lead to im-provements in the quality of the service. This section examines whether theintroduction of participatory irrigation management resulted in an improvementin the quality of irrigation service. The analysis is based on data obtained fromNachchaduwa and Hakwatuna Oya schemes. Changes in the quality of irri-gation service were assessed by computing relative irrigation supply (RIS) andrelative water supply (RWS), and farmer perceptions about changes inthe adequacy, timeliness and fairness of water distribution, and incidence ofirrigation-related con¯ icts among farmers before and after turnover.

RIS is the ratio of irrigation supply to demand and can be considered as anindicator of ef® ciency and adequacy in targeting water delivery at the schemelevel. RWS is the ratio of total water supply (irrigation plus rainfall) to demand.RIS and RWS were computed for both wet (® rst) and dry (second) seasons fora 10-year period. The estimates were based on the norms used by the IrrigationDepartment for determining water demand for rice and other ® eld crops grownin the dry zone irrigation schemes.5 Figure 4 gives the trend of RIS and RWS forthe period 1985± 95. In both schemes there is no obvious change in RIS and RWSin the years before and after turnover. An exception is that in Nachchaduwathere appears to be excess irrigation in the wet seasons of 1994 and 1995. Thiswas due to the high rainfall experienced in these years and more water beingreleased into the canals.

Figure 5 displays farmer perceptions about the quality of irrigation servicebefore and after turnover. Most farmers in both schemes consider the watersupply to be adequate before and after turnover. However, in Nachchaduwaabout one-third of the farmers in the head-reach and about 25% in the middle-and tail-end areas reported that water supply in both seasons had worsenedafter turnover. Farmers attributed the worsening of water supply to the poorquality of work done during rehabilitation prior to management transfer. Theresponses of the majority of farmers in both schemes were similar with regardto the timeliness of water supply, fairness of distribution, and the frequency ofcon¯ icts over water distribution in that these had not changed signi® cantly aftertransfer. What was negative or positive before remained so afterwards.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Table 3. Annual irrigation costs to farmers after IMT (1994± 95)

Hakwatuna Oya

Cost component Unit Nachchaduwa scheme scheme

Cash cost per hectarea US$/ha 6.34 (36)b 6.58 (50)

Value of unpaid family labourcontributions for canal maintenance US$/ha 8.18 (67) 9.00 (74)

Total irrigation costc US$/ha 14.52 (47) 15.58 (54)

Source: Farm Survey (July and November, 1996).Notes: a Irrigation cash costs include cash payments plus the monetary value of payments made in kind;b ® gures in parentheses are the coef® cients of variation in percentage terms; ctotal irrigationcost 5 irrigation cash costs 1 monetary value of family labour.

Figure 3. Farmer perception about changes in irrigation costs.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Fig

ure

4.

Rel

ati

ve

irri

gati

on

sup

ply

and

rela

tive

wate

rsu

pp

ly1985

±95,

Nac

hch

ad

uw

aan

dH

ak

watu

na

Oy

asc

hem

es.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Fig

ure

5.F

arm

erp

erce

pti

on

sof

the

qu

ali

tyof

irri

gati

on

serv

ice,

bef

ore

and

aft

er

IMT

.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Impact on Agricultural Production

Although irrigation schemes contribute about two-thirds of the national riceoutput, there is growing concern about low cropping intensities and stagnationof rice yield in the schemes. Problems related to irrigation are considered to bea major reason for the stagnation of agriculture in the schemes (NationalDevelopment Council, 1996). If the shift of primary responsibility for waterdistribution to farmer organizations leads to an improvement in the quality ofirrigation service, one could expect cropping intensities to improve and farmersto use more inputs as a result of greater con® dence in the irrigation servicewhich, in turn, would lead to higher yields. This proposition is tested byexamining the trend in rice yields and cropping intensities in 50 schemes overa 10-year period (1985± 95). The analysis was done separately for rehabilitatedand unrehabilitated schemes with and without IMT.

Trends in Rice Yields

The trend in rice yield during the period 1985± 95 is estimated using equation 1.Table 4 shows the estimated coef® cients. Figures 6a to 6d give the yield trendsfor each group.

The results indicate that in the pre-IMT period, rice yields in the rehabilitatedschemes, irrespective of whether they were transferred or not, show a decliningtrend ( 2 b 1). The decline is statistically signi® cant in the schemes with IMT andrehabilitation. During the same period, yields in the unrehabilitated schemeshow a statistically signi® cant upward trend ( 1 b 1). In the post-IMT period,there is a statistically signi® cant upward shift in rice yields in the group showingthe effects of both rehabilitation and management transfer ( b 2 5 245.54). There isno statistically signi® cant change in trend in the schemes that had been rehabil-itated but not transferred and those that had been transferred but not rehabili-tated. In the post-IMT period, rice yields in the group without the two forms of

Table 4. Estimated regression coef® cients explaining trends in rice yield in theselected schemes, 1985± 95




Constant 12941 5163 2 1761.38 2 3558.15

Trend in rice yield in the 2 98.79 2 6.32 61.14 89.83

pre-IMT period ( b 1) ( 2 2.875)* ( 2 2.219) (2.338)* (3.088)*

The change in trend rice yield 245.54 2 0.70 2 52.09 2 93.66

in the post-IMT period ( b 2) (3.799)* ( 2 0.219) ( 2 1.06) ( 2 1.728)*

Adj. R2 0.113 0.01 0.038 0.076

F. stat 7.81* 0.124 5.18* 7.72*

Notes: *Signi® cant at or less than 10% con® dence level; ® gures in parentheses are t values.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Table 5. Estimated regression coef® cients explaining trends in cropping intensitiesin the selected schemes, 1985± 95




Constant 2 34.16 242.63 372.87 2 27.21

Trend in cropping intensities in the 1.797 2 1.356 2 2.49 1.57pre-IMT period ( b 1) (0.578) (0.551) ( 2 1.158) (0.496)

The change in trend in cropping 5.878 5.545 7.026 2 0.375

intensities in the post-IMT period (0.937) (1.133) (1.645) (0.058)( b 2)

Adj. R2 0.11 0.01 0.01 0.01

F. stat 4.31 1.041 1.511 0.424

Notes: Figures in parentheses are t values

intervention show a statistically signi® cant declining trend when compared withthe pre-IMT period. The conclusion that emerges from the analysis is that therehas been a signi® cant improvement in yield in the schemes that have undergoneboth management transfer and rehabilitation. There is no statistically signi® cantchange in yield trends in schemes with only one type of intervention, and thosewithout either of the two forms of intervention show a signi® cant decline inyield. These ® ndings are consistent with the results from the Gal Oya scheme inSri Lanka (Amerasinghe et al., 1998).6

Cropping Intensities7

The regression model 1 was used to analyse trends in cropping intensities in thedifferent groups of schemes. The estimated regression coef® cients are given inTable 5. The analysis indicates that there are no signi® cant differences in thetrends in cropping intensities in any of the four groups of schemes in the periodsbefore and after transfer.

Economic Returns per Unit of Land and Water

This section examines the value of agricultural production over a 10-year periodof ® ve years before and ® ve years after transfer. Gross values of output per unitof land and per cubic metre of water diverted were estimated. Rice is the majorcrop grown in the irrigation schemes in Sri Lanka. Although in recent yearsthere has been an increase in cultivation of non-rice crops, particularly in the dryseason, there is a lack of reliable data on the area and the yield of other cropsgrown in the schemes. Therefore, an estimate of the gross value of output perunit of land and water is based solely on the output of rice.

To permit international comparisons, the total value of the crop was standard-ized in terms of the international price of rice, and expressed in terms of constant1995 US dollars.8 The trends in the gross value of production were analysed

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Figure 6. (a) Trend in rice yield in schemes with IMT and rehabilitation; (b) trendin rice yield in schemes without IMT and with rehabilitation;

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Figure 6, continued. (c) trend in rice yield in schemes with IMT and withoutrehabilitation; (d) trend in rice yield in schemes without IMT and without

rehabilitation.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Table 6. Estimated regression coef® cients explaining trends in the productivity ofwater in the selected schemes, 1985± 95




Constant 0.181 0.135 Ð Ð

Trend in GVO/m3 in the pre-IMT 2 0.001 2 7.6512 Ð Ðperiod ( b 1) ( 2 1.323) 2 (0.400)

The change in trend GVO/m3 in 0.0033 0.0053 Ð Ð

the post-IMT period ( b 2) (1.710)** (1.693)**

Adj. R2 0.014 0.11 Ð Ð

F. stat 1.54 0.011* Ð Ð

Notes: Figures in parentheses are t values; * statistically signi® cant at the 10% level; ** statistically

signi® cant at 5% level.

using the regression model 1. As a standard price is used to value the output ofrice, the trend in the gross value of output per unit of land corresponds closelywith the trend in rice yields noted earlier, with schemes that have beentransferred and rehabilitated showing a signi® cant change in the gross value ofoutput in the post-IMT period compared with the pre-IMT period.9

Returns per Unit of Water

Returns per unit of water were estimated in terms of gross value of output perunit of water diverted. As most of the unrehabilitated schemes did not haveaccurate time-series data on irrigation releases, the analysis is con® ned to theschemes that had undergone rehabilitation. Table 6 gives the estimated re-gression coef® cients of the parameters used to estimate trends in the gross valueof output per unit of water diverted (GVO/m3). Figures 7a and 7b display thetrend in GVO/m3 during the period 1985± 95. The results indicate that there is adeclining trend in the productivity of water in the pre-IMT period in bothcategories of schemes but it is not statistically signi® cant. In the post-IMT periodthere is a signi® cant reversal in the declining trend irrespective of whether theschemes had been transferred or not. The results suggest that rehabilitationrather than IMT may be the major contributing factor for the improvements inthe productivity of water experienced in the post-IMT period.

Conclusions

The purpose of this study was to apply the methodology developed by IWMI toassess the impacts of irrigation management transfer. The methodology wasapplied to analyse the effects of Sri Lanka’s participatory irrigation managementprogramme on the performance of irrigation schemes. The results of the analysislead to the following conclusions regarding the impact of the participatoryirrigation management programme on the performance of irrigation schemes:

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Figure 7. (a) Trend in productivity of water in schemes with IMT andrehabilitation; (b) trend in productivity of water in schemes without IMT and with

rehabilitation.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


· There has been a substantial decline in government expenditure on irri-gation, beginning before transfer. The declining trend is not con® ned toschemes where IMT had occurred but is common to non-IMT schemes aswell.

· The reforms have not generated an appreciable increase in the costs ofirrigation to farmers. Farmers generally make fewer direct payments (in cashand kind), but contribute more labour for canal maintenance.

· Management transfer alone did not bring about signi® cant changes in thequality of irrigation services.

· Management transfer alone did not result in signi® cant improvements inagricultural production levels or in the gross value of agricultural productionper unit of land or per unit of water diverted. Neither did rehabilitationalone create signi® cant effects. However, in schemes where both manage-ment transfer and rehabilitation occurred, signi® cant effects on agriculturalproductivity levels and economic returns were observed.

Application of the impact assessment methodology to the irrigation systems inSri Lanka has shown that the combination of performance measures, comparedbefore and after transfer, and with and without the intervention, can yield acomprehensive picture of the impacts of management transfer. The selectedperformance measures cover ® nancial, hydrological, agricultural and economicaspects. The impact assessment model adopted for this study was found to beeffective in measuring the trends in the selected performance indicators. Adif® culty with the proposed methodology and type of analysis that has beenattempted here is the demand for reliable data. In most irrigation systems indeveloping countries, time-series data, particularly relating to irrigationreleases and crop yields, may not be readily available. Even where somedata are available, there may be a lack of correspondence of information fordifferent schemes and time periods. In such situations a comprehensive `before’and àfter’ comparison of performance may not be feasible. An alternativewould be to limit the analysis to a `with’ and `without’ analysis of currentperformance.

Acknowledgement

The authors would like to thank Upali Amarasinghe for assisting with thestatistical analysis.

Notes

1. At the ® rst major international conference on irrigation management transfer held in Wuhan, PRChina in September 1994, over 100 papers were submitted on wide range of issues relating to

irrigation management reforms.2. The proposed methodology is described in detail in Vermillion et al. (1996).

3. It has been estimated that participatory management has been introduced in about 85% of the200 schemes included under these three major government-sponsored programmes: the Inte-

grated Management of Irrigation Schemes (INMAS), Management of Irrigation Systems(MANIS), and the Mahaweli Development Project (IIMI/HKARTI, 1997).

4. For sampling purposes, we de® ne rehabilitation herein as being restoration or improvementof irrigation scheme infrastructure (i.e. canals and water control structures) wherein annual

expenditure levels exceed the average annual O&M budget by at least 50%. Where expenditurein a scheme was less than this, it was not considered as rehabilitation.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


5. According to the Irrigation Department, water demand for rice and other ® eld crops for majorirrigation schemes is 15 000 m3/ha in the wet season (maha) and 17 000 m3/ha in the dry season

(yala) (Irrigation Department, personal communication). RWS was estimated on the basis of 60%effective rainfall in the wet season and 80% in the dry season.

6. See also the article by Murray-Rust et al., this issue.7. Cropping intensity 5

Area cultivated in ® rst (maha) season 1 area cultivated in second (yala) season 3 100

Cultivable area 3 2

8. The method of estimating the standardized gross value of output is explained in Molden et al.

(1998).9. The details of the analysis are reported in Samad & Vermillion (forthcoming).

10. Standardized gross value of output

(SGVO) 5 S Ocrops

AiYi

Pi

PbD Pworld

where Yi is the yield of crop i; Pi is the local price of crop i; Pworld is the international price of thebase crop; Ai is the area cropped with crop i; Pb is the local price of the base crop. For a detailed

explanation of SGVO see Molden et al. (1998).

References

Abeywickrema, N. (1986) Government policy in participatory irrigation management, in: Participa-

tory Management in Sri Lanka’s Irrigation Schemes (Colombo, Sri Lanka, International IrrigationManagement Institute).

Amerasinghe, Upali A., Sakthivadivel, R & Hammond Murray-Rust (1998) Impact Assessment ofRehabilitation Intervention in the Gal Oya Left Bank, Research Report 18 (Colombo, Sri Lanka,

International Irrigation Management Institute).Brewer, J.D. (1994) The participatory irrigation system management policy, Economic Review, 20(6),

pp. 4± 9.Geijer, J.C.M.A., Svendsen, M. & Vermillion, D.L. (1996) Transferring Irrigation Management Responsi-

bility in Asia: Results of a Workshop,Short Report Series on Locally Managed Irrigation, Report No.13 (Colombo, IIMI/FAO).

GOSL (Government Service of Sri Lanka) (1991) Agrarian Services (Amendment) Act, No. 4 of 1991(Colombo, Sri Lanka, Government Publications Bureau).

IIMI/Hector Kobbekaduwa Agrarian Research and Training Institute (HKARTI) (1997) Monitoringand evaluation of the participatory irrigation management policy, Final Report to Ministry of

Irrigation Power and Energy, and Asian Development Bank, Contract TA 1705 SRI (Colombo,Sri Lanka, International Irrigation Management Institute).

Johnson III, S.H., Vermillion, D.L. & Sargadoy, J.A. (Eds) (1995) Irrigation Management Transfer:Selected Papers from the International Conference on Irrigation Management Transfer, Wuhan, China,

20± 24 September (Rome, IIMI/FAO).Meinzen-Dick, R., Mendoza, M., Sadoutlet, L., Abiad-Shields, G. & Subramaniam, A. (1997) Sustain-

able water users association: lessons from a literature review, in: A. Subramaniam, V. Jaganathan& R. Meinzen-Dick (Eds) User Organizations for Sustainable Water Services, World Bank Technical

Paper 354 (Washington, DC, World Bank).Molden, D., Sakthivadivel, R., Perry, C.J., de Fraiture, C. & Kloezen, W.H. (1998) Indicators for

Comparingthe Performance of Irrigated AgriculturalSystems, Research Report 20 (Colombo, Sri Lanka,International Water Management Institute).

National Development Council (1996) Report of the National Development Council Working Groupon Agricultural Policy (Colombo, Sri Lanka), mimeo.

Ratnayake, R. (1995) Irrigation management transfer in Sri Lanka, in: J.C.M.A. Geijer (Ed.) IrrigationManagement Transfer in Asia, papers from the expert consultation on irrigation management

transfer in Asia, Bangkok and Chiang Mai, 25± 29 September (Bangkok, FAO/IIMI).Samad, M. & Vermillion, D. (Forthcoming) Participatory Irrigation Management in Sri Lanka: Partial

Reforms, Partial Bene® ts (Colombo, Sri Lanka, International Water Management Institute).Svendsen, M. & Vermillion, D. (1994) Irrigation Management Transfer in the Columbia Basin: Lessons

and International Implications, Research Paper (Colombo, Sri Lanka, International IrrigationManagement Institute).

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14


Vermillion, D.L. (1997) Impacts of Irrigation Management Transfer: A Review of Evidence, ResearchReport 11 (Colombo, Sri Lanka, International Irrigation Management Institute).

Vermillion, D.L. & Garces-Restrepo, C. (1996) Results of Management Turnover in Two Districts inColombia, Research Report 4 (Colombo, Sri Lanka, International Irrigation Management Institute).

Vermillion, D. L., Samad, M., Amerasinghe, U. & Narayanamurthy, S.G. (1996) A Standard Method-ology to Assess the Impacts of Irrigation Management Transfer (Colombo, Sri Lanka, International

Irrigation Management Institute) Draft.

Dow

nloa

ded

by [

The

Uni

vers

ity o

f M

anch

este

r L

ibra

ry]

at 0

5:03

25

Nov

embe

r 20

14

Documents

An Assessment of the Impact of Participatory Irrigation Management in Sri Lanka