Report No. 15863-PAK Pakistan Impact Evaluation Report On-Faim WVater Ma nagemunt Prolt( t i( wdi 1 I PAr Irrigation Systems Rehaiitation ProJe< ( d re 1 1-;, Command ater NAmnagement Poje I H -T Second On-Farm Water Management An' l ' June 28, 1996 Operation Ev,lumhon Dep,rmen Document of the World Bank Currency Equivalents (annual averages) Currency Unit Rupee (Rs) 1980 US$1.00 9.90 1981 US$1.00 10.55 1982 US$1.00 10.55 1983 US$1.00 12.75 1984 US$1.00 13.50 1985 US$1.00 15.90 1986 US$1.00 16.13 1987 US$1.00 17.60 1988 US$1.00 17.60 1989 US$1.00 18.90 1990 US$1.00 18.90 1991 US$1.00 19.2 1992 US$1.00 23.8 1993 US$1.00 25.9 1994 US$1.00 25.9 Abbreviations and Acronyms ACOP Alluvial Channels Observation Project (WAPDA) ADP Annual Development Programme CWMP Command Water Management Project ERR Economic Rate of Return FGW Fresh groundwater areas GOP Government of Pakistan IBIS Indus Basin Irrigation System IDA International Development Association IER Impact Evaluation Report IFAD International Fund for Agricultural Development ISRP Irrigation Systems Rehabilitation Project ISS Indus Special Study M&E Monitoring and Evaluation MAF Million acre feet MFAC Ministry of Food, Agriculture and Cooperatives MRES Mona Reclamation Experiment Station NWFP Northwest Frontier Province O&M Operation and Maintenance OFWM On-Farm Water Management OFWMD On-Farm Water Management Directorate OED Operations Evaluation Department PAD Provincial Agriculture Department(s) PCR Project Completion Report PID Provincial Irrigation Department(s) PLL Precision land leveling PR President's Report RAP Revised Action Programme for Irrigated Agriculture RERR Re-estimated Rate of Return SAR Staff Appraisal Report SGW Saline groundwater areas SMO Sub-project Management Committees USAID United States Agency for International Development WAPDA Water and Power Development Authority WUA Water User Association Glossary chak the command area of a watercourse check the means of blocking of the full flow of a watercourse to divert water into a farmer's field or channel through a "turnout" dikka a temporary means to restrict flow (used by irrigation departments) hari ladles people (laborer, sharecroppers, serfs) in Sinned kareze a system of underground channels, between shafts to the surface, to bring water to irritable land katcha local/traditional/earthen/unimproved/not brick and concrete (the opposite of puke) khal committee the traditional committee of farmers on a watercourse (Punjab) khanif summer mogha the irrigation department's engineered uncontrolled turnout into a watercourse, designed to deliver a constant flow proportional to the level of the supply canal (hence designed to allocate water fairly between all watercourses on a canal, as canal flows vary) pukka nucca improved permanent concrete turnouts and checks on watercourses, comprised of a concrete opening set in the watercourse and a removable concrete lid to close it rabi winter sarkari khal government channel, applied to the main stem of a watercourse turnout the diversion point, apukka nucca where modernized, but otherwise a earthen cut in the watercourse side, through which water is passed from the watercourse to a farmer's channel or field warabandi "fixed turn," the roster whereby each farmer on a watercourse receives the full water flow in turn at a fixed time each week and for a fixed period watercourse the small channel from the mogha to farmer channels and fields zamindar a landlord in Sindh Fiscal Year Government: July 1-June 30 The World Bank Washington, D.C. 20433 U.S.A. Office of the Director-General Operations Evaluation June 28, 1996 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Impact Evaluation Report on Pakistan: On-Farm Water Management Project (Credit 1163-PAK)) Irrigation Systems Rehabilitation Project (Credit 1239-PAK) Command Water Management Project (Credit 1487-PAK) and Second On-Farm Water Management Project (Credit 1603-PAK) Attached is an impact evaluation report on four irrigation projects in Pakistan. The IDA credits were all approved between FY81 and FY85 and closed by FY92. The evaluation was based on a review of Bank files, including the project completion reports, secondary sources, a sample survey of beneficiaries, an engineering inspection of a sample of canals, field visits to a large number of project sites, where farmers were interviewed, and discussions with government and Bank staff. The World Bank has been a major source of external funds for the very large and important irrigated sector in Pakistan for over 30 years, and has also been associated with government's periodic reviews of irrigation and power sector strategy and investment plans. During the 1980s the government's Revised Action Program for Irrigated Agriculture (RAP), drawn up with Bank assistance, emphasized improving system efficiency through rehabilitation and upgrading of irrigation infrastructure, and small-scale physical investments (mainly watercourse improvements and private tubewells). Previously large-scale dams for irrigation power and public tubewell fields had dominated public sector investments. The main purpose of the RAP strategy was to reduce system losses and thereby make more water available to crops in the field, while avoiding the high cost of the next major storage scheme. These four projects were the first to be supported by the Bank under the new strategy. They provided investments throughout the four provinces of the country, with on-farm water management activities modeled on an earlier project which was assisted by USAID. The main components were physical investments in canals, drainage and watercourses, and institutional initiatives to improve irrigation management and services to farmers, including promoting water users associations at the watercourse level. The outcomes of all four projects were rated satisfactory at implementation completion with high re-estimated economic rates of return, because all comprised rehabilitation investments with low costs per hectare, to correct system deficiencies, especially water losses. In a water deficit irrigation economy the additional water supplies realized good benefits, being used mainly to expand the area cropped. Total water saved exceeded what would have been available from the next large scale storage reservoir, and at much lower unit cost, thereby confirming the main thrust of the RAP strategy. Institutional initiatives largely failed, however, due to poor design. Single projects are not able to change long-term traditional patterns of behavior in major institutions, and there was a lack of appreciation of traditional local institutions. However, during implementation the projects departed from important parts of the RAP strategy and the projects' designs such that outcomes were not as strong as they could have been. Chiefly, they failed to give highest priority to rehabilitation and renovation in saline groundwater areas, as RAP intended, where the returns to controlling watertables and to supplying more surface water- since there is no alternative source-are highest. Also, countrywide project packaging did not allow designs to be tailored to the needs of individual provinces and localities, and this resulted in sub-optimal investments and unnecessary resource transfers to the non-poor. Watercourse renovation especially was driven by Bank inspired physical targets which overrode the careful fostering of participatory institutional growth and encouraged important development priorities to be disregarded. Vested interests and the pecuniary gains available from project activities distorted incentives to participants, just as the efficient management of the system as a whole was undermined by political influences and rent seeking. Even so the evaluation found that the social impact of the projects was positive in terms of improving the lives and welfare of farm families. This was especially so where supply canal improvements were associated with watercourse improvements-by design in the case of the Command Water Management Project and fortuitously in cases where watercourses on a canal improved under the Irrigation Systems Rehabilitation Project were improved under the On-farm Water Management projects. The main effect was that cropping expanded leading to increases in farm incomes. Watercourse improvements reduced labor demand for irrigation management and watercourse maintenance, but other field tasks often performed by women increased. Health problems decreased and social cohesion increased due to fewer disputes over water supplies. The evaluation confirms the high priority accorded watercourse improvements, especially in saline groundwater areas, subject to more attention being given to equity and participation; advocates the unbundling of assistance packages to allow designs to be tailored better to local conditions; and recommends that attention be given to establishing a "Water Resources Institute," or the like, to provide planners and policymakers with continuous and reliable information on Pakistan's scarce water resources. A new government strategy for this critical sector of the economy is being designed to overcome long-standing institutional and operational problems. It involves decentralizing irrigation management to autonomous public utilities at the command level and promoting farmer organizations to operate and maintain the system at lower levels. The findings of this evaluation are relevant to this new approach since much remains to be done to improve the system's operating efficiency, physical maintenance and responsiveness to diverse local conditions, and to meet farmers' needs, especially those of small farmers-which are issues highlighted in this report. Attachment Contents Prefac ............................................................................................................................................ 5 Basic Data Sheet ............................................................................................................................ 7 ExecutiveSummary .................................................................................................................... I5 1. Indus Basin Irrigation Development and the Four Bank-Assisted Projects.....................25 2. Impact of the Projects on Watercourse Losses and Groundwater Levels........................36 3. Social and Economic Impacts ................................................................................................ 47 4. The Impact of Irrigation Systems Rehabilitation................................................................60 5. Institutional Im pact ............................................................................................................... 67 6. Future Strategy-Addressing the Irrigation Crisis ...........................................................77 7. Conclusions and Recommendation ...................................................................................... 81 Boxes 1.1 World Bank-Assisted Irrigation Projects Implementing RAP Recommendations.........28 1.2 Watercourse Lining: An Understatement .......................................................................... 32 2.1 The Impact Evaluation Farmer Survey ............................................................................... 38 2.2 Water SavingsEstimate ....................................................................................................... 44 3.1 Watercourses as Status Symbols: The Bigger, the Better.................................................53 3.2 Economic Rates of Return to OFWM-I and II, and CWMP.............................................54 4.1 Canal Rehabilitation Design Criteria .................................................................................. 61 4.2 Sustainability of Canal Rehabilitation-The Engineering Survey ................................... 64 5.1 Irrigation Management Malpractices-The Collapse of Discipline.................................68 5.2 Institutional Aspects From the Evaluation Survey Findings.............................................72 5.3 WUA Formation-Intentions and Reality ..........................................................................73 5.4 The On-Farm Water Management Directorate in Sindh Province..................................74 B.1 Indus Basin Irrigation System-Effects of Partition in 1947...........................................98 B.2 The Indus Special Study-Proposals for Surface storage and Public Tubewells.........101 B.3 The Indus Special Study-Issues and Solutions...............................................................102 B.4 The Revised Action Program for Irrigated Agriculture (RAP) .....................................103 H.1 Balochistan's Invovement ................................................................................................. 168 This report was prepared by Julian Blackwood (Task Manager), Peter Naylor and Ty Mitchell (consultants) who visited project sites in September/October, 1994. Constance Frye and Megan Kimball provided administrative support. The report was issued by the Agriculture and Human Development Division (Roger Slade, Chief) of the Operations Evaluation Department (Francisco Aguirre-Sacasa, Director). 2 Annexes A Description of the Four Bank-Assisted Projects..................................................................87 B. Investment Strategies for the Indus Basin Irrigation System............................................97 Appendix 1: The Indus Basin Irrigation System (IBIS): Vast Size, Elegant Design, But Incomplete ................................................................................... 105 Appendix 2: Waterlogging andSalinity .................................................................... 109 C. Watercourse Improvement - Evolving Policies and Practices......................................... 111 D. Monitoring and Evaluation of Watercourse Conveyance Efficiency..............................117 E. Precision Land Leveling-A Declining Priority ...............................................................127 F. Irrigation System Rehabilitation and Sustainability (Survey Results)...........................131 G. The Impact Evaluation Farmer Survey ...........................................................................141 Appendix 1. Survey Design Considerations .............................................................. 159 Tables 2.1: Gains and Loss Effects from Reduced Channel Seepage................................................37 2.2: Cost of Watercourse RenovationinOFWM- ............................................................... 42 B.1: Groundwater Balance (MAF) ........................................................................................... 110 F.1: Canals Observed by ISRIP............................................................................................... 136 F.2: Conditions of Berm ........................................................................................................... 137 F.3: Heigh ofFreeboard .......................................................................................................... 138 F.4: Other Indicators of CanalCondition .............................................................................. 139 F.5: ConditionofOutlets .......................................................................................................... 140 G.1: Survey Cells in a Three-Way Stratification...................................................................141 G.2: Survey Watercourse Profile, By Province......................................................................142 G.3: Watercourse Distribution by Groundwater Quality....................................................148 G1.1: Watercourse Distribution inPakistan .......................................................................... 160 G1.2: Survey Cells in a Three-Way Stratification.................................................................161 G1.3: Watercourses Sampled by the Impact Evaluation......................................................162 H.1: Cost Distribution by Province ......................................................................................... 170 Figures 1: Schematic Diagram of Indus Basin Irrigation Layout .......................................................30 2: Indus Basin Irrigation System-waterflows, "losses" and recoveries..............................31 3: An Example of Varying Water Flows in Watercourses on One Distributary..................91 4: Distribution of WCs By LiningShare ................................................................................ 146 5: Impact of Lining on Water toFarm ................................................................................... 146 6: Location of FO Leaders ....................................................................................................... 153 7: Farmer Relations with Government Agencies................................................................... 155 Photos 1: An unimproved watercourse with its mogha destroyed......................................................33 2: An unimproved watercourse in a chasm of excavated silt .................................................33 3: An improvedwatercours ...................................................................................................... 33 3 4: Pukka nuccas, one open, oneelosed ...................................................................................... 33 5: An am enity w hich is lost with channel lining ......................................................................52 6: An am enity gained w hen included in lining plans...............................................................52 7: A reconstructed mogha serving an improved channel........................................................65 8: A channel rehabilitated under ISRP running high.............................................................65 Bibliography ............................................................................................................................... 171 Map: IBRD 27702 5 Preface This report examines the impacts of four Bank-assisted irrigation projects on irrigation efficiency, crop production, the lives of irrigator families and on the institutions which support the system. The four projects were the first to support a change of investment strategy away from new large dams and tubewells fields towards improving the efficiency of the existing irrigation infrastructure. The four projects approached the systems inefficiencies in three different ways: Irrigation System Rehabilitation sought to repair dilapidated and stressed secondary distribution canals, On-farm Water Management invested in the farmers' tertiary distribution system and Command Water Management did both for specific commands. The report seeks to explore the efficiency of these approaches, especially as perceived by the beneficiary farm families. The timing of the report is significant in that the Revised Action Programme for Irrigated Agriculture, which set the new course, has now been largely abandoned in favor of pilot activities of a new far reaching strategy to decentralize management of the irrigation system, including transferring responsibility for the tertiary distribution system to farmers' organizations. However, the continued shortfall in operation and maintenance suggests that investment of the types examined here will continue to be needed as part of a successful decentralization strategy, either as a precondition or as part of a package to implement the strategy. This report seeks to make a contribution to the current strategy by examining the relative impacts of the four projects. The report is based on review of the staff appraisal reports, the credit documents, the project completion reports, project files, and a large number of reports prepared in connection with the projects (by the implementing agencies, Bank staff, bilateral donors and consultants), and reports relating to the issues by the Bank and other agencies (see the Bibliography). The projects have been discussed with key Bank staff (current and retired) involved with their design and implementation. A mission visited Pakistan in September/October 1994 where it held discussions with government staff, at the federal level and in the four provinces, and visited a sample of subprojects. The project works on 85 watercourses were inspected and structured discussions were held with small to large focus groups of farmers. In addition, a Pakistani consulting firm, which was familiar with irrigation sector issues, conducted a survey of project beneficiaries, men and women in groups and individually, on a sample of watercourses in Punjab, Sindh and Northwest Frontier Provinces. The consultants provided the survey data, while analysis and interpretation was done by the evaluation team. A second Pakistani consulting firm, of engineers, was engaged to make a visual inspection during the canal closure period (December/January) of the condition of a sample of distribution canals which were rehabilitated and improved under two of the projects. Copies of the draft report were sent to the relevant government officials and agencies concerned for their review and comments. No comments have been received. The mission gratefully acknowledges the valuable assistance provided by the survey consultants, and the many project beneficiaries, government staff and others who patiently responded to the mission's inquiries and provided welcome hospitality. u 7 Basic Data Sheet ON-FARM WATER MANAGEMENT PROJECT (CREDIT 1163-PAK) Key Project Data (in US$ million) Item Appraisal Actual Actual as % of Appraisal Estimate Estimate Total Project Costs 111.58 59.23 53 Credit Amount (SDR m) 33.40' 25.92 78 Cofinancing (SDR m) IFAD 10.50 8.91 85 Date Physical Components Completed 06/30/84 06/30/85 Proportion then Completed (%) 100 116 Economic Rate of Return (%) 45 29 64 Cumulative Estimated and Actual Disbursements FY82 FY83 FY84 FY85 FY86 FY87 Appraisal Estimate (SDR m) 5.7 15.9 28.9 33.4 Actual (SDR m) 2.1 5.8 13.8 21.1 25.8 25.9 Actual as % of Appraisal 37 36 48 63 77 78 Final Disbursement August 22, 1986 Project Timetable Item Original Revisions Actual Identification Reporte 06/04/80 Negotiations 04/81' 04-05/81' Board Approval 06/16/81 06/16/81 Credit Signing 03/80 08/19/81 Credit Effectiveness 11/19/81 02/18/82' 03/30/82 Credit Closing 12/31/84 12/31/85 12/31/85 Staff Inputs (in staff weeks) FY79 FY80 FY81 FY82 FY83 FY84 FY85 FY86 FY87 Total Preappraisal 17.0 51.2 31.8 0.1 - - - - - 100.1 Appraisal - - 99.0 - - - - - - 99.0 Negotiation - - 16.2 0.5 - - - - - 16.7 Supervision - - 0.2 29.2 30.4 27.7 21.6 32.4 13.9 155.4 Other - 0.1 1.3 - - - - - - 1.4 Total 17.0 51.3 148.5 29.8 30.4 27.7 21.6 32.4 13.9 372.6 8 Mission Data Project Cycle Month/ No. of Days Specialization Performance Rating Type of Year Persons in Represented Statu? Trend Problems' Field Identification 04/80 4 93 2A,E,I Preparation 06-07/80 3 49 A,21 Appraisal 11-12/80 6 165 2A,2E,21 Supervision 1 09-10/81 I 9 A 2 - T,F,O Supervision 2' 05/82 I 4 A Supervision 3' 10-11/82 2 4 AJI Supervision 4 03-04/83 3 37 A.21 2 1 M Supervision 5 11-12/83 I 15 I 2 I M Supervision 6' 06-07/84 2 5 A,J Supervision 7 09-10/84 4 20 2A,E,l 2 I M Supervision 8 04-05/85 1+1 26 l+A I I F Supervision 9 10-11/85 2 25 2A,l Supervision 10 04-05/86 2 17 A.E PCR' 04/86 3 66 A,E,I Other Project Data Fiscal Year of Borrower July 1 - June 30 Name of Currency (Abbreviation) Rupee (Rs) Exchange Rate (Rs/USS) Appraisal Year Average (1980-81) 9.90 Intervening Year Average (1981-82 to 1984-85) 13.02 Completion Year Average (1985-86) 16.13 Follow-on Project Name Second On-Farm Water Management Project (OFWM-II) Credit No. 1603-PAK Credit Amount (SDR m) 34.8 Date of Board Approval June 6, 1985 Date of Credit Signing September 30, 1985 Date of Effectiveness February 14, 1986 a. Equivalent to US$41 million at time of negotiations. b. Project brief identifying the project as a spin-off of the watercourse cleaning and maintenance component of a proposed nationwide water management project. c. Start April 20. d. Start April 28. e. First of two revisions. f. Most supervision missions worked on several assignments but did not specify in all cases the time spent on each. In such cases, supervision time has been apportioned based on the recollection of mission members or. when this was not possible, on the assumption that equal time was spent on each assignment. g. A agriculturist; E = economist; I = irrigation engineer. h. 1 = problem-free or minor problems; 2 = moderate problems. i. I = improving. j. F = financial; M = managerial; T = technical; 0 = other. k. Partial; no Form 590 prepared. 1. FAO/IBRD Cooperative Program. m. 10.55 for 1981-82, 12.75 for 1982-83, 13.50 for 1983-84. 15.30 for 1984-85. 9 Basic Data Sheet IRRIGATION SYSTEMS REHABILITATION PROJECT (CREDIT 1239-PAK) Key Project Data Item Appraisal Estimate Actual Actual as % of Appraisal Estimate Total Project Cost (US$ m) 118.10 164.30 139 Credit Amount (US$ m) 40.00 40.3 101 Cofinancing - USAID 45.0 103.0 229 Date Physical Components Completed 06/30/85 06/30/88 Economic Rate of Return (%) - 50 Cumulative Estimated and Actual Disbursements FY FY FY FY FY FY FY 83 84 85 86 87 88 89 Appraisal Estimate (SDR m) 1.3 10.7 30.1 35.5 - - - Appraisal Estimate (US$ m) 1.5 12.0 34.0 40.0' - - - Actual (SDR m) - 5.0 11.6 19.4 27.5 32.5 35.0 Actual (US$ m) 5.3 11.8 20.4 30.4 37.0 40.3' Actual as % of Appraisal (SDR m) - 47 39 55 77 92 99 Actual as % of Appraisal (US$ m) - 44 35 51 76 93 101 Date of Final Disbursement October 19, 1988 Project Timetable Item Original Revisions Actual Identification 04/80 Preparation 06/80-03/81 Appraisal 06/81 Negotiations 02/82 03/82 Board Approval 04/06/82 05/04/82 Credit Signing 06/03/82 Credit Effectiveness 09/01/82 05/17/83' 05/11/83 Credit Closing 12/31/85 12/31/86' 12/31/87' Project Completion 06/30/85 06/30/88' Staff Inputs (in staff weeks) FY79 FY80 FY81 FY82 FY83 FY84 FY85 FY86 FY87 FY88 FY89 Total Preappraisal 6.4 25.5 10.4 - - - - - - - - 42.3 Appraisal 40.0 67.0 - - - - - - - 107.0 Negotiations 12.9 - - - - - - 12.9 Supervision 2.7 26.3 22.0 25.4 30.9 26.8 8.3 7.09 149.4 Other 0.3 0.2 1.7 - 0.1 - - - - - 2.3 Total 6.4 25.8 50.6 84.3 26.3 22.1 25.4 30.9 26.8 8.3 7.0 313.9 10 Mission Data Project Cycle Month/ No. of Staffdays Specialization Performance Rating Type of Year Persons in Represented h Status' Trend ' Problems Field Pre-identification 04/79 1 4 Ag Identification 09-10/79 2 30 Ag,Eg Identification 2 1 04/80 3 61 Ec,IEg,EEg Preparation 09/80 1 4 Ec Preparation 2 03-04/81 1 4 Ag Appraisal 05-06/81 7 192 2Ag,2lEg,EEg,Ec,LO Negotiations 05/82 1 2 Ag Supervision I 10-11/82 2 3 Ag,IEg Supervision 2 01/83 2 3 Ag,IEg Supervision 3 03-04/83 4 33 21Eg,Ag,PO 2 1 F,M,T,O Supervision 4 11-12/83 2 37 IEg,PO 2 1 M,O Supervision 5 06-07/84 2 5 Ag,1Eg Supervision 6 10-11/84 I 7 lEg Supervision 7 10-11/84 2 52 IEg,PO 2 1 M,O Supervision 8 03/85 1 2 Ag Supervision 9 04-05/85 2 31 IEg,PO 2 1 M,O Supervision 10 10-11/85 4 23 2Ag,lEg,PO Supervision II 04-05/86 2 30 IEg,PO Supervision 12 07/86 2 3 Ag,lEg Supervision 13 07/87 m 11 Completion 05/89 2 30 Ec,lEg Other Project Data Fiscal Year of Borrower July I - June 30 Name of Currency (Abbreviation) Rupee (Rs) Exchange Rate (Rs/US$) Appraisal Year Average 11.0 Intervening Year Average (1982-83) 12.7 Completion Year Average (1987-88) 17.6 Follow-on Project Name Irrigation Systems Rehabilitation II Credit No. 1888-PAK Credit Amount (US$m) 79.50 Date of Board Approval 03/29/88 Date of Credit Signing 06/09/88 Date of Effectiveness 06/01/89 a. US$/SDR exchange rate: 1.12676. b. US$/SDR exchange rate: 1.15143. c. Last of five revisions. d. First of two revisions. e. IDA agreed, as an exception, to extend the period for receipt of final withdrawal applications to nine months beyond the closing date. f. Refers to IDA credit. g. Partly estimated. h. Most supervision missions worked on several assignments but did not specify the time spent on each. The number of staff/days therefore has been estimated, assuming in most cases that equal time was spent on each assignment. i. Ag - agriculturist; EC - economist; EEG - equipment engineer; lEg - irrigation engineer; LO - loan officer; PO - project officer. j. 2 = moderate problems. k. I = improving. 1. F = financial; M = managerial; T = technical; 0 = other. m. Not available. Action letter says project was reviewed along with ISRIP-l appraisal. 11 Basic Data Sheet COMMAND WATER MANAGEMENT PROJECT (CREDIT 1487-PAK) Key Project Data Item Appraisal Actual Actual as % of Estimate Appraisal Estimate Total Project Costs (US$ m) 81.9 100.1 122 Credit Amount 46.50 56.9 122 Cofinancing - USAID 25.0 22.9 92 Date Physical Components Completed 10/31/86 12/30/92 Economic Rate of Return 24 23 Cumulative Estimated and Actual Disbursements FY85 FY86 FY87 FY88 FY89 FY90 FY91 FY92 FY93 Appraisal Estimate - SDR million 1.32 6.48 14.10 24.90 38.53 43.70 - USSS million 1.40 6.90 15.00 26.50 41.00 46.50 Revised Estimatesh - US$ million (1/86) 0.25 5.00 13.50 26.00 41.00 46.50 - US$ million (7/89) 0.25 3.30 7.00 14.20 16.00 30.00 44.00 46.50 Actual ' - SDR million 0.26 2.57 6.60 11.34 12.73 18.15 24.93 34.41 42.69' - US$ million 0.25 2.81 7.82 14.18 16.05 23.00 32.31 45.27 56.93 Actual as % of Appraisal -SDR 20 40 47 46 33 42 57 79 98 -US$ 18 41 52 57 39 49 69 97 122 Date of Final Disbursement: May 19, 1993 Project Timetable Item Original Revisions Actual First Final Identification 06/80 - - 06/04/80 Preparation 05/15/81 03/31/83' 09/15/83h 08/83 Appraisal 07/81 06/83' l1/83h 10-11/83 Negotiations n.a. n.a. 01/84h 04/09-17/84 Board Approval 05/84 n.a. 05/29/84h 05/29/84 Credit Signature n.a. n.a. 06/84h 06/13/84 Credit Effectiveness 03/30/85 09/11/841 03/28/85 03/28/85 Credit Closing 03/31/87 12/31/891 12/31/92i 12/31/92 Staff Inputs (in staff weeks) Item Pre- FY FY FY FY FY FY FY FY FY FY FY FY FY Total FY82 82 83 84 85 86 87 88 89 90 91 92 93 94 Preappraisal 3.3 6.1 7.4 8.1 - - 24.7 Appraisal - - 0.2 73.9 - - 74.3 Negotiations - - 1.2 9.9 - - 11.1 Supervisionk - - - 5.5 19.5 14.8 20.4 27.0 23.2 11.2 9.0 11.8 4.8 147.2 PCR! - - - - - - - - - - - - - 13.4 13.4 Total 3.3 6.1 8.8 97.4 19.5 14.8 20.4 27.0 23.2 11.2 9.0 11.8 4.8 13.4 270.7 12 Mission Data Project Cycle Month/ No. of Days in Specialization Performanc Type of Year Persons Field Represented" e Problems Rating" Identification 04/80 n.a. n.a. n.a. - - Preparation I 03-04/81 I 18 AG - - Preparation 2 09-10/81 1 36 AG - - Preparation 3 01-02/82 I 16 AG - - Preparation 4 05/82 1 18 AG - - Preparation 5 10-11/82 1 14 AG - - Preparation 6 01/83 2 8 AG,EC - - Preparation 7 08/83 1 17 EC - - Appraisal 10-11/83 5 27 IDA: 2AG,EC,2EN - - 3 27 USAID: IS,SO,EN - - Supervision I " 06-07/84 1 15 AG Not rated Not rated Supervision 2 4 09-10/84 2 25 AG,IS 2 (2)M,F,C,l Supervision 3 " 03/85 1 12 AG Not rated Not rated Supervision 4 09/85 2 n.a. PS,DS n.a. n,a. Supervision 5 q 10-11/85 5 25 IDA: 2AG,EN 2 M3,F2,D2 USAID: EC,EN Supervision 6 P 07/86 I 8 AG Not rated Not rated Supervision 7 q 06-07/87 4 25 IDA: AG,EN,EC,AE(PfT) 2 (2)M,D USAID: EC Supervision 8 4 05-06/88 5 29 IDA: AG,EN,EC 2 (2)M,D USAID: EN,EC Supervision 9 q 05-06/89 6 32 IDA: AG,IS,EN 2 (2)M,D USAID: 3 (Pff(?) Supervision 10 4 11-12/90 3 33 2EN,AE 2 (2)M,D Supervision 11 q 11/91 2 25 EN,EC 2 (2)D,C,P PCR 01-02/94 3 22 EN,EC,AG Other Project Data Fiscal Year of Borrower July I - June 30 Name of Currency (Abbreviation) Rupee (Rs) Exchange Rate (Rs/USS) Appraisal Year Average (FY84) 13.5 Intervening Years Average (FY85-92) 19.2 Completion Year Average (FY93) 25.9 Follow-on Project None a. All SDR. UIS$ and percentage figures have been rounded. b. Revisions made in supervision reports issued on 01/07/86 and 07/26/89. c. IDA disbursement information. d. SDR 1,006,494.64 cancelled as of May 19, 1993. e. Initiating Project Brief based on findings of the April 1980 identification mission. f. Preparation (called PREID in internal report) mission which visited Pakistan in January/February 1982. Revised scheduling for subsequentt steps is not known because timetables for project processing are not available in project files. g. As per Development Credit Agreement. h. Revised Project Brief based on findings of the June, 1983 preparation follow-up mission. i. Last of two extensions of the terminal date for effectiveness. j. Last of three one-year extensions. k. Excluding PCR. 1. Including time spent by FAO/CP. m. AE = Agricultural Engineer; AG = Agriculturist; EC = Economist; EN = Engineer; IS = Institutional Specialist; SO = Sociologist; PS = Procurement Specialist; DS = Disbursement Specialist; P/T = Part-time. n. Rating: 2 = Moderate problems; 3 = Major problems. o. M = Project management; F = Availability of funds; C = Compliance with covenants; I = Institution building; D = Development impact; P = Procurement progress. p. The mission resulted in a Back-to-Office report (mission worked only part-time on the project during its stay in the country). q. A full supervision report was prepared as a result of the mission. 13 Basic Data Sheet SECOND ON-FARM WATER MANAGEMENT PROJECT (CREDIT 1603-PAK) Key Project Data (in US$ million) Item Appraisal Actual PCR Estimate/Actual Estimate Estimate Total Project Costs 57.8 95.6 165 Credit Amount (SDR m) 34.5 44.0 126 Economic Rate of Return (%) 22 22 - Cumulative Estimated and Actual Disbursements FY86 FY87 FY88 FY89 FY90 FY91 FY92 Appraisal Estimate - SDR million 8.5 19.9 32.0 34.8 - - - USS million 8.5 19.8 31.7 34.5 - - Actual 1.3 10.6 21.3 29.8 33.0 34.7 34.8 - SDR million 1.5 13.2 27.6 38.9 43.0 45.3 45.5 - USS million Actual as % of Appraisal - SDR 15 53 66 86 95 99 100 - USS 18 67 87 113 125 131 132 Date of Final Disbursement September 14, 1991 Project Timetable Item Original Revisions Actual Identification 09/82 n.a. n.a. Preparation 08/83 06/84 08/84 Appraisal 10/83 10/84 " 09-10/84 Negotiations 06/84 04/85 04/85 Board Approval 09/84 05/85 06/06/85 Credit Signing n.a. n.a. 09/30/85 Credit Effectiveness 12/84 n.a. 02/14/86 Credit Closing 12/31/88 12/31/91 12/31/91 14 Staff Inputs (in staff weeks) FY FY FY FY FY FY FY FY FY Total Preappraisal 0.1 - - - - - 100.1 Appraisal - - - - - - - - 99.0 Negotiation - - 0.5 - - - - - 16.7 Supervision - - 29.2 30.4 27.7 21.6 32.4 13.9 155.4 Other - - - - - - 1.4 Total 29.8 30.4 27.7 21.6 32.4 13.9 372.6 Mission Data Project Cycle Month/ No. of Days in Specialization Performance Rating Type of Year Persons Field Represented' Status Problemsl Identification n.a. n.a. n.a. n.a. - Preparation 08/84 n.a. n.a. n.a. - Appraisal 09-10/84 4 n a. AG,AG,EC,EN - Supervision I 10-11/85 4 49 AG,AG,EN(2) I Mgt(2) Supervision 2 04-05/86 2 10, AG,EC I Mgt(2) Supervision 3 I 1-12/87 3 30 AG.EN 2 Mgt(3) Supervision 4 01-02/89 3 25 AG.EN.AD 2 Mgt(3) Supervision 5 03/90 3 25 AG.AG(P/T).EN(P/T) 2 Mgt(3) Other Project Data Fiscal Year of Borrower July I - June 30 Name of Currency (Abbreviation) Pakistani Rupee (PRs) Exchange Rate (Rs/US$) Appraisal Year Average (1985) 15.9 Intervening Year Average (1986-90) 18.9 Completion Year Average (1991) 23.8 Follow-on Project Name Third On-Farm Water Management Project (OFWM-Ill) Credit No. 2245-PAK Credit Amount (US$ m) 47.30 Date of Board Approval 05/21/91 Date of Credit Signing 07/31/91 Date of Effectiveness 05/14/92 a. Pre-Project Brief, February 1982. b. Project Brief, June 1984. c. In addition to the five full supervision missions listed above, there were three short missions (July 86, March 87 and June 88). d. AD = Auditor; AG = Agriculturist; Ec = Economist; EN = Engineer; P/T = Part-time participation. e. I = No significant problems; 2 = Moderate problems. f. Mgt(2) = Despite some shortcomings, project management is capable of doing its job with only moderate problems. Mgt(3) = Major deficiencies in project management which will affect its ability to do the job, but acceptable measures are being taken to strengthen managerial capacity. g. Estimated, because mission worked on several assignments but did not specify the time spent on each. 15 Evaluation Summary Introduction 1. The Bank has played a central role in the development of Pakistan's Indus Basin Irrigation System, the largest integrated irrigation network in the world. In the 1950s the Bank's good offices assisted the negotiation of the Indus Waters Treaty (1960) between India and Pakistan, which settled the division of the waters of the Indus Basin between the two countries following Partition in 1947. The Bank then administered the Indus Basin Development Fund (IBDF) which financed the physical works to implement Treaty provisions. Thereafter the Bank played a catalytic role in periodic major reviews of irrigation and power sector strategy and subsequently funded parts of the resulting investment programs. 2. In the 1960s and early 1970s strategy and investment in irrigation and drainage in Pakistan focused on three activities: (a) major storages, barrages and link canals, financed by the IBDF, to compensate for loss of surface supplies to India; (b) control of increasing waterlogging and salinity through drainage (public tile and tubewell drainage projects); and (c) expanding water supplies by new storages and public tubewell schemes. During the 1970s, however, three factors led to a change of direction: (a) the high unit costs of additional stored water; (b) the faltering performance of the surface irrigation system (especially inadequate maintenance leading to operational problems); and (c) sustainability problems with the public sector tubewell fields. 3. Through the government's Revised Action Programme for Irrigated Agriculture of 1979 (RAP) the emphasis switched to improving system efficiency through rehabilitation and upgrading, and to small-scale physical investments (private tubewells, watercourse improvements and soil reclamation). Large and long gestation projects for drainage and surface water reservoirs were deferred. The new strategy focused on better utilization of existing infrastructure, especially at the watercourse level, the "saving" of surface water "losses" from channels, and improved management of water from the irrigation command level down to the farm. The four projects evaluated in this report were the first projects, out of ten, supported by the World Bank Group under these new priorities. The projects are the first and second On-farm Water Management projects, the Command Water Management Project and the Irrigation Systems Rehabilitation Project. Focus of the Evaluation 4. Since the prime objective of the RAP strategy was to reduce the need for the construction of incremental storage, by raising system efficiency mainly by reducing channel "losses," the central focus of the study was to determine the impact of the projects at the farm level on water supplies and their reliability. Secondary impacts discussed include the effects on waterlogging and salinity, as influenced by changes in groundwater levels, the impacts on cropping and the welfare of irrigator families, and economic and institutional aspects. Given that the RAP strategy has since been replaced by a far reaching new approach, on a pilot basis to start with, to transfer responsibility for the irrigation system to a combination of farmers' groups and public utilities, this evaluation also discusses issues arising from the project experience which relate to future strategy. 16 Objectives of the Projects 5. The four projects were approved from 1981-85 and were designed: (a) to make up for deferred maintenance of critical parts of the irrigation and drainage system; (b) to improve rudimentary watercourses at the farm level, mainly by constructing masonry conduits to reduce water "losses," (c) to establish more water user associations at the watercourse level; and (d) to overcome endemic institutional inefficiencies by bringing irrigation and agriculture staff closer together in implementation units. Primary direct benefits were expected to be increased agricultural production, resulting from expanded cropping with "saved" water, and a reduction in the depredations of waterlogging and salinity through reducing seepage from channels. The avoided costs of new storage reservoirs, in the never-ending search for expanded surface supplies, were a major secondary benefit. In contrast to channel lining programs elsewhere, there was no mention of reduced pumping costs as a benefit, with increased surface supplies at the farm substituting for tubewell supplies in fresh groundwater areas, because the strategy concentrated on achieving water savings in saline groundwater areas. 6. The strategies adopted for the three project types were quite different. The On-Farm Water Management (OFWM) projects aimed to improve the final distribution of surface supplies to individual farms through watercourse upgrading and better water management at the farm level, without reference to related supply canal conditions and the reliability of supplies to watercourses. The Irrigation Systems Rehabilitation Project (ISRP) provided for critical rehabilitation of selected surface irrigation and drainage systems which had badly deteriorated because of inadequate maintenance, but without any improvements in the lower reaches of the system. In the areas served by ISRP, the project was to provide more reliable and more equitable water supplies, reduce crop and property losses from canal breaches, strengthen operation and maintenance (O&M) programs and serve as a model for similar investments elsewhere. The Command Water Management Project (CWMP) took a broader view and aimed to provide complementary investments in the physical and institutional aspects of water management for the entire irrigated area of selected canal commands, including improving O&M and cost recovery. Thus CWMP combined elements of ISRP and OFWMP. All four projects were national in scope, with activities in each of the provinces, which gave rise to scheduling and design problems in that the specific different characteristics of each province could not be fully affected in the design implementation arrangements. Evaluation Methods 7. Except for CWMP, which covered large parts of whole commands, the subprojects financed under these projects were scattered widely throughout the system. Hence the evaluation had to rely mostly on information from the farm level rather than existing system information. A number of secondary sources provided limited information on water "savings" in particular, but it was outside the scope and resources of this study to measure flows in the field. Given the large scale of these investments and their dispersion throughout the Indus Basin, and a dynamic agricultural sector, it was impractical to attempt to quantify the output or farm income changes during the period, especially since establishing causality would have been very difficult. Instead, qualitative responses to the evaluation team's inquiries on 85 watercourses and a survey of farmer and women's focus groups were used. 8. The 85 watercourses visited by the team in the four provinces, were sometimes chosen by the mission at random and without notice, to inspect the state of the watercourses and talk to 17 available farmers, and others by arrangement, including interviewing assembled farmers in small to large groups. In addition, a sample of project irrigation channels were inspected by engineering consultants while they were closed and empty for maintenance during December and January, 1994/95. The OED team's field work validated or otherwise information available from extensive secondary sources, and assisted with the design of the surveys and the interpretation of the results. 9. The Pakistani consultants for the farmer survey used a questionnaire developed jointly with the OED team and field tested by the consultants. The 240 watercourse sample was both random and purposive covering 120 Punjab watercourses and 105 in Sindh Province, the bulk of the irrigated area, plus 15 watercourses in North West Frontier Province for descriptive purposes. Of this sample, 67 were "control" watercourses which had received no project investments. In addition, groups of women were interviewed by female consultants on a sub-sample of 152 watercourses using a separate questionnaire. The consultants conducted the field surveys in late 1994 and were responsible for data entry. OED cleaned the data and did the analysis. The engineering consultants for the channel surveys during closure relied on visual inspections, photographs and simple measurements to report on the condition of project channels, especially maintenance standards. Major Impacts (a) Farm Water Supply 10. Monitoring and evaluation studies provide limited evidence of the impact of OFWM watercourse improvements on conveyance losses in the watercourses. Despite long-standing concerns about system losses and large investments in system upgrading, a consistent methodology and monitoring program to track water flows and system conveyance efficiencies has not been established. Measurement of watercourse flows is not often attempted. Furthermore, the situation is confused by lack of agreed terminology on how to measure and report both channel "losses" and water "savings" following improvements. 1. Although conditions vary markedly throughout the system, it is concluded that watercourse upgrading reduced losses on average from about 40 percent of flows at the mogha, at watercourse head, to 25-30 percent, a "saving" of 10-15 percentage points. This is based on field inquiries and a review of available monitoring results. The upper estimate is equivalent on average to an extra 188 acre feet (230 cubic meters) reaching farm turnouts per improved watercourse per year. This is only a little over half the expectation of the Revised Action Programme of 325 acre feet per improved watercourse, but which had anticipated that a much higher proportion of a watercourse's length would be lined. 12. The water "saving" impact for the 9,856 watercourses improved under three of the four projects, plus that from canal lining under CWMP, totals about 2.0 million acre feet (MAF), or 2.3 billion cubic meters. This is ten percent of the estimated 20 MAF of total watercourse seepage losses, and four percent of total field and channel seepage losses from the Indus Basin Irrigation System as a whole (46 MAF). Adding the watercourse renovation carried out under all the other OFWM projects since 1978, the total reduction in conveyance losses would amount to about 4.7 MAF. Although this is only a little more than half the savings anticipated under RAP, it is nevertheless more than a new surface storage dam at Kalabagh would provide (though that 18 would have power benefits as well). This next proposed main storage site would provide an estimated 3.5 million acre feet at the mogha at a cost of US$3.5 billion (in a 1985 estimate). (b) The Economic Impact of Watercourse Renovation 13. At appraisal the estimated economic rates of return (ERR) for OFWM-I and II and the OFWM component of CWMP were strong (45, 22, and 24 percent respectively) and robust in the face of changed assumptions. Despite lower crop prices, the ERRs re-estimated in the Project Completion Reports (PCRs) were still strong. This might be expected for what are relatively small, but vital, renovation investments in a very large and complex system. Without attempting to re-estimate the RERRs again, in the absence of reliable data for such dispersed projects, this report discusses some methodological issues. The most important is the need to take into account the role of tubewells in fresh groundwater areas and of the benefits from lowered drainage costs in saline groundwater areas. On the basis of the available estimates this report concludes, however, that the most likely rate of return to these projects is around 20 percent, indicating that they were sound investments. 14. Comparing the value of water "saved" by watercourse investments with the costs of incremental surface supplies from a new main storage, watercourse renovation is the heavily favored option. The OFWM strategy is relatively low cost, it is very popular with farmers and government staff, and it is technically effective. However, to become fully efficient, the program needs to move away from subsidies and toward greater private sector involvement at market prices. (c) Groundwater Recharge 15. The immediate impact of the projects on groundwater recharge is identical in both fresh and saline groundwater areas (i.e. recharge is reduced and therefore watertables rise more slowly, or even fall), but the effect of this impact is quite different in the two areas. In fresh groundwater areas these changes had two offsetting effects: (i) farmers relied less on tubewells, thus reducing overall pumping costs (an economic benefit); but (ii) the fall in the level of the watertable increased unit pumping costs. In saline groundwater areas the impact would be wholly beneficial since any reduction in recharge slows the rise in the watertable, even if it does not actually decrease it. The farmer survey results confirm these conclusions: almost all watercourses in fresh groundwater areas reported a decrease in waterlogging. In the saline groundwater areas of the Sindh waterlogging and salinity still increased, but to a lesser extent than on the control watercourses which had not been improved. (d) Canal Rehabilitation and Lining 16. The OED team's visits to a sample of channels and discussions with irrigation staff confirmed reports that, contrary to the intentions of the project designs, canal rehabilitation became canal enlargement. This occurred apparently to allow the provinces to absorb Tarbela water and to establish rights to that water ahead of the formal allocation agreement in 1991. Furthermore, and more seriously, because this took place without any overall system plan, the drainage works to cope with extra flows were not put in place. This resulted in increased waterlogging and salinity problems, including widespread abandonment of affected farms and villages in some of the worst areas. Some have argued that the waterlogging and salinity problem may not be that serious. But, given the huge size of the whole system, the losses from 19 waterlogging and salinity are enormous and the human misery is readily apparent over large areas. 17. Comments from irrigation staff, and the engineering survey confirm that maintenance remains inadequate and that rehabilitated canals have gone back to, or are returning to, their former run-down state. The pattern of periodic and high cost rehabilitation, instead of regular lower cost maintenance, is continuing. 18. The responses of farmers in the survey confirmed field inquiries that ISRP and the canal components of CWMP did not result in improved reliability of canal deliveries at watercourse heads. (e) Institutional Impact 19. The projects attempted to improve both the identified problems at the government level and to introduce new institutions at the watercourse level. The OFWM and CWM projects attempted to tackle the institutional problem of lack of coordination between the Provincial Irrigation Departments (PIDs) and Provincial Agriculture Departments (PADs), but the project- specific arrangements were not able to overcome long-standing and deep divisions between the departments. The ambitions of the projects in this respect were unrealistic, since the degree of institutional change expected was too extreme to be achieved in the limited context of individual projects. The ISRP had more modest objectives of strengthening the PIDs' operations through their direct implementation of the project works, and specifically to improve their operation and maintenance capability. In that, the project was reasonably successful in the institutional sense, but continued lack of funds has restricted maintenance such that the outcome is less positive. 20. On watercourses, the major institutional initiative of establishing water users associations (WUAs) ignored the existence of traditional local institutions. The project designs failed to recognize the existing watercourse committees, and hence the WUAs were not the innovation they purported to be and had a limited effective life-mostly functioning only while the watercourses were being renovated. Furthermore, field inquiries established that an unknown number of WUAs were no more than token institutions formed merely to satisfy the prerequisite that they be established before watercourses could be renovated. A few WUAs are reported to continue to function, but in reality it is often the traditional canal committee which is functioning still, and little incremental institutional growth is attributable to the WUAs. The design of institutional components of such projects needs to be based more closely on existing local institutions and have more modest objectives. (f) Economic and Social Impacts 21. Economic and social impacts are mainly changes in work loads for men and women, health benefits, changes in amenities and communications, and adjustments in equity and social cohesion. 22. Both in the farmers survey and in field discussions with the OED team, farmers reported a large fall in labor demand. following renovation of watercourses. The substantial labor savings on irrigation management and watercourse maintenance outweighed the increase needed to expand cropping. Watercourse disputes were also much reduced, since a lined channel makes 20 illicit activities more difficult. For women the picture is less clear: labor demands increased for field tasks, because cropped areas increased, but the domestic amenity value of improved watercourses was much appreciated by women's groups. 23. The projects had various impacts on local communities. Watercourse renovation was credited with improving communications since it became easier to cross a small channel than an overgrown and embanked unimproved watercourse. However, the re-engineering, renovation and lining of large irrigation channels often resulted in some dislocation of the local communities through which they passed, particularly where inadequate bridging was provided and access to the canals for domestic and livestock purposes was limited. A greater concern for the welfare of local communities is needed when designing canal upgrading works. 24. Lining of watercourses had health benefits in terms of less malaria and lower fly populations in nearby villages, since stagnant swampy areas from seepage and overflow were drained. Even smaller lengths of lining, than the current 15-30 percentage formula, without regard to irrigation benefits, would be worthwhile when such health related problems can be reduced. Future Strategy 25. In 1992 the Bank temporarily ceased lending to Pakistan's agriculture sector given the poor performance of the sector portfolio. The four evaluated projects were a key part of the Bank's assistance strategy for the irrigation subsector, and had been followed by three more similar projects. That strategy was seen as failing to address the causes of the ills of the Indus Basin Irrigation System. Remedial efforts through projects such as this set were seen as not producing the desired results with respect to institutional effectiveness and the long term sustainability problems of the irrigation system. 26. Following extensive discussions between the Bank and the government and other institutions involved in the irrigation and drainage sector, a new approach was agreed in 1994. It consisted of five inter-linked parts: (a) legislation to formalize water trading and recognize individual property rights to water; (b) division of provincial irrigation department functions into public utilities at the irrigation command level and a provincial water authority at the provincial level; (c) develop the provincial water authorities as administratively autonomous agencies to implement the provincial governments' responsibilities in water resource management; (d) active promotion of farmer organizations, which would: (i) represent users in dealing with public utilities; (ii) purchase water delivery services wholesale from the public utilities and deliver them to members; (iii) collect water charges; (iv) be responsible for irrigation system management at the distributary/minor level; and (v) organize O&M of irrigation and on-farm drainage by members; 21 (e) strengthen federal agencies [specifically the Water and Power Development Authority (WAPDA) and the Ministry of Water and Power] to take the lead in planning, provision and O&M of inter-provincial facilities for canal water capture, storage and delivery, and for drainage and flood protection. 27. Understandably, the government and the Bank are approaching reform cautiously and are beginning with a transition program, initially in drainage, through a proposed National Drainage Transition Program (NDTP). The new investment strategy for drainage is a 25 year National Drainage Plan, the centerpiece of which is a national surface drainage system with large disposal drains discharging to the sea. Progress on institutional and policy reforms has been linked by the Bank with future investment strategy for the subsector. Given adequate progress the Bank would then support a more ambitious investment program, including for successor projects to those reviewed in this IER.1 Findings and Lessons Learned 28. At the time the four projects under review were approved they supported some of the most important priorities established in the Revised Action Programme for Irrigated Agriculture (the problems of the public tubewells were addressed through other projects). The impacts of the projects should have been correspondingly relevant and substantial were it not that in the course of implementation, without objection from the Bank, the projects strayed from pursuing the original strategy. The Bank, on its part, was an unhelpful adviser and development partner in not insisting on implementation of the best strategy, established by the Revised Action Programme, against the pressures of special interests to do otherwise. The Bank allowed short term implementation progress to overshadow longer term development impact. Specifically, the opportunity to implement the strategy of the Revised Action Programme, of concentrating OFWM in saline groundwater areas, was abandoned before it had a chance to show its worth. Hence, some of the substantial benefits that RAP foresaw from OWFM investments were foregone. 29. The four projects provide good examples of the need for greater attention to the specification of clear development objectives at the start of project design. The strategy of the Revised Action Programme was to increase water supply at the farm level without the need for investment in new surface storage. The objective which came to dominate the Bank during implementation was the achievement of watercourse improvement targets, regardless of their effect on increasing water supply and on reducing waterlogging and salinity problems. 30. The OFWM projects, and the watercourse component of CWMP, were highly popular with farmers and implementing staff alike, and had strong impacts on the lives and well-being of those farm families who were lucky enough to be included in the programs. The canal rehabilitation and lining work facilitated some significant capacity expansion which was not part of the strategy, and if poor operation standards and low maintenance levels continue, the rehabilitation component of the investments will not be sustained and further rounds of emergency rehabilitation will be needed in due course. 1. (a) Pakistan-irrigation and Drainage: Issues and Options, Report No. 11 884-PAK, March, 1994. (b) Pakistan: National Drainage Transition Program, Executive Project Summary, June 14, 1995. 22 31. Beyond the findings relating to the impacts of the four projects, some significant findings of operational importance are summarized. These relate to the extent that OFWM failings are attributable to rent seeking behavior and the implications for irrigation system investments. 32. (a) Watercourse renovation. The current privatization and public utility strategy for the irrigation system does not obviate further substantial investment at the watercourse level, driven by public policy and funding, in the interests of increasing water supplies to farmers. In a water scarce economy, current system inefficiencies permit enormous benefits to the economy and to rural communities to be achieved by increasing the efficiency of the irrigation system at the watercourse and farm level. To the extent that watercourse improvement can be community based and sustained, this investment is to be preferred over other methods of improving irrigation system efficiency in the publicly managed system. The current pattern of successful, sustainable and popular watercourse renovation provides a suitable, but flawed, model on which to base further investments, with some changes on equity and cost-effectiveness grounds. Without compromising the current pilot strategy for privatizing the lower end of the system, a rejuvenated program of watercourse and on-farm initiatives needs to take into account the following points. 33. The bundling of assistance for the four provinces together in these four projects may have been administratively convenient, and clearly facilitates the Bank's wholesaling of development assistance, but is likely to have reduced the overall impact of the assistance. Operations more closely tailored to physical and social circumstances on the ground in each province, with appropriate timing, would have had better results. Bundling up a time slice of a single uniform program for all provinces, for convenient financing by the federal government and external agencies, risks imposing sub-optimal solutions on very different areas with differing problems. Resources, social factors and the physical state of the irrigation system vary too much from place to place to accommodate such an approach without something being lost. Too many important differences between provinces, and also between areas within provinces, are pushed aside in the search for a common formula that will work reasonably well everywhere, without adequately addressing the precise needs of any one situation. Watercourse lining by a percentage formula is an example. Similarly, recent projects have had some worthwhile and widespread poverty alleviation impact, but have also provided at the same time, without any justification, large transfers of public funds to many of the rural elite. Differentiation would permit, among other things, a more efficient allocation of scarce resources, taking relative needs into account. 34. A new approach is required, disaggregated at least to the provincial level. This should take into account the Bank's current re-emphasis on poverty alleviation as the primary objective, and include equity and the fullest participation of stakeholders in design and implementation. Furthermore, if future Bank financing is to take the poverty, equity and stakeholder points into full account, it will be necessary to break out areas within individual provinces, and even within canal commands, into separate monitorable packages with their own development impact indicators for tracking purposes. 35. An unbundled approach should allow for individual subproject designs for each watercourse, within agreed guidelines for the area, command and province and with full farmer participation. Lining by formula should give way to individualized designs for each watercourse according to what the people want and are prepared to finance and implement. Targeted 23 subsidies may be tailored to real need in terms of holding size and water supply, for example to favor watercourses located far down the supply channel with a less reliable supply. Lining of watercourses primarily for health and amenity purposes should be a valid objective where demanded. Designing investments specifically for individual watercourses would allow for social differences between watercourse, and within watercourses, to be better accommodated, provided a transparent participatory approach is adopted. 36. The long history of institutional experience which attests to the short lives of rural organizations when they are imposed by some outside authority, was borne out again by the WUA experience under the OFWM projects. Such social engineering must be fostered rather than imposed so that organizations arise from within communities in response to their needs. Institutional growth and impact on a sustainable basis in the OFWM program is also undermined by incentives for implementation staff being closely tied to physical progress, watercourse reconstruction, rather than to development impact and institutional progress. The Bank helped to further this distortion of objectives by making it plain that construction progress was the highest 2 priority. Moreover, in the case of the OFWM program, the brushing aside of healthy, functioning, traditional watercourse committees, in favor of frequently token WUAs, was an added mistake. Participation rates worldwide in community based common interest groups are surprisingly low, such that both expectations and the approach adopted, to support future irrigation system investments and management, needs to be carefully based on local realities and not dictated by a broad master plan and physical implementation targets. 37. While the sustainability of the improvements made to watercourses under the projects is not in doubt, the method of financing adopted, with its heavy reliance on external aid and subsidization, is not sustainable. The government cannot afford to extend OFWM to all the remaining unimproved watercourses on that basis. The Bank bears some responsibility for promoting such an unsustainable financing strategy for so long. The Revised Action Programme advocated a credit approach, arguing that OFWM profitability would make it attractive for farmers to borrow. The choice seems to have been between short-term results and establishing a long-term sustainable program, and the Bank chose the short-term objective and persuaded the government to drop the RAP approach. 38. (b) Canal rehabilitation. The social and amenity aspects of canal improvements, in relation to the communities through which canals pass, need to be taken more fully into account during design. The social and economic disruption caused by inadequate bridging is obvious, but other amenity losses seem to occur too frequently when canals are rehabilitated and lined. Irrigation departments should take community concerns more into account by closer consultation. 39. (c) Water resources monitoring and analysis. Monitoring of development impact should be given higher priority, and on a continuous and consistent basis throughout the irrigation system. It is unsatisfactory that after twenty years of investments in conveyance efficiency improvements, to reduce wastage of Pakistan's vital scarce resource, the amounts of water saved remains a matter for analysis and desk estimation, rather than a matter of routine and continual record. Development indicators are sometimes difficult to identify and measure, but in this case 2. The Bank's traditional quite rigid project cycle, which was originally the child of massive engineering projects, dictates that physical implementation be scheduled and the schedule obeyed, whereas community based participatory implementation has a pace of its own. Progress may be uncertain, but it will be more sustained in the long run. Regrettably the rigid engineering schedule is still seen in project designs that purport to be participatory. 24 irrigation water, as a proxy in a water scarce economy for farm income and family welfare, is readily measured. The very large funds involved on a national basis, together with the implications for irrigation and power sector policy and strategy, suggests that more effort to monitor water related matters is fully justified. 40. Development of efficient strategies for water resource investments requires a current and fully independent information base on the system's performance, the parameters involved and the probable impact of the options under consideration. An example of an issue which ought not to be a matter of opinion is provided by the running debate about whether investments in the canal system should be given priority over OFWM activities, or the reverse. The major agencies and individuals in the public and private sectors who have influence over these important decisions have often quite legitimate vested interests to protect, but also, as described in the report, there are other vested interests at work which can have negative impacts on strategies and investment decisions at the local level. There is an urgent need to move the monitoring, analysis and reporting of water resource issues into a highly skilled and disinterested institution outside the influence of political, regional and special interest. Such a detached body could make a substantial contribution to solving some of the hard decisions that will be required as the new strategy for irrigation sector operation and management is gradually put in place. A "Water Resources Institute," comprising in part at the beginning at least sections from existing bodies, and possibly in a not-for-profit NGO format, could become a center of excellence for all matters relating to this national resource, but without any direct interests in the management or financing of the subsector. 25 1. Indus Basin Irrigation Development and the Four Bank-Assisted Projects Introduction and Background 1.1 This impact evaluation report (IER) is concerned with a series of four Bank projects which set out to help implement a new strategy developed in the mid-1970s for improving the efficiency of Pakistan's surface irrigation system, and reducing risks of canal failures by breaches. The Bank has been closely involved with strategy and investment in the vast Indus Basin Irrigation System since the 1950s. In the mid-1970s a change of direction was needed because of the surface irrigation system's faltering performance, high costs of planned new storages, and sustainability problems with the public sector tubewell fields. The Revised Action Programme for Irrigated Agriculture (RAP, 1979) switched emphasis towards rehabilitation of irrigation infrastructure and small-scale investments to improve system efficiency. The four projects evaluated in this report were the first to be financed by the World Bank Group in support of the new strategy. They focused mainly on rehabilitation of collapsing and overburdened irrigation channels and upgrading of watercourses to reduce system inefficiencies. A separate OED impact evaluation report examines the SCARP Transition Pilot Project which addressed public tubewell problems through a strategy based on farmers owning and operating tubewells. 1.2 OED's emerging style of impact evaluation has moved on from the initial pattern in the late 1970s of "re-auditing" an agricultural project some years after an original performance audit. In those days slow maturing irrigation and tree crop projects were chosen for impact evaluation to check whether the original audit's conclusions were borne out by time as projects matured, particularly as the projected benefit streams became actual benefits. In keeping with the Bank's renewed focus on poverty alleviation, impact evaluations are now designed to give more emphasis to the social impacts on the direct beneficiaries, that is income, consumption and general welfare changes. However, these four projects pre-date the explicit identification of poverty alleviation as a prime objective of what are, most simply, engineering infrastructure projects. The project designs did not concern themselves with either income or its equitable distribution. From an accountability perspective, judging the impacts of these projects in hindsight primarily on these grounds is therefore too harsh. But in an irrigation economy which is short of water, every increment of water available to farmers translates into increased income. Water can therefore be taken as a proxy for increased welfare in the Indus Basin Irrigation System, with a caveat with respect to who receives increased water supplies-on equity grounds. 1.3 For lesson learning purposes the evaluation explored welfare impacts through both its field inquiries with groups of farmer beneficiaries and more formerly through the beneficiary survey conducted by consultants. Both of these inquiries sought to establish in qualitative terms (quantitative recall questions being impractical) the impact chain of events from changes in water supplies, to the resulting cropping changes, to uses of increased income. Clear patterns of responses emerged which verified the initial assumption of the evaluation team that incremental water is a reasonable proxy for increased income and community welfare in Pakistan's water- constrained irrigated areas. 1. SCARP Transition Pilot Project (Credit 693-PAK), Impact Evaluation Report, OED, forthcoming. 26 The Sector Setting 1.4 Pakistan is a low income, semi-arid country of about 120 million people with average per capita income of US$440 in 1992-93. The country was rated in 1991 as 120th out of 160 developing countries in terms of UNDP's human development index. About 70 percent of the population live in rural areas. Although the share of agriculture in the economy has been falling, it remains the dominant sector. It produces over a quarter of GDP, employs over half the labor force and contributes 70 percent of exports. About 90 percent of agricultural production is dependent on irrigation which covers about three-quarters of the cultivated area, but the system is generally regarded as being in crisis because of under-investment, poor maintenance and serious operational problems. A healthy irrigated agriculture system is vital to the economy as a whole, to poverty alleviation and to social stability. The Indus Basin Irrigation System (IBIS) 1.5 Agriculture in the Indus Valley has from ancient times depended on irrigation, given low and unreliable rainfall. In the nineteenth century the construction of vast and sophisticated irrigation infrastructure started the gradual and continuing expansion of what is now by far the world's largest integrated system of dams, reservoirs, barrages, link canals and irrigation channels.2 That farsighted colonial strategy was designed to stabilize and ensure food supplies for a traditionally famine-threatened land at the same time as it raised revenue from new irrigation settlements. More recently, in the second half of this century, the system was augmented by thousands of modem tubewells to exploit the abundant groundwater in the Indus Plains. And most recently work has started on construction of the vital main stem drains, without which the system would gradually decline as soil salinity and waterlogging continued to reduce output and even to destroy agricultural communities in many areas. 1.6 Annex B describes the system and its history in greater detail, placing special emphasis on the succeeding investment strategies over the last forty years, in which the World Bank has played a prominent and at times dominant role. This helps to place in context the four projects which are the subject of this impact evaluation. The rest of this introduction summarizes the series of investment strategies and the Bank's role. 1.7 The first half of this century saw IBIS gradually expand as public and private initiatives added new irrigation works to command more barren land. This provided a more secure livelihood to the existing people and established new communities with migrants from overcrowded areas elsewhere. Then the end of the colonial period brought a major change. The division of the sub-continent into India and Pakistan at Partition in 1947 placed the headwaters of the three eastern tributaries of the Indus in India. Eventually the serious political, financial and engineering problems that Partition had brought to IBIS were resolved with the help of the good offices of the Bank in the form of the Indus Waters Treaty and a fund, administered by the Bank and to which it was a major contributor, to finance the works needed to give physical effect to the treaty provisions. 1.8 The post-Partition problem was well in hand by the mid-1960s, with the construction of some of the largest canals, five barrages and Mangla Dam, although a shortage of funds had 2. A commanded area of 16 million hectares, which is a little more than the land areas of Florida and England. 27 postponed the construction of a second huge dam, at Tarbela, and the drainage components of the plan. These proposals then came under further scrutiny as the Bank assisted the government in a new master planning process. The resulting integrated program for water and power development proposed to expand surface supplies with more storage (Tarbela), develop hydro- power and deal with the emerging drainage constraint. Lack of drainage infrastructure was threatening the safe absorption of additional surface irrigation supplies and was contributing to the relentless increase of waterlogging and salinity problems. Project-by-project implementation of the water plan was delayed, despite funding from several donors and the Bank, and that problem, together with operational and cost problems, led by the late 1970s to another master planning exercise to reorder priorities. The Revised Action Programme for Irrigated Agriculture (RAP) and Bank Support 1.9 The Revised Action Programme for Irrigated Agriculture was drawn up with JNDP funding and execution by the Bank. It was generally adopted by the Government in 1980 and provided first for small scale investments, including canal rehabilitation and watercourse renovation, to improve the efficiency of the system. Large, long gestation investments in drainage and new surface storage were deferred to the second half of the 1980s. 1.10 Bank lending for Pakistan irrigation changed significantly as a result of the recommendations of the Revised Action Programme. Starting around 1980 the Bank supported four main activities corresponding to RAP recommendations: (i) watercourse improvement; (ii) irrigation and drainage system rehabilitation; (iii) strengthening of the provincial irrigation and agricultural departments; - and (iv) transition from public to private tubewells. 1.11 The four projects covered by this Impact Evaluation Report (IER) included combinations of the first three items. The Irrigation Systems Rehabilitation Project was designed to renovate neglected supply canals (and some drainage); the two On-Farm Water Management Projects were designed to reduce watercourse flow losses by improving watercourses, and the Command Water Management Project improved both the supply channels and the watercourses. Since there is a continuing debate in Pakistan between those who favor canal investments over watercourse investments, or vice versa, as the most cost-effective way of improving water deliveries to farms and reducing problems caused by losses in the system, this set of projects provides a convenient opportunity to compare the impacts of the two approaches. The full list of projects which stemmed from RAP, divided according to their primary objectives, is shown in Box 1.1. In addition, the Bank continued to support the construction of new drainage, as well as financing minor irrigation in Balochistan, repair of flood damage, and the improvement of the Water and Power Development Authority's (WAPDA) reservoir maintenance facilities. 1.12 A clarification is needed here. "Lost" and "losses" are words frequently used in technical reports and in Bank documents to describe the difference between the volume of water entering and leaving a channel. But only in saline (unusable) groundwater areas is this water "lost" to the system, and where it becomes a liability by contributing to waterlogging. In fresh 28 groundwater areas it recharges the aquifer and is a valuable form of over-season storage, since it can be pumped later when surface supplies are scarce.3 This is an important distinction which figures prominently in this report's findings. Figure 2, with recent data, demonstrates the main flows and reflows within the Indus Basin Irrigation System. Box 1.1: World Bank-assisted Irrigation Projects Implementing RAP Recommendations a Project Title L/C Number Commitment Financial Year (US $ million) Approved (a) Watercourse improvement projects: On-Farm Water Management Project C1163 41.0 1981 Second On-Farm Water Management Project C1603 34.5 1985 Third On-Farm Water Management Project C2245 47.3 1991 1 L3327 36.3 1991 (b) Irrigation and Drainage Rehabilitation Projects Irrigation Systems Rehabilitation Project C1239 40.0 1982 Command Water Management Project C1487 46.5 1984 Second Irrigation Systems Rehabilitation Project C1888 79.5 1988 Fordwah Eastern Sadiqia Project C2410 54.2 1993 (c) Public/Private Tubewell Transition Projects SCARP Transition Pilot Project C 1693 10.0 1986 Private Tubewell Development Project C2004 34.4 1989 Second SCARP Transition Project C2257 zQ 1991 384.2 a. Projects in bold are those evaluatcd in this report. Early Watercourse Improvement Work-USAID's Role 1.13 The external initiative in promoting and financing watercourse improvements in Pakistan was originally taken by the United States Agency for International Development (USAID). In 1968 USAID assisted WAPDA to start a water management research project at the Mona Reclamation Experiment Station (MRES),4 with expatriate research staff provided by Colorado State University. Studies of watercourse conveyance efficiencies revealed a greater problem than previously appreciated (Annex D). Results showed that 44 percent of the water entering the 3. However, a reviewer from the Bank's regional office queries this report's emphasis on this point, noting that seepage in fresh groundwater areas may be less than generally assumed. A footnote to para. 2.13 below discusses this point further. 4. MRES was set up in 1965 in the first Salinity Control and Reclamation Project (SCARP I) area to find out why SCARP programs were not as effective in lowering the watertable and increasing agricultural production as originally expected. 29 watercourse head was lost before it reached the crop (Figure I shows the location of a typical watercourse in the system). Further studies, distributed across the country, were undertaken by WAPDA in the mid-1970s as part of the RAP studies. They found considerable variation in "losses" between provinces but confirmed the original findings: about 45 percent of water entering watercourses did not reach the crops. Furthermore, since RAP estimated that saving water by increasing watercourse efficiency (i.e. reducing "losses") cost about a quarter that of developing new irrigation supplies, on-farm water management activities took on a new urgency. 1.14 In 1976 USAID followed up its support of the work at Mona by agreeing with Government on a five-year nationwide OFWM Pilot Project. This project introduced two major physical improvements to watercourses-masonry lining and improved concrete control structures, orpukka nuccas (further details are in Annex A). Land leveling was another substantial physical component of the early appioach.5 Although there were a number of problems, particularly shortages of adequately trained staff, there was an enthusiastic response from farmers to watercourse renovation (Box 1.2), an increased awareness of the need for water management at the official level and, by the last year of the project (1981), 1,300 of the planned 1,500 watercourses had been improved. (Photos 1-4 show typical examples of unimproved and improved watercourses, although not financed by this USAID-assisted project, and ofpukka nuccas.) Implementation was dominated by watercourse construction work in Punjab and two other components were less successful (there was only limited land leveling, mainly in Sindh with larger farmers, and extension activities for improving on-farm water management were minimal).6 1.15 In 1979 there was a temporary cessation of the US aid program to Pakistan, and thus of USAID support for OFWM. The Bank had been interested in these OFWM developments and had participated with USAID in 1979 in an evaluation of the Pilot Project, which was called "a limited but significant success." Thus, when the Government asked the Bank to finance a project that would build on the experience of the Pilot Project, an identification mission was sent in April 1980. It recommended that there should be two projects, an On-Farm Water Management Project and a Command Water Management Project (CWMP). 5. The initial OFWM concept believed that improving the conveyance efficiency of the watercourse is of little benefit unless it is accompanied by land leveling and to a greater level of precision than the traditional technology can manage. In the original USAID project the watercourse team consisted of four land leveling specialists in a team of eight who introduced a sophisticated laser-based land leveling technology-Precision Land Leveling (PLL), with only two of the team involved with watercourse improvement. There was only one specialist per team for irrigation agronomy. PLL has since declined dramatically in importance in the program, for good and well documented reisons, while watercourse improvement has grown in popularity. PLL is not needed everywhere, not in rice areas for instance. But it is beneficial in cotton-growing areas, especially in those cotton areas with fresh groundwater where it is higher priority than lining watercourses. Annex D discusses the declining priority of PLL. 6. In parallel with this project, the Government of Punjab initiated a "crash program" for "heavy cleaning" of watercourses using technology developed at MRES (principally desilting, remodeling to a uniform channel cross- section and bank compacting). About 9,800 watercourses were cleaned in a year. Attempts to provide credit so farmers could buy their own pukka nuccas largely failed. This program used farmer-donated labor, following the practice started under the.OFWM Pilot Project. 7. Following passage of the Symington Amendment in Congress in connection with Pakistan's nuclear program. ：꽈：.’· 31 February 26,1996 Figure 2: Indus Basin Irrigation System-water flows, "losses" and recoveries million acrefeet (MAF) per year- boxes are proportional Total inflow from rivers Potential irrigation supply 147 "Losses" from seepage, etc. Evaporation Lost supplies Offtake to irrigation systems Canal 106 Dams ftepage '19 15 18l Watercourse inflows Rain Watercmirse seepage 39 39 Evaporation 37% Rain Total ch,4annel and 2 field seepage "losses" 45 63',', Flow onto farms groundwater 67 30 Field seepage Available to crops 69 Riverbed losses Recovered by tubewells Flow to 30 Arabian Sea 36 Source: Data from WAPDA Uanuary 1996) and SAIAN documents. IBRD/OEDD1 02/96 32 Box 1.2: Watercourse "Lining"-An Understatement "Lining" is the word usually used to describe the main investment at the watercourse level in these projects, but this understates the investment to a misleading degree. Although "lining" primarily attacks the problem of seepage losses, it is possible to speculate on a technical seepage barrier (chemical soil treatment, an impervious membrane) applied to the existing earth channels that would eliminate seepage, but not deal with any other problems that the typical well-engineered construction of bricks and mortar conduits solves as well. To get over this semantic problem this report has preferred other words to describe what is done: "renovation, (re)construction," improvement, but even so a description of a typical before and after situation is illuminating. Existing earthen watercourses were put in place by farmers decades ago to distribute the full flow at the mogha in turn to all land holdings (and perhaps also to some domestic and livestock needs in a village). As such they originally reflected old land holding patterns and power structures, with land holders at the head being able to take water at will, if they are powerful enough to ignore the warabandi, leaving farmers at the tail completely at the mercy of the head farmers. The original village may have been built alongside the channel, but by now it has expanded so that the watercourse serves as the main waste disposal system through the heart of the village. Other watercourses were not engineered for hydraulic stability such that excavated silt deposits have over the years been thrown into steep banks on either side of the channel, as much as ten to 15 feet in height. Channel cleaning now requires laborious carrying of silt up and over these high banks, or along the chasm in which the channel runs to a dump site. The channel banks may be overgrown with grasses, shrubs and trees such that the yard-wide trickle of water in the small chasm looks like a natural stream, is an excellent wildlife habitat, but is not an efficient piece of micro-infrastructure in a vast irrigation system. Such an earthen (or katcha) channel can be improved without the use of bricks and mortar as was done in the Accelerated Programme, but when the masonry channel is constructed over a part of the watercourse length a number of other improvements can be introduced: (a) the channel can be engineered properly to all but eliminate siltation; (b) provided farmers agree, a new route may reduce its length or otherwise result in a more efficient channel (the land tenure pattern may have changed since the earth ditch was laid out and village expansion may have engulfed the channel making it an unsanitary village drainage ditch); (c) the high banks of silt can be leveled and brought under cultivation; (d) damage from animals can be eliminated; and (e) pukka nuccas can be installed. These improvements lower the amount of labor needed during irrigation, and maintenance labor is also greatly reduced. Typical masonry watercourses are about 45 centimeters or more deep, and are usually constructed of a single course of brick rendered with mortar. Several masonry watercourses , visited by the mission, by appointment or at random, were over ten years old and had held up well, having received or needing only minor spot repairs to restore them to original condition. 33 444 detoe exavte sl 3. An uimproved watercourse wit itsk moh n uimproe wtore n casm of 34 The (First Bank-assisted) On-Farm Water Management Project (OFWMP-1) 1.16 The unexpected termination of USAID assistance had left a financing gap in the on- going program. The Bank OFWM operation was accelerated to minimize this gap and to get a project into the FY81 lending program, while the CWMP took much longer to prepare and appraise. During preparation and appraisal of OFWMP-I, the project design increasingly departed from a critical recommendation of RAP, that watercourse lining be confined to saline groundwater areas.9 The only deference to the RAP recommendation was that only up to 15 percent of a watercourse length was to be lined in a fresh groundwater area, whereas up to 30 percent of the length could be lined in saline groundwater areas. The project was essentially an engineering project designed to "save" water through infrastructure investments and improved water management practices. The project had three institutional objectives: (i) strengthening the planning and implementation capacity of the OFWM directorates in the four provinces, and the Feder Water Management Cell; (ii) improving coordination between the OFWM directorates and the Provincial Extension Services of Das; and (iii) improving water management (farm demonstrations, water management extension specialists), including the organization of Water User Associations (WUAs) as a condition of the lining investment on each watercourse, with legal identity under new provincial legislation for the purpose. The Second On-Farm Water Management Project (OFWMP-II) 1.17 The rationale for IDA continuing to finance the on-farm water management program was that it would consolidate the gains achieved under earlier projects; changes in the program could be made so that water savings could be translated efficiently into increased agricultural production; maintenance of renovated watercourses could be improved, and a broader range of OFWM practices could be supported. The emphasis was on helping the provinces to establish sustainable OFWM program. Main changes in the design of OFWMP-II were strengthening the water management capability of the extension service and promoting greater farmer participation in WUAs. (Further details on the projects are in Annex B and in the Basic Data Sheet after the Preface of this report, and the relevant outcomes of the projects are discussed in Annexes C and D, and in the following chapter.) The (First) Irrigation Systems Rehabilitation Project (ISRP-1) 1.18 This project originated out of a proposal to support the flood protection activities of the provincial irrigation departments (PIDs) following on the Bank's two flood rehabilitation and restoration projects subsequent to the disastrous 1973 and 1976 floods. These floods had highlighted the crucial need for improving routine maintenance of flood protection and drainage facilities if future flood damage was to be avoided. As preparation proceeded it became clear that there was equal need for similar improvements in the irrigation operations of the PIDs. RAP had pointed to the problems created by the lack of adequate maintenance and stressed the vital 8. "On-Farm" is itself an intriguing misnomer: although farm management advice is part of the package at time in different projects in the series, little was done and the main investment is always off-farm in construction/renovation of the watercourse, which conveys water to the farms. 9. The case for including FGW areas without access to tubewell water in project coverage was not strengthened by using farm models of fresh groundwater areas with tubewells to calculate project benefits. Although it is not specified in the main text, the footnote in the SAR to the cost table, Annex I, Table 8, assumes the ratio of FGW:SGW to be 75:25 in Punjab, 50:50 in Sindh and 70:30 in NWFP. 35 need to strengthen the PIDs in the area. The project scope was thus broadened to cover the activities of PIDs as a whole, rather than giving emphasis to one particular aspect of their operations. ISRP was also seen as an important step in evolving a new balance in the Bank's lending to the water sector generally and providing an opportunity for a continuing dialogue with the Government of Pakistan (GOP) on major policy issues as they affected the water sector, particularly those given emphasis in RAP. ISRP is also important in this set of four projects because it represents one school of thought in Pakistan's planning circles: that investments in the upstream end of the irrigation system (i.e. in major canals and structures) are more worthwhile and should have higher priority than downstream investments (i.e. on the watercourse). The Command Water Management Project 1.19 Command water management was concived as a basin-wide comprehensive water management program made up of sub-projects of 100,000 acres (45,000 ha) or less within canal commands.10 These sub-projects would include improvements throughout the whole irrigation distribution system of a canal command, both upstream of the mogha in that part of the distribution system within the jurisdiction of the PIDs, as well as below the mogha within the watercourse command, along with programs for strengthening agricultural inputs and services. This was the concept originally articulated in RAP which had envisaged the integration of specialized agencies at the canal command level in order to strengthen the management of irrigated agriculture. It built on the experiences of OFWM-I and ISRP, but was innovative in attempting to bring together on a pilot basis, for the first time, the other major policy and institutional recommendations of RAP. The problem was clear: fragmented institutional arrangements resulted in inefficiencies in the delivery of water and non-water inputs to farmers. The solution was not clear: the arrangements provided for in the project were an attempt to improve coordination between PIDs, OFWM directorates of PADs and extension departments. PIDs would be responsible for canal improvements and OFWM directorates for on-farm works. Coordination would be at three levels: federal, provincial and subproject. It was hoped to develop a system and organization that could be replicated. Subsequent Projects 1.20 Although the subsequent projects are still under implementation and are not covered by this impact evaluation, it should be noted that the Second OFWM project was followed by a third (Ln. 3327/Cr. 2245-PAK), which became effective in May 1992, and that the ISRP was also followed by a second ISRP (Cr. 1888-PAK), which became effective in January 1989. The follow-up to the CWM project was canceled following the cessation of all new Bank lending to the sector in 1992, as discussed in Chapter 6. 1.21 The rest of this report examines: (a) how successful the projects were in achieving the water saving objective; (b) the impact this had on reducing waterlogging and salinity problems, in so far as they are caused by channel losses; (c) the social and economic impacts of these changes; and (d) the institutional impacts. The report concludes in Chapter 6 with a discussion of the new strategy for the future. 10. Compared with the RAP recommendation of 500,000 acres. 36 2. Impact of the Projects on Watercourse Losses and Groundwater Levels Introduction 2.1 The Revised Action Programme for Irrigated Agriculture (RAP), emphasized the conservation of irrigation supplies through conveyance system improvement, and on producing more output per unit of water through improved management and agricultural extension advice. Fundamental to this emphasis was the conclusion that the cost of water saved through watercourse improvement was about 25 percent of the cost of developing an equivalent volume of new water supplies (by capturing more of the uncontrolled summer flood discharges in a new storage reservoir). Hence water management projects were "an efficient vehicle for supplying additional water to the chronically water-short lands of the Indus Basin."II As a substantial bonus to the primary water saving objective, RAP also recommended that investments be concentrated in saline groundwater areas. Here channel losses were (and still are) a major cause of rising watertables leading to increased waterlogging and salinity, and all the hardship and social disruption that occurs with falling production and eventual abandonment of villages as buildings collapse. 2.2 To clarify later discussion Table 2.1 sets out the main gains and losses experienced by the system and farmers when channel seepage is reduced, according to whether a channel is in a saline or fresh groundwater area. The table shows why RAP emphasized channel renovation in saline groundwater areas as the first priority, and also demonstrates the strong private short term incentives in fresh groundwater areas for watercourse renovation (especially from local leaders who are likely to be tubewell owners), which resulted in the RAP strategy being disregarded. 2.3 The findings of the monitoring under OFWM-I and II and the results of the impact evaluation farmer survey with respect to watercourse conveyance efficiencies are presented in detail in Annex G. The survey is described in Box 2.1. The social and economic impacts of the projects are examined in the next chapter. Water Losses and Savings 2.4 Water flows in the downstream part of the system are not routinely monitored and there are few measurements available of watercourse flows. Hence qualitative information from farmers is the more usual means of investigating changes in water flows to farms. Although the equipment for measuring small flows of 1-3 cusecs is portable and quite simple, to track change it would be necessary to take a large number of measurements over at least a cropping season since canal flows can vary widely over short periods. This would be necessary in both project watercourses and "control watercourses" which had not received the investments. Such a major measuring exercise is beyond the scope and resources of this impact evaluation, but should be a part of an adequate monitoring system, on a sample basis, for the whole system. 11. WAPDA, November 1977, An Initial View of Water and Associated Power Developmentsfor RAP and Perspective Plan, (quoted in the SAR of OFWM-I1, p. 5). 37 Table 2.1: Gains (in italics) and Loss Effects from Reduced Channel Seepage ffects bY Groundwater QuaijU Qf Location Those Affected Saline Fresh 1. Surface flow to farm Increase Increase Most farmers, during fixed irrigation especially at tails. turn (i.e. not on demand) 2. Tubewell pumping costs: (a) Total (less pumping-given Not applicable Down Tubewell owners substitution) (b) Unit costs (lower watertable) Not applicable Up Tubewell owners 3. Watertable levels Down/slower rise Down/lower Tubewell owners & (reduced waterlogging recharge all farmers in saline and salinity risk) areas. 4. Watertable resource Not applicable Reduced Tubewell owners and (inter-seasonal their customers storage/on-demand supplies) 2.5 To obtain information from beneficiaries at the village level on project impacts, OED commissioned a survey of groups of farmers and women on 240 watercourses, of which 210 benefited from the projects and 30 were on control watercourses which had not benefited. The main survey was confined to Punjab and Sindh provinces. The survey compared four different types of project treatment, depending on whether it was above (ISRP and part of CWMP) or below (OFWM) the mogha, with a control group (called "none/none", meaning there was no treatment above/below the mogha). These were ISRP/none (i.e. canal improvements/no treatment below the mogha); ISRP/OFWM; none/OFWM and CWMP/CWMP. 2.6 Water Supply upstream of the mogha: virtually all farmers on watercourses supplied by channels rehabilitated or lined under ISRP or CWMP reported no increase in the supply level of these channels and no improvement in the reliability of water supplies (conversely, it seems likely that without JSRP and CWMP improvements the supply situation might have deteriorated). However, where ISRP canal improvements coincided with watercourse renovation under OFWM-I and II, and under CWMP, more farmers reported increases in water supply to their farms, and in supply reliability, than where OFWM was the only type of improvement. The 38 main cause of unreliability is erratic flows above the distributary head, which neither ISRP nor CWMP addressed. Box 2.1: The Impact Evaluation Farmer Survey Objectives To obtain information at the watercourse level on the impacts of the projects under study as they have developed during the years since the projects were completed, the IER mission retained the services of Enterprise & Development Consulting (Pvt) Limited, a local consulting firm based in Islamabad, to carry out a farmer survey on its behalf. The survey complemented the IER field mission and the large volume of secondary sources by directly soliciting impressions of the projects' impacts from both male and female beneficiaries; the interviews were carried out in their local languages. Special objectives that affected the design of the survey included providing an unbiased sample, providing built-in control observations, and focusing on project impacts (rather than socio-economic indicators). The lack of credible benchmark M&E data necessitated extra attention to the design of survey controls. The survey also was intended to contribute data to the ongoing debate as to whether rehabilitation of major supply channels or investment at the watercourse level (i.e. the OFWM program) is the preferable strategy when resources are scarce. Impacts Measured by The Survey The survey attempted to collect farmers' perceptions of impacts directly. The indicators included in the survey are intended to reveal qualitative directions of change in selected areas of project impact without incurring the cost and data problems of a more comprehensive socio- economic investigation. This design strategy limited the IER's descriptive narrative of project beneficiaries, but made possible a more focused investigation of project impacts. Design The details of the survey design are provided in Appendix I to Annex G and a copy of the survey form is available on file in OEDD1. The sampling frame was designed to allow comparisons to be made between provinces, between agro-climatic cropping zones, and between type of project intervention. It was also designed to allow testing of hypotheses with three different units of observation-watercourse, watercourse reach, and individual farmers. Since Punjab and Sindh contain over 80 percent of the watercourses in Pakistan, the survey was restricted to these two provinces. A small number of watercourses was sampled in NWFP for descriptive purposes only. In total 240 watercourses were included in the sample, 120 in Punjab, 105 in Sindh and 15 in NWFP. On 152 of these watercourses a group of women were interviewed with a separate questionnaire. 2.7 Changes in water supply at the mogha: changing mogha discharges was not a project objective and only 12 percent of project farmers reported that their sanctioned discharge had 39 been altered, equally divided between increases and decreases. Nevertheless as many as one- third of farmers experienced changes in actual mogha discharges, with two-thirds being increases. There was little evidence of tampering with mogha from the survey, but this was in marked contrast to the field observations that on some watercourses most moghas had been tampered (i.e. the masonry apertures increased in size-usually under water out of immediate sight). Reticence to admit to tampering seems the obvious explanation. 2.8 Changes in water supply to the farm: farmers were equally divided as to whether water supply to their farms had increased or decreased. A third of the farmers on control watercourses reported that the supply to the farm had increased (a result of increased availability of tubewell water). The survey evidence that increased supply to the farm was the result of improved conveyance efficiency (reduced losses) is surprisingly slight: only 22 percent of project watercourses which had neither increased supply at the mogha nor increased tubewell water reported increased water at the farm. As would be expected, projects which provided improvements of both the supply channels and the watercourses had a slightly higher proportion of reported supply increases than projects which were limited to improvements on only one side of the mogha (55 to 43 percent). 2.9 Watercourse renovation and lining percentage: although the survey provided no evidence concerning the relationship between variations in water loss reduction and the proportion of the watercourse that was lined, it does provide data on the variability of the proportion lined. The original formula was that 15 percent of the watercourse would be lined in areas underlain by fresh groundwater and 30 percent in areas with saline groundwater---4he difference being related to the greater (non-quantified) benefits of reducing losses in saline groundwater areas.12 In Punjab the criterion was ignored during implementation, there being no significant difference in the lining proportion between the two groundwater zones. In Sindh the planned proportions were more or less adhered to. 2.10 To support the qualitative findings of the survey, the results of earlier monitoring of watercourse projects are reviewed to arrive at an estimate of water savings. The conclusion of the review of OFWM monitoring results in Annex D (d) is that the type of watercourse improvement adopted in all the OFWM projects increases the conveyance efficiency of the watercourse by at most about 15 percentage points, from a pre-project level of about 60 percent to a post-project level of 75 percent of mogha flows.13 This represents a substantial 25 percent increase in the amount of water delivered to the field, but also a substantial amount of seepage continues because only the first part of a watercourse is lined.14 These data are based on results from only 56 watercourses out of about 25,000 watercourses improved to date (36 from the evaluation of the OFWM Pilot Project and 20 under OFWM-Iy-details are in Annex D (d). This is an extraordinarily scanty information base after all these years, and a severe criticism of 12. However, the actual lengths renovated varied greatly because watercourses vary in total length. The farmer survey indicated the following averages: Punjab-5,700 meters; Sindh-3,000 meters; and NWFP-1,850 meters. 13. Conveyance efficiency is defined as the ratio of the water delivered to the field to that diverted at the mogha. 14. In a similar project in India (Punjab Irrigation and Drainage Project, World Bank Report No. 8056), it is reported that measurements of water savings show an increase of 20 percent in the amount of water reaching the plant after lining. 40 the monitoring effort, which could produce so little on this major impact. For example, it is impossible to refine the figures for losses in lined watercourses according to the amount of lining done per watercourse since no such records are available, and it is not known what affect an increase in the proportion of lining has on losses.16 2.11 The percentage of a watercourse's length that has been lined under the projects was determined by the across-the-board decisions for saline or fresh groundwater areas. With a heavy subsidy and enthusiastic response from the farmers, the lining percentage as decided upon were rationing decisions in the face of excess demand, since taking that decision for individual watercourses, based on the circumstances of each, would not be an objective process given rent seeking behavior. From field enquiry it was clear that there could in many cases be an engineering, public health and economic solution which was different from the set lining percentage. This is especially captured in the views of one senior Pakistani official, interviewed for this report, as detailed in Annex D (b). These views may be summarized as advocating an approach which can be characterized as "every case on its technical merits," rather than the uniform formula used to date. In contrast the lining formula approach was a pragmatic reaction to the subsidy on lining creating excessive demand and the difficulty of overcoming corrupt influences on individual lining decisions. It should be noted also that there is ncthing to prevent farmers agreeing amongst themselves to line more of a watercourse using there own resources, although no such cases were observed. 2.12 Turning back to water savings, based on the watercourse monitoring data, the three projects with watercourse components are together estimated to have "saved" 2.0 MAF per year. Details of these estimates are in Box 2.2. This compares with estimates of total seepage losses in IBIS as a whole which range from 47 MAF (including from fields)'7 to 69 MAF.18 With 25,000 watercourses renovated to date from all programs, not just those of the Bank, the total water saving of the whole OFWM program would be 4.7 MAF (5.8 Bcm) per year, assuming pro rata results. This is a little over half the RAP expectation of about 8.0 MAF of annual savings from the regular OFWM program by 1990.19 More significantly, it is more than the amount of additional surface water (3.5 MAF) that would have become available at the mogha from the construction of the Kalabagh dam, the proposed next major main stem storage. 15. The evaluation team is concerned that such a fundamental part of the Indus Basin water balance is not kept under closer and routine review and notes that there is no suitable independent institution in the country charged with such a responsibility. 16. Nevertheless, the proportion of watercourse length lined increased over time because the maximum permitted increased with each project. But even after controlling for project, the percentage of total length lined increased each year and in each province. After controlling for project and fiscal year, watercourses in Punjab had the highest average coverage of lining in all provinces (Annex F on the fanner survey results). 17. Ahmad, Masood and Kutcher, Gary P, Irrigation Planning with Environmental Considerations, A Case Study of Pakistan's Indus Basin; World Bank Technical Paper No. 166, April 1992. 18. WAPDA communication. 19. This figure is based on RAP's estimates of 85 percent loss reduction per watercourse, which was estimated to represent 325 acre feet, and 24,000 watercourses lined by 1990. 41 Box 2.2: Water Savings Estimates The average annual water consumption per watercourse in the Indus basin is about 1,250 acre feet (equivalent to about 1,500 cubic meters). This figure is well above what aggregate data would suggest, and was derived from the general estimate of approximately 105,000 watercourses in the Indus Basin (Table 1.1 of the SAR of OFWM-III). The average surface water releases at watercourse head are 14 MAF. This is made up of 81 MAF irrigation releases plus 30 MAF from tubewells. This gives an average of 71,060 acre feet per watercourse at the head (before losses), equivalent to 530 cusec days per year. Assuming a watercourse runs in most cases for 290 days per year, a discharge of about 1.76 cusecs would be required. This compares well with observed discharges of watercourses. The average discharge of the sample watercourses in the evaluation data reviewed by the evaluation team is about 1.8 cusecs, equivalent to say 1,250 acre feet per year. (There is not enough surface water to provide this much to 105,000 watercourses, and hence distributary operation is on a rotation system.) With 1,250 acre feet as the average release at the mogha, increasing the watercourse delivery efficiency from 60 to 75 percent is equivalent on average to saving 188 acre feet, or 230 cubic meters, per watercourse per year. On this basis the water saved by OFWM-I works out to be 0.68 million acre feet (MAF; 0.84 Bcm) per year, and by OFWM-II, 0.97 MAF (1.12 Bcm) per year. In addition, it is estimated that the water savings under CWMP are 0.4 MAF per year. This is made up of about 0.13 MAF from canal lining preventing water losses in transit upstream of moghas (more than twice the SAR estimate as the length of canals that were lined doubled), and another 0.27 MAF per year from the watercourse improvement program, for a total of 0.40 MAF. This is 41 percent more than the estimate in the SAR, as follows:20 Water Saving from CWMP (thousand acre feet) SAR IER Canal Lining 50 130 Watercourse Renovation 217 214 Total 287 404 Adding this saving to the 1.65 MAF already estimated as saved by the two OFWM projects, the total water saving from all the projects included in this impact evaluation is just over 2.0 MAF per year. 2.13 Unfortunately, much of the "saved" water is in areas of fresh groundwater, where the "saving" has contributed to the dramatic lowering of the watertable that has occurred in recent 20. The SAR estimates of the likely water savings are acceptable, given the lack of firm data on which to base them at the time of appraisal. The Monitoring and Evaluation (M&E) results came too late to influence these assumptions, except for OFWM II. The various assumptions made in the projects under review are summarized in Annex C (a). 42 years.21 Had the projects concentrated on the saline groundwater areas, as recommended in RAP, the OFWM program could have been rated a major success. The lesson of this experience has been taken into consideration most recently in the design of the proposed National Drainage Transition Project which includes a component for the lining of watercourses, but in saline groundwater areas only.23 2.14 On tubewells and groundwater quality, the farmer survey found that over 40 percent of project watercourses surveyed reported increased reliance on tubewell water. There were major provincial differences. In Sindh, over half the farmers reported a decreased reliance on tubewells to supplement surface water. This was possibly a reflection of increasing waterlogging and salinity problems, even though in Sindh there was a marked bias towards the renovation of watercourses in fresh groundwater areas, although the next paragraph casts doubt on whether this bias was more in reporting than in reality. In Punjab, by contrast, farmers reported a near universal reliance on additional tubewell water where watercourses had not been renovated, but, with renovation, only about 40 percent used additional tubewell water. The clear implication of these findings for Punjab is that watercourse renovation was a partial substitute for reliance on tubewells. This was a logical farmer response in the Punjab since water from renovated watercourses does not cost farmers any more, whereas tubewell water has pumping (or purchase) costs. 2.15 Unit costs are used to compare the efficiency of the watercourse improvement. The estimated cost per annual acre foot saved is based on the cost of watercourse renovation given in the SAR of OFWM-III, which is in 1991 prices (Table 2.2). 2.16 The project cost for civil works on watercourses, including the farmer contribution, 24 averages Rs 332,000 per watercourse. The capital cost per acre foot saved per year is Rs 21. A reviewer has commented that "the recent lowering of groundwater tables is mainly due to increased groundwater pumping by private tubewells and not because of water savings in the surface system. The groundwater pumping would be even higher without watercourse improvement to offset the high operational losses and for supplying the same quality of water at the root zone. The real reason is the growth in private tubewells which are increasing by more than 6 percent annually over the past two decades and the pumping levels are reaching their potential now. This means that regulation of groundwater use is becoming more important than before." 22. For a comparable project in India, the SAR estimated that the economic rate of return to watercourse lining in saline groundwater areas was twice that in fresh groundwater areas. Another paper prepared by the HR Wallingford Group, U.K., outlining a method for the economic evaluation of canal and watercourse lining, estimated that the modal value of the IRR for watercourse lining in India was less than one percent, in both fresh and saline groundwater areas. In the Indian projects the proportion of the watercourse lining is about 75 percent compared with 15-25 percent in Pakistan. Studies in India have also shown that linings deteriorate much more rapidly than in Pakistan. Within a period of six years in some areas seepage rates from lined distributaries and watercourses have risen to the same or even higher levels than those from unlined channels. (See Chancellor, Felicity, Economics ofCanal Lining-An Evaluation Method, in Proceedings ofa Workshop on Canal Lining and Seepage, 18-21 October, 1993, International Waterlogging and Salinity Research Institute, Lahore, and HR Wallingford, U.K.) 23. A reviewer from the Bank's regional office has suggested that seepage losses from watercourses are not such a significant factor as this report assumes, compared with operational losses. The factor of concern is the wetted area of the watercourse, both its area (mostly determined by length) and the duration of wetting, which decreases with distance from the mogha. This reviewer notes that watercourses are often dry for long periods and the lower end of watercourses is seldom wet long enough to connect with the watertable. According to this reviewer, this report overemphasizes the significance of watercourse seepage as a contributor to recharge of fresh groundwater. 24. This figure agrees quite closely with recent actual costs. The average cost of civil works in current rupees per watercourse improved in the three fiscal years 91-93 was Rs 208,756 (OFWM-III, Federal Coordination Unit, data of 43 1,766. This represents an annual cost of Rs 215 for adding an additional acre foot of water at the farm outlet.25 No account is taken in this figure of future maintenance costs, since maintenance is needed with or without renovation, and maintenance costs tend to be less with renovation than without. This cost of water "saved" is towards the low end of estimates of the prices charged for irrigation water within IBIS. Prices in informal local water markets show large variations, with privately pumped tubewell water ranging from Rs 100 to 400 per acre foot and informal water sales that take place "illegally" along a watercourse) ranging from Rs 100 to 700 per acre foot.2 To the farmer, the cost of watercourse renovation has been only half this Rs. 215 figure, so that it is no wonder that the OFWM projects are so popular. The cost recovery system, which explains why the farmer pays only half the cost, is described in Annex D (c).27 Table 2.2: Cost of Watercourse Renovation in OFWM-lI (Rupees thousand per watercourse) Punjab Sindh Civil Works 204 252 Farmer Cash Contribution 34 - Donated Labor 96 76 Total 334 328 The Impact on Groundwater Recharge and Tubewell Use 2.17 This section will relate the dismal tale of how the RAP emphasis on reducing channel losses in saline groundwater areas, and not in fresh groundwater areas, was abandoned during design and implementation of the Bank-assisted projects, despite that emphasis remaining the highest priority to this day. Thus a 1992 Bank report by Ahmad and Kutcher,28 based on re- running the Indus Basin Model, stated: "Canal losses through fresh groundwater areas is not of concern: they serve to offset tubewell pumping and hence dampen the process of aquifer mining. Of course, there is a delay in achieving the ultimate benefits from use of the water as it goes into temporary storage instead of immediate use [which inter-seasonal transfer is a benefit]. Canal losses through saline areas are much more destructive. The water is not only lost forever for purposes of irrigation as it mixes with highly saline groundwater, it is environmentally detrimental as it contributes to waterlogging" (page 80, parenthetical remark and emphasis added). The economic and financial costs to Pakistan resulting from the failure to follow January, 1995), with the farmer's contribution in cash and kind (estimated at Rs 74-125,000), the total cost is Rs 283,000-334,000. 25. Annualized over 18 years at ten percent. 26. World Bank, March 1994, Pakistan: Irrigation and Drainage: Issues and Options, para. 3.25 and footnote. 27. Annex C (c) is based on a larger, more detailed review of how vital parameters (lining percentage, cost recovery and the areas to qualify for watercourse renovation) were determined for the Bank-assisted projects in what now appears to have been an ad hoc way, with little reference to priorities and the optimizing strategy of RAP. For the interested reader, that review is at Annex B. It demonstrates how the USSI billion, or more, program for watercourse renovation wandered away from the sound criteria and principles which had been established under RAP. 28. See Ahmad and Kutcher, op. cit. 44 through on a major strategy imperative of RAP (giving priority to saline groundwater areas) have been, and will continue to be, very substantial. 2.18 Another quotation from this report is also pertinent: "The largest percentage losses in the irrigation system occur below the level of canals on watercourses and fields. It has long been recognized that so-called on-farm water management (OFWM) projects have great promise for increasing the delivery efficiency and reducing losses to groundwater" (page 84, emphasis added). 2.19 While the reduction of watercourse losses was initially seen primarily as a substitute for additional surface storage, it had an almost equally important role, in the eyes of the authors of the RAP, in assisting in the control of waterlogging and salinity. Masood and Kutcher estimate that 28 percent of all surface water deliveries seep into the groundwater through watercourse and field losses,29 and that in Sindh two-thirds of all groundwater recharge comes from watercourse seepage and field losses. If even a proportion of this seepage could be eliminated by watercourse lining, the drainage problem, if not solved, would be greatly relieved and at a much lower cost.30 Surface drainage costs an estimated US $1,200 per acre (approximately Rs 4,000) at present prices compared with watercourse lining at about US $175 (Rs 700) per acre. 2.20 At the time watercourse lining was being proposed by RAP, the drainage problem in areas underlain by fresh groundwater was being tackled by a combination of surface and vertical drainage, the latter directly by the installation of fields of public tubewells and indirectly through the rapid spread of private tubewells. However, the growing problems of public tubewells were also becoming known and a policy to phase them out in favor of more private tubewells was strongly recommended in the RAP report. Either way, the need to put more resources into drainage of fresh groundwater areas was given a much lower priority than tackling the problem of drainage in saline groundwater areas, where a sustainable solution had still to be found. Additional supplies of surface water from Tarbela had recently become available in 1976. Despite the recommendations of the Indus Special Study, that these additional supplies not be allocated to saline groundwater areas unless drainage had been provided, saline areas were receiving extra supplies, thus adding to the drainage problem. Watercourse renovation to cut seepage, if not the long-term solution to the drainage problem, was seen at least as an interim supplementary measure which would help to reduce the rise in groundwater levels. The RAP report therefore recommended that the watercourse renovation program should initially be confined to saline groundwater areas. There was the additional factor that reducing seepage losses in fresh groundwater areas, where there was a heavy concentration of private tubewells, would be counter-productive as it would reduce the recharge available for pumping. 2.21 The farmer survey produced a big difference in farmer responses to questions relating to waterlogging and salinity between provinces. In Punjab, 80 percent of project farmer groups reported a decrease in waterlogging; all the control watercourse reported no increase in waterlogging, and there was no reported impact on salinization. In Sindh, a quarter of project 29. Recent WAPDA figures raise this estimate to 32 percent (26/81) -see Figure 1. 30. Ahmad and Kutcher were pessimistic that watercourse lining in Sindh saline areas, although having a high ERR, would reduce the drainage problem. They concluded that only surface drainage could prevent abandonment of much of this part of Sindh. 31. World Bank, Executive Project Summary, Pakistan: National Drainage Transition Project, June 1995. 45 farmers reported a decrease in waterlogged areas, almost all in fresh groundwater areas, but at the same time almost one-third reported an increase in salinization. Every control watercourse reported increased waterlogging. Thus the projects had a limited impact on waterlogging and salinity in Punjab and Sindh fresh groundwater areas, but practically no impact in Sindh saline groundwater areas, where 93 percent of farmer groups reported that salinity had increased. 2.22 The first OFWM program, assisted by USAID, began in 1976 before the implications of locating it in either areas of fresh or saline groundwater had been fully appreciated. The main problem was to assure farmer acceptance of the new program, a problem that was successfully overcome by not recovering costs from farmers-a substantial subsidy. The watercourses for lining were selected without reference to groundwater quality, but by the time preparation began on the Bank-assisted OFWM project in late 1979 the Bank must have been well aware of the RAP recommendations, since they were within months of being accepted by the Federal Government. 2.23 The Bank's original concept of the project followed the RAP recommendations: proposing that watercourses in fresh groundwater areas only should receive the "heavy cleaning" treatment, while watercourses in saline groundwater areas should receive 50 percent lining. The concept was revised during the identification mission of May 1980 with fresh groundwater watercourses receiving 10 percent lining and saline ones being cut to 30 percent. There is nothing in the files to indicate the reasons for this change, or even any comment that it was seen as significant. These percentages remained until negotiations when, at GOP's request, the lining percentage for watercourses in fresh groundwater areas was raised to 15 percent, and the project cost was raised to accommodate this request. Again, this is indicative that GOP had also lost sight of one of the two main purposes of the OFWM program and that the Bank was no longer supportive of it either. 2.24 Although preference was supposed to be given to lining watercourses in saline groundwater areas, the SAR for OFWM I implicitly assumed that in Punjab 75 percent of the watercourses would be in fresh groundwater areas, and in Sindh 50 percent. During implementation, this recommendation was ignored, at least in Sindh. The results of the farmer survey discussed in Annex G show that in Punjab only 25 percent of the project watercourses were in fresh groundwater areas whereas in Sindh no less than 65 percent of the project watercourses were in fresh groundwater areas. The guidelines as to the proportions to be lined do not seem to have been followed in Punjab since the farmer survey showed no significant difference in the proportion lined between fresh and saline groundwater areas,32 but in Sindh the proportions were 23 and 34 percent respectively. The reasons for this discrepancy are clear from supervision reports. The Bank was more interested in meeting the physical targets for numbers of watercourses to be renovated than in their location. The OFWM Directorates, at least in the early stages, were selecting watercourses on a first-come, first-served basis (as confirmed by field enquiries), again with little attention to the groundwater criterion. The OFWM projects therefore threw away the opportunity to ameliorate drainage problems. They also threw away the other main benefit-water savings as a substitute for additional costly surface storage-since losses prevented in fresh groundwater areas merely substitute for private tubewell pumping (when needed) by farmers and do not represent a net gain as they do in saline groundwater areas. 32. One reasonable explanation for this, from a well informed source, was that there were political pressures to pursue the OFWM program "uniformly" throughout the Punjab. 46 2.25 The watercourses renovated under CWMP were those in the commands selected, and groundwater quality was not one of the selection criteria for commands-only one of the five commands in Punjab and Sindh was underlain by saline groundwater. Again the opportunity to use watercourse improvements to reduce drainage needs and as a storage substitute was lost. 33 Moreover the conclusion of an earlier OED study, that the long run solution for saline areas remains to be found, has been reinforced nine years later. 2.26 The survey threw up a strange anomaly: 48 percent of the farmers groups on OFWM watercourses and 77 percent of those on CWMP watercourses were classified as underlain by saline groundwater (by definition too saline to use for irrigation) reported tubewells. The most likely explanation is that OFWM Directorate staff, urged on and supported by farmers, recorded the groundwater as saline in order to allow a greater lining percentage. This suggests that the opportunity to use the watercourse renovation program as a contribution to solving drainage problems in real saline groundwater areas was never seriously pursued. 2.27 The combined reduction in "losses" achieved by the four projects is estimated at just over 2.0 MAF annually, which is 3.0 to 4.5 percent of the range of annual channel seepage losses in the Indus Basin as a whole of 47-69 MAF (para. 2.9). On the basis of the number of renovated watercourses in fresh and saline areas and their relative contribution to seepage losses, the projects reduced seepage losses in saline areas by about 30 percent in Punjab and one percent in Sindh. It is impossible to estimate their impact in fresh groundwater areas since it not known what proportion was recovered by tubewell pumping. From the farmer survey only about 40 percent of the renovated watercourses reduced their reliance on tubewells after the renovation. Hence, contrary to the intentions of RAP, the Bank has been an unhelpful adviser and development partner by not insisting on implementation of the best strategy. The more recent report of Masood and Kutcher has confirmed the vital importance of concentrating limited investment resources on reducing channel losses in saline groundwater areas. 33. See OED report-World Bank in Pakistan: Review ofa Relationship 1960-1984 (Report No. 6048 of January 27, 1986) In para. 2.45, and also in the PAR on Tarbela, Report No. 6398, there is a discussion of the difficulty of arriving at a technical and institutionally viable system for controlling waterlogging and salinity. "After two decades of controversy and delay, and many studies, there now appears to be substantial agreement between the Government and the Bank on strategy covering areas that are not saline: public tubewells would be gradually phased out and replaced by privately managed tubewells, plus there would be a program of improved canal design and more efficient on-farm water management. It appears, however that no long-run program has yet been agreed on for areas with saline water." This was being written some twenty years after the Indus Special Study (ISS) report. Even today there is still no agreement. The Bank's 1994 report Pakistan: Irrigation and Drainage: Issues and Options "Although Pakistan has considerable experience in drainage, there still is no agreement on what is the best technology, particularly for achieving a desirable salt balance in the loot zone" (para. 2.26). 34. According to the farmer survey 56 percent of the renovated watercourses were in areas which were reported as being underlain by saline groundwater, 90 percent of which were in Punjab (although, for reasons given in Box 3.2, this is believed to be an overestimate). About 1.1 MAF of this saving would thus be in saline groundwater areas, say 1.0 MAF in Punjab and 0.1 MAF in Sindh. According to Ahmad and Kutcher seepage from watercourses is 3.4 MAF in Punjab saline areas and 9.7 MAF in Sindh. Thus the projects saved 30 percent of Punjab saline area losses and one percent of Sindh's. 47 3. Social and Economic Impacts Introduction 3.1 From ancient times to the beginnings of the modem system in the nineteenth century, irrigation in the Indus Plains has been dominated by subsistence food production. The enormous expansion and modernization of irrigation that began in the nineteenth century continued to be directed primarily towards food production to avoid catastrophic famines. The water savings under these four projects have also contributed to food production, but high value cash crops such as fruits and sugar cane (in the neighborhood of factories), were frequently mentioned in 35 the field as having expanded as a result of the projects. Farmers groups interviewed by the evaluation team were explicit on the uses to which they put the resulting increased income, with general agreement on improving houses, electricity supply, education (including for girls), transport (motorcycles) and better clothes, including for women. Questions of equity aside (which are discussed further below), the projects clearly contributed to accelerating income and consumption growth in an agricultural sector which has experienced comparatively high growth rates for some time. 3.2 Overall, there can be little doubt that the wide popularity of the OFWM program is due not only to the subsidy and immediate production benefits it delivers to land owners, but also to 36 a number of other social factors that benefit the whole family. These social impacts, such as changes in work loads, health benefits, amenity and communications changes, and equity and social cohesion aspects are discussed below, based on field inquiries and the farmer survey results (Annex G). The farmer survey included interviews conducted by female enumerators of groups of women on 152 watercourses (paragraphs 37-43 of Annex G). These interviews mainly explored social issues. For ISRP and CWM (canal improvements component), despite the promise of more water and a more reliable supply, the outcome is less clear because generally social factors have been inadequately addressed during design and implementation of canal works. Workloads 3.3 In common with experience elsewhere, to the extent that the projects accelerated agricultural expansion beyond that occurring generally, the projects have increased the work load of women who work in the fields. A half to three quarters of the women's groups interviewed on project watercourses reported increased demand for agricultural labor, especially for harvesting, weeding and crop processing and storage. (Even greater labor increases on control watercourses is attributable to increased use of tubewell water.) The survey also found that more women's groups on control watercourses reported an increase in "other business" activities (i.e. off-farm) than project watercourse women, suggesting a general move in rural areas towards diversifying family income from non-agricultural sources. These results suggest that during a period when women in irrigated areas have generally become more economically active, the projects contributed to this trend. 35. Other analysis has reported that the sugar cane crop is a net consumer of economic resources, and therefore expansion is undesirable (see Pakistan-A Strategy for Sustainable Agricultural Growth, op. cit., Chapter III). 36. The impacts of canal rehabilitation and improvements under ISRP and CWMP are more difficult to track since they did not so clearly result in benefits to the farmers. 48 3.4 In other respects, however, women's workloads improved in that renovated channels and watercourses reduced domestic chores. Women's groups generally reported improved conditions for washing babies, clothes and utensils, except for those on unimproved watercourses served by ISRP channels. Women were particularly pleased where laundry facilities had been provided as part of the improvements (but engineering designs have not always been so thoughtful). Similarly women felt that crossing watercourses was easier once they had been improved, given that unimproved watercourses may be surrounded by stagnant swampy areas or surrounded with large mounds of silt from watercourse cleaning. This response is in contrast to the problems often experienced in crossing improved canals-see below. 3.5 The survey results and field interviews present a clear picture of the projects greatly reducing the demand for male labor, despite expanded cropping. Generally, the labor savings outweighed any labor increases for extra cropping, but since the interviewees were generally land owners and tenants, and not landless labor who are less evident at such meetings, a different perspective on labor changes may have been missed. The main labor savings result from lower needs for managing an irrigation turn, less silt to be removed from the watercourse, and fewer repairs. All sources of opinion agree on a dramatic reduction in the labor required for irrigation management following watercourse improvement. Typically farmers reported that irrigation management now takes one person where three had been needed in the past. Patrolling of watercourses upstream of the farm is hardly necessary when the watercourse is not likely to breach and water theft is so much more difficult now, that it has all but ended. The introduction of pukka nuccas, with their positive open or shut positions, and thereby the elimination of bank breaking to create an outlet, is a large part of this labor reducing impact. Moreover, irrigation management within farms is easier when the full flow is being received and irrigation coverage is more even. The sharp fall in desilting chores after renovation is a further benefit to farmers hard pressed for labor at certain times of the year. This is so whether farmers themselves did this work or it was done by hired labor provided instead by farmers. Examples quoted in the field were typically that for unimproved watercourses where 40 people would work on desilting once a month, after renovation only four people needed to do this work every few months. 3.6 The dramatic drop or absence of disputes over irrigation supplies, as reported by farmer groups everywhere, also contributes to reducing calls on farmers' time. Theft of water from a lined channel is more difficult and easier to detect than with an unlined watercourse. However, this is not to say that water use is now more equitable in general. There was much evidence and reports of widespread water "theft" on a much grander and routine scale by the watercourses on upstream stretches of channels, where influential land owners could make arrangements to suit themselves. The design of renovated (pipe) moghas under CWMP, where distributaries have been lined, actually makes it easier for farmers to interfere with the outlet, and more difficult for PIDs to detect tampering. Health Benefits 3.7 The appearance to the visitor of many unimproved watercourses and the surrounding land, particularly close to villages or within villages, suggests that there should be substantial health gains to watercourse renovation. Unimproved watercourses often act as both the main village water supply and drain, and may in places be bordered by piles of rotting refuse and stagnant standing water. Seepage, overflows and lack of any attempt at channel repair along earthen lengths, results in stagnant swampy areas, or even ponds of evil looking liquid (to call it "water" would be misleading). In contrast, after renovation, stagnant ponds and swampy areas 49 are gone, and may be converted to orchards (as OFWM directorates show in their slide presentations), and the water flows free of village drainage and refuse. Provided that buffalo wallows, washing sites and crossings were included in the design, there are obvious social benefits. Indeed, there would be substantial social benefits to lining only those stretches of watercourses within or close to villages, apart from any water savings, as advocated as a high priority in an early report discussing strategy for watercourse investment. This would have spread benefits more widely than was the case (see below on equity aspects). 3.8 The farmer survey sought to get information on these aspects both from groups of men and separately from women's groups. Over half the women's groups reported the improvements described above-including reduced ponding near villages (but an unexplained anomaly is that even more of the groups on control watercourses reported such a change for the better). As a direct consequence, there was a perception that the incidence of mosquitoes and malaria was not as severe as it might have been otherwise during a period of rising incidence. That is to say, increased mosquitoes and malaria were reported on improved watercourses by around a quarter of the women's groups, but women's groups on control watercourses reported increased incidence in two-thirds of the groups. About a quarter of the survey groups interviewed thought that houseflies had decreased. A quite startling finding was that these public health benefits were restricted mostly to Punjab watercourses, which, since the benefits are related to absence of standing water and swampy conditions, ties in with the reported and observed excess of surface water supplies in many areas of Sindh. These survey findings were in line with the evaluation team's discussions with farmer groups. 3.9 Attempts in the survey to investigate changes in drinking water quality (six questions) produced conflicting results, partly because the source of the drinking water was not recorded (tubewells or channels), and the project connection, or not, was not clearly specified. An important point that did come out, however, is that where seepage is reduced in mildly saline groundwater areas, where unpolluted water from tubewells is tolerable for domestic use, that water may become intolerably salty as seepage recharge is reduced. Communications and Amenity Aspects 3.10 Most communities were in danger of suffering quite serious disruptions of communications, plus some loss of amenity, when canals were improved under ISRP or CWMP. From the survey responses on this point the Punjab Irrigation Department is the subject of most criticism, compared with the Sindh ID, for being insensitive to the needs of communities through which canals pass. In particular, the failure to provide adequate crossings, animal watering places and washin and bathing facilities was of concern and disrupted the daily live of canal- side communities. The evaluation team saw many cases in the field where in the absence of an engineered crossing to link communities to each other, to the main road, or to their fields, rickety local structures spanned the canals. There were also cases seen where earlier engineered crossings had not been replaced when canals were widened or lined under ISRP or OFWMP. It 37. Whether the problem is any less in Sindh would warrant further investigation, as the difference may reflect, as discussed elsewhere, that given land tenure patterns the farmers group responses in Sindh are more likely to be representative of relatively large landowners, with tubewells, and not necessarily those living in canal-side communities who would be most affected by re-engineering of canals. 50 was also clear that, if facilities had not been provided, the various domestic uses of canals are harder when canal banks are lined since it is more difficult to reach the water level. 3.11 The lining of canals has generally made life harder for livestock owners who had previously relied on access to canals to water and wallow their animals (a concern to engineers because of the damage done to earthen channels-Photo 5). Except in the few cases where livestock facilities were included in designs, lined channels can no longer be used safely by livestock. Cases were reported of animals, seeking water, slipping into lined canals and drowning. While excluding animals from using canals may reduce damage, it has made life more difficult for herders when alternative facilities are not provided. On some watercourses buffalo wallows have been included. The lesson here is that engineering designs need to take into account the needs of communities through which channels pass, as much as they seek to optimize water conveyance efficiency (Photo 6). Community participation at the design stage, coupled with more responsive guidelines to IDs, are needed in such situations. Equity Issues 3.12 Considerations of equity, especially poverty alleviation targeting with respect to the project beneficiaries, were not a high priority in the design of the projects or the criteria for selecting canals or watercourses for investment. Watercourses selected were supposed to serve small farmers, but this intention was not strictly applied and could not be enforced in much of Sindh. Even if equity had been a more prominent criterion, in the free wheeling spirit in which the projects were implemented with little reference to priorities, and given the financial, social and political forces at work, it is unlikely that the equity outcome would have been any different or more satisfactory. Had the principle of concentrating on saline groundwater areas in OFWM operations been followed there would have been a more equitable outcome since these areas are less fortunate than the fresh groundwater areas. However, it is difficult to imagine how an equity based strategy for these projects could have been applied to the whole of the Indus Basin Irrigation System, given the great variations in land holding patterns throughout the system (see the discussion in Chapter 6 on national versus provincial projects). 3.13 As it was, field inquiries confirmed well informed intuition that in the selection of watercourses for improvement it was not the most in need nor the poorest which got the benefits of the subsidy (a common finding for all subsidies). Rather the implementation targets for each OFWM unit ensured that it was those watercourses which were more cohesive, better organized already, and led by well connected farmers, which came to the front of the line for selection. This natural self-selection process aided rapid and trouble free implementation, but resulted in many desperately poor areas being left out. Typically these included areas at the tail reaches of minor canals, as seen by the evaluation team, for example, in the Nara in Sindh. 3.14 In contrast, at the other extreme in Sindh there was "competition" between owners of large land holdings, and of several watercourses, to have the largest and longest lined watercourse (Box 3.1). Equity considerations during project design have to face the realities on the ground, and it would have been difficult for disparate country-wide projects such as these to attempt to ameliorate some of the more extreme inequities of the centuries old pattern of rural societies. However, on equity grounds, in much of Sindh, as well as some of the more recently developed irrigated areas of southwest Punjab, the very large land holdings should not have qualified for OFWM investments on the same terms as the typical multiple owner watercourses. For example, the fact that in Sindh the vast majority of farm families are excluded from 51 membership of WUAs, since they are landless labor or sharecroppers, raises questions about equity aspects of the primary institutional model for the OFWM program. 3.15 On individual project watercourses the cause of equity has, however, been well served with respect to evening out water supply levels, as well as calls on their labor, between farmers at the heads and tails of watercourses. This is evident from half of the farmer groups reporting more water available in watercourses and the widespread evidence (from both the survey and field interviews) that disputes over water supplies on improved watercourses, which were common before improvement, had all but ceased. With greatly reduced labor required for desilting after renovation, tail end farmers especially have been relieved of a heavy and disproportionate labor contribution. This can be deduced, since traditionally farmers are required to work in gangs, or supply labor, for desilting the watercourse from the mogha to their own inlet, but no further. Hence, tail end farmers are especially burdened by the laborious desilting work needed on unimproved watercourses. Returns to Watercourse Renovation 3.16 The economic evaluations at appraisal of the three projects involving watercourse renovation-On-Farm Water Management I and II, and Command Water Management-were based on (i) the research results from the Mona Reclamation Experiment Station; (ii) a farm level agricultural economic survey conducted by WAPDA for the Revised Action Program; and (iii) the World Bank's computerized Indus Basin Model. The first results of any M&E studies did not come out until after the last of the three projects had been appraised. All three analyses thus have basically similar sets of assumptions about increased production, and the follow-on projects did not refine the analysis or use better data. 3.17 Most of the project-induced increases in agricultural production were expected to come from increases in cropped area brought about by increased water supply. Modest yield increases induced by the additional water and the better husbandry and increased input use demonstrated by the extension service were also assumed. The assumptions made about likely increases in crop production were related to what was known about farmer response to extra water. The resulting ERRs were high, ranging from 22 percent for OFWM-II to 45 percent for OFWM-I. The switching values were also high, making them seem robust in the face of possible increases in cost or reductions in benefits. The RERRs38 in the PCRs supported the original estimates, although changes in crop prices made some adjustment necessary. Thus these OFWM projects were confirmed as economically sound investments. A more detailed review in Box 3.2 suggests that there is concern about the reliability of the estimates, and their consistency one with another. There are also conceptual problems which are discussed below. 38. Re-estimated economic rate of return. 52 瓔‘ 5. An 비ㄲenity wllich 15 lost w<h channeI lining 튑 닙｀편＊· ::, 쟈J－하 T，죤씩’른 「，.. ‘」r F 6. An amenity gained when included in Iinjng plans 53 Box 3.1: Watercourses as Status Symbols: The Bigger, the Better During a visit to the Sukkur Right Bank Command, the evaluation team learned from the OFWM staff how large zamindars are competing with each other to have a bigger masonry watercourse than their neighbors, regardless of whether this is needed to meet crop water demands. A masonry watercourse has become a status symbol in that area, no zamindar should be without one, and its size is an indicator of the zamindar's power. Water distribution is under the absolute control of zamindars, who take as much water as they want, regardless of sanctioned discharges. The OFWM Directorate is then forced to renovate the watercourses, to carry that larger amount, on the instructions of the zamindar. The largest watercourse seen by the mission was owned by a member of the Legislative Assembly, and had a design discharge of over 26 cusecs and had 4,000 m of lining. Another zamindar had had all three of his watercourses renovated under the OFWM program. This situation is not confined to the Sukkur Right Bank, norjust to the watercourses visited by the team. One Bank supervision mission to Sindh reported that "all watercourses have been designed on reported actual measured discharge which is sometimes three or four times the sanctioned discharge." Following this up, the supervision mission was told by OFWM staff that they were under pressure to design masonry watercourses on the basis of actual flows. They were unable to resist such pressure despite the fact that it would aggravate the equity problems and disrupt the rights of other farmers using the same minor or distributary. A survey of the extent of this problem was proposed for OFWM-111, but was not followed up. In fact, the guidelines for field teams renovating watercourses, while specifying in detail the length of watercourse to be lined, are silent on the-discharge to be adopted for the design. The Bank's legal agreements should specify the distance to be used for deciding on the lining dimensions. 54 Box 3.2: Economic Rates of Return to OFWM-I and II, and CWMP The economic rates of return for the two OFWM projects and the OFWM component of CWMP were very satisfactory, both in the staff appraisal reports and when re-estimated in the PCR:s. Economic Rates of Return (percent): OFWM I OFWM II CWMP SAR 45 22 24 PCR 29 22 23 However, there are concerns about the methods used which affect reliability of these estimates and their consistency, both between projects and between the SAR estimates and the PCR re-estimates. The economic analysis of OFWM-I' was based on ten farm models representing different combinations of cropping patterns, groundwater quality, tubewell water availability and province. For most farm models, and without explanation, there was no differentiation in water savings between watercourses that would get 15 percent lining and those that would get 30 percent. Cropping intensity was assumed to increase by an average of 18 percentage points, from 123 percent to 141 percent at full development. Some shift in cropping pattern towards higher value crops was anticipated, and modest (five to ten percent) yield increases were assumed. With these assumptions, the overall ERR for renovated watercourses was 45 percent. Benefits would need to fall by 40 percent, or costs increase by 66 percent, before the ERR fell below 12 percent. It thus seemed that, even if the water savings assumptions were over-optimistic, the project design was robust and the project would remain economically viable within a wide range of alternative circumstances. By the time the PCR for OFWM-I was issued, the results of the monitoring of both the USAID project and OFWM-I were known. According to the PCR, these showed that the increases in water delivered to the field resulting from lining were about half those expected at appraisal. However, the PCR, in re-estimating the ERR (RERR), says that the monitoring reports were inadequate for measuring project impact with reasonable accuracy. It nevertheless recalculates the ERR at 29 percent, versus 45 percent in the SAR. The PCR attributes this to "significantly lower commodity prices now expected to prevail," rather than to any problems with project outcome. OFWM-II was prepared before much more information on likely water savings from watercourse renovation was available than for the first OFWM project. It thus makes similar assumptions about increases in crop area and yields. The resultant ERR is 22 percent, half that of OFWM-I, but still respectable. The ERR varies between provinces, from a low of 15 percent in NWFP, to a high of 25 percent in Sindh. The SAR does not give details on the variation of ERR between farm models or explain the reason for the difference from the first project. a. This was done only for the watercourse renovation component. Regrettably, no attempt was made to calculate benefits attributable to the accelerated improvement watercourses, nor was their ERR estimated, other than to say that, because the costs of the technology were so low, "even small and short term production increases... would generate very high returns... Since benefits and life expectancy of the accelerated watercourse improvement are difficult to predict, the economic benefits of this technology ...have not been quantified" (SAR, para. 7.02). This omission now seems surprising considering that as many as 16,500 watercourses were to be improved. (continued) 55 Box 3:2 (continued) Although the OFWM-II RERR is exactly the same as the SAR estimate, 22 percent, the PCR comments that the likely increase in water supply is only about half that expected in the SAR. It also comments: "although physical achievements are commendable, the question of whether or not the primary project objective, of increasing agricultural production through effective use of irrigation water saved, has been achieved cannot be answered properly due to absence of reliable data." The CWMP was also appraised before the results of the evaluations of either the USAID or OFWM-I projects were published. The SAR used basically the same assumptions as the earlier OFWM projects, although making some distinction between different lining proportions. It assumed an increase of water available of 15 percent in the case of watercourses with 15 percent lining, and "over 25 percent" for watercourses with 40 percent lining. The overall increase in water supply resulting from watercourse improvement was taken as 18 percent. As there were also additional project components increasing water supply, it is impossible to separate out what increases were attributed to watercourse improvement and what to other improvements. The PCR comments that "water savings from canal improvements and watercourse renovation are reported to be equal to or more than assumed at appraisal" (para. 6.2) and adds a footnote that "local officials expressed their interest in a systematic survey and evaluation of the project, because of their general feeling that this project's success deserves to be well documented." Its RERR was 23 percent, against 24 percent at appraisal. 3.18 Groundwater balance estimates indicate that in fresh groundwater areas all the recharge (and more) is re-circulated by tubewells. Reducing seepage recharge by watercourse renovation can be expected therefore to have the following main effects for a given (constant) cropped area and crop mix: (i) it will reduce farmers' reliance on tubewells since the additional water will be available in the watercourse, and this will reduce oeerall pumping costs; but (ii) it will accelerate the lowering of the watertable, since recharge is reduced, but since tubewell supplies will still be needed at peak times, lowering the groundwater level will increase unit pumping costs.39 The results of the farmer survey given in Annex G show that the first type of effect was predominant only in Sindh, where 55 percent of farmers groups on renovated watercourses reported a decrease in tubewell use after renovation. In Punjab the second type of effect predominated and in fact almost 60 percent of watercourses with tubewells reported an increase in their use after renovation, presumably because renovation made possible an increase in crop area which created a higher peak period demand that could be met only by an increase in tubewell pumping. The implication of this situation for the economic analysis is that, in Sindh at least, not all the decrease in recharge brought about by watercourse renovation results in an increase in water supply at the farm turnout. It becomes a substitute for tubewell pumping. The calculations of the ERR in both the SARs and the PCRs are thus flawed in their assumption that all the decreased discharge is a net gain.40 39. A reviewer suggests, however, that lower groundwater tables recently are more the result of increased groundwater pumping than of reduced seepage (see footnote to para. 2.13). 40. In the Indian State of Haryana, where the World Bank financed a project that included watercourse lining, the economic analysis did not make the same mistake. The only benefits of lining in fresh groundwater areas used to calculate the ERR were the avoided costs of not having to pump seeped water back again, plus the cost of the water 56 3.19 Taking account of such overestimation of the additional water available would, however, not greatly affect the RERRs calculated in the completion reports. If it is confined to 55 percent of the Sindb watercourses, this represents only 10 percent of the almost 10,000 watercourses renovated under the three projects. And even if tubewell pumping was reduced by as much as 40 percent on these watercourses, this is a reduction of only four percent in total project benefits, and it would be offset by a reduction in pumping costs. The impact on the RERR would be at most a reduction of one-percentage point. 3.20 The major factors determining project benefits is the amount of additional water resulting from watercourse renovation and what is done with it. The factor was examined in detail in Chapter 2, which concluded, that renovation, on average, increased the amount of water reaching the farm inlet by 25 percent over the pre-renovation level. Results of the farmer survey focus on two aspects of how farmers reacted to this additional supply of surface water: risk taking and the use of the water (Annex G). Risk taking is linked with farmers perceptions of the reliability of supplies. In both Punjab and Sindh, farmers reported that the reliability of surface supplies improved, but this was influenced more strongly by watercourse improvements than by canal improvements under ISRP and CWMP. Indeed farmers served by ISRP channels (but with no improvement of their watercourses) reported decreased reliability. Increased reliability gave farmers confidence to take risks and expand their cropping: 46 percent of project farmers were risk takers, compared with only 12 percent of farmers in the control (without) sub-sample. 3.21 A group of survey questions focused on what farmers had done with the extra water: whether it had been used to cultivate a larger area, to grow more water-intensive crops or to increase the delta on an unchanged area and crop mix (the least risky option). Project farmers with extra water were much more likely to report spreading the water on more land (particularly in the rice-wheat zone), while control farmers were more likely to grow more water-intensive crops. This differs slightly from the IER field mission's observations that nearly all project farmers reported a shift and expansion into higher value crops, which were also more water- intensive, typically cotton, sugarcane or orchards. These findings on water use are confirmed by farmer group reports elsewhere in the survey that farmers shifted into fodder (primarily during kharif season) vegetables, sugarcane (except NWFP), and "other kharf season crops, largely by reducing fallow. These cropping shifts are likely to be the result of the combination of additional water and greater reliability of supply. There was a significant and important difference between farmers' perceptions of external risks in Punjab and Sindh, with only three percent of Punjab farmers reporting that such risks had increased compared with 55 percent of farmers' groups in Sindh. This is in accord with the mission's impressions of the relative state of the irrigation and drainage systems in the two provinces. 3.22 There are also a number of other benefits and costs which, although hard to quantify, need to be considered in assessment of RERRs in completion reports. On the positive side are: (a) reduction in recharge: in saline groundwater areas every unit of water prevented from reaching the groundwater reduces future damage that will result lost to non-beneficial evapotranspiration during pumping. Even this, however, assumes that farmers will substitute the additional surface water by reducing pumping. The results of the farm survey reported in Annex F indicate that in the (Pakistan) Punjab this did not happen: lining was accompanied by an increase in tubewell pumping. The chain of causation is not clear, whether the additional pumping would have occurred anyway or whether the additional surface water created the demand for additional tubewell pumping in order to balance water supply with crop water requirements. 57 from additional waterlogging and salinity and the degradation of water quality. Thus the benefits of watercourse renovation in saline groundwater areas are underestimated in available models (See Annex B, Appendix 2); (b) improved water use efficiency: low irrigation flows are difficult to distribute evenly around a farm. By increasing the size of the farm irrigation stream, there is a reduction in both water losses arising from over-irrigation at the head of the field, and in crop losses resulting from under-irrigation at the tail; (c) improved equity: tubewell water is not available to those without access to tubewells; these are usually the poorer farmers who are least able to influence the distribution of surface supplies. Water saved by watercourse renovation is generally accessible to all farmeln since it increases the volume of each warabandi proportionately, in particular there is a greater equity of supply as between head and tail farmers; (d) a reduction in labor costs: after renovation less time is required to irrigate, less time is needed to patrol the watercourse, and less time is required for maintenance; (e) reduction in disputes over water; (f) deposition of silt within the field rather than in the watercourse; (g) reduction in non-beneficial evaporation during the cycle of recharge and pumping. 3.23 On the negative side: (a) the extra cost not taken into account is the additional pumping lift in fresh groundwater areas as the watertable level falls. The amount involved is, however, small. On the basis of the figures in Chapter 2, the Bank projects have reduced groundwater recharge by at least one MAF annually in fresh groundwater areas, compared with a total annual inflow to the fresh groundwater of about 35 MAF. The effect of this on the ERR cannot be significant. (b) an offsetting factor, when evaluating seepage reduction in fresh groundwater areas ( as discussed in Chapter 2 ) is that seepage is the main source of recharge. Hence, to the extent that the fresh groundwater "reservoir" is a valuable "on demand" source of water at times of peak demand, converting that recharge water (by channel improvements) to a mandatory supply year round involves an indirect cost. 3.24 Although it is not possible to place precise numbers on the benefits listed above, they are not negligible. This was borne out by the field observations. However, calculations of the ERRs of OFWM projects in the Bank's SARs and in the completion reports do not capture these benefits. The RERRs (ranging from 22-29 percent) thus tend to underestimate the true rates of return. And even when the double counting of benefits in fresh groundwater areas and the additional pumping costs from lowering watertables is considered, the ERRs are likely to remain above 20 percent. The conclusion of this report is, therefore, that the RERR of the OFWM and 58 CWM projects is unlikely to be below 20 percent.41 Given the shortcomings of the basic data, particularly on the amount of reduction in groundwater recharge in FGW areas, any attempts to refine this estimate further are unlikely to change the basic conclusion.42 Furthermore, as noted below, ERRs vary considerably between watercourses and a project average is a poor indication of project accomplishments. On the basis of the economic evaluation made in an Indian project (see footnote 30) one would also need to know the proportion of watercourses in FGW and SGW areas in order to arrive at the weighted project RERR. Such data are not available. 3.25 There can be no doubt that watercourse renovation, whatever its shortcomings in execution, has been an extremely popular program with farmers. Almost every observer who has visited any of the renovated watercourses, including the impact evaluation team, has come away impressed by the long list of benefits which farmers attribute to the program, including such accolades as: "This is the only good thing the government has ever done for us." This is confirmed by OFWM officials, who say that the strongest demand for more lining comes from those watercourses that have had some lining done already. 3.26 An examination of the costs and benefits of the OFWM program, outside the context of a formal ERR calculation, provides some explanation of the farmers' strong response. This has been done by making comparisons of the cost of water saved with the benefits of additional water, calculated from the Indus Basin Model.43 This model provides values for the marginal product of additional water, that is for water added to relieve shortages at peak periods. These range from over Rs. 2,000 per acre foot (in 1988 prices, equivalent to about Rs 3,200 or US$100 in 1995 prices) for fresh groundwater areas in Punjab, down to Rs 360-380 per acre foot for saline groundwater areas in Sindh (Rs 575-600/US$19-20, in 1995 prices). These values reflect the value of tubewell water, which can be provided on demand at peak periods, as an average for canal regions, but not of additional water from watercourse renovation savings. This is valued on the basis of average returns to water as discussed next. 3.27 A recent Bank document has further estimates of rates of return to water: "The average financial return to water is 700 rupees per acre foot, but ranges from zero to 2,000 rupees.,,44 These figures are based on a sample of farm budgets assembled by World Bank staff. A footnote comments that "variations in returns can be observed both for different crops within one region and for the same crop between regions. For example, returns to water in Gujranwala are as low as 11 rupees per acre foot of water for Basmati rice and 973 rupees for sunflower seed. The return to water in sugarcane production in Faisalbad is even slightly negative (-10), while in Mardan it is 182 rupees per acre foot." The favorite crop to benefit from additional water from renovation was frequently sugarcane. This range of values, compared with the Rs 215 per acre 41. A reviewer from a central vice presidency of the Bank has noted that the findings of this report contrast with those of a concurrent impact report from OED on six projects in Southeast Asia (Irrigation O&M and System Management in Southeast Asia: An OED Impact Study, June, 1996). The four Pakistan projects had satisfactory economic rates of return, whereas the Southeast Asian projects were all judged to be uneconomic investments. This difference appears largely to be attributable to the low cost per hectare of the Pakistan rehabilitation projects, but good benefits in such a semi-arid location, compared with new schemes in the humid tropics of Thailand, Vietnam and Burma. 42. A reviewer from the Bank's South Asia Regional Office has taken issue with this report's conclusion that OFWM's benefits have been reduced by watercourse improvements in fresh groundwater areas, but agrees that priority should be given to OFWM activities in saline groundwater areas. 43. See Ahmad and Kutcher, op. cit. 44. World Bank, March 1994, Pakistan: Irrigation and Drainage: Issues and Options (para. 3.25). 59 foot cost of saving water through renovation (see para. 2.13), shows that there must be many instances where farmers use the extra water for less than the cost of producing it. This is a natural result of the distortions produced by the 50 percent subsidy implicit in the watercourse renovation program.45 This suggests that, when disaggregated to smaller areas, there must be a wide range of ERRs, which will vary from watercourse to watercourse. Alternative costs of increasing surface water supplies 3.28 The calculations made for the Revised Action Programme indicated that it cost about four times as much to increase the supply of water at the mogha through additional surface storage as it did through a program of watercourse renovation and lining. This relationship, now that it is possible to calculate the actual cost of water saved through lining, still seems to hold good. In Chapter 2, the capital cost of saving water through the OFWM program was estimated at Rs 1,766 (US$57) per acre foot. This figure would be relevant in only saline groundwater areas since a reduction in watercourse losses in fresh groundwater areas is not a true saving. The approximate capital cost of additional storage is Rs. 9,500 (US$310) per acre foot at the mogha. This figure is subject to a very wide margin of error as it has to be based on a number of 46 questionable assumptions, but the comparative orders of magnitude are not in doubt. Most of the uncertainties tend to move the cost estimate of new surface storage upwards, increasing the difference compared to the cost of extra water from OFWM. It therefore still seems to be a valid conclusion that the least cost method of increasing surface water supplies to farms in the Indus Basin Irrigation System is by reducing seepage to groundwater. 45. The Bank's Regional Office has commented that significantly higher levels of capital cost contributions by the farmers have been agreed in the proposed Punjab Private Sector Groundwater Development Project. It has been agreed with the government that farmers would contribute 77 percent of the capital cost in net present value terms with an annual interest rate of 12 percent, of which they will contribute all the skilled and unskilled labor cost and 30 percent of the materials cost up-front. 46. The assumptions made to calculate the cost of additional surface storage are: i) The 1985 cost of Kalabagh dam was US$3,500 million, including contingencies, equivalent to Rs 55,800 million in 1985 prices (exchange rate Rs 15.5 to US$1.0). ii) In 1995 Pakistani rupees this is equivalent to Rs 134,000 million based on the relative GDP deflators in 1985 and 1995 (an increase of 240 percent). iii) 25 percent of the cost of Kalabagh is attributed to irrigation and 75 percent to power, based on the relative benefit shares as calculated in the original 1985 Project Preparation Report. iv) Kalabagh would produce an additional4.7 MAF of irrigation water at the canal head, based on calculations in The Agricultural Impact ofKalabagh Dam, World Bank Report No. 6884 of August 1986, Table 5.4, run #4, less run #13. v) Losses from canal head to mogha are 25 percent of canal head discharges (a mean of estimates from the above report), making 3.5 MAF additional mogha deliveries from Kalabagh. vi) Capital cost per acre foot is Rs 9,600. This calculation ignores (a) O&M costs of Kalabagh; (b) additional canal remodeling costs to convey Kalabagh water to moghas; (c) the impact on drainage requirements, and (d) additional water available for tubewell pumpage. The first three of these factors would increase the cost of providing water through surface storage, reinforcing the conclusion that watercourse renovation is the cheaper alternative. 60 4. The Impact of Irrigation Systems Rehabilitation Design Issues 4.1 Canal rehabilitation criteria were an overriding issue of concern to the Bank during the design of the Irrigation System Rehabilitation Project (ISRP) as described in Box 4.1. While emphasizing that the project was principally concerned to remedy the deteriorating state of the canal system because of inadequate O&M funding, the project documents give less emphasis to the overriding reason for the urgency to correct deterioration. This was the additional surface water which the storages of Mangla and Tarbela had made available, and the scramble by provinces to establish a claim to this water, by diversions into their canals, before an inter- provincial accord on water distribution was signed (as it was in 1991). Consequently, the project design specified three criteria which the provinces were to follow to avoid ISRP investments contributing to this competition for incremental water rights: (i) that rehabilitation should be carried out on complete systems or subsystems; (ii) that no rehabilitation of main canals would be financed, except in exceptional circumstances with the concurrence of IDA; and (iii) that after rehabilitation canals should not be operated above the designed freeboard. Even so, the account which follows indicates that the physical need to accommodate Mangla and Tarbela water in the system, and the political pressures and vested interests brought to bear on the allocation of that valuable resource, led to the project being implemented without regard to the above design criteria. The Farmer Survey Responses 4.2 Impact on Water Supplies. Field inquiries confirmed that farmers are keenly aware of the supply situation above the mogha in the PID channels. In part this may be because the position and state of repair of the mogha is the subject of negotiations between farmers and PID staff. In the survey farmers groups were asked a number of questions about the supply channel, but with mixed results (Annex G). Two points stand out: (a) that farmers groups in Sindh more frequently reported an increase in supply channel levels than in Punjab (56 percent to 10 percent), but this reflected increased relative diversions rather than any project impact; and (b) the reliability of supplies was perceived as improved more in Sindh, than in Punjab, and was much more evident to farmers whose watercourses had also been improved, than to farmers on ISRP channels without watercourse improvements. Thus, despite the fact that canal diversions did increase, farmers cannot benefit to the full extent if their watercourse remains unimproved. This is logical and points to the importance of watercourse renovation investments, over and above the main concern to reduce losses, in making the system more efficient (Photo 7). 47. A similar issue concerned the Bank in the design of the canal rehabilitation component of the Command Water Management Project, although this project is not referred to in the SAR text for ISRP. 61 Box 4.1: Canal Rehabilitation Design Criteria The main ISRP design issues were: (i) which canals to rehabilitate, and (ii) the extent of the rehabilitation and/or enlargement. The PIDs could select canal systems for rehabilitation according to broad criteria agreed with the Bank. Priority was to be given to systems with inequitable deliveries because of sedimentation, or where rehabilitation needs were urgent, and there was a high risk of breaching. The Bank was concerned that ISRP rehabilitation did not become a controversial issue because of inter-provincial conflicts over diversions. The project was being prepared at a time when additional surface supplies had become available from Mangla and Tarbela reservoirs, but before there was an inter-provincial allocation agreement. As a result, most of the canals in both Punjab and Sindh were carrying more than their design discharge. This was placing great stress on the system and increased the risk of canal failure. "These larger flows have been made possible by an improved system of barrages and the storage in Mangla and Tarbela," causing the canals to be operated "beyond the limits of safe operation."a The project design had to choose between either restricting rehabilitation to design, thus bringing the canals back within safe operating limits, or allowing the rehabilitation criteria to be to "present operating capacity," accepting canal enlargement as afait accompli. Restricting rehabilitation to the original design was impractical, since it would not be acceptable to provincial governments. Therefore, the project permitted rehabilitation to whichever was larger, design or present operating capacity. But since the Bank was c9ncerned that the project would lead to even higher diversions, in the race to secure the available extra supplies, the Project Agreements with the provinces stated that canal operations, "shall maintain the integrity of the minimum designed freeboard in rehabilitated channels." a. SAR, para 2.05. An indication of the stress placed on the system is clear from the size of the increase in surface diversions during this period. Prior to the inception of Mangla (1967) annual canal diversions averaged 83.5 MAF (105 Bcm), while during the post-Tarbela period (1976-88) average diversions increased by 21 percent to 101 MAF (124.7 Bcm), but with virtually no enlargement or improvement of the canal system Present diversions are around 106 MAF (132 Bcm). 4.3 One explanation for the apparent failure of ISRP sub-projects (on their own) to increase reliability could be that the direct benefits of ISRP were too dispersed to register with farmers, since breaches had been spread over a wide area.48 Another explanation is that canal renovation does nothing to reduce variations in supply that result from erratic flows above the distributary head. These erratic flows, not breaches, are the primary source of unreliability in flow at the mogha.49 48. See Gleason, J.E. and Wolf, J.M. Evaluation ofthe Impact of Canal Rehabilitation on Hydraulic and Agricultural Economics Indicators, Irrigation Systems Management Project Phase II, USAID, Islamabad, Pakistan, April 1993. 49. It was beyond the resources and scope of this evaluation to measure flows in a representative way, but an indication of this variation is provided by daily measurements by IIMI during kharif 1993. On six selected distributaries and minors monthly flows (averaged from daily measurements) varied from May to October, as a percentage of design discharge, from 83-103 percent, 70-86 percent, 66-106 percent, 37-66 percent, 28-65 percent and 56-117 percent. Standard deviations also indicated that there was "considerable daily variations in the actual 62 Canal Operating Levels 4.4 Project implementation agencies ignored the criterion of rehabilitating only degraded complete systems or subsystems, and systems with inequitable deliveries. The PCR endorses this change by making approving comments as follows: "Project guidelines emphasized rehabilitation of complete systems or sub-systems and systems with inequitable water deliveries. In practice, however, early rehabilitation efforts were concentrated on canal reaches in most danger of being breached. Hindsight shows that this was a justified approach because a somewhat greater area than envisaged at appraisal was protected from immediate risk" (although this was not taken into account in recalculating the ERR).50 Although "Rehabilitation of complete systems or sub-systems" (DCA, Schedule 2) was a legal requirement, no formal agreement to this substantial change was traced in the project files. It may be argued that the piecemeal selection of canals for rehabilitation lessened the impact of the investments, but that 51 would be hard to prove. 4.5 It also seems clear, though beyond proof, that the PIDs did not adhere to the criterion, which was a legal commitment, to maintain the integrity of the design freeboard: that is to say, the rehabilitated canals continued to be run at discharges above the revised design. This was in breach of the Development Credit Agreements. Figures prepared by the Alluvial Channels Observation Project (ACOP, WAPDA), which was monitoring the rehabilitation, show that in head reaches discharges of 150 percent of the rehabilitation discharge design were being passed, and in tail reaches almost three times the rehabilitation design discharge was being passed. The ACOP monitoring report (November, 1988), perhaps unaware of the legal restriction, comments that ACOP's measurements show that "due to sufficient strengthening of the banks or restoration of the free-boards, channels even passed safely more water than the rehabilitation discharge figure." 4.6 A subsequent survey of a sample of the rehabilitated canals, commissioned as part of this IER, confirms that canals are being run at discharges above design freeboard. The results of this survey are discussed in Annex F and are summarized in Box 4.2. The main conclusion from the survey is that, while rehabilitation may have reduced the number of breaches, its other benefits are not being sustained because of inadequate maintenance due to shortage of O&M funds. The rehabilitated canals are reverting to their pre-project state of poor maintenance. On the specific issue of discharges above design, the findings of the survey are that, although for many of the channels the amount of water being discharged is not known, they are being run at full supply level as long as water is available. At many control points there are no gauges to measure water discharges, and where there are gauges the relationship between gauge levels and discharge is often not known. The survey findings suggest that it is not possible to make any aggregate observations about whether the surveyed canals were being run within the design limits and that the DCA requirement was unrealistic. discharge." (Bandaragoda, D. J. and Rehman, Saeed ur, Warabandi in Pakistan's Canal Irrigation Systems, IIMI Country Paper, Pakistan No. 7, 1995 (page 32 et. seq.) 50. One senior irrigation engineer commented in explanation that the project was in response to an "'emergency" (such was the dilapidation of the system) and that it successfully corrected some of the worst problems in the system. 51. Neither did the project agencies adhere to the second criterion of not rehabilitating main canals. Three Punjab link canals (Sidhnai-Mailsi-Bhawal, Qadirabad-Balloki, and Rasul-Qadirabad), all taking off from major rivers, were rehabilitated. This may have been necessary, but was not part of the original design for good reasons relating to the water allocation issue. 63 4.7 In the field the evaluation team was shown examples of canals which had been enlarged under both ISRP and CWMP beyond their original design, and which were being run even above the enlarged design discharge (Photo 8). It was, however, unable to obtain data on how many canals had had their design discharges altered as a result of the project or how many were now being operated above safe limits. 4.8 One particularly dramatic example visited by the evaluation team was the Muzaffagarh Canal in Punjab. It was designed for a discharge of 6.37 cusecs per 1,000 acres, but could not carry even this amount before rehabilitation without danger of breaching, which occurred regularly, with the consequence that tail reaches were short of water and O&M funds were used to repair breaches to the detriment of maintenance elsewhere. After rehabilitation its design discharge was raised to 8.57 cusecs per 1,000 acres, an increase of 34 percent. Since much of the lower end of this command could not safely absorb additional water, large areas, including whole villages, have become derelict over recent years as a result of waterlogging and salinity. This situation reportedly arose because the Irrigation Department could not resist demands to increase supplies coming from influential land owners on the head reaches of distributaries at the lower end of Muzaffagargh Canal. 4.9 Canals shown to the evaluation team in Sindh included the Thul and Dad branches from the Rohri, which were strengthened to allow an increase of one-third or more in discharge, with the Rohri banks having to be raised to accommodate this level of supply. These canals are in the command of Sukkur Barrage where the design discharge of perennial canals has been raised from 2.92 cusecs per 1,000 acres to 3.90 cusecs, an increase of one-third, with help from ISRP. 4.10 Another objective of the ISRP rehabilitation was to improve water distribution to tail watercourses. Unfortunately, it was not possible to tell from the engineering survey whether this had been achieved. At best the survey shows that distribution remains unsatisfacto57 on many of the canals surveyed and that tail watercourses still suffer from inadequate supplies. In Punjab two-thirds of the canals surveyed reported water shortages at the tails. In Sindh, despite the massive tampering of moghas, only one of the canals reported shortages at the tail, and that was because it was partially blocked by an abandoned railway bridge. This result may reflect the general abundance of surface supplies in Sindh, if not over-abundance given the widespread waterlogging and salinity problems. 52. Detailed studies of some channels desilted under ISRP have shown that improvements in water supply to tail reaches are highly dependent on both how much desilting took place, on its location on the channel, as well as on the level of head discharge in relation to design. Excessive desilting in head reaches may result in tail watercourses receiving more than their fair share when head discharge is at or above design. What is also clear is that lining tail reaches (as was adopted in ISRP and CWMP) does not improve equity between head and tail reaches unless it is accompanied by desilting in the head reaches. (See Murray-Rust, D. Hammond, and Vander Velde, J., Changes in Hydraulic Performance and Comparative Costs ofLining and Desilting ofSecondary Canal in Punjab, Pakistan, in Proceedings ofa Workshop on Canal Lining and Seepage, 18-21 October, 1993, International Waterlogging and Salinity Research Institute, Lahore, and HR Wallingford, U.K.) 64 Box 4.2: Sustainability of Canal Rehabilitation-The Engineering Survey A survey to investigate the state of selected canals rehabilitated under ISRP and CWMP was undertaken on behalf of OED during the canal closure period of 1994/95 by the International Sedimentation Research Institute of Pakistan (ISRIP). The most significant finding of the survey was that the effects of rehabilitation last only about five years.a If canals are not brought under the regular O&M regime of the irrigation departments, they return to their previous state. The survey found, at least for Punjab and Sindh provinces, that shortages of O&M funding were preventing the irrigation departments from carrying out adequate maintenance, and that some of the rehabilitated canals were in a poor state. Various indicators of the channel condition pointed to inadequate maintenance; several of the canals had silted up since rehabilitation and had had to be cleaned, but inadequate funding had meant the sediment had to be left on the canal banks; the state of the berms (canal banks), though generally good, left much to be desired on some channels; in several cases the canal inspection path was impassable for considerable lengths; seepage was still a serious problem in Sindh, less so in Punjab. Examination of structures showed their condition to be satisfactory in Punjab, but unsatisfactory on most of the Sindh canals. In NWFP the structures were mostly in satisfactory condition. The condition of outlets was satisfactory in Punjab, with only three percent of the outlets examined requiring repair. In Sindh the outlets on the CWMP canals was found to be in reasonable condition, but on the ISRP canals 89 percent of outlets required repair. The main cause of this was believed to be tampering by farmers, over which the Irrigation Department no longer has control. The survey confirmed the PCR finding that breaches were no longer a regular problem. In the past, breaches were mostly due to sending more water down canals than their design discharge. The survey attempted to collect data on canal discharges but they are insufficiently complete to make aggregate observations about whether the canals are being run within design limits, whether they are now carrying more water than pre-rehabilitation, or whether they are just carrying more but with reduced risk of breaching or overtopping. The survey revealed that the relationship between gauge levels and discharges (rating curves) is not known for many channels and that the rating curves are not properly developed, maintained and updated by the PIDs. At some of the control points there are not even any gauges against which to measure discharge. Thus for many of the channels the amount of water being discharged is not known, and channels are run full as long as water is available. The height of the freeboard was measured during the survey to ascertain whether the canals were being run safely. The survey showed that safe limits were being generally disregarded. The situation was worst in Sindh Province. This would indicate that the PIDs are in breach of the covenant in their respective Agreements for ISRP, which required them to maintain the integrity of the minimum designed freeboard. a. Murray-Rust and Vander Velde (op. cit.) say their findings suggest that "the economic life length of five years may be realistic for the major desilting intervention." 65 - * 7 A reconstructed mogha serving an improved channel r¶ 8. A channel rehabilitated under ISRP running Iigch 66 4.11 Hence, ISRP ended up enlarging a number of canals so that they could take Tarbela water less riskily, but this was achieved at the expense of the project design, and the strategy behind this defacto allocation of Tarbela water is not known. That this was not an explicit objective of the project appears to have resulted from the Bank's reluctance to get too close to the inter-provincial water allocation issue, a problem which was resolved as a potential cause of inter-provincial dispute by the agreement reached in 1991. Even so, the implications of Tarbela water allocations, with respect to resource efficiency and environmental impacts, are beyond the scope of this report. Canal Renovation (ISRP) 4.12 In calculating the ERR for ISRP at appraisal, the benefits taken into account were those arising from the reduction in damage caused by canal breaches, and reversing the deterioration in yields and cropping intensities in the absence of the project (despite the large increases in water supplies that were becoming available). In terms of the reduction in breaches, the project impact was considerable, as confirmed in the field and by other reports. The PCR, for instance, notes (para. 6.1) that "evaluation of nine of the systems monitored indicates no breach in any of the 163 channels in the systems after rehabilitation, 155 showing improvements in head reaghes, 132 in tail reaches, and 80 in equitable distribution. Water carrying capacities of drains rehabilitated under the project were found to have improved considerably, thus reducing, if not altogether eliminating, crop losses by flood waters."53 There was also a significant increase in surface water deliveries to tail-end watercourses. The PCR made similar assumptions to the SAR about changes in cropping intensities, but assumed, probably unrealistically, that yields would not deteriorate without the project. Both sets of results indicate the ERR in excess of 50 percent. These results tend to be confirmed in essentials by project monitoring, although monitoring activities, as with all the monitoring arrangements for all the projects, were terminated before they could provide conclusive results. 53. A senior engineer remarked to the evaluation team that this was the first time the 44 drains concerned had been rehabilitated, and that the physical improvements were excellent. 67 5. Institutional Impact Introduction 5.1 The vast Indus Basin Irrigation System was designed originally to spread water to the largest possible number of beneficiaries, who had previously suffered periodic famines, but the design and construction went down only to the mogha serving 100-200 ha of land with 20-50 farmers. Development and management of the system below the mogha was left to farming communities. Scarce engineering and administrative staff resources, who were mostly expatriate officers, could not deal with the details of tens of thousands of individual farms, spread over thousands of acres, needing hundreds of miles of micro-engineered field layouts. The policy then was to get the benefits of extension of irrigation facilities to as many potential famine victims as capital budgets would allow, and to meet operating costs from irrigation charges tied to land areas. Irrigation and Agriculture Departments 5.2 From the earliest nineteenth century pioneering schemes to this day, the government split responsibilities for irrigation system operation and performance between separate engineering and agricultural staffs in different departments. The staff appraisal report for CWMP saw this as a major constraint on raising irrigated agricultural production, referring to "fragmented institutional arrangements which have proven inefficient in delivering water and non-water inputs to farmers on a timely and assured basis." Hence the irrigation staff who managed the water supply were interested in the wholesale delivery of water at the mogha, and not in how the water was used to produce crops on farms. That aspect was the responsibility of agricultural departments which had no responsibility for water delivery (the limiting resource) and, arguably, no influence over water delivery. The disadvantages of splitting responsibility for water delivery from responsibility for agricultural production with the water has been reinforced by the long-standing coolness between the staffs of the two departments, and by the collapse of discipline within IDs caused by rent seeking. Extracts from a recent survey report, Box 5.1, describe some of the malpractices involved. 5.3 Thus, traditionally the two staffs seldom work closely together to the benefit of the ultimate purpose of the irrigation system-providing a secure and growing standard of living to cultivator families. This phenomenon can be Lxperienced by visitors in terms of the two departments usually preferring to take responsibility for itineraries on different days, at the expense of covering the same ground twice with both departments, rather than coordinate a single joint visit to an area. Of course, the relationship between the two departments varies from province to province, and from place to place within provinces. There are places where the traditional distance between the departments is narrowed by personal affinities between senior staff, but they are the exceptions. 5.4 In contrast to the situation described in Box 5.1, working in the agricultural service is relatively less attractive, and this has affected its stature in the countryside and with its sister agencies. All of this changed with the formation of the OFWM directorates in the PADs, including recruitment of engineers by PADs to staff the directorates, since their primary function was improvement of the watercourses by construction of masonry channels. In the field, the 68 PADs, through the OFWM directorates, are now seen for the first time to have a real role of benefit to farmers, and correspondingly the directorate staff benefit from being in demand and having such a solid objective. However, the transfer of PAD's attention away from production towards rural construction is sometimes criticized in Pakistan. Box 5.1: Irrigation Management Malpractices-The Collapse of Discipline Much reference in Bank reports and other studies to the collapse of discipline caused by the financial relationships between irrigation staff and farmers is captured by a description in a recent monitoring report.! "Currently the [study watercourse] has 'enough' water to cultivate land that is not waterlogged or saline. But in the past, when water was scarce, zamindars used to steal water from other watercourses by breaking the mogha of the watercourse. This practice has a long history and is well institutionalized particularly at the head-end and middle watercourses along the minor. "The zamindars broke the floor of a mogha of their watercourse to deepen it by few inches to get more water to their watercourse. They often bribed the Abdar not to report this illegal water transfer to the PID. If reported, the PID collects money from all zamindars of the watercourse to repair the mogha. Or, the PID officials may collect money from zamindars not to repair it for the season as such an action will destroy the additional land already cultivated. The size of the bribe in such an instance depends on the additional area cultivated using the 'stolen' water. "Once this deal between the PID officials and zamindars is done, it is the practice among officials, according to farmers, to contact the zamindars at the beginning of next cultivation season to 'find out' whether they are interested in keeping the mogha in its 'more-open-than-usual' status. If the zamindars are interested then they have to bribe the officials again. If not, the officials will build a barrier called dikka not only to correct the discharge but also to reduce it! For example, if the designed depth of the mogha is 12" and the zamindars deepened it by 2", the PID may repair the mogha by making the depth 10", 2" less than the designed depth. This creates water scarcity and the zamindars are compelled to bribe the officials to remove the dikka! "From the zamindars' viewpoint, the most difficult situation arises when the officials threaten them in each season that a dikka will be placed in the mogha unless they pay bribes. In such a situation, the zamindars cannot go to higher authorities as they have already obtained the support of officials illegally. " When the WUA was re-established in 1991 to comply with requirements to enable the (On-Farm Water Management Directorate (OFWMD) to line their watercourse, the WUA Chairman advised his colleagues not to break the mogha to obtain more water. He explained to them that the amount of water flows through the mogha is sufficient to cultivate the land, as a significant amount of cultivable land was abandoned due to waterlogging and salinity and the cultivated land needs water due to shallow watertable. Since then, farmers claimed, there was no incident of breaking the mogha or the PID placing a dikka at the mogha." a. op. cit. Planning and Development Department, Government of Sindh, April 1993 69 Project Institutional Performance 5.5 The projects in this evaluation had varying institutional objectives and components, with CWM and the two OFWM projects attempting to tackle the institutional problems discussed above. ISRP was intended only to strengthen the operation and maintenance capability of PIDs and, by giving implementation responsibility to PIDs rather than perhaps to WAPDA or some contractor arrangement as had been the prior pattern, to provide PIDs with resources and further experience of major rehabilitation works. To that extent ISRP was reasonably successful according to the PCR (the evaluation did not investigate that aspect of impact). 5.6 The other three projects attempted to make greater inroads into the dual problems of system management and services above and below the mogha, as follows: (a) Command Water Management-Project Management 5.7 The project was to overcome the major institutional constraint by the "authority exercised at the Secretariat level of the Provincial Governments through the Provincial Policy Committee," who would set up Sub-project Management Committees (SMO) staffed from the line departments, including from both PIDs and PADs. The project design viewed the proposed project management structure as a pilot operation, which allowed for several different approaches at the sub-project level in the different provinces. In essence these differences consisted only of the Sub-project Management Office (SMO) in Punjab (for four sub-projects) being located in Lahore, and administratively linked to the Agriculture Department, whereas in the other three provinces each sub-project had its own SMO reporting to the provincial Planning and Development Department. The overall strategy was that within the SMO the staff from both the PIDs and the PADs would work together and, for the project at least, overcome the long- standing distance between the two agencies. This was to be achieved by the head of the SMO chairing a Sub-project Program Coordinating Committee (SPCC) on which PID and PAD representatives sat. Above this SPCC committee a Provincial Policy Committee brought together, in the design at least, the Secretaries of Agriculture and Irrigation under the chairmanship of the P&D Department. 5.8 However, in the event none of this institutional design functioned as intended. As the PCR states: "The project failed to develop a model and capability of the relevant provincial agencies for integrated planning and management of irrigated agriculture." This was confirmed by the field discussions, which in summary concluded that it was unrealistic to expect that a single project's institutional implementation plan could reverse decades of remoteness between the two provincial technical departments. From all accounts the project was implemented independently by the two departments, with the various institutional arrangements hardly functioning. (b) Command Water Management and On-farm Water Management Projects- Watercourse Management 5.9 This aspect of the institutional problem can be discussed jointly for these three projects since they generally followed the lead of the original USAID-assisted program by establishing formal Water Users Associations, except that under the earlier USAID-sponsored initiative more effort was made to help the WUAs to become all-round farmers' organizations. Indeed under OFWM-I, which set the pattern, it was agreed that WUAs would be set up as a precondition to 70 "installing permanent structures" (i.e. building the masonry watercourses and installing pukka nuccas). Furthermore, the water users had to be willing to improve, on their own, the earthen lengths of watercourse branches and field ditches, and follow recommended practices on their farms. It was supposed to be a condition that this earthen renovation was completed before OFWMD undertook the lining; but this was frequently not adhered to (see Annex D, para. 10). Box 5.2 contains a summary of findings from the farmer survey on institutional points, including relationship with PIDs, PADs and OFWMDs. 5.10 To facilitate the WUA approach to below-the-mogha organization, legislation had been passed by each province to establish a framework for WUAs in each province.54 However, outcomes were somewhat different from the long term institutional change and "empowerment" of farmers that the project documents intended. Field inquiries confirmed the findings of others who have studied the Pakistan WUA "movement," that in these three projects the formation of WUAs was more a means to obtain the benefits of the subsidized investment program, than it was a movement to organize farmers to better themselves more generally through continuing initiatives. For example, Byrnes, 1992 (op. cit., page 60) refers to WUAs as "short lived "paper" organizations" which were interested in improving the watercourses, but that much remains to be done if they are to evolve into self-sustaining organizations. 5.11 Thus, in Bank and other reports, a common conclusion is that WUAs have mostly been an ephemeral institutional feature that operated while watercourses were being rehabilitated, that is while there were literally "concrete benefits," but then mostly went dormant. The principal explanation is that with no long term broader function to perform, once the construction period was over, other than watercourse maintenance, the WUAs were bound to become dormant. However, although field inquiries confirmed that most WUAs were not fully active the way they were during construction, and that very few had any broader role to perform (only a few of those WUAs formed during the USAID phase were broadly based), a quite different and extreme situation was also evident. Thus, discussions in the field and inspections of the OFWMD files revealed that some WUAs were only token institutions in the form of a list of land owners offered to the OFWMD staff or even drawn up by them, in symbolic fulfillment of the project conditionality. 5.12 An even more critical stance queries the need for WUAs at all, given the existence of traditional watercourse institutions. Thus, common sense suggests that it would have been impossible, when public irrigation supplies first became available, for groups of dozens to scores of farmers to organize, construct and manage an irrigation system below the mogha without some form of social order and structure being present to coordinate activities and settle disputes. It is also to be expected, and was confirmed in the field, that such "indigenous institutions" are usually based on the existing hierarchy and power structure. Hence, in the Sindh we might expect the institution arrangements at the watercourse level to reflect the semi-feudal system of landlord-tenant relationships that dominate Sindhi agriculture. In Punjab province, NWFP and Balochistan similar local institutions reflect the more varied social hierarchies. The indigenous watercourse institution was widely acknowledged everywhere that field inquiries were made, and was known by different names in different localities (for example-khal committee in Punjab), all of which may be translated conveniently as "watercourse committee." The earliest USAID- assisted OFWM work had tried to work through these existing committees, but soon abandoned 54. Byrnes, 1992, op. cit., page 16, comments that the legislation varied in quality between the provinces, with the preferred NWFP model not being followed in the weaker Punjab and Sindh ordinances. 71 that approach in favor of introducing a new WUA-based system, with the result that the traditional committees are now generally overlooked by outsiders. 5.13 These indigenous khal committees have functioned since the watercourses were first constructed and continue to function today, principally to coordinate maintenance work and to resolve disputes. But even so, there is little if any mention of the indigenous committees in documents on the CWMP and two OFWM projects (both by the Bank and other agencies). -Furthermore, when the OFWM initiative started up under USAID assistance, watercourse committees were soon seen as ineffective, with the WUA movement being promoted as an essential innovation instead. The introduction of WUAs was seen as filling an institutional vacuum, whereas the truth is quite different.5 Many other sources, not relating to these projects or Pakistan, have written about the role and importance of local organizations and the tendency 56 to ignore them in project design, and such comments accord with inquiries in the field. 5.14 Thus references to the WUA movement introduced through OFWM operations ring hollow. The reality in the field, as told to the mission by many groups of farmers on the watercourses, was to the effect that the establishment of the WUA was in essence the renaming of the watercourse committee, with the WUA officers often being the same people as those on the existing watercourse committee. Similarly, the functions of the WUA were seen as the same as those of the watercourse committee, except that by forming the WUA the watercourse qualified for the subsidized improvement of the watercourse. Renaming an indigenous institution seems a small price to pay to reap the subsidy. Unfortunately, the rigid programmed approach to OFWM watercourse construction, with targets for each PAD/OFWMD, led to the type of token institutional development activity described above. Preparation of a list of officers and members' names was often all that was required to fulfill the WUA prerequisite (Box 5.3). In the Sindh the continuing feudal society has further undermined the institutional design of the OFWM program, as described in Box 5.4.57 5.15 Since this is the reality of institutional development on watercourses, references in numerous reports to WUAs no longer functioning, once the construction period is over, are specious. A more accurate interpretation given by farmers is that the watercourse committee has always performed its traditional functions, just like the WUA on paper. The WUA was superimposed during the project implementation period, but throughout the watercourse committee continued to function-"you could call it the executive arm of the WUA" was one way of putting it. Hence, except for isolated cases visited where WUAs had moved into broader activities, which WUAs were possibly selected for a mission visit for that reason, the mission concludes that the impact of the WUA institutional components of those projects was spurious to transitory, at best. The WUAs were themselves only a veneer on top of existing, and still functioning, indigenous institutions. In Sindh the feudal power structure made even the indigenous committees superfluous (see Box 3.1), and the formation of WUAs was often no more than an empty ritual. 55. And this oversight continues. See for example Box 6 of Participation in Irrigation, op. cit., a recent Bank report, which fails to recognize, in discussing inadequate empowerment of WUAs, that traditional "participation" in watercourse committees long pre-dates the imposed WUA movement and that these committees continue to function. 56. See for example Cernea, M. M (editor), Putting People First, 1985, page 33-"Ignoring Local Irrigation Organization," which describes two cases in the Philippines and Sri Lanka. 57. This box also contributes to the discussion towards the end of Chapter 1 that the differences between the provinces require that projects are designed and implemented separately, for Punjab and Sindh at least. 72 Box 5.2: Institutional Aspects From the Evaluation Survey Findings Water Users Associations: these had been formed on all the project watercourses as a prerequisite to participation in the projects (except for ISRP where there was no investment in watercourses). In addition half the control watercourses in the survey reported that they had a farmer organization for watercourse maintenance. The sustainability of the superimposed WUAs in contrast to the indigenous watercourse committees has proven slight: at the time of the survey only eight percent of the official WUAs were fully active; 50 percent were fairly active and 42 percent had become dormant, while all the committees were either fully or fairly active (that is they still served a usual function from time to time as needed-mainly for maintenance and conflict resolution). Membership of the WUAs had declined on 36 percent of the project watercourses by the time of the survey, while it had increased in 25 percent of the khal committees on control watercourses. The committees were also reported to meet more regularly than the officially recognized WUAs. The most obvious reason, noted by most observers, for the decline in the WUAs is their lack of any substantial functions after project completion. This is especially understandable where there is a longer-standing traditional organization through which watercourse maintenance is organized as before and disputes are settled. Only 10 percent of the WUAs were reported by farmers interviewed as having started any new activities. Grassroots leadership: the survey provided evidence that the WUAs were as dominated by influential farmers as were the watercourse committees on control watercourses, and that there was little genuine grassroots participation. Some 54 percent of those regarded as influential persons on project watercourses lived on the head reaches of watercourses, compared with only 19 percent on control watercourses, indicating a greater concentration of local power and status in the head reaches of project watercourses than on other watercourses (what this difference implies is a matter for speculation, including what effect such concentration had on watercourses being included in the program). WUA benefits: the most frequently mentioned benefit was the reduction in disputes over water. Almost all (81-88 percent) of farmers on renovated watercourses reported disputes having decreased while only two percent of control farmers reported fewer disputes. The implication is that watercourse renovation greatly reduces disputes over water, while supply channel improvement, without watercourse renovation, has the perverse effect of increasing disputes. The reduction in disputes is an important contributor to reducing labor requirements, since there is less need to guard watercourses during the irrigation turn. Relationships with Government Agencies: individual project farmers in Punjab generally reported a worsening of their relationships with all three government agencies covered by the survey: Irrigation and Agricultural Departments, and the OFWM Directorate. Sindhi farmers were much more likely to report no change in their relationships with government or to report improvements if reporting any change. This is a reflection of the differences in land tenure and social systems between the two provinces. In Sindh most interaction with officials would be through politically influential zamindars who are usually able to influence water distribution arrangements, by one means or another, in their own favor. In Punjab, farmers usually had to follow whatever instructions were handed out by departmental officials. OFWM project farmers were the most likely of all project farmers to report changes and exhibit inter-provincial differences in their relationships with government When responses from Punjabi farmers from all projects were pooled, they were more likely to report a worsening in their relationship with their provincial Irrigation Department than with their Agriculture Department. By either action or inaction, the Punjab project agencies disappointed Punjabi farmers. 73 Box 5.3: WUA Formation-Intentions and Reality According to the Water Users' Associations Acts of 1981, a water users' association (WUA) must be formed with the signatures of at least 51% of the land owners served by a given watercourse. All land owners were eligible to be members of the WUA, but only dues-paying members would have voting rights under the legislation. Once a WUA was formed, the PC-I for OFWM I and 11 2 allowed a watercourse to be renovated under the respective project if at least 66% of the land owners so desired. Given this legislation, and the rhetoric of Bank documents and government officials interpreting the legislation, the mission expected to find documentation in the field that the OFWM directorate had complied with the legislation requiring the majority and super-majority described above. In fact, each of the OFWM directorates visited had signatures and membership forms which WUAs had been required-to complete prior to their involvement with the project. But a selective sampling of watercourse files in each of the four provinces revealed that rarely had all reported members of the WUAs actually signed the required paperwork.a In a majority of sampled cases only a small minority, frequently just the recorded officers, of members had signatures in the file. To reconcile the files with the published legislation (about which all staff appeared informed), various government/OFWM employees noted that: * all of the paperwork (articles of incorporation, etc.) had been signed on one day and not all farmers were present: * a small number of influential owners traditionally made decisions for the larger community; * even when fragmented family holdings are listed under many different names, there may be only one individual making the major decisions for the block of holdings; * often it was possible to convince a minority of land owners to commit all the required labor resources for the entire lining (e.g. if the lined portion of the watercourse was perceived to benefit them sufficiently); and * the implementing OFWM officials simply were not required, and therefore did not care, about requiring a majority. While each of these rationalizations may be viewed as acceptable from an administrative efficiency standpoint, accepting these explanations sidesteps the project objectives of participation, community organization, and cost-recovery. Not requiring a majority virtually guarantees that in areas where top-down decision-making prevails, top-down decision-making will persist. Creating "paper" organizations undermines the goal of encouraging joint activities by the formation of a viable community organization. And when cost-recovery shares are based on land ownership and not voting power, a variety of skewed incentives are created for voters at decision-making and non-voters at cost-recovery. a. Thumbprints were acceptable to the Directorates in lieu of written signatures, and the mission treated them accordingly. 74 Box 5.4: The On-Farm Water Management Directorate in Sindh Province "The OFWMD has to play a pivotal role in improving structures and management at the on-farm level and for this, it has to work closely with the PID. But the relationship between the OFWMD and the PIDa is distant and there is hardly any dialogue or joint planning with regard to irrigation water management. The OFWMD has its own budget, procedures and working rules. It independently carries out precision land-leveling and watercourse lining. It has its own water management officers who design and supervise watercourse lining. Currently, the OFWMD is under severe financial strain and as a result, it has to reduce its work to which it responded by reducing the length a watercourse is usually lined. At the beginning, it lined the first 30 percent of the selected watercourse. But now it does only 10 percent. The engineers rarely consult project beneficiaries in such matters and solicit their support. This seems an odd decision on how to make cuts in budget allocation. It might have been more sensible to cut the number of watercourses rather than the length of lining. It would be interesting to compare the rationale that led to the original 30 percent lining decision and the rationale that led to the new 10 percent decision. We have not been able to obtain information to make this comparison. The OFWMD apparently does not have a clear understanding of the role and functions of WUAs in Sindh agricultural communities. It prescribes that farmers would organize into WUAs at the watercourse-level when the zamindars expect its assistance in lining watercourses. In this it does not take account of how the traditional semi-feudal social system supports or hinders group activities with regard to irrigation water management at the farm-level. As we found in the [unlined study watercourse], the labor contributions towards any O&M activity come from the haris who are not members of WUAs. The prevailing practice is that the zamindars mobilize their haris enthusiastically to contribute their free labor towards watercourse cleaning or maintenance, only if the watercourse provides irrigation water to cultivate sufficient amount of their lands so that they could earn their livelihood from agriculture. Otherwise, as the small zamindars do in the [unlined study watercourse], they simply ignore the O&M of the watercourse. When the zamindars are not interested in O&M activities it is difficult to obtain haris'support in such work. The OFWMD does not mobilize local skills in physical works although it is supposed to bring in the beneficiaries' participation in OFWM activities. It brings inputs and skilled laborers from outside although farmers claim they have masons in their village. The officials merely asked the farmers to provide free unskilled labor. As experience in many countries has shown, such procedure will not encourage effective beneficiary participation in subsequent irrigation development and management." Source: P&D Department, Government of Sindh, Project Beneficiaries and On-Farm Water Management, Volume 1, April, 1993. a. Provincial Irrigation Department. b. Landless sharecroppers. Lessons on Watercourse Institutions 5.16 The lesson for the future- is that lasting institutional development needs first to recognize and understand existing institutions, and whenever possible build on them, rather than go through the process of apparently constructing a whole new institutional arrangement which unknowingly merely duplicates the existing indigenous institutions. Second, if an institution is to be sustainable it must have important sustained functions or acquire new functions. Since the WUA, real or just a veneer, was really necessary only during the watercourse reconstruction 75 period, expectations that it would continue to function were unrealistic.58 For much of Sindh, furthermore, the concept of WUAs as obligatory project institutions is often an irrelevancy given the concentrated land ownership patterns on most watercourses. Monitoring and Strategy Development 5.17 Since the four projects represented a major part of the investment strategy of the Revised Action Programme, related institutional aspects are reviewed briefly. RAP was only the latest, in its time, of periodic master plans and programs for resolving the physical and technical problems of the Indus Basin Irrigation System. In the past few decades these have typically been drawn up by largely expatriate consultant groups with greater or lesser local participation. Such studies were generally financed by UNDP, the Bank, USAID and other bilateral donors, and were often under the management of WAPDA. RAP was more indigenous in character, but now has fallen by the wayside, through lack of impact in certain areas, and has been replaced by the emerging consensus between the government and the Bank on a new strategy, as described in Chapter 6. 5.18 There are two other elements in the institutional structure that contribute to both the periodic changes of direction and, most importantly, allow the adverse impacts identified by the evaluation to persist. 5.19 First, the system as a whole is not adequately monitored. With routine measurements by field staff and modern means of communication and data storage, the condition of the Indus Basin Irrigation System's infrastructure should be routinely available to politicians, senior officials and technical staff-local or visiting. An objective and transparent monitoring system is needed, outside the influence of budget and water allocation pressures that could corrupt it at the provincial and agency level. This leads to the second concern in this area. The reduction in water losses through lining and the increase in crop production resulting from these water savings, which are fundamental to any assessment of the impact of the whole OFWM program, remain largely questions of judgment some 20 years after such investments began. Given total projected costs of watercourse renovation (as currently practiced) of at least US$1 billion (team estimate), some further investment in routinely and regularly measuring results and impacts is justified. 5.20 In the past, on a broader front, a degree of objectivity in developing strategies and plans for the system was provided by the practice of putting such work out to large consortia of internationally recognized consultants. WAPDA in the past was also seen as a technical agency with enough power and distance from competing concerns within the system to provide objective guidance for initiatives affecting the system. But that situation has changed and it is concluded that another institutional source of water sector monitoring, analysis and strategy formulation is 58. The Bank's South Asia Regional Office has commented on institutional impact as follows: "The limited institutional growth in watercourse command and inactivity of water users association (WUAs) after watercourse improvement is mainly because of the limited functions assigned to the WUAs after watercourse improvement. After the watercourse improvements the need for O&M reduces, water supply is more reliable, therefore, the WUA becomes inactive. This would also be the situation even if the traditional khal or watercourse committees were used to improve the watercourse. The WUAs could only stay active if after the improvements they are assigned roles essential for irrigation-distribution of water and collection of revenue. The government agencies have to move out in order to create space for WUAs to continue their functions. The institutional impact of ISRP and CWM projects have been presented rightly (in the impact study). These lessons from these projects were the basis for our moving towards the institutional reform package proposed under the new strategy for Pakistan's irrigation sector." 76 now appropriate. Such an institution should have clearly defined powers to obtain the information it would need, a set of firm objectives and outputs, and a budget which was independent of parochial or partisan influence. Ideally it would have long term affiliations with recognized overseas agencies which could help to provide it with a fast start in addressing critical concerns. An NGO, with a set of directors who are clearly independent, technically qualified and represent a range of disciplines and interests, appears to be the institutional format deserving first consideration. Part of the initial capacity of such an institution might come from reassigning units from within WAPDA. 77 6. Future Strategy-Addressing the Irrigation Crisis 6.1 This final chapter briefly reviews the new strategy for the irrigation subsector, as developed with the government over the past three years, and then proposes that future project assistance be disaggregrated to the provincial or sub-provincial level. This would permit investments to meet local needs more precisely and would result in greater impact on poverty alleviation and lasting institutional growth for future development. 6.2 In the late 1980s there was a further updating exercise of the irrigation investment plan. The Water Sector Investment Plan (WSIP) of 1991 provided a medium term investment plan for 1990-2000, but suffered from preceding the settlement later in 1991 of the water allocation issue between the provinces. Consequently, the plan was a composite of provincial investment proposals drawn up by provincial planning cells, and coordinated by consultants, rather than a strategy to optimize resources for the whole basin. Of interest to this report, the plan endorsed OFWM activities, but without subsidies; proposed to initiate systems to optimize water use; and made several suggestions for improved planning, monitoring and analysis of water resources and related issues. These points continue to be important. 6.3 The WSIP did not, however, influence the Bank's lending program for irrigation as in 1992 the Bank temporarily ceased lending to Pakistan's agriculture sector as a result of dissatisfaction with the results of the approach which had been followed since the early 1980s. The four projects which are the subject of this IER were a key part of that approach with respect to irrigation, and were followed by six other projects. The strategy was failing to address the causes of the ills of the Indus Basin Irrigation System. Remedial efforts were not producing the desired results, the portfolio under implementation was not tied to a coherent strategy and was performing badly, the government either side-stepped or only half-heartedly implemented difficult policy and institutional reforms, and performance continued to decline. 6.4 The Bank then changed its assistance strategy quite remarkably, away from supporting the rehabilitation of a faltering public sector irrigation system towards promoting a quite different solution. The Bank's sector report Pakistan, A Strategy for Sustainable Agricultural Growth, November, 1994, summarizes the proposed new approach to the irrigation sector as below. "To address the present crisis in irrigation management and ensure optimal use of a scarce resource, market pricing of water must be allowed to direct the irrigation system. Pilot programs and experimentation will be useful in phasing in market-based reforms. The government needs to give priority to drawing up enforceable property rights to water. Administering institutions will be restructured along commercial lines. The best option is to develop autonomous, commercially-oriented public utilities on a canal command basis [down to the head of minor canals], ensuring cost recovery of all current operation and maintenance and future capital expenditures. Farmers themselves will become responsible for operations and maintenance [from minor canals onwards]. Water user associations will be vital to the new market-oriented irrigation system." [clarifications in parentheses added] 6.5 Following extensive discussions between the Bank and GOP and other institutions involved in the irrigation and drainage subsector, details of the new approach were formulated and agreed. The strategy consisted of five inter-linked steps: 78 (a) legislation to formalize water trading and recognize individual property rights to water; (b) division of provincial irrigation department functions into public utilities at the irrigation command level and a provincial water authority at the provincial level; (c) develop the provincial water authorities as administratively autonomous agencies to implement the provincial governments' responsibilities in water resource management; (d) active promotion of farmer organizations, which would: (i) represent users in dealing with public utilities; (ii) purchase water delivery services wholesale from the public utilities and deliver them to members; (iii) collect water charges, and (iv) organize O&M of irrigation and on-farm drainage by members; (e) strengthen federal agencies [specifically the Water and Power Development Authority (WAPDA) and the Ministry of Water and Power] to take the lead in planning, provision and O&M of inter-provincial facilities for canal water capture, storage and delivery, and for drainage and flood protection. 6.6 This approach owed much to a world-wide movement throughout public sector irrigation subsectors, starting in the Philippines in the mid-1970s, to reduce the role of governments in what have long been seen as poorly functioning and unsustainable command irrigation systems. The "turnover" process, as it has been dubbed in some countries, has been widely assisted by the Bank, amongst other agencies,1-fotably in Mexico, Philippines, Turkey, Nepal and Indonesia. This focus of the Bank on giving irrigators responsibility for the lower end of the system and converting government irrigation authorities into commercial water supply agencies ("public utilities") dates from work done in the mid-1980s (see Jones, W. I., 1995, page 106 et. seq.). 6.7 These are radical proposals which will affect all those who have any connection, directly or indirectly, with irrigated agriculture and the irrigated areas. While there is overall support for the strategy at high levels of government, there is also understandable skepticism and doubt around the provinces. The established patterns of social, political and commercial relationships will change markedly as the proposals are implemented, and there will be those who hesitate to support the strategy or take time to find their place in the new order. Such lingering doubts, as confirmed by field inquiries, are strongest amongst those who will be most affected by the new strategy, principally the provincial IDs.. Conscious of this, the government and the Bank have held back from promoting wholesale reform and are beginning with a transition program and pilot commands, initially in drainage, through a proposed National Drainage Transition Program (NDTP). The first steps in the implementation of this new strategy are: (a) legislation to legalize water rights and permit water trading, and to establish farmer organizations, public utilities, Provincial Water Authorities and Provincial Regulatory Commissions, and to enable them to operate in the water sector; (b) pilot projects to establish farmer organizations and public utilities in selected canal commands and to establish Provincial Regulatory Commissions; and 79 (c) institutional strengthening, restructuring and capacity building of the key irrigation and drainage institutions, to enable them to carry out their functions efficiently. Some of the pilots have already been financed through on-going projects, and NDTP would start new pilots. The new investment strategy for drainage is a 25 year National Drainage Plan, the centerpiece of which is a National Surface Drainage System to discharge drainage flows into the sea. The continued rapid deterioration of the irrigation and drainage system has made the need to start on this a matter of critical urgency. 6.8 Given that the new strategy evolved out of the lapse in Bank support, on performance grounds, for the agriculture sector from 1992, the Bank's willingness to support the new strategy is explicitly linked to performance against undertakings. Thus the documents59 indicate that if GOP implements the reform agenda with determination, allocates the needed O&M budget for two years, and exhibits a new and higher level of determination to implement the investment program, then the Bank would confirm its commitment to support GOP in its implementation of a more ambitious investment program by appraising a number of other irrigation and drainage projects, including successor projects to those reviewed in this IER. If NDTP is unsuccessful due to lack of government commitment, then these new projects would not go ahead, and any incremental lending would be scaled back. 6.9 Based on the findings of this impact evaluation the following considerations are relevant to the pursuit of the new strategy for the irrigation subsector: * system condition and modification-"turnover" to irrigators of the tertiary part of the irrigation system and conversion of government agencies into public water supply utilities presupposes that the facilities concerned are in good repair and can function efficiently to ensure the strategy can work, or that the transition includes arrangements to ensure that the needed investments are made. Thus it seems self- evident that this strategy has to be on the basis of starting the new era with "going concerns", and not the present situation of "irrigation and drainage crisis" as it is termed in recent Bank reports. Furthermore, not only are there repairs and rehabilitation backlogs, but the system design will need major modification if a water supply utility is to respond to market-based demand as intended. * from strategy to local designs-the strategy requires disaggregation into designs suited to the different circumstances throughout the system. As discussed further below, the irrigation system is so large and contains such great variations in physical and social characteristics, from province to province and place to place within provinces, that a wholesale strategy may not be applicable without substantial modification to suit different realities on the ground. The pilot commands should contribute to highlighting the differences. * design and implementation partnerships-departures from important strategy and design criteria during implementation, as experienced by the four evaluated projects, suggests that more thorough examination of the intentions and interests of all parties is needed before agreements are reached. The superficial agreements of 59. (a) Pakistan - Irrigation and Drainage: Issues and Options, Report No. I 1884-PAK, March, 1994. (b) Pakistan: National Drainage Transition Program, Executive Project Summary, June 14, 1995. 80 the past, as seen in the reviewed projects, need to be replaced by working partnerships pursuing joint objectives. A repeat of this failure to follow an agreed strategy would rapidly undermine external support for the radical changes now sought. National Versus Provincial Project Designs 6.10 The scope of these projects is of concern to this evaluation because the main focus is how the projects now look in terms of their impact on direct beneficiaries-tli farm families. The project designs, particularly of the three projects involving watercourse improvements, resulted in less impact on poverty alleviation and the welfare of the most needy than would have been possible if more focused projects, for discrete homogenous areas, were designed, in particular to take local social conditions into account. It is clear from this evaluation that projects such as these designed on a national basis cannot take into account local social conditions, and nor was that the intention. More recently the Bank's renewed emphasis on poverty alleviation as the main priority suggests that irrigation subsector investments are broken down into smaller design units so that this priority can be taken into account. 6.11 Annex H discusses the case for such a renewed approach, based on the poverty targeting and other considerations which affected the impact of the four projects evaluated. These points are that implementation was difficult for the provinces since they were one step removed from the Bank; that project operations were delayed for some provinces where there was follow-up financing, since a project had to wait for the slowest province; differences in institutional and social conditions between provinces (and even within provinces) do not allow one project design to properly respond to those differences; physical conditions within the irrigation system vary such that project designs need to be tailored to those differences (the obvious example being fresh and saline groundwater areas); vested interests and the widespread rent seeking behavior within the system require that effective designs are tuned to local realities rather than being broad strategies which are undermined locally; and finally, the needs of smaller provinces, and of specific areas within provinces, should be reflected rather than being dominated by the larger provinces. 81 7. Conclusions and Recommendations Introduction 7.1 The Bank has played a central role in the development of Pakistan's Indus Basin Irrigation System, both as the largest source of funding for the subsector and by its catalytic role in periodic major reviews of irrigation and power sector strategy. This report has described and evaluated a phase of that assistance during a critical period when a new investment strategy was adopted to correct for operational problems and increase the system's efficiency. 7.2 The government's Revised Action Programme for Irrigated Agriculture of 1979 (RAP) switched investment emphasis towards improving system efficiency through rehabilitation and upgrading, and small-scale physical improvements. RAP deferred the large and long gestation projects, which had previously dominated the irrigation investment program, in favor of a smaller scale approach through the evaluated projects and other similar ones that followed. At the time the four projects under review were approved they supported some of the most important priorities established in the RAP (the problems of the public tubewells were addressed through other Bank-assisted projects). 7.3 The four projects were a successful part of the early implementation of the RAP strategy, but in a number of ways described earlier implementation drifted away from the strategy such that results and impact were not as great as RAP projected. The Bank, on its part, was an unhelpful development partner to the government in not insisting on implementation of the best strategy, established by the Revised Action Programme, against the pressures of special interests and rent seekers to do otherwise. The Bank allowed concerns for short term physical progress to overshadow longer term development impact. The most obvious missed opportunity was the failure to concentrate the watercourse renovation campaign in saline groundwater areas, as recommended by RAP. 7.4 Other main concerns relate to equity and design issues. While the projects had some worthwhile and widespread poverty alleviation impact, they also provided, without any justification, large transfers of public resources to some of the rural elite. Also the national scope of the OFWM projects, in particular, tended to demote local differences and priorities. Operations more closely tailored to physical and social circumstances on the ground in each province, with appropriate timing and implemented in accordance with the agreed strategy, would have had better results. 7.5 The institutional outcomes of the projects were not up to expectations: both the attempt to organize farmers into Water Users Associations (WUA) and to improve farm services by bringing together the irrigation and agricultural staffs under CWMP. However, the WUA movement deserves most attention. The long history of institutional experience, which attests to the short lives of rural organizations when they are imposed by some outside authority and have limited functions, was borne out again by the WUA experience under the OFWM projects. Such social engineering must be fostered rather than imposed so that organizations arise from within communities in response to their needs. Institutional growth and impact on a sustainable basis in the OFWM program is also undermined by incentives for implementation staff being closely tied to physical progress-watercourse reconstruction-rather than to development impact and institutional progress. The Bank helped to further this distortion of objectives by making it plain 82 that construction progress was the highest priority. Moreover, in the case of the OFWM program, the brushing aside of functioning, traditional watercourse committees, in favor of frequently token WUAs, was an added mistake. 7.6 While the sustainability of the improvements made to watercourses under the projects is not in doubt, the method of financing adopted, with its heavy reliance on external aid and subsidies, is not sustainable. The government cannot afford to extend OFWM to all the remaining unimproved watercourses on that basis. The Bank bears some responsibility for promoting such an unsustainable financing strategy for so long. The Revised Action Programme advocated a credit approach, arguing that OFWM profitability would make it attractive for farmers to borrow. The choice seems to have been between short-term results and establishing a long-term sustainable program, and the Bank chose the short-term objective and persuaded the government to drop the RAP approach. (A reviewer has noted that adopting the RAP approach may have presented quite a challenge given later performance problems with rural credit programs.) 7.7 Even so, these four projects were generally successful in achieving their main physical objectives, but the greater institutional and social problems affecting system performance remained and may have increased. Chapter 6 described briefly how the government and the Bank have reached a new understanding towards tackling the longer term problems of the system through a decentralization strategy, but the need for more packages of physical investments such a those evaluated will not be any less as a result. 7.8 The combination of the field inquiries by the evaluation team, the farmer survey and reviews of the many recent reports which address aspects of the irrigation system's problems, have led to two overarching findings: (i) the need for disaggregated investment packages which are tailored closely to the specific physical, social and economic characteristics of the very different areas within the system, rather than broad scale national projects which lack the design specifics which would make them most effective in any one situation, and (ii) the need for a much improved and independent analytical base on water sector issues, routinely maintained in current status rather than being periodically updated when another study or survey tackles the latest high priority concern. Major Results and Impacts 7.9 (a) Farm Water Supply. The investments in watercourse renovation under the On-farm Water Management projects are popular with farmers and with the staff in the OFWM directorates. They are particularly attractive to farmers because they are subsidized--more than intended because cost recovery is low--as well as giving substantial and immediate benefits. Farmers receive increased water supplies with the bonus of saving labor on irrigation operations and there are far fewer disputes over water. When water savings are added from similar later projects, total water savings from RAP projects exceed increased surface irrigation supplies that the (proposed) Kalabagh dam would have provided, thus confirming the validity of the RAP strategy with respect to increasing irrigation supplies at the farm. 60. The Bank's traditional quite rigid project cycle, which was originally developed for massive engineering projects, dictates that physical implementation be scheduled and the schedule obeyed, whereas community based participatory implementation has a pace of its own. Progress may be uncertain, but it will be more sustained in the long run. Regrettably, the rigid engineering schedule is still seen in project designs that purport to be participatory. 83 7.10 (b) Economic Rates of Return of Watercourse Renovation. Available estimates indicate that the economic rates of return of these water saving projects is around 20 percent, indicating that they were sound investments. Comparing unit costs for increasing farm water supplies, watercourse renovation is the heavily favored option. The OFWM strategy is relatively low cost, but remains an unsustainable financial burden so long as subsidies are maintained. Greater private sector involvement, and at market prices, is desirable. 7.11 (c) Groundwater Recharge. Although the immediate impact of watercourse improvement is identical in both fresh and saline groundwater areas (groundwater recharge is reduced), the secondary effects are different. The unit costs of pumping rise in fresh areas as the watertable falls, but less pumping may be needed, while in saline areas the beneficial effect on (high and rising) watertables level is a major bonus on top of increased irrigation supplies. Hence the reduction of seepage losses in saline areas remains of highest priority. (At a late stage in completing this report a reviewer suggested that seepage losses from watercourses were not such a significant factor as the report assumes. Consequently, recharge flow from watercourses in fresh areas is lower than usually assumed, and hence that the loss of the benefit of recharge in these areas is less significant than this report maintains. The further point is that this clarification is not evident in the technical reports reviewed, and thus the need for improved monitoring and analysis of water issues is reinforced.) 7.12 (d) Canal Rehabilitation and Lining. The departures from the RAP strategy of note for this study is that canal rehabilitation became canal enlargement, much as feared by the Bank. Since drainage works to cope with extra flows were not put in place, waterlogging and salinity problems increased in some areas. The farmer survey also suggests that ISRP and the canal components of CWMP did not result in improved reliability of canal deliveries at watercourse heads. Given that canal maintenance remains inadequate, periodic capital investments of this type, substituting for operational expenditures, will continue to be needed. 7.13 (e) Institutional Impact. Project objectives for institutional change were misdiagnosed and too ambitious in consequence. The Bank's record in the institutional arena is not strong and these projects maintained that trend. CWMP's intention to solve the problems of the separated irrigation management and farmers' services functions was unrealistic, and the Water Users Association concept is flawed so long as it attempts to replace local institutions and is imposed rather than promoted. Imposing WUAs as a condition of receiving a large subsidy for watercourse renovation is probably the least likely way of achieving lasting institutional change, and in this case led quickly to tokenism as physical targets took priority. The ISRP had more modest and achievable objectives of strengthening the irrigation departments' operations through their implementation roles. 7.14 (f) Economic andSocial Impacts. Water savings discussed above translate into the greatest economic gains of these projects, largely through expanded cropping, but other effects have indirect economic impacts. These are mainly changes in work loads for men and women, health benefits, changes in amenities and communications, and adjustments in equity and social cohesion. The negative changes relate mainly to increased work for women, resulting from cropping expansion, and the loss of amenities with some forms of canal improvements when canals are the main community water source. Workloads for men seem to have reduced significantly as a result of the projects, not least because water-related disputes, and the need to police the watercourse in one's own interests, have greatly reduced. Lining of watercourses had 84 health benefits which could be replicated with health-focused watercourse projects which restrict work to watercourse runs within or close to villages. Recommendations 7.15 The new directions now being taken towards decentralized system operations with a split of responsibility--between irrigation departments (as a form of public utility) and farmers associations-at the head of the minor canal, does not imply that public sponsorship and financing of irrigation investments will no longer be required. The system is not now what could be considered a "going concern" in business terms, in the sense of being able to cover all its costs and make a profit, without eating into its capital. Also, like any modern business it will need a continuing line of finance. This suggests that investments of the type evaluated in this report will continue to be publicly financed, but with the following changes the results could be much improved in terms of the kinds of impacts discussed above. 7.16 (a) Watercourse Renovation. In addition to its economic merits, watercourse improvement can be community based and sustained, which is a distinct advantage given the operation and management problems of the public irrigation system. Hence, from this perspective watercourse investments are to be preferred as the main means of reducing water losses and thereby increasing irrigation supplies at the farm. However, to increase the social and economic impact of such projects, design changes on equity and cost-effectiveness grounds are needed. 7.17 A rejuvenated program of watercourse and on-farm initiatives needs to take into account the following points: (a) better results would be achieved with operations closely tailored to physical and social circumstances on the ground in each province and different areas of a province; (b) farmer participation in the broadest sense (not just their representatives) in design and implementation of each watercourse subproject; (c) cost sharing formulae need to be strictly enforced (possibly by accepting farmer contributions up front) with attention to relative needs in the interests of greater equity, particularly to end large transfers of public funds to many of the rural elite, (d) physical implementation targets should not be allowed to supplant institutional and social objectives, and (e) lining of watercourse stretches, primarily for health and amenity purposes, could be a valid objective in some cases. 7.18 The case for taking social differences into account in designing participatory interventions has been well-stated in a recent Bank report: "In cases of very hierarchical social structures and inequitable distribution of assets (for example, Sindh in Pakistan) it may be unrealistic to expect fully equitable and democratic local organizations. Therefore, the Bank and government-need to recognize their role in controlling vested interests and acting as advocates for the poor. The differential incentives of different categories of farmers ("head"/"tail", small/large) should be recognized and dealt with in program design (for instance, in defining water rights), along with the resulting problems of achieving collective action."61 Participation rates worldwide in community based common interest groups are quite low, such that both expectations and the approach adopted, to support future irrigation system investments and management, needs to be carefully based on local realities and not be dictated by time-bound master plan and physical implementation targets. 61. Participation in Irrigation, Environment Department, February, 1995, page 10. 85 7.19 ( Canal Rehabilitation. The canal rehabilitation and lining work facilitated some significant capacity expansion which was not part of the strategy, and this led to increased waterlogging and salinity in some areas. Drainage investments, especially in these areas, remain a high priority. Poor operation standards and low maintenance levels continue, the rehabilitation component of the investments will not be sustained and further rounds of emergency rehabilitation will be needed in due course. The social and amenity aspects of canal improvements, in relation to the communities through which canals pass, need to be taken more fully into account during design. The social and economic disruption caused by inadequate bridging is obvious, but other amenity losses seem to occur too frequently when canals are rehabilitated and lined. Irrigation departments should take community concerns more into account by closer and continuous consultation with communities. 7.20 (g) Design and Funding Packages. A new approach to project packaging is required, disaggregated at least to the provincial level. Operations more closely tailored to physical and social circumstances on the ground in each province, with appropriate timing, would have had better results. Such an approach should take into account the Bank's current re-emphasis on poverty alleviation as the primary objective, and include equity and the fullest participation of stakeholders in design and implementation. Differentiation would permit, among other things, a more efficient allocation of scarce resources, taking relative needs into account. Furthermore, if future Bank financing is to take the poverty, equity and stakeholder points into full account, it will be necessary to break out areas within individual provinces, and even within canal commands, into separate monitorable packages with their own development impact indicators for tracking purposes. 7.21 Bundling up a time slice of a single uniform program for all provinces, for convenient financing by the federal government and external agencies, risks imposing sub-optimal solutions on very different areas with differing problems. Resources, social factors and the physical state of the irrigation system vary too much from place to place to accommodate such an approach without something being lost. Too many important differences between provinces, and also between areas within provinces, are pushed aside in the search for a common formula that will work reasonably well everywhere. The danger of such an approach is that it will inadequately address the precise needs of specific situations. An unbundled approach should allow for individual subproject designs for each watercourse, within agreed guidelines for the area, command and province, and with full farmer participation. Lining by formula should give way to individualized designs for each watercourse according to what the people want and are prepared to finance in large measure and to implement. Targeted subsidies may be tailored to real need in terms of holding size and water supply, for example to favor watercourses located far down the supply channel with a less reliable supply. Lining of watercourses primarily for health and amenity purposes should be a valid objective where demanded. Designing investments specifically for individual watercourses would allow for social differences between watercourse, and within watercourses, to be better accommodated, provided a transparent participatory approach is adopted. Disaggregation of project designs would permit, among other things, a more efficient allocation of scarce resources, taking relative needs into account. 7.22 (h) Water Resources Monitoring and Analysis. Monitoring of development impact should be given higher priority, and on a continuous and consistent basis throughout the irrigation system. It is unsatisfactory that after twenty years of investments in conveyance efficiency improvements to reduce wastage of Pakistan's vital scarce resource, that the amounts of water saved remain a matter for analysis and desk estimation, rather than a matter of routine 86 and continual record. Development indicators are sometimes difficult to identify and measure, but in this case irrigation water, as a proxy in a water scarce economy for farm income and family welfare, is readily measured. The very large funds involved on a national basis, together with the implications for irrigation and power sector policy and strategy, suggests that more effort to monitor water related matters is fully justified. 7.23 Development of efficient strategies for water resource investments requires a current and fully independent information base on the system's performance, the parameters involved and the probable impact of the options under consideration. An example of an issue which ought not to be a matter of opinion is provided by the running debate about whether investments in the canal system should be given priority over OFWM activities, or the reverse. The major agencies and individuals in the public and private sectors who have influence over these important decisions have often quite legitimate vested interests to protect, but also, as described in the report, there are other vested interests at work which can have negative impacts on strategies and investment decisions at the local level. There is an urgent need to move the monitoring, analysis and reporting of water resource issues into a highly skilled and disinterested institution outside the influence of political, regional and special interest. Such a detached body could make a substantial contribution to solving some of the hard questions that remain as the new strategy for irrigation sector operation and management is gradually put in place. A "Water Resources Institute," comprising in part at the beginning at least sections from existing bodies, and possibly in a not-for-profit NGO format, could become a center of excellence for all matters relating to this national resource, but without any direct interests in the management or financing of the system. 87 Annex A Description of the Four Bank-Assisted Projects Introduction 1. This annex describes the four projects which are the subject of this evaluation. First, however, a paragraph on the USAID-assisted earlier initiative in watercourse improvements sets the scene for the Bank's first On-Farm Water Management Project. 2. Before the Bank was involved, the Government and USAID agreed on a five-year nationwide OFWM Pilot Project in 1976. The project was executed by the Ministry of Food, Agriculture and Cooperatives (MFAC) through a Federal Water Management Cell. Implementation was through field teams under OFWM directorates within each Provincial Agriculture Department (PAD).] It was intended to improve 1,500 watercourses with compacting of banks, ten percent lining and introduction of pukka nuccas (improved concrete control structures). In addition 175,000 ha were to be precision leveled and 40,000 farmers were to adopt improved water management practices. Formation of a watercourse committee was a precondition of participation in the project and watercourses had to have not less than three quarters of their land holdings not more than 25 acres in size. Three training institutes were to be established. 1. The (First) On-Farm Water Management Project (OFWM-I) 3. Responsibility for OFWM project preparation rested with MFAC and the provincial agriculture departments (PADs). The original proposal called for heavy cleaning of 33,600 watercourses and lining of 2,869, with ten percent lining in fresh groundwater areas (FGW) and 30 percent in saline groundwater areas (SGW). The RAP recommendation, that lining be confined to saline groundwater areas, was ignored. However, selection priority was to be given to watercourses in saline groundwater areas without access to tubewells, and to those cultivated by low income farmers.2 In addition there would be precision land leveling (PLL) of 40,000 ha.3 4. The final OFWM project design halved the number of watercourses for heavy cleaning, and cut the number to be lined to 2,065, though raising the lining amount in FGW to 15 percent of the watercourse length. It de-emphasized PLL, cutting the proposed area to about 15,000 ha because there were insufficient local funds to maintain the 50 percent subsidy levels provided 1. The later Bank-assisted project followed this model, although it ignored the RAP recommendation that such works should be implemented by autonomous Provincial Minor Works Corporations to be created for the purpose, in order to leave the agricultural departments free to concentrate on their existing functions. 2. The case for preferring FGW areas without access to tubewell water was not strengthened by using farm models of fresh groundwater areas with tubewells to calculate project benefits. In addition, although it is not specified in the text, the footnote in the SAR to the cost table, Annex I, Table 8, assumes the ratio of FGW:SGW to be 75:25 in Punjab, 50:50 in Sindh and 70:30 in NWFP. 3. The initial OFWM concept believed that improving the conveyance efficiency of the watercourse is of little benefit unless it is accompanied by land leveling. In the original USAID project the watercourse team consisted of four land leveling specialists in a team of eight, of whom only two were involved with watercourse improvement. There was one specialist for irrigation agronomy. PLL has since declined dramatically in importance in the program, while watercourse improvement has grown in popularity. PLL is not needed everywhere, not in rice areas for instance. But it is essential in cotton-growing areas, especially in those cotton areas with fresh groundwater where it is higher priority than lining watercourses. Annex D-discusses the declining priority of PLL. Annex A 88 earlier and, without subsidy, it was believed the demand would be small. The subsidy was maintained in Balochistan. The project had three institutional objectives: (i) strengthening the planning and implementation capacity of the OFWM directorates in the four provinces, and the Federal Water Management Cell; (ii) improving coordination between the OFWM directorates and the Provincial Extension Services of the DAs; and (iii) improving water management (farm demonstrations, water management extension specialists), including the organization of Water User Associations (WUAs). WUAs are formed of farmers on a single watercourse, with legal identity under provincial legislation. 5. This first OFWM project was essentially an engineering project whose basic objective was to increase agricultural production by saving water through infrastructure investments and improved water management practices. Raising farmers' incomes was not mentioned as an explicit objective in the Staff Appraisal Report (SAR), but was clearly implicit. Heavy cleaning was estimated to save six to eight percent of the water 'lost' between the mogha and the farmer's field. Lining was estimated to save 25 percent [see Annex C(d)]. 6. WUAs had to be established before any lining work could start, but WUAs were not required where only heavy cleaning was to be undertaken. In practice, at least in the Punjab, farmers on such watercourses were made to form WUAs.4 Each province promulgated a WUA ordinance to make such bodies legal entities. The essential reason for having a WUA, which was stressed in the SAR and the President's Report (PR), was to overcome the problem of poor maintenance experienced in the Pilot Project.5 WUAs had to express their willingness to improve their own watercourse branches and to maintain the other improvements made by the OFWM directorates, before renovation work started. In addition, the possibility of WUA federations being formed along a distributary or canal was foreseen, and WUAs would be able to obtain loans from the Cooperative Bank, the Agricultural Development Bank of Pakistan and commercial banks. 7. An International Development Association (IDA) credit for SDR 33.4 million (US$41 million) was approved on June 16, 1981. The International Fund for Agricultural Development (IFAD) also provided a loan of SDR 10.5 million for the project. The total project cost was estimated to be Rs 1,104.6 million (US$111.6 million). An economic rate of return (ERR) was calculated for only the renovated watercourse component, but the heavy cleaning under the accelerated program was said to have an ERR of over 50 percent. The Basic Data Sheets after the Preface to this report contain more details. The ERR of PLL was also not calculated on the grounds that it was difficult to separate out the benefits of PLL from the associated changes in farming practices. The renovated watercourse component accounted for 60 percent of project costs and the overall ERR on this was estimated at 45 percent, ranging from 21 percent in the Punjab Cotton/Wheat model with FGW and a tubewell, to 77 percent in the Punjab Rice/Wheat model with FGW and a tubewell. As would be expected, ERRs were generally found to be higher for models representing SGW areas or FGW areas without tubewells. The high ERR for the Punjab Rice/Wheat model is due to the high value of Basmati rice. The re-estimated ERR in 4. In Sindh, NWFP and Balochistan the number of watercourses lined was almost the same as the number of WUAs established, while in Punjab the number of WUAs established was almost four times the number of lined watercourses. This was because in Punjab about half the watercourses with accelerated improvement had WUAs while in Sindh, NWFP or Balochistan, only lined watercourses had WUAs. 5. The Pilot Project showed that earthen renovation deteriorated rapidly without maintenance, while the lined portion required less maintenance than before renovation. The role of WUAs in watercourse maintenance is discussed later. 89 Annex A the Project Completion Report (PCR) was 29 percent, the reduction being due to significantly lower commodity prices than those assumed at appraisal. Sensitivity tests showed that these results were likely to be sustainable. Implementation experience is discussed later for all the four projects together. 2. The (First) Irrigation Systems Rehabilitation Project (ISRP) 8. This project originated out of a proposal to support the flood protection activities of the provincial irrigation departments (PIDs) following on the Bank's two flood rehabilitation and restoration projects subsequent to the disastrous 1973 and 1976 floods. It was not originally part of the Bank's strategy to implement RAP's recommendations; it only became so during preparation. These floods had highlighted the crucial need for improving routine maintenance of flood protection and drainage facilities if future flood damage was to be avoided. As preparation proceeded it became clear that there was equal need for similar improvements (caused by inadequate recurrent funds and deteriorating facilities) in the irrigation operations of the PIDs. This had also received additional emphasis in the RAP, which had pointed to the problems created by the lack of adequate maintenance and stressed the vital need to strengthen the PIDs. The project scope was thus broadened to cover the activities of PIDs as a whole, rather than giving emphasis to one particular aspect of their operations. A project to strengthen the irrigation operations of PIDs was also seen as an important step in evolving a new balance in the Bank's lending to the water sector generally and providing an opportunity for a continuing dialogue with the Government of Pakistan (GOP) on major policy issues as they affected the water sector, particularly those given emphasis in RAP. A consultant firm was engaged by GOP to prepare a feasibility report which became the principal reference document of an IDA appraisal mission which visited Pakistan in June 1981. This first ISRP was approved in April 1982, for an IDA Credit of SDR 35.5 million (US$40.0 million) out of total costs of US$118.1 million. It was co-financed by a grant from USAID, which had by then resumed its aid program, originally for US$45.0 million, but eventually raised to US$103.0 million. 9. Although the SAR gave primary emphasis to inadequate funding for Operation and Maintenance (O&M) as the major cause for the accumulation of deferred maintenance and the resulting problems affecting the irrigation, drainage and flood protection facilities, it clearly placed much of the responsibility for the need for rehabilitation on the excessive amounts of water being released into the canal system. Diversions were well beyond the limits of safe operation, and the problem was compounded by the additional flows just then being made available from Tarbela reservoir. The SAR described how most canals were operating with inadequate or no freeboard, how protective berms had been scoured away, how service road widths had become inadequate, outside embankments deteriorated and how sedimentation (caused by running canals below full design levels) was severely reducing the capacity of many minors and distributaries. Running the canals above capacity was resulting in frequent breaches, which caused enormous losses of infrastructure, private property and crops. Sedimentation was also causing maldistribution along canals, with watercourses in upper reaches withdrawing above their design, with corresponding reductions in deliveries to watercourses in lower reaches. (Figure 3 shows a recent example of the range of inequitable supply as measured by IIMI.) Inadequate maintenance of surface drainage systems was resulting in crop losses through flooding after heavy rains. 6. The International Irrigation Management Institute, Colombo, Sri Lanka. Annex A 90 10. The project was unable to tackle the basic problem of excessive discharges in the canal system, but aimed to contain the side effects by providing funds to rehabilitate approved surface irrigation and drainage systems that had deteriorated badly "because of deferred maintenance." It excluded rehabilitation of tubewells (private and public), subsurface drainage and most flood protection facilities. Its stated objectives were: (a) to increase agricultural production by providing a more reliable and equitable water supply, and reduce crop losses by improving the surface drainage system; (b) to strengthen the capability of the PIDs to implement the O&M program; and (c) to serve as a model that could be replicated in future projects. As part of the effort to improve O&M, the Project Agreements required the provinces to increase water charges periodically so as to achieve full recovery of O&M by the time schedules agreed under earlier projects, ranging from 1988 for Sindh to 1992 for Balochistan. This topic is reviewed in Box A. 1. I1. It was planned that the project would provide earthwork, stone pitching, and repair of hydraulic structures for about 14,000 km of canals, as well as cleaning and re-sectioning about 3,500 km of surface drains. The total area commanded by these systems was 3.8 million hectares. Canals would be rehabilitated to carry safely either the maximum discharge that they had ever conveyed, or their design capacity if that was greater. Hydraulic structures would be repaired in line with existing or enlarged capacities. The project documents clarify that rehabilitation would generally be limited to minors, distributaries and branch canals, with main canals being rehabilitated only in exceptional circumstances and with IDA concurrence. Canals to be rehabilitated in the first year were selected at appraisal, while those for future years would be proposed by PIDs in their Annual Work Plans. Criteria for selection were laid down in the SAR, including systems or subsystems where breaches were most prevalent and sedimentation was causing inequities in water distribution. Unlike the OFWM project, where provincial agricultural departments were responsible for implementation, this project was the responsibility of PIDs under the overall coordination of the Federal Ministry of Water and Power. 12. Project benefits were calculated on the basis of farm income increases resulting from reduced crop losses following improved maintenance of both the irrigation and drainage systems. It was estimated that about 12 percent of the canal command area was adversely affected by canal breaches and closures and that without the project this would increase to 15 percent in the next four years. It was also estimated that system deterioration over the same period would reduce both yields and cropping intensities against present levels by two to five percent over a much wider area. The project was estimated to reduce losses resulting from breaches to two percent annually and to benefit a large part of the canal command area leading to yield and cropping intensity increases of two to five percent, again over a four year period. Considerably higher increases were expected under the different conditions existing in Balochistan. The resulting ERRs ranged from 22 percent in Balochistan to 65 percent in Punjab, the higher ERRs reflecting the greater extent of deterioration, with an overall average of over 50 percent. 91 BRO 2770, FIGURE 3. AN EXAMPLE OF VARYING WATER FLOWS IN WATERCOURSES ON ONE DISTRIBUTARY PERCENT OF DESIGN FLOW. NOT SAMPLED <50 50-100 S01-150 - 151-200 >200 MANAWALA DISTRIBUTARY Source: IIMI Sample D& 1990-91, Mancwalo D;stributary Command- 19 Sample Watercourses, Kharif 1991 -------- > rr othe mírmoron how N | -d orld Bank- Grup" Sb1o dore deno,,rerir, ed 0V as-I ether,e~rmi,en ho~MARCHF 1996I Annex A 92 Box A.1: Water Charges and Operation and Maintenance The saga of the Bank's attempts to cajole GOP into raising water charges to a level at which they cover the operation and maintenance costs of the irrigation system is long, complicated and with no end in sight, other than to hand much of the system over to farmers to operate as under the current initiative. The four projects under review were but a chapter in this saga. They were a means of sustaining the dialogue with GOP on the level of water charges, a dialogue which had begun with Khairpur Tile Drainage Project (Cr. 648-PAK) which was approved in July 1976, but without any cost recovery covenants, continued through successive irrigation and drainage projects, through the 1982 First Structural Adjustment Loan (Loan 2166) and then the 1988 Agricultural Sector Adjustment Loan, and which is still continuing under the proposed National Drainage Transition Program. Despite considerable increases in water rates over the years, some of which were attributable to these projects under review, although not to the extent that they met the legal covenants, the gap between O&M expenditures and recoveries through water charges has been increasing. "Between FY88 and FY92, cost recovery fell short of O&M expenditures by an average of 38 percent, and would average 57 percent if recoveries were compared with requirements instead of expenditures." This issue is discussed further in the Operations Evaluation Department's (OED's) Audit of the Agricultural Sector Adjustment Loan (Report No. 12613 of December 17, 1993). Another reason for not reviewing the question is that OED studies show that the Bank policy on water rates that was in force at the time these projects were approved, of raising water rates to cover O&M, was based on a false premise. OED's most recent study on irrigation experienceb "confirms earlier studies by OED that there is normally no link between higher water charges and better operation and maintenance. Revenue from water charges generally goes to the Treasury and is not ear-marked for O&M." Thus, even had the level of water rates been raised to cover O&M, there is no reason to believe that this would have had any impact on the amount spent on O&M. The Bank's fresh approach in Pakistan to improving O&M, handing these functions over to farmer groups and to commercially-oriented public utilities, holds the best long-term promise of a solution but runs counter to all the Bank's experience in India. This new strategy builds on the experience gained by the Bank in similar turnover experiments in the Philippines, Mexico and Indonesia. The Bank's experience is summarized in the OED's report on irrigation as follows: "The Bank has promoted irrigators' groups and turnover of systems to them in a variety of ways, but disappointment has been more common than success. Success requires more than covenants in which borrowers agree to set up users' groups. Empirical studies have identified the conditions that lead irrigators to create users' groups that endure (Executive Summary, page 6)." a. National Drainage Transition Program: Executive Project Summary, para. 17. b. Jones, W. I., The World Bank and Irrigation, A World Bank Operations Evaluation Study, OED, July 1995. 13. Although the genesis of the project was the need to reduce water losses in the surface irrigation system, no attempt was made to estimate what the project might achieve in this respect. However, as the account in Chapter 4 will show, this oversight is of little consequence alongside the fact that the project was implemented largely without regard for the main design criteria and related conditionality. 93 Annex A 3. The Command Water Management Project (CWMP) 14. The recommendations of RAP for water management projects had been reviewed by a Bank identification mission in April 1980. This mission proposed separate projects for on-farm water management and command water management. On-farm water management projects would be confined to improvements within the watercourse command, below the mogha or outlet on the distribution channel. This led t6-the Bank's first OFWM project. Command water management was conceived as a basin-wide comprehensive water management program made up of sub-projects of 100,000 acres (45,000 ha) or less within canal commands. These sub- projects would include improvements throughout the whole irrigation distribution system of a canal command, both upstream of the mogha in that part of the distribution system within the jurisdiction of the PIDs, as well as-below the mogha within the watercourse command, along with programs for strengthening agricultural inputs and services. This was the concept originally articulated in RAP which had envisaged the integration of specialized agencies at the canal command level in order to strengthen the management of irrigated agriculture. It built on the experiences of OFWM-I and ISRP, but was innovative in attempting to bring together, for the first time, the other major policy and institutional recommendations of RAP. 15. Project preparation began in July 1982, financed under a UNDP project (PAK/001/84), Investment Project Preparation in the Water, Agriculture and Energy Sector. The Bank was Executing Agency and WAPDA provided the consultancy services. In addition USAID, as a potential co-financing agency, provided some expatriate technical assistance. The feasibility report was completed in August 1983 and formed the basis for appraisal in October 1983. An IDA credit for SDR 43.7 million (US$46.5 million equivalent), out of a total cost of US$81.9 million, was approved by the Board on May 29, 1984. USAID provided US$25.0 million co- financing. 16. As described in the Decision Memorandum after the appraisal mission,' the project "by simultaneously providing complementary investments in the physical and institutional aspects of water management for an entire irrigation command area,.....is expected to demonstrate the high returns to be obtained from a comprehensive approach to water management." It was intended that this approach would be able to be replicated throughout the distribution system. The project was also to be used as a means of making progress in the Bank's policy dialogue with the Government, especialy on O&M and cost recovery. In the Loan Committee Review of the project the Chairman said it was seen as a pilot project.9 It was thus particularly important that monitoring and evaluation should be started promptly. 17. In the SAR, the project concept (para. 3.02) was stated to be the "strengthened management of the existing institutions, infrastructure, agricultural inputs and services, so as to efficiently remove major constraints to irrigated agricultural production in selected subproject areas.... Improvements to and strengthened management of the irrigation infrastructure are emphasized in association with improved deliveries of non-water inputs and services." 7. Compared with the RAP recommendation of 500,000 acres. 8. December 2, 1983. 9. Memo to Files, April 3, 1984. Annex A 94 18. Along with the impact of such improvements on agricultural production, a major project objective was institutional development: "to build within the provincial agencies a continuing capability for planning, implementing, operating and maintaining integrated and efficient programs of irrigated agriculture." In addition farmer participation was to be strengthened through "formal water user associations to improve their overall water and non-water input management and provide them an opportunity to have a stronger voice in decision making." Informal federations of WUAs were to be formed "to facilitate implementation of project works at the minor and distributary canal levels," and PIDs or OFWM directorates were to seek the guidance of WUAs about proposed improvements and "obtain their concurrence prior to formulating detailed plans" (para. 5.07). The role of WUAs in providing free labor for watercourse improvement was stressed more strongly in this SAR than in the SAR on OFWM, as was their role in collecting the required initial cash contribution. 19. The watercourse remodeling program under CWMP had one significant difference from that of the OFWM. Instead of watercourses being selected on a first-come, first-served basis, the project would improve at least 80 percent of all watercourses within the subproject areas (for a total of 1,050 watercourses), thus allowing the water savings to be concentrated in a command. In addition the length of the watercourse to be lined would be increased to 40 percent of the total in SGW, compared with the 30 percent allowed in OFWM-I. This recognized the greater importance of reducing seepage in SGW than in FGW areas, where the lining percentage would remain at 15 percent. In one subproject area, Las Bela in Balochistan, where the soil was highly permeable, 50 percent was to be lined. 20. The water saving from watercourse lining was estimated to range from about 15 percent of water delivered to the watercourse head in FGW areas where 15 percent lining would be installed, to over 25 percent where 40 percent lining was installed in SGW areas, and 30 percent in Las Bela. It was estimated that in total about 236,000 acre feet per year, or 18 percent of the water delivered to watercourse head, would be saved by the project. 21. The canal rehabilitation and remodeling component of the project included not only lining of selected reaches, remodeling of moghas, and improvements to structures, but also rehabilitation. Rehabilitation was justified, on the same logic as that which was used to justify the ISRP, since it was needed to restore subproject canals to safe operating conditions based on the greater of design or present operating capacities. This later became a way in which additional surface water supplies were introduced into subproject areas, and indeed an examination of project files indicates that the Sindh PID was already asking for additional supplies even before negotiations.i0 Water savings from canal lining were estimated at about 50,000 acre feet per year. 22. The arrangements for project management are explicitly described in the SAR as a pilot undertaking. The problem was clear: fragmented institutional arrangements resulted in inefficiencies in the delivery of water and non-water inputs to farmers. The solution was not clear: the arrangements provided for in the project were an attempt to improve coordination between PIDs, OFWM directorates of PADs and extension departments. PIDs would be responsible for canal improvements and OFWM directorates for on-farm works. Coordination would be at three levels: federal, provincial and subproject. It was hoped that a system of organization that was able to be replicated could be developed. 10. Telex from Resident Mission to Bank headquarters, dated February 23, 1984. 95 Annex A 23. The estimates of the agricultural impact of the project were based on the water savings that would result from project actions, with farmers using 70 percent of the additional water to increase cropping intensity and 30 percent to support yield increasing practices. As a result crop areas would increase by between nine and 19 percent in Punjab/Sindh subprojects, 26 percent in Northwest Frontier Project (NWFP) and almost sevenfold in Balochistan. The assumed yield increases were substantial: 53 percent for rice, 59 percent for cotton and 55 percent for wheat. 24. The physical achievements of the project were considerable. A combination of the devaluation of the rupee and three one-year extensions of the project closing date allowed a major increase in civil works. Canal lining increased to well over twice the original estimate; canal rehabilitation/remodeling, outlet remodeling, and surface drainage construction all went up by 20-25 percent, and the number of watercourses renovated increased by 39 percent over the original expectations, to 1,459. 4. The Second On-Farm Water Management Project 25. An IDA Credit of SDR 34.8 million (US$34.5 million equivalent), for total costs of US$57.8 million, was approved in June 1985. The rationale for IDA giving high priority to continuing to finance on-farm water management was that it provided an opportunity to consolidate the gains achieved under earlier projects; to make changes in the program so that water savings could be efficiently translated into increased agricultural production; to improve maintenance of renovated watercourses, and to emphasize a broader range of OFWM practices. It recognized that other donors were also funding OFWM projects, but believed that these were unlikely to correct the identified weaknesses. "Thus a second IDA-financed project is required to help the provinces establish strong sustainable OFWM programs." 26. The project objectives were essentially the same as those of the first OFWM project. The main differences were increased emphasis on strengthening the water management advisory capability of the extension services and greater farmer participation in WUAs. The main investment was a renovation program for 2,400 watercourses in all four provinces over three years (FY86-88)." The accelerated technology method of watercourse improvement was dropped.12 It was replaced by a pilot component, "renovation without pukka lining"--on 110 watercourses, which differed from the accelerated technolo only in that it provided for full design and engineering. It was to be rigorously monitored. The project also included PLL on 16,000 ha, 70 water storage tanks and demonstration plots on each renovated watercourse. 11. In total the Bank has supported, through OFWM-I and II and CWMP, about 40 percent of the renovated watercourses-9,856 out of 24,830 (Annex F, Appendix 1). 12. The PCR explains that the accelerated program was not well accepted by both farmers and OFWM staff, mainly because of the inadequate engineering associated with the installation of the pukka nuccas. During the preparation of OFWM-I Bank staff had agreed with USAID that pukka nuccas should be installed only after an engineering survey had established the optimum vertical location. Despite this they were included in OFWM-I without requiring an engineering survey. Improper location ofpukka nuccas led to farmer dissatisfaction. In addition the farmer contribution to either technology was about the same, while the benefits of the full technology were much higher. So farmers preferred to wait until they got the full package. 13. There is no indication from the files that this monitoring was ever carried out. Annex A 96 27. Changes in the institutional arrangements were intended to address the weaknesses revealed during implementation of the first project: (a) unsatisfactory maintenance of renovated watercourses; (b) inefficient translation of water savings into increased agricultural production; and (c) inadequate application of OFWM practices at the farm level. This was to be achieved by: (a) assigning more responsibility to the Water Management Officer (Agriculture) who was part of the OFWM directorates; (b) developing the capability of the extension services in water management at the farm level to take over from OFWM directorates after the completion of watercourse improvement; and (c) strengthening the coordination between OFWM directorates and extension services. 28. Project benefits were based on the increased agricultural production that would result from a 20 percent increase in water supply to the watercourse brought about by canal lining, combined with other improvements (PLL, demonstration plots, better water management). This would increase crop yields (assumed increases were mostly in the range of five to ten percent), cropping intensities (increases of six to 12 percent), and farm incomes. Farmers would also benefit from: (a) reduced time and labor for irrigation; (b) easier watercourse maintenance; (c) more water reaching tail-enders; and (d) fewer disputes over water distribution. The ERR was calculated to be 22 percent. 29. Physical execution of the project expanded greatly such that the command area benefiting more than doubled, and the number of farm families benefited almost doubled. But the PCR could not say whether the project met its objective of increasing agricultural production by improved on-farm water management practices with saved water. This was because monitoring and evaluation results were of limited use, although the data did show that water savings (i.e. reduction in losses) were much less than projected in the SAR (five percent for unlined and 10 percent for lined watercourses, against 20 percent for lined watercourses in the SAR). Despite the lack of data, however, the PCR re-estimated the project's economic rate of return at 22 percent, the same as at appraisal. Project farmers reported to the PCR mission that the project works helped to reduce stealing of water and that less time and labor was needed for irrigation because of the installation of turnouts. WUAs were reported as functioning well during implementation, but failing to assure post-implementation responsibilities, often including maintenance of the works. Even so, maintenance was carried out. Lessons learnt included that more rigid criteria should be used to select watercourses for improvement in the interests of increasing productivity from water savings and that WUAs should be helped to function as intended. Annex B Investment Strategies for the Indus Basin Irrigation System Introduction 1. Excavations over the past 70 years have shown that around the middle of the third millennium BC, contemporaneous with the start of pyramid building in Egypt, a prosperous urban civilization arose at several sites in the Indus River Basin, lasted about a thousand years and then vanished. Over the next two millennia, waves of invaders from the north swept through the area. Few clues remain as to the economies of these old civilizations but in what was largely a region of scant and undependable monsoon rains, and extremes of temperature, these urban concentrations depended heavily on irrigation. Over the intervening centuries, empires have come and gone, but an industrious fanning population has continued to use the river waters, particularly when the melting snows of the Himalayas brought bountiful floods in summer (kharif), and where groundwater was near the surface and easily lifted for irrigation and domestic use. 2. From the middle of the nineteenth century, the waters of the Indus tributaries were progressively brought under control by the construction of barrages across the various rivers, to divert high summer flood flows into canals to produce kharifcrops. Low river flows in winter meant that many areas received no water, or very limited water, to irrigate the rabi crops-in the so-called non-perennial areas. 3. This canalization was a remarkable technical achievement, and the technologies of diverting flows from large alluvial rivers and of canal design in erodible soils were born and developed to high sophistication in the Indus Basin. Ironically, little attention was paid by engineers and administrators to the agricultural use of the water delivered (and this tradition has continued). Water was simply released through an outlet (mogha) onto areas of land of some 100-200 ha, and the farmers were left to fend for themselves in conveying the water from the mogha to their individual fields. The canals were designed to provide the minimum amount of water needed for a basic living for the farmers, many of whom were new settlers, so as to relieve famine risks, to maximize the number of people that could be settled, as well as to maximize the tax base and ensure the financial viability of the schemes. 4. In recent decades reservoirs have been added to store surplus summer flows for use in the rabi period of low river flows (about 85 percent of total river flows occur in khan, only 15 percent in rabi). Gradually the agricultural economy of the Indus Basin has become almost totally dependent on the river and its abundant waters. Today irrigated land supplies more than 90 percent of agricultural production, and agriculture supplies most of the country's food, accounts for 26 percent of GDP and employs 54 percent of the labor force. It is also the source of raw materials for major domestic industries, particularly cotton products which account for 80 percent of the value of exports. Irrigation is thus the lifeblood of the economy. 5. Despite the extensive irrigation system, with its vast network of reservoirs, barrages, canals and watercourses, there is still not enough water to supply the available land with adequate flows for intensive modern agriculture. Over the years the layout has been continually expanded and modified, but the original design concepts which governed the operation of the system have remained. In perennial areas, all canals and moghas are designed to run at full Annex B 98 discharge all the time (except for a brief closure period for maintenance), with the full watercourse flow being rotated in timed turns once a week to each farm (with each turn proportional in time to individual farm sizes). However, by the 1970s, age, overuse and poor maintenance, had combined to make the canal system increasingly inefficient and increasingly prone to breaches. The main features of the Indus Basin Irrigation System (IBIS) are set out in Appendix 1. To understand how these projects fit into the broader picture, however, reference to the post-Partition history of the investment strategy for the irrigation system is helpful, starting with the immediate effect of Partition and the Bank's role in resolving them (Box B. 1). Box B.1: The Indus Basin Irrigation System-Effects of Partition in 1947 The Partition of the Indian subcontinent in 1947 resulted in the border between India and Pakistan being drawn across the north-eastern corner of the Indus Basin. Pakistan became the downstream riparian on all the major tributaries and India was in a position to divert these waters for its own uses. It was not until 1960 that final agreement on how to divide up the waters was reached with the signing of the Indus Waters Treaty. It was agreed that India would have the sole use of the waters of three eastern rivers (the Ravi, the Beas and the Sutlej), and would make a contribution to the cost of the replacement works, while Pakistan would have the use of the western rivers (the Indus, the Jhelum and the Chenab). The system of works envisaged to implement the treaty originally consisted of two large dams, one on the Jhelum River (Mangla) and another on the Indus River (Tarbela), about 400 miles of new link canals, (to transfer Pakistan's western flows to its downstream portions of the eastern tributaries) and three major and two minor barrages, the remodeling of three existing link canals, and a program of groundwater development and drainage by tubewells and surface systems. The World Bank contributed to, and also served as Administrator of, the Indus Basin Development Fund that was set up simultaneously with the Indus Waters Treaty to finance the works. The original plan had to be modified later since funds were insufficient to construct everything proposed. Drainage works were dropped from the master plan and the construction of the huge Tarbela dam was postponed pending the findings of a study in the mid-i 960s of the water and power sectors in the then West Pakistan. 6. Once the post-Partition problems were in hand the Bank turned to assisting the government to plan the development of the IBIS further, for both irrigation and power generation. The irrigation part of the resulting study, the Indus Special Study (ISS) of 1966 was principally focused on alternative methods of increasing irrigation water supplies.14 It examined four types of main investments: interseasonal surface storage; canal remodeling and enlargement; public tubewell pumping; and private tubewell pumping. Somewhat surprisingly, especially compared with later concerns, reducing the substantial irrigation water "losses" from both canals and watercourses was not one of the options considered in any detail.'5 The option 14. The Indus Special Study was later published by the Bank in abbreviated form in its Water and Power Resources of West Pakistan: A Study in Sector Planning, Johns Hopkins Press, 1968. 15. "Lost" and "losses" are words frequently used to describe the difference between the volume of water entering and leaving a channel. But only in saline (unusable) groundwater areas is this water "lost" in the system, and where it becomes a liability by contributing to waterlogging. In fresh groundwater areas it recharges the aquifer and is a valuable form of over-season storage, since it can be pumped later when surface supplies are scarce. A further confusion arises by various Bank and other reports referring to percentage losses and savings (through lining) inconsistently as to whether figures are percentage point changes or percentage changes in losses [see Annex C (a)]. 99 Annex B of lining canals to reduce seepage, and thus lower pumping costs, was touched on but was dismissed for the immediate future.16 The ISS pointed out that at that time almost none of the canals in the country were lined because one of the main advantages of canal lining, that it permits higher water deliveries by allowing increased velocity, was generally slight owing to the extreme flatness of the land. It would also be a more expensive way of saving water than any other, although there would be an offset in saline areas by reduced drainage pumping (conversely, in fresh groundwater areas pumping costs would rise with the fall of watertables as seepage is reduced). There is the additional problem that lining could be carried out only after the groundwater had been lowered, because otherwise the lining would not stick, and it would be extremely difficult to line existing canals because of the complications of interrupting canal supplies. Other main proposals of ISS are outlined in Box B.2. 7. The ISS strategy is presented at greater length in Box B.3 because the issues addressed by ISS find echoes 30 years later in both the findings of this impact evaluation and in matters arising. 8. Disillusionment with the public tubewells promoted by ISS grew during the early 1970s, particularly their effectiveness in controlling waterlogging and salinity in saline groundwater areas. Also, research from 1968 by staff from Colorado State University (under USAID financing) measured water tosses at different points in the distribution system. It was found that much less water was arriving in the field to be productively used than had been assumed previously. Particularly alarming was the finding that public tubewells actually increased losses in the watercourses by increasing the hydraulic gradient between watercourse and watertable. This largely offset the drainage effect of the tubewells. (Subsequent research has tended to indicate that this phenomenon became less important once the watertable had been lowered.) The overall strategy then began to swing towards improving the efficiency of the irrigation system (i.e. getting more of the diverted water to crops) by canal and watercourse improvements. The waterlogging problem could be reduced by improvement of watercourses through lining and pukka nuccas (concrete checks and turnouts), and by precision land leveling (PLL) to make better use of water in fields. Improving farmer crop and water management practices and regular cleaning and maintenance of watercourses were also identified as critical components of any effort to increase irrigation system efficiency. 9. Furthermore, by the time Tarbela came on stream in 1975, the magnitude of the management difficulties of large-scale public tubewell schemes were becoming apparent and the wisdom of proceeding with the integrated plan of the ISS was being called into question. The O&M problems of public tubewells were proving to be much greater and much more expensive than anticipated. The impact on the irrigation budget was starting to loom ever larger. Wells were going out of commission faster than anticipated. In addition the growth of private tubewells was going much faster than the ISS had expected, thus undercutting its justification for public tubewells in FGW areas. How to control waterlogging in SGW areas was still an open question. The first pilot horizontal drainage project in the country was just then being appraised by the Bank (Khairpur II, Cr. 648-PAK) and public tubewells solely for drainage in saline areas were also being tried. The possibility of reducing recharge of the watertable resulting from 16. op. cit. volume 1, page 51. 17. For further information on these tubewelI and drainage aspects see (a) PAR: Khairpur Tile Drainage and Irrigated Farming Development Project (Credit 648-PAK), Salinity Control and Reclamation Project (SCARP) VI (Credit 754-PAK), and the SCARP Transition Pilot Project (Credit 1693-PAK), June, 1995 (Report No. 14769), (b) Annex B 100 channel seepage was scarcely yet considered an alternative. Appendix 2 discusses the waterlogging and salinity problem further. 10. This developing situation led to the recognition in the mid-1970s that a more direct approach to the problems of management, maintenance and efficiency in the operation of the irrigation system was required, and further, that such an approach would need to be more closely attuned to the immediate constraints on agricultural production than in the past. To help evolve the policies and program necessary to implement such a strategy, a UNDP-financed and World Bank-executed study was mounted to prepare the Revised Action Programme for Irrigated Agriculture (RAP). This was essentially an updating of the ISS and led to the series of projects included in this impact evaluation. The major recommendations of RAP were generally adopted at a meeting of the National Board for Irrigated Planning and Development in July 1980 (Box B.4), and the Bank supported RAP implementation with substantial funding as shown in Box 1.1. PAR: Salinity Control and Reclamation Project (SCARP) Mardan (Credit 877-PAK), OED Report No.15224, December 1995; and IER: SCARP Transition Pilot Project (Credit 693-PAK), OED, forthcoming. 18. Revised Action Programme for Irrigated Agriculture, Master Planning and Review Division, Water and Power Development Authority, April, 1979. 101 Annex B Box B.2: Indus Special Study-Proposals for Surface Storage and Public Tubewells When the surface irrigation system was installed, the depth to groundwater was considerable and drainage was not considered an immediate problem. However, the original designers realized drainage would eventually be necessary as continuous seepage from unlined earthen channels and deep percolation of irrigation water would in time cause the watertable to rise. The main surface storage proposed, and later constructed with minority financial assistance from the Bank, was for a new reservoir on the Indus River at Tarbela with a live storage capacity of 6.4 MAF.' It was recognized that this option was more expensive in the short-run (although cheaper in the long-run) than pumping from the groundwater aquifer, but it could provide benefits more rapidly, and was more flexible and more consistent with the integrated pattern of water resource development that was planned.b The canal remodeling and enlargement option was rejected on two grounds: (i) the present withdrawal capacity of the canal system already equaled or exceeded the combined natural mean flows of the river system in all months except June, July and August; additional water just for three months was not long enough to mature a crop; (ii) the watertable had already risen to high levels in extensive areas, especially in some of the older irrigated areas; it was doubtful if these areas could cope with additional surface supplies without major investments in drainage and the drainage option (at least by means of surface drains) had been dropped for lack of funds, even though in some areas, particularly those underlain by saline groundwater, such drainage was needed before even Tarbela water could be prudently absorbed. The ISS came down strongly in favor of using public tubewells, in areas underlain by fresh groundwater, as the main means of increasing irrigation supplies, particularly as they would simultaneously help to control the watertable. They provided the flexibility of making water available in those months when river flows were already fully committed, with the added benefit of raising the value of additional Tarbela water. In the final plan nearly 50 percent of public sector expenditures on irrigation and drainage were to be on public tubewells. The ISS was cautious on the role of private tubewell pumping. Private tubewells appeared to offer a rapid, but uneven and unpredictable, means of overcoming the constraints set by limited canal capacities and limited river flows. But because of their haphazard location, and their operation according the needs of individual owners, they were of limited use as a dependable way of lowering the watertable. They did not fit neatly into the carefully integrated plan that was developed and, apart from allowing the number of public tubewells to be reduced, they were not given much attention. Even before the ISS report was issued, this strategy of using public tubewells in fresh groundwater areas was being contested in some quarters in Pakistan, arguing that public tubewells should be concentrated in saline groundwater areas while encouraging the development of private tubewells in fresh groundwater areas through the provision of credit and the extension of the electric grid.c a. Subsequently audited by OED; see PAR, Report No. 6398, October 1986 b. op. cit., page 226. c. Mohammad, Ghulam (1965), Private Tubewell Development and Cropping Patterns in West Pakistan, The Pakistan Development Review, Vol. V, No. I (Spring): pp. 1-53.. Annex B 102 Box B.3: The Indus Special Study-Issues and Solutions Canals in perennial areas have a smaller delivery capacity than canals in nonperennial areas.' The nonperennial/perennial distinction came about as a rationing device in the 1930s, and the distinction was maintained even though river flows were seldom adequate to maintain deliveries as originally allocated. The ISS found that this distinction made little difference to actual rabi cropping intensities between the perennial and nonperennial areas in the Punjab, or equivalent areas in Sindh, and in Punjab there was no noticeable difference in yields of rabi crops. Superficially therefore, canal enlargement to provide sufficient delivery capacity to the best agricultural areas, to enable cropping intensities to be raised nearer to their potential of around 150 percent (against the original design of around 80 percent, or the 100 percent being attained in practice), appeared an attractive option. As a solution, however, it had two drawbacks. The present withdrawal capacity of the canal system already equaled or exceeded the combined natural mean flows of the river system in all months except June, July and August. Using this surplus to expand the cropped area would not provide water for long enough to mature a crop. The second drawback was that the watertable had already risen to high levels in extensive areas, especially in some of the older irrigated areas. The report concluded that it was doubtful if these areas could cope with additional surface supplies without major investments in drainage. However, extensive areas already existed with canal capacities sufficient to convey large increases in irrigation supply fed from tubewells. These areas tended to be nonperennial since almost all the recent nonperennial development was in areas where the groundwater was of irrigation quality, while perennial surface supplies had been allocated to areas with poor quality groundwater. The proposed tubewells would provide the drainage, and where the groundwater was usable for irrigation, they would also provide irrigation water in the months when the river flows were already fully committed. This would make additional canal deliveries in other months far more useful than they would otherwise be. This was particularly important in the Sutlej Valley which constituted a large proportion of the nonperennial area in the basin. Furthermore, as most of the groundwater came from seepage, nonperennial areas had higher groundwater recharge in the kharifseason, when canals are normally full, than in the rabi. By concentrating pumping in those months of the kharifseason when canal capacities were the constraint, and in the rabi season when surface supplies were scarce, a significant increase in cropping intensity would be possible. Canal remodeling was the only way to gain the same benefits in areas not underlain by fresh groundwater; and these were areas that already tended to have less canal capacity per hectare of canal command. Public tubewells (or some form of horizontal drainage) would be needed to control the groundwater level when additional surface supplies were made available. Such areas would be relatively small in Punjab, but extremely important in Sindh. This would be a complex matter, involving the interdependent enlargement of canal capacity and increased supply of surface water in rabi season, as well as provision of drainage. a. Perennial irrigation projects were generally designed for a rabi cropped acreage between 1.5 to twice the kharif cropped acreage; with an annual average of some 80 percent, the kharf intensity would be between about 27 percent and 32 percent. The non- perennial areas, by way of compensation for their right to canal supplies being limited to one season of the year, were generally given canal capacities sufficient to support a somewhat higher kharif intensity, in the order of 30 to 40 percent. 103 Annex B Box B.4: The Revised Action Program for Irrigated Agriculture (RAP) The RAP strategy was to maximize output growth during 1980-1990 so as to build up the domestic capability needed to finance development needs beyond 1990. This involved concentrating on small-scale physical investments (private tubewells, watercourse improvements, land reclamation) in the first half of the decade and deferring to the second half major physical investments in long gestation projects (drainage and surface storage). It believed that the earlier strategy, of concentrating on major infrastructural works and the development of usable groundwater, had resulted in unbalanced development with the neglect of drainage for marginal and saline areas, of watercourse improvement possibilities and of support to private tubewells (and also to the neglect of agricultural extension and research, and cost recovery policies). The past emphasis on capital investments in new infrastructure had also resulted in largely ignoring the physical interdependence between increased irrigation supply and drainage. The need for adequately funded programs for operation and maintenance of existing and new facilities and the higher utilization of existing investments, had received little attention. Previous investment plans had ignored conservation of present water supplies through conveyance system improvements, and management and extension improvements that could produce more output per unit of water, as well as the availability of private capital for usable groundwater development. RAP recommended that the present SCARP tubewells be gradually phased out of the public sector and replaced by smaller privately owned and operated tubewells (financed through a major credit program). It proposed renovating the 29,000 watercourses in areas overlying unusable saline groundwater, as the priority target group. The main investment was lining with bricks and mortar for the first 50 to 70 percent of the watercourse's length. This would cut seepage losses by 85 percent. The relatively high cost of this meant that it would have to be fully borne by the beneficiaries. The potential returns from lining were estimated to be high enough to make this feasible (incremental returns per acre were expected to be about Rs 80 compared to the annual cost of Rs 45, although it is unclear how the capital cost was to be met). Given the high returns obtainable from watercourse lining, RAP recommended the fastest possible rate of implementation, with 24,000 out of the 29,000 watercourses being lined by 1990. For watercourses in areas where the groundwater was usable there was scope for water saving from a watercourse improvement program involving heavy cleaning, shaping, compacting and the provision of pukka nuccas , with partial lining in some cases along the lines of the USAID Watercourse Demonstration Program that was then being implemented. RAP proposed that about 48,000 out of the estimated 60,000 watercourses in usable groundwater areas be improved by 1990. Another component of RAP was an Irrigation and Drainage System Rehabilitation Program. The attribution of system deterioration to inadequate expenditures on O&M, caused by inadequate water charges combined with past emphasis on using available funds on new projects, led to the recommendation to increase O&M expenditures (and water charges to finance them) from Rs 10 per acre to Rs 25 per acre within three years. A phased basin-wide program was recommended, since nearly the whole system was equally badly deteriorated, with irrigation facilities serving 30 million acres (12 million hectares) being rehabilitated by 1990. The RAP stressed that it was vital for the implementation of these proposals that the Provincial Irrigation and Agricultural Departments be substantially strengthened and their functional objectives be reoriented to satisfying the needs of modern irrigated agriculture. The extension service would be expanded and the creation of a Water Management and Reclamation Extension service was also recommended. This was to be combined with redistributing control of land and water resources by giving as much control as possible to the farmer, particularly over water supply. The target group was farmers with less than 12.5 acres (5 ha). All these water sector infrastructure projects were to be conceived and managed as area-based development projects; with the country being divided into Command Area Management Zones, not exceeding 500,000 acres (200,000 ha) in size. Over time the functional responsibilities of the present Irrigation and Agriculture Departments were to be reduced. 105 Annex B Appendix 1 The Indus Basin Irrigation System (IBIS): Vast Size, Elegant Design, But Incomplete The Extent of the System The Indus Basin covers an area of 566,000 square kilometers, 70 percent of Pakistan. Its irrigation system (IBIS) commands over 14 million hectares and encompasses the Indus River and its major tributaries, three major reservoirs (Tarbela, Mangla and Chasma) with a combined storage capacity of about 18 billion cubic meters (14.5 million acre feet, equivalent to about 16 percent of total diversions), 19 barrages or headworks, 12 link canals, and 43 canal commands with about 107,000 watercourses (village level channels). The length of the canals is about 61,000 km. Communal watercourses, farm channels and field ditches stretch another 1.6 million km (twice the distance from the earth to the moon and back). In addition there are 14,000 km of drainage channels. The IBIS is the largest integrated irrigation system in the world, and its intricately connected system of reservoirs, barrages, link canals and diversion canals provides almost complete control of diverted water. Only two other countries have larger irrigated areas, (China, 47 m ha; India, 46 m ha). The irrigated area of the USA is of a similar size to IBIS. It is twice as large as the irrigated area of Indonesia, and larger than the whole of the combined irrigated areas of Bangladesh, Philippines, Thailand and Mexico put together. But in all these countries the irrigated areas are in scattered systems, the largest of which is rarely more than one million hectares; the next largest integrated system is in Egypt, but with its 2.6 million hectares of irrigation it is dwarfed by the 14 million ha of IBIS. Water Distribution Arrangements The annual diversion of water from the rivers into the canal system is 106 MAF [131 billion cubic meters (Bcm)]. In addition about 12,500 public tubewells and 225,000 private tubewells pump around 33 MAF (41 Bcm) from the groundwater. River water is diverted by barrages and weirs into main canals and subsequently into branch canals, distributaries and minors. From these distributaries and minors, water then flows through concrete or brick outlets (moghas) into watercourses, which carry the water down to the individual plots. The average watercourse is about five km long and was "constructed" by farmers (being originally just a small earthen ditch) clearly acting in cooperation many decades ago. Hence, to the extent that these farmers' watercourses were not engineered (and may silt up unduly, or run through areas of high permeability without any lining), the system is incomplete in terms of its engineering design and construction. Figure 1 is a schematic diagram of the layout of IBIS channels. On a distributary or minor canal all the moghas were scaled and positioned (vertically) to divide the total water supply fairly between all the watercourses being served, with the discharge of the moghas varying uniformly in proportion to delivery canal flow levels. Within the watercourse command, which varies in size from some 100 ha to 300 ha, farmers are supposed to receive water proportional to their land holding, by varying the length of time that a farmer receives the fully watercourse supply. Thus the entire discharge of the watercourse is given to one farm for a fixed period at a fixed time on a seven day rotation. The rotation schedule, called warabandi-literally fixed (bandi) turns (wara}--is established by the Annex B 106 Appendix 1 Irrigation Department. This arrangement is said to date from the earliest days of controlled irrigation schemes and was initiated by farmers. The great and elegant advantage of this "full flow system" was that in the early days when communications were slow there was minimal need for adjustments of control structures by government staff. Of course, on watercourses, outlets to farms must be continually opened and closed by farmers. If the system is working properly (without malfunction or illicit interference) each farmer knows exactly when he will receive water, and, given that his "full flow turn" is timed, that he is receiving his fair share. More recently the corruption of the management of the system, combined with inadequate maintenance, has resulted in watercourse full flows varying greatly from the designed amounts. Typically, watercourses on the upper end of irrigation channels receive more than their share, and the lower down receive less. Figure 3 from the International Irrigation Management Institute (IIMI, Lahore office) illustrates the point for one minor canal. A similar pattern of decreasing flow is typical down traditional unlined watercourses, such that farms in the lower reaches receive much less than their fair share. The effect, as both between watercourses and-within watercourses, of this uneven distribution is two- fold: reduced flow both reduces the crops that can be grown and increases water distribution problems within farms. Thus it is more difficult to irrigate evenly over the entire farm when the flow is reduced, and under and over-irrigation in different areas tends to result. Over-irrigation contributes to field seepage losses. Drainage Flat topography and lack of well-defined natural drainage in the Indus Plain, create a surface drainage problem, which has been compounded by the construction of railways, roads, flood embankments and canals which have obstructed such natural drainage as there was. Massive seepage from the canal system has also created a major sub-surface drainage problem. Thus an irrigation system with these physical characteristics may be considered incomplete, and thus unsustainable, over a large part of its area, so long as a full drainage and disposal system is not in place. However, delaying drainage installation was the right economic decision so long as rising watertables and increased salinization were not depressing production. That time has long passed. Crops and Crop Areas The total gross commanded area in the IBIS is about 16 million ha, the culturable command area is almost 14 million ha. Potentially this land is capable of producing two crops per year, one in the monsoon or kharif season and one in the winter or rabi season. In practice, however, available water supplies are insufficient for such double cropping. The proportion of the culturable command area which cultivated in a season is termed the cropping intensity; the kharif cropping intensity for the IBIS is presently 116 percent, varying from 151 percent in Punjab, to 61 percent-ifSindh; the rabi cropping intensity is 69 percent overall, with 92 percent in Punjab, and 32 percent in Sindh. The main kharif crops are cotton and rice (about 2-2.5 m ha 107 Annex B Appendix 1 each); the other important khari crops are maize, millets and sugarcane. The main rabi crop is wheat (about 6 m ha irrigated). 19. Source: Agricultural Statistics of Pakistan, 1987-88. 109 Annex B Appendix 2 Waterlogging and Salinity When the surface irrigation system was installed, the depth to groundwater was considerable and drainage was not considered an immediate problem. However, the original designers realized drainage would eventually be necessary as continuous seepage from unlined earthen channels and deep percolation of irrigation water would in time cause the watertable to rise. Although it may be argued that Pakistan "mortgaged" its agricultural future by this neglect, there is little doubt that the strategy was correct since it enabled irrigation to proceed for many decades without having to incur the heavy investment of drainage. By the 1950s, however, the rise in watertables began to have a measurable impact, and by the 1960s it was recognized as a major national problem. Although there are no accurate figures available, approximately 30 percent of the gross canal commanded area now has a watertable within 150 cm of the surface, high enough to have adverse affects on crop yields, reducing them by an estimated 25 percent. The degree to which crop yield are affected varies with the type of crop, the salinity of the groundwater and the amount of evaporation from the soil surface. At least 8 percent of the Basin soils are "severely" affected and another 6 percent "moderately" affected. Most of these areas are in Sindh. The situation is deteriorating rapidly; about 24 million tons of salt are being added to the soil or the groundwater annually from irrigation water. The following table shows the estimated groundwater balance for the Indus Basin as of 1988. Seepage from channels (i.e. from canals, branches, watercourses) and fields accounts, on average, for 83 percent of all inflows to the groundwater. A further 15 percent comes from recirculation of water pumped from tubewells. On average 45 percent of all surface water diverted into the canal system ends up in the groundwater. The situation is radically different in areas where the groundwater is fresh enough to be reused for irrigation (less than 1,500 ppm), compared with areas where it is saline. In the fresh areas, seepage from canal inflows accounts for only 70 percent of inflows to the groundwater and 24 percent comes from recirculation from tubewell pumpage. In the saline areas, seepage from canal inflows accounts for 96 percent of inflows to the groundwater, and there is no recirculation from tubewells. The result, taking into account non-beneficial evaporation, is that the groundwater level is declining in fresh areas and rising in saline areas. This rise is rapid in saline areas and drainage is the only means of controlling it. The situation is again different in fresh and saline groundwater areas (of which there are 13.5 m ha and 5.4 m ha, respectively). In the fresh areas, with their relatively deep watertables, tubewell pumping has the effect of bringing up salts from the groundwater and thereby increasing surface soil salinity. Seepage from canals and watercourses has the beneficial effect of flushing these salts down again, offsetting the effects of tubewell withdrawals. The salinity of the groundwater is inevitably rising over time but, if its initial salinity is low, this situation can be sustained over a long period. Anything that reduces seepage (e.g. watercourse lining) has the effect of upsetting this equilibrium. In saline areas it is the high evaporation from groundwater whicb brings the salts to the surface through capillary action, until in the end the watertable is close enough to the surface for all recharge to be balanced by evaporation. Where the watertable is still relatively deep any measures that slow its inevitable rise will help to check the rate of Annex B 110 Appendix 2 salinization. In the medium term only drainage can bring inflows and outflows into balance, but measures such as watercourse lining bring short term benefits and help to sustain production levels. Where the groundwater is already high (i.e. about 100 cm) even palliative measures like watercourse lining are of little benefit without drainage. Table B.1: Groundwater Balance (MAF) Fresha Saline Total Inflow (seepage) Rainfallb 1.5 0.8 2.4 From tubewell water 9.8 0 9.8 From canals 13.4 8.4 21.8 From watercourses & fields 16.8 14.9 31.7 From rivers -.0 -_Q - 11 Total inflow 40.5 23.9 64.4 Outflow by tubewells 38.9 0 38.9 Balance 1.6 23.9 25.5 Evaporation from groundwater .A5 _LL2 152 Balance after evaporation - 2.9 12.7 9.8 a. "Fresh" is groundwater with less than 1,500 parts per million of dissolved salts. b. One reviewer suggests a higher figure. Source: Adapted from World Bank, The Agricultural Impact ofthe Kalabagh Dam, Report No. 6884-PAK, August, 1986 (Revised June, 1987), Appendix H, page 82. Figure 2, which reflects more recent data, shows a lower outflow from tubewells Ill Annex C Watercourse Improvement, Evolving Policies and Practices 1. The watercourse improvement program has changed in concept, direction aiid emphasis over the years in not always very logical or consistent ways. This is especially so with respect to which watercourses should be renovated, what level and type of investment should be promoted (particularly the length of lining), and how it should be financed (cost recovery, subsidies and self-help). The following account is complex, but provides essential background to an understanding of current concerns as they affect the impact of the projects. 2. When the USAID-assisted OFWM program started in 1976 with an experimental pilot project, it was regarded as a significant achievement that farmers were donating free labor for watercourse renovation (representing about 40 percent of total cost), even though the project financed all the material costs (about Rs 70,000 per watercourse). The precision land leveling component was 50 percent subsidized. Material cost included about ten percent of the main watercourse channel to be reconstructed of bricks and mortar and 40 pukka nuccas per watercourse.20 3. The Punjab "crash" program that followed in 1980 adopted the same procedures, except there were no bricks and mortar (for "lining"), just heavy cleaning and maintenance. Cooperative societies were organized on all watercourses to be cleaned. Farmers donated their labor and in theory the societies borrowed the money from the Cooperative Bank for installing the pukka nuccas. The provincial government provided the precast pukka nuccas to the societies free of charge and underwrote the interest charges for the loans to the societies.21 4. The Revised Action Program report of 1979 concluded that, on the basis of a mathematical model, the most economical proportion to line was between 50 and 70 percent. It concluded, furthermore, that the income potential of the lining technology was sufficiently high for farmers to bear the full cost of lining. "Subsidization of capital or operating costs by the public sector is not only unnecessary, but also wasteful ofpublic resources" (para. 10.108). The report went on to add that "the substitution of credit for subsidies and the introduction of the principle of full cost recovery have profound effects on private savings and the public sector budget." It recommended channel lining only in saline groundwater (SGW) areas, concluding that full cost recovery would allow about 80 percent of the watercourses in SGW areas to be renovated by 1990. It recommended that the renovation be done by the public sector with 20. A memo to files (09/80), reporting a meeting between Bank staff and the consultants preparing the feasibility report for OFWM 1, says that, regarding USAID experience, there is no clear guidance on how much to line: the program had initially lined six percent, then ten percent and finally 20 percent. The matter is considered in more detail in the Feasibility Report for OFWM III, prepared in 1990, quoting several references to research, which conclude that the optimal length to line is about 55 percent if there are no branch watercourses, reducing to 40 percent if the number of branches is more than one. Incredibly, the present criteria are not based on any engineering or economic analysis, or evidence, theoretical or empirical Reports say that 30 percent lining in SGW areas is justified, over 15 percent in FGW areas, because of the greater water savings, but no evidence is presented to indicate what the difference in water savings is, if any. Surprisingly, most Bank reports assume that water savings are the same, regardless of the length or percentage of the watercourse that has been lined. In practice the amount lined in OFWM-1 appears to be about 18 percent in Punjab, and 14-15 percent in Sindh and NWFP. In Sindh this is well below the SAR criteria. 21. The same memo reports that 400,000 nucca lids were ordered but only 1,800 had been installed at that date, with the conclusion that there was a major problem with installing them. There was also a major quality control problem with the lids that had been produced. Annex C 112 repayment over 15 years in equal annual installments with an interest of 12 percent and a one year grace period. The heavy cleaning technology, with pukka nuccas, would be suitable for use in fresh groundwater (FGW) areas. 5. This policy was adopted by the Federal Government. At a meeting on May 15, 1980, chaired by the Minister for Food, Agriculture and Cooperatives, it was agreed that the subsidy for the watercourse renovation program should be removed over a four year period, at the rate of 20 percent for the first two years and 30 percent for the next two years. Under the "crash" program nuccas would be provided free, farmers would pay for everything else.22 The subsidy on PLL would be withdrawn as well and be replaced by loans, interest free to farmers with under 12.5 acres. 6. The Bank began to have reservations about this new policy almost at once. An Aide Memoir (06/80), which the OFWM-I preparation mission left with the Bank's Resrep in Islamabad, (06/80) reports GOP's desire to shift the burden of financing construction materials to farmers and the need this will create for legislation to set up WUAs with authority to borrow funds and collect revenue. It continues: "this shift offinancing for watercourse materials may be too abrupt. Conceivably this transition may be too fast and result in severe reduction in project accomplishment for this new program that is just now beginning to receive farmer acceptance " (emphasis added). In the "back-to-office" memo three weeks later, this thought was not mentioned, merely that there was a possible policy issue of "interest free credit" and whether it would be applicable to this project, citing the Punjab interest free loans for nuccas under the "crash" program. A hand-written note from a Bank director asked, "What is the proposed approach to the handling of interest free lending?" No written reply was found on file.23 7. It is curious that, more or less simultaneously with these discussions, the Bank was deciding on the lining percentages to be applied under the project. In SGW areas it would be 30 percent. Lining the whole sarkari khal would be "prohibitively expensive at the present," and even lining ten percent would cost several times as much as there are funds available (but if the farmers are to provide funds, where is the limitation?) Shortage of trained manpower is given as another reason why it would not be possible to complete the program in a reasonable time (ten years is suggested). For the first time, despite the shortage of funds, lining in FGW areas (ten percent) is mentioned (presumably this was following the USAID precedent). Apparently the Bank had no difficulty reversing the RAP conclusions (para. 4 above) in this respect. 8. At negotiations GOP ask for the FGW lining percentage to go up to 15 percent, and the Bank agreed. This increased the total project cost by US$5 million. All talk of eliminating the 22. A memo (07/80) from the Project Preparation Mission for OFWM I says that the opinion of USAID engineers, "in which we concur, is that the 'crash' program should not involve the installation of pukka nuccas, or water control structures. The installation of such permanent structures should always follow a complete engineering survey to establish the optimum vertical location." Despite this, in the Project Brief dated less than two months earlier (06/80) "structures at the principal junctions" were included. They were later included in OFWM-I, only to be dropped in OFWM-II, in favor of a pilot program which included a complete engineering survey. 23. These doubts must have been reinforced in the discussions with the consultants mentioned in footnote I to the previous page, where the consultants said they doubted if the government's intentions to withdraw subsidies would be (fully) implemented. The Bank representatives said GOP planned to provide interest-free loans to farmers to pay for watercourse improvements. 113 Annex C subsidies disappeared. Cost recovery was to be 20 percent of material cost in Punjab, with farmers paying the cost of masons, and 25 percent in Sindh, with GOS paying the masons. No cost recovery was sought in Balochistan or NWFP. 9. It is clear that the subsidy percentage had an impact on the amount of lining, at least in Sindh. Without informing the Bank, GOS put a limit (in its PC I), of ten percent lining of watercourses, even in SGW areas, and a maximum cost per watercourse of Rs 80,000, compared with the agreed figure of 30 percent lining and an estimated cost (without farmer labor) of Rs 115,000 in SGW areas. There is no sign in the files that this was ever changed. It could explain why the target number was exceeded. From the PCR for OFWMP-I,24 the lining per watercourse was 640 meters on average, while in Sindh it was only 280 meters.25 10. When the second OFWM project was under preparation, a Bank manager questioned the concept of cost recovery as it was being implemented under the first project. In a memo in June, 1984, following a pre-appraisal Issues Meeting, he explained his view: that in the first OFWM project, as in the three other projects where the Bank had OFWM components, the Bank had, implicitly or explicitly, treated OFWM as if it were a normal irrigation project from which the government was recovering cost of O&M (through farmer labor) and 20-25 percent of the cost of materials (in cash installments from farmers over 5-7 years). He argued that this was not a correct approach since all the works were on farmers' lands. "What we really have here is a farmers' self-help scheme to which the government provides: (a) technical advice, layout and supervision (a normal departmental free service); (b) a subsidy of 75-80 percent of the cost of materials; (c) a subsidy for hired artisans; and (d) an interest-free deferred payment scheme for the balance of cash costs." He thought that there would be a number of benefits to recognizing this situation, among which would be: (i) that it provided a logical reason for the inclusion of the accelerated program (lower subsidy traded off against low water savings); and (ii) explain farmers' negative reactions to that program ("sordid self-interest"). It would also allow simple comparisofs with subsidies paid for other on-farm works, both in Pakistan and other countries. The change from OFWM-I could be explained by the move from a pilot project to financing a time-slice of a huge program. "Improving our product is far more important than precedent." 11. The point is analyzed in the Issues Paper of January, 1984: it was described as a choice between farmer self-help schemes with government subsidy, or government-sponsored irrigation projects with cost recovery. The premise for raising the point was that public financed civil works were being carried out on private land. "Public funding of OFWM civil works, however, is limited to improvement of communal watercourses. In Pakistan, communal watercourses are located on both private and public lands. In Punjab, all communal watercourses are located on public rights-of-way. Since 60 percent of the watercourses to be improved in OFWM-II would 24. PCR, Annex 2, Table 2. 25. -The Feasibility Report of OFWM-II quotes the length lined in OFWM-II as 1,124 meters in Punjab and 426 meters in Sindh, apparently much higher than in OFWM-1. What this is as a percentage is hard to determine. The same feasibility report says that the average length of a watercourse is 6,000 meters in Punjab and 3,000 meters in Sindh, but the very rounded numbers are of concern. It also says watercourse areas are larger in Sindh. In any case, these data indicate lining was 19 percent of the watercourse length in Punjab and 14 percent in Sindh. There is also evidence of very large cost variations between watercourses within the same province and between provinces. The PCR for OFWM-1l quotes (para. 6.12) the average cost for Sindh (materials only) as Rs 644 per acre (range from Rs 133 to 2,233) and for Punjab as Rs 329 per acre (range from Rs 21 to Rs 933), only half as much as in Sindh for twice the length. Annex C 114 be located in Punjab, a major part of such work would be carried out on public lands.26 Whether located on public or private lands, communal watercourses serve groups of farmers (about 30 on average in Pakistan). GOP has decreed that prior to receiving public assistance, farmers along a watercourse must organize a Water User Association (WUA) as authorized under Provincial Ordinance. WUAs are described in the Ordinance as corporate bodies... GOP and Provincial officials view watercourse improvements as a cost-sharing arrangement between the Provincial government, which supplies technical services and materials, and WUAs, which supply the required labor." However, the Bank could have found equally convincing arguments to prove that it was indeed a self-help program, but it would have been politically damaging at that point. It would also have upset the lending program, though that was not something that usually concerned the particular Assistant Director. In addition, the statement above does not deal with the 40 percent of the project watercourses outside Punjab, the majority of which are almost certainly in private ownership, and frequently in the hands of a small group of landowners. The Decision Memorandum (11/85), says the programs "embody concepts of self help" but that the meeting agreed to seek a larger percentage of cost recovery (five percent more). This successfully kept the issue from being reviewed at a higher level since the SUP's office merely noted (01/85) the encouraging improvements in cost recovery and asked that cost recovery be emphasized as an important issue. Thus the position during OFWM-II was settled in favor of the government-sponsored irrigation project approach. 12. It is clear that increasing the rate of cost recovery collection was a high priority during the implementation of OFWM-II, with particular efforts being exerted to get the provinces to collect what little was due. As at the end of 1990 the collections rate (of amounts due) was 40 percent in Punjab and three percent in Sindh. 13. Fiscal constraints remained a major problem to the rate of implementation of OFWM-II, with inadequate allocation in the provincial Annual Development Plans holding up the rate of watercourse improvement. The preparation of the third OFWM project provided the opportunity to review alternative ways of financing the program so that it could be accelerated. Unfortunately the feasibility study7 dealt very superficially with the subject. At one point it calculated that farmers could easily bear the full cost28 but does not follow through with this to the conclusion that they should bear it. It comments that it would be desirable to put collections into a revolving fund which could be used to finance further improvements, but it does not elaborate on how this would work. 14. It was agreed at negotiations for OFWM-III that the percentage of cost to be recovered would be raised in Punjab and NWFP, that part of the farmers' share would be paid as advance, and that the grace period would be eliminated and the number of installments would be cut from ten over five years to six over three years. The result of all these changes was that farmers would be paying, net of their labor contribution, the following percentage share: 26. One reviewer of a draft of this report cast doubt on this interpretation that watercourses were public property, believing that they were the common property only of the shareholders. A legal opinion on the exact position would be useful to clarify the issue. 27. National Engineering Services Pakistan (PVT) Limited, On-Farm Water Management III Project: Feasibility Study, Lahore, May 1990. 28. The same consultants had more or less simultaneously produced a report for the Government of Balochistan which showed that farmers could not afford to pay 25 percent cost recovery. 115 Annex C Punjab about 25 percent Sindh 22 percent Balochistan 19 percent NWFP 18 percent 15. The SAR comments that these increases, of between about five percent and nine percent over the current rates, were an important step toward the eventual goal of making farmers fully responsible for the financing of construction. "Greater increases in farmers share would have been desirable, but GOP provinces do not consider this feasible at present as program acceptance would be jeopardized." 16. The "eventual goal" of full farmer financing has been given a new meaning in the context of the Bank's latest strategy for Pakistan's irrigation development.29 Under this Farmer Organizations (WUAs) would take over full responsibility for water distribution at the minor/distributary level including responsibility for O&M, and planning and development of the system within its area. The WUAs would become fully responsible for the investments, with some technical guidance for technological improvements beyond their capability. The exact arrangements envisaged for providing the finance are not clear, but what is clear is that the government would cease to provide subsidies for watercourse improvement. The cost of the improvements would be incorporated within an overall water charge which covered all the WUA costs for water purchase and distribution, of which watercourse improvement would be one element. 29. Issues and Options, op. cit. 117 Annex D Watercourse Aspects (a) Water Loss Estimates In Bank Appraisal Reports 1. Four appraisal reports on projects with OFWM components (OFWM-I, II, III, and CWMP) have had to make assumptions about the amount of water that will be saved by project actions. Somewhat surprisingly, these assumptions vary from project to project. Confusions also arise, when savings are discussed, from failure to distinguish between savings as a proportion of losses and savings as a proportion of the flow at watercourse head. The key definition is that of watercourse delivery efficiency: the ratio of flow at the point of delivery (the field outlet) to the flow at the watercourse head (mogha). Losses are the converse (i.e. 100 minus watercourse delivery efficiency). 2. At the time of the preparation of OFWM-I (1980) the only data available to assess the likely result of watercourse improvements was from the research at Mona. This showed that for the type of masonry channel being proposed up to 90 percent of water losses could be saved, i.e. that watercourse delivery efficiency could be increased from about 55 percent to over 90 percent, provided that a sufficient length of the watercourse was upgraded. Under the project, the maximum percentage lining was 30 and the SAR estimates (para. 3.0) that this would result in savings of 25 percent. In the detailed calculations for the economic rates of return loss savings of 17 percentage points were used. Given the available data at the time, these seem realistic assumptions. 3. The next project chronologically is Command Water Management (prepared during 1983). This design had the results of the WAPDA evaluation of the USAID Pilot Project to go on. These show that, immediately after improvement, the amount of water delivered to the field outlet increased by 18 percent, coming down to 15 percent after about two years. The SAR makes the assumption that savings will vary from about 15 percent of the surface and public tubewell water delivered to the watercourse head in FGW areas with 15 percent lining, to "over 25 percent" in one of the project sub-areas (6-R Hakra Branch) with 40 percent lining. On average 18 percent savings are expected. The report gives the savings in acre feet per subproject which work out at 225 acre feet (275 cubic meters) per watercourse on average, ranging from 170 to 240 acre feet (210-270 cubic meters), presumably on the basis of different lining percentages. Given the available evidence, these assumptions, while on the high side, are not unreasonable. 4. The second OFWM project was appraised in September-October 1984, when only the results of WAPDA's evaluation of the USAID project were available, not the results of the evaluation of OFWM-I. In the SAR the average (of FGW and SGW with their different lining percentages) saving is taken as "about 20 percent of the water supplied to each chak" (para. 2.05). The 20 percent saving assumption is confirmed by para. 5.01 (agricultural impact sub- section) where the additional irrigation water available from reduced conveyance losses is estimated at about 25 cubic meters per second for 2,500 watercourses. This is equivalent to a 315 cubic meters per watercourse, which is slightly over 20 percent of the average watercourse delivery of about 1450-1500 cubic meters/year. 5. The M&E evaluation of OFWM-I of 1985 showed that watercourse improvement raised the delivery efficiency of the watercourses under study from 64 to 73 percent, or by-9 percentage Annex D 118 points, equivalent to increasing the amount of water delivered to the field outlet by 14 percent. The evaluation report describes this as a 25 percent reduction in watercourse losses. This M&E result is considerably less than the 18 percent assumption which had been made in the projects financed to date. 6. The PCR of OFWM-I, which was prepared during the first part of 1988 and used the M&E evaluation results, concluded that there was a 9 percent increase in watercourse delivery efficiency from renovation. This PCR came out just in time to provide an input to the SAR of OFWM-III, but the SAR ignored the evaluation findings. It summarized the work to date as that "W/C renovation reduces conveyance losses between 35 and 50 percent," which converts to a 20 percent increase in delivery efficiency. 7. The PCR of OFWM-1l, prepared in 1992, confused decreases in losses with changes in conveyance efficiencies, saying at one point (para. 6.1) that WAPDA's evaluation results show that the saving in water conveyance losses due to project works are 10 percent, and commenting that this was much less than the 20 percent assumed in the SAR. Finally the SAR commented (para. 6.9) that "it can probably be concluded from these results [the M&E findings] that delivery efficiencies increased 10 percent in the case of watercourses with lining." (b) Lining, Free Choice or the Ministry Knows Best? 8. A senior Agricultural Department official told the mission that there are absolutely no technical grounds for the lining criteria now being applied, and the technical solution that is right for one watercourse may not be right for another. More than half the losses along a watercourse were operational losses: overtopping of the watercourse bank, leakage at the field offtake, losses down rat holes and tree roots. What sort of technology, he asked, do we need to control these kinds of losses? In his experience one gets different answers depending on who one asks. If one asks the "expert," he will tell you his technical solution, watercourse lining, which is the same for every watercourse. But, if one asks the farmer, one gets quite another set of answers. We, who have to implement the projects, talk about participation, but we never ask the farmer what he wants. We impose our technical solution on him. Farmers accept it because they have no other choice and, anyway, it's subsidized. Farmers should be given the choice of technology, told the costs of alternatives, and be allowed to select what fits their watercourse. 9. The official believed that there is no need to line to a formula. On some watercourses, even lining a few problematic meters will be beneficial. On others, such as the Punjab rice- growing areas, lining is not needed, farmers are not short of water. They have tubewells, and their watercourses provide only 20 percent of their water needs, but even so the Field Assistant has a target for watercourse lining he must meet, and meet it he does, at least on paper. Nor is lining needed in the Sindh rice-growing areas: they have more surface water than they need already. Unnecessary lining is not confined to rice-growing areas. In much of the fresh groundwater area lining is also unnecessary. On the other hand, there may be a case for exceeding the imposed criteria, for instance in very sandy areas. If the farmer is allowed to choose, and is faced with the costs of his choice, there will be far less wastage from unnecessary lining. The watercourse improvement program needs to be site-specific. The original Bank Appraisal Report laid down some selection criteria, but they are not followed. 10. The AD official said that in many areas the demand for watercourse improvement is far larger than the department can meet. When he was in charge he controlled the excessive number 119 Annex D of applications by insisting that he would not sanction any lining until the rest of the watercourse, which was not to be lined, had been improved by the farmers. After all, this was only what the Bank had laid down, and it was right. But donor pressure to achieve targets was leading to claims that far more watercourses had had their unlined sections improved than was true. On over half the watercourses now claimed as improved, the non-lined portion had not been touched. 11. The official believed that there was a close linkage between the social situation on the watercourse and the appropriate technical solution. Social situations vary enormously across the country and therefore so should technical solutions. Seepage losses vary for social reasons. On watercourses with little social cohesion, O&M is bad and losses are high. Where social cohesion is strong, O&M is good. But when watercourse renovation is offered on a subsidized first-come- first-served basis, it is those social cohesive watercourses, where improvement is least needed, which come forward first. Watercourse improvement does not alter social cohesion amongst farmers on a watercourse, and where it is weak O&M will remain poor. The creation of a WUA will rarely change things. (c) Cost Recovery for Watercourse Improvement 12. The watercourse renovation (or lining) program is very popular with farmers, and one of the reasons for this is that cost recovery is low. Rates have varied over the years, but remain low. When the first watercourses were renovated and lined, the technology was still experimental and benefiting farmers were not asked to contribute to the cost of the works. It was the USAID Pilot Project which introduced the concept of the beneficiaries volunteering their labor for the renovation. This was valued at Rs 37,000 per watercourse in 1976 prices, representing 12.75 percent of project cost. No other contribution was required, except for PLL, the cost of which was shared 50:50. A similar arrangement was followed under the Punjab "crash" program. 13. The Bank's OFWM-I, in line with normal Bank policy, was the first project to require a farmer contribution towards the cost of materials, in addition to providing free labor. The amount required varied between provinces: 20 percent of the nominal cost of construction materials in Punjab, with farmerspaying the cost of masons, 25 percent in Sindh, with the project paying masons, and nothing in NWFP and Balochistan. Costs were to be recovered by OFWM directorates in ten bi-annual installments with a two-year grace period. 14. The CWM project kept the same arrangements, except for Balochistan, where the same rates as in Sindh were introduced. Under OFWM-1I the cost recovery rates for Punjab and Sindh were raised to 25 percent and 30 percent, respectively. The rates for the other provinces remained unchanged except that NWFP proposed to introduce a recovery rate of ten percent, later changed to a ten percent lump sum advance. The value of the donated labor was estimated in the SAR of OFWM-II, at between Rs 35,000 and Rs 40,000 per watercourse in 1985 prices, equivalent to 20-25 percent of total cost. Farmer contribution is thus between 50 and 55 percent of total cost. In many parts of Sindh the landowners require their tenants to provide the labor, so to the many such Sindhi landowners even "voluntary" labor is free. 15. At the time OFWM-II was being reviewed within the Bank, the treatment of these projects as normal irrigation projects, with the government recovering the cost of O&M and part of the materials, was queried on the grounds that they were self-help projects, with all the works Annex D 120 on farmers' lands. If they were such the high subsidy element would be much more transparent, particularly what amounted to interest-free deferred payment for the balance of cash costs (contrary to normal Bank policy) would be highlighted. After considerable discussion this position was rejected on the grounds that most watercourses were public property (at least in Punjab), and that the 1873 Canal Act allowed government to undertake improvements on watercourses and to recover the cost from farmers. 16. These recovery rates imposed under the project are still very low. The RAP had estimated that beneficiaries were capable of bearing the full cost and recommended they be required to pay in full, but in installments over a 15-year period with one year of grace. However, when the Bank tried to raise the proportions substantially under OFWM-III, the central government did not consider this feasible "as program acceptance would be jeopardized" (SAR, para. 5.16). 17. The long waiting lists, in areas where watercourse renovation is highly profitable, contradict this argument. Many farmers would be willing to pay substantially more if it increased their chance of being included in the program. It might well jeopardize the program in areas where renovation is not needed, but this would be a good thing. In practice, the rates of collection are so low that to many farmers the watercourse renovation is virtually a free good. 18. A reviewer of a draft of this report stressed the importance of cost-sharing as a test of farmer interest by relating the experience not too far away in India: 19. "Up to late 1970s, farmers in Haryana and Punjab paid for a large proportion of the cost of watercourse lining (i.e. 50-70 percent or so). As a result of this financial interest, the process was extraordinarily participatory (though we did not know it at the time, because the catch- phrase had yet to be invented). Farmers elected a committee to oversee the finalization of the alignment, contract award, and construction supervision, jointly with the ID. Records were kept of bricks and cement use, the farmers posted guards on the materials at night! Farmers paid fully for O&M within the chak (in accordance with the Act). Then the government decided that the farmers of Haryana-the most prosperous in India-could not afford to repay these costs... Progressively [the farmers] lost interest in construction quality, stopped doing O&M ('The watercourse belongs to the Government now-they lined it') and deterioration set in... The important point of this story is that significant cost recovery is fundamentally important in deciding whether the farmers want the investment, and retaining their interest in it. Farmers in Haryana were clamoring for lining while they were paying for it. If the farmers are not prepared to make a serious contribution to costs, don't invest. This means resisting the politicians (who want to give something to the farmers, resisting the ID, who want an investment project, and resisting the contractors, who are often in 'contact' with these last two parties." (d) Monitoring and Evaluation of Watercourse Conveyance Efficiency 20. Prior to the first experimental measurements of watercourse losses, made in 1973 by WAPDA in collaboration with Colorado State University (CSU) at Mona Reclamation Experimental and Research Project, most studies has assumed a conveyance efficiency of 90 percent, i.e. that only ten percent of the water delivered at the mogha was lost during transit to farmers' fields. The results at Mona indicated that 44 percent of the water delivered at the mogha was lost in transit (i.e. a conveyance efficiency of 56 percent). Two further studies in the field confirmed these findings. The first was the "Forty Watercourse" survey (1975-76) in 121 Annex D Punjab and Sindh carried out by WAPDA and CSU. The second was the "Sixty-One Watercourse" survey (1976-77) carried out by WAPDA with the financial assistance of UNDP and the World Bank, which was designed to provide the basis for formulating the Revised Action Programme (1979). The first survey measured conveyance losses as 47 percent; the second as 45 percent in rabi and 44 percent in kharif 21. Watercourse improvement technologies were developed at Mona to reduce losses in transit. The reduction in losses varied according to the type of technology adopted: Type of Improvement Water Saving.% Heavy cleaning and maintenance30 25 Earthen Improvement 50 Conventional Brick Lining (bed and sides) up to 90 Low Cost Cement Block Lining (sides only) 65-85 22. The costs of type of improvement varied from Rs 0.25 per linear foot in 1977-78 prices for heavy cleaning up to Rs 56 per linear foot for full conventional lining. No attempt seems to have been made to calculate which of these techniques was most economic, in terms of cost of lining versus value of water saved. 23. Monitoring of watercourse losses on a project basis began with the USAID Pilot Project. This was evaluated by WAPDA's Watercourse Monitoring and Evaluation Directorate (WM&ED) or its predecessor agencies. The main finding of its four volume final report on the results, which came out in 1984, was that in the first year after improvement lining had increased conveyance efficiency by 11 percentage points, with overall watercourse conveyance efficiency, defined as the ratio between the water delivered to the field and that diverted from the outlet at the mogha, rising from 61 percent before improvement to 72 percent after improvement. This is equivalent to an 18 percent saving in losses; this is the average of losses both in the main watercourse, only part of which was lined, as well as in the farmers' channels, which were unlined and only partly improved. By the third year after improvement losses had increased again so that the improvement in conveyance efficiency was only 9 percentage points (15 percent reduction in losses). This decrease took place in the farmers' branches, which had been improved by strengthening their earthen banks and indicated inadequate attention to maintenance.31 24. WAPDA's WM&ED also evaluated the Bank's first OFWM project from 1982 to 1985, with a random sample that included both 20 watercourses improved under the 'regular' program (i.e. brick and cement lining of a portion of the watercourse) and 25 watercourses improved under the 'accelerated technology' (i.e. earthen improvement only),in the three provinces of 30. Equivalent to the heavy cleaning adopted in the early OFWM projects. 31. This pre-improvement figure of 61 percent efficiency is much higher than the efficiencies which the early research work at Mona and the RAP surveys had found, and which had created the original interest in the OFWM program: these had shown that efficiencies were only about 53 to 56 percent, i.e. losses were 44 to 47 percent. The work being done on OFWM-1II shows an even higher overall pre-improvement efficiency of 67 percent in fresh groundwater areas and 63 percent in saline. (Baseline Report.) The high savings, which the original Mona research had demonstrated were possible, were what really created the interest in watercourse lining. Annex D 122 NWFP, Punjab and Sindh. Unfortunately the evaluation stopped before the project closed and failed to provide data on other than the immediate impact of the watercourse improvements. (e) The Impact of Watercourse Improvement on Conveyance Efficiency: OFWM-I and II Monitoring Results32 25. A separate study on OFWM in Balochistan is not used here. The OFWM-II monitoring covered 12 lined watercourses (the accelerated technology was dropped in OFWM-II).33 The time-scaled evaluation done under OFWM-III covered eight lined watercourses, four each from OFWM-I and II, with six farms on each watercourse for OFWM-I and nine on each for OFWM- II. The time-scaled analysis aimed to track conveyance efficiency over time to assess the useful life of the improvements. For the OFWM-Ill evaluation, the length of each watercourse was sub-divided into head, middle and tail, with 2 farms on each reach for OFWM-1 and 3 farms on each for OFWM-II. Flow rates were measured with cut-throat flumes at (i) the mouth of the mogha; (ii) the end of the sarkari khal; and (iii) upstream of the field nucca. 26. Overall conveyance efficiency is defined as the percentage of water delivered from the mogha that reaches the field nucca, and includes losses in the main watercourse (sarkari khal), both the lined and unlined sections, as well as in the farmers' unlined branches. The length of each watercourse was sub-divided into head, middle and tail, with 2 farms on each reach for OFWM-I and III farms on each for OFWM-I1. Flow rates were measured with cut-throat flumes at (i) the mogha outflow; (ii) the end of the sarkari khal; and (iii) upstream of the field nuccas. The data are as overleaf: 32. References: (a) WAPDA, April 1994, First Time-Scaled Evaluation Report ofFarm Water Management Program, IDA Financed OFWM-I & II Projects, Watercourse Monitoring and Evaluation Directorate, Planning and Investigation Organization, Planning Division.(b) Waryam Ali Mohsin, "The Impact of Watercourse Improvement on Conveyance Efficiency". Paper prepared for presentation at The National Conference on On-Farm Water Management, May 29-30, 1994. Mona Reclamation Experimental Project, WAPDA, Bhalwal.(c) Punjab Economic Research Institute, Planning and Development Department, GOPunjab: Evaluation of On-farm Management Programme in Punjab, December, 1985 and February, 1993. 33. Monitoring of the second OFWM project was given to a private firm initially, but their contract was canceled after they had collected the baseline data. The work was then entrusted to WM&ED, but it was found that the base-line data were unreliable, and thus the reduction in watercourse losses could not be evaluated. Unfortunately, no fresh selection was made, so this vital data remains unavailable. 123 Annex D Watercourse Conveyance Efficiencies Reported in Monitoring Studies of OFWM-I and H Conveyance Efficiency (%) Tail Mid Head Average OFWM-I Before 46 66 85 62 After 65 76 90 77 Improvement (% points) 19 10 5 15 After ten years 64 76 92 75 Improvement (% points) 18 10 7 13 OFWM-II Before 65 71 84 73 After 71 78 90 80 Improvement (% points)- 6 7 6 7 After four years 68 74 92 78 Improvement (% points) 3 3 8 5 27. The observations for OFWM-I made ten years after lining show small changes in efficiency at the head and tail, and a small overall fall in efficiency of two percentage points, which is attributed to poor maintenance. Thus the benefits of watercourse improvement have been largely sustained for ten years and are expected to last for some more years. 28. It is difficult to explain an improvement of only seven points after improvement, or half that of the OFWM-I watercourses, for the OFWM-II sample. Four years later there is some decline in efficiency in the middle and tail reaches, but an increase in the head or lined reach. Again, this is attributed to poor maintenance of the unlined lower reach. One possible explanation of such low water savings is that the pre-project data collected by consultants, who could not complete the work, is from unrepresentative watercourses. On the basis of improved overall efficiencies found in other studies, it is probable that the increase in OFWM-ll efficiencies is comparable to the OFWM-I results, on the assumption that the "Before" data is unreliable. 29. Another study34confined to Punjab, found that the delivery efficiency on a sample of61 improved watercourses was 68 percent compared with 61 percent on a sample of 61 unimproved watercourses, an increase in efficiency of 7 percentage points. This is a lower figure than the WM&ED data but the sample of improved watercourses includes some that had been improved up to 15 years earlier and may have deteriorated over the intervening years. The WM&ED data are also based on actual measurements of discharges and therefore to be preferred as a base for estimating aggregate water savings. 34. Saleem, Mahamood Ali, Amin, Muhammad Sajid and Mazar-ul-Haq, February 1993, Evaluation of On-Farm Water Management Program in Punjab, Punjab Economic Research Institute, Publication No. 281. Annex D 124 30. Another source of information is provided by two farmer surveys in the Punjab in 1985 and 1993 by the Punjab Economic Research Institute (PERI).c Both surveys compared data from interviews on samples of improved and unimproved watercourses, in 1985 from 143 farmers on 16 improved watercourses (out of a total of 177 improved watercourses) and in 1993 from 549 farmers on 61 improved watercourses out of a total of 10,833 improved. The first survey related to only the Bank-assisted OFWMP-I, while the second covered watercourses improved with funds from USAID and ADB, as well as the Bank. Neither survey measured water flows (indeed without baseline flows this would have been of limited use), and one report even criticized measuring methods as problematic. Changes in flows are deduced indirectly from cropping changes. These results suggest that watercourse flows increased on average with improvement ("lining") by around 20 percentage points, which is substantially higher than the measured increases discussed above. 31. The increase in conveyance efficiency found by the different studies thus varies between nine and 11 percentage points in the evaluation of the USAID project and around 15 percentage points in the evaluation of the Bank projects. This is equivalent to a percentage reduction in conveyance losses of between 15-18 percent for the USAID project and 24 percent for the Bank projects. To some extent this difference may be a reflection of the greater length of lining under the Bank projects, 16 percent of the average length under OFWM-I and 23 percent under OFWM-II, compared with less than ten percent under the USAID project. Unfortunately there is no experimental evidence, or monitoring results to indicate the relationship between lining proportion and water savings. On the other hand there is also some reason to suspect that the time-scaled evaluation may be exaggerating the savings under the Bank projects. The results for OFWM-I are for a sample of only four out of the original sample of 20 watercourses improved under the regular technology (and 25 under the accelerated technology). The original monitoring results for the sample of 20 watercourses show an increase in conveyance efficiency from 64 percent pre-improvement to 73 percent about one year after improvement, an increase of only nine percentage points (a reduction in losses of 14 percent). It should be noted, however, that only 13 percent of the length of the sample watercourses was lined, compared with the minimum required under the project of 15 percent. The results for the 25 watercourses improved under the accelerated technology show an initial improvement from 67 percent to 72 percent, an increase of only five percentage points, and by the time of the final evaluation the increase in efficiency was not statistically significant. 32. Given the very small sample size, and the wide variability in measured improvements in conveyance efficiency, either set of monitoring results is likely to be subject to a wide margin of error. There is, however, a high probability that the improvement in conveyance efficiency will not exceed the highest measured level of 15 percentage points (or a reduction in losses of 24 percent). It may well be less but it is impossible with the available data to say how much less. 35. The report selected only four because it found the others to be unrepresentative of the universe. Many selected watercourses in the original sample were close to cities, agricultural universities, etc. 36. WAPDA, December 1985, Monitoring and Evaluation of On-Farm Water Management Program-I; Final Report, Watercourse Monitoring and Evaluation Directorate, Planning and Investigation Division, Planning Division. 37. The proportion of the watercourse length lined varies considerably. For the four watercourses selected for the time-scaled evaluation the proportion was 16 percent; for the original sample it was 13 percent. There is a major variation between provinces. The Punjab watercourses had 15 percent lining, as per design, the Sindh ones had only ten percent lining, but the average change in efficiency is reported as being the same, as far as can be detected in the final report. The efficiency actually fell after improvement on 20 percent of the sample watercourses. 125 Annex D Therefore the figure of 15 percent improvement is used to calculate the impact of the Bank projects on water savings. This will probably overestimate the actual impact but the scale of the divergence from the original expectations of what the whole OFWM program would achieve is so great that minor downward adjustments would make little difference to the conclusions. The original RAP proposals were based on an 85 percent reduction in conveyance losses. Whether the actual is 24 percent or 18 percent is of little consequence. 33. The factors affecting conveyance efficiency are the extent of lining, the watercourse discharge at source and the distance from the source to the nucca. The extent of lining differs considerably between watercourses and projects. For OFWM-I watercourses the average length lined is 16 percent, ranging from nine percent to 32 percent; for OFWM-II, the average is 23 percent, ranging from 14 to 36 percent. The mean discharge at source for the OFWM-I watercourses is 1.8 cusecs pre-project and between 1.5 and 1.7 cusecs post-project, depending on the season. For OFWM-II watercourses the mean post-project discharge is 1.3 cusecs The distance from the so4ce to the nucca on OFWM-I watercourses is 1,100 m pre-project, and ten years later-1,000 m. Using these figures the water loss rate per unit length of the watercourse was calculated as follows: Water Losses (cusecs/300 m. of watercourse) OFWM-I Before 0.19 After 0.15 Decline 0.04 Ten years later 0.12-0.16 Decline 0.07 OFWM-II Before No data After 0.08-0.1 34. This quantification of water loss rate tends to support the sustainability of improved overall conveyance efficiency, with a somewhat better result, so far, for OFWM-II. 35. _ Conclusions from the analysis of conveyance efficiency versus discharge are that before renovation high water losses are directly related to high discharges, while post-renovation, high discharges (as would be expected) can be accommodated without high losses. 36. The following data demonstrate the decline in conveyance efficiency with distance from source: Annex D 126 Conveyance Efficiency with Distance from Mogha (Percent of Mogha Flow) OFWM-I Watercourses OFWM-II Watercourses38 Before After After 10 After Years Up to 300 meters 92 90 92 86 300-600 83 80 84 88 600-900 71 69 74 86 900-1,200 62 70 74 83 1,200-1,500 47 69 74 81 1,500-1,800 45 56 61 82 Overall mean 62 78 76 80 37. Statistical analysis shows that the influence of distance down a watercourse on conveyance losses is lower after renovation, and OFWM-II data show similar results. 38. Average of five seasons. 127 Annex E Precision Land Leveling-A Declining Priority 1. Research studies at Mona estimated that low water application efficiencies, i.e. the ratio of water stored in the crop root zone to water applied on the field, were resulting in water losses through deep percolation of about 30 percent of water released into the field. This was mainly caused by over-irrigation, the result of farmers allowing water to flow into fields until the highest spot was covered, thus applying too much water in the low spots. Only if the land was very precisely leveled was it possible to apply precise amounts of water.39 Land leveling thus formed an important component of the original "on-farm water management" (OFWM) package, and indeed was possibly the reason for the "on-farm" part of the title. The experts believed that the traditional system of land leveling, using a plank pulled by a bullock or a tractor, had to be replaced by modem equipment that would allow more precise leveling. So a Precision Land Leveling (PLL) project was started through the provincial directorates of agriculture, with technical assistance from the US Soil Conservation Service. 2. This was then expanded under the USAID-funded pilot OFWM project. It was planned to level 430,000 acres (about 200,000 ha) with a 50 percent subsidy for up to five acres for farmers owning less than 25 acres. Only about 15 percent of this target was achieved. The main beneficiaries were big farmers: small farmers were reluctant to accept the technology, and the farm size limits written into the project document had to be ignored to get any response. The bulk of the land leveling thus occurred in Sindh where land holdings are large. The government later officially de-emphasized this component in order to concentrate resources on watercourse improvement.4 It had also been hoped that contractors specializing in PLL services would emerge, but despite the subsidy this did not happen, probably because other uses for tractors were more profitable. A joint USAID/Bank evaluation of the project in late 1979 recommended that market forces should be allowed to determine the rate at which PLL developed, without any increase in subsidy.41 In the middle of 1980 the government decided to remove the subsidy. 3. Thus when the Bank-assisted OFWM-I project was under preparation it was unclear what the demand for unsubsidized PLL would be, as experience with the USAID project was not encouraging, even with subsidy. At the same time it was also evident that PLL was not attractive to small farmers, so only a small (15,000 ha) PLL component was included under the OFWM-I project. No attempt was made to target small farmers. Project assistance was limited to the provision of equipment for rental by farmers and to technical assistance. 4. By the end of the project, according to the PCR, the leveling targets were exceeded in both Punjab (by almost double) and Sindh (by about one-third). The probable reason for this 39. There are two additional benefits from land leveling other than saving water: fertilizer is wasted when excess water leaches it from the soil root zone, and seed germination is poor in excessively wet or dry spots so the crop stand is uneven. The mission was told by a former Director of CWMP of how the original concept of watercourse improwgment was like a tripod, with three legs, one for watercourse engineering, one for PLL and one for agronomy. The last two legs had withered over the years. In the beginning half the members ofeach watercourse improvement team were PLL specialists. 40. USAID, The On-Farm Water Management Project in Pakistan, Project Impact Evaluation Report No. 35, June 1982, page 16. 41. Mission BTO of 11.7.79. Annex E 128 was the extension of the project life. Much of the land leveled was in the head reaches of watercourses where large amounts of silt had been deposited, by farmers in the process, over decades, of cleaning the watercourse. Much of the work was thus more in the nature of land grading than PLL. 5. A similar component was included in OFWM-II, with the exception that laser-controlled leveling equipment was to be provided (it had been tried out experimentally in Punjab in OFWM-I). The target area to be leveled was only 16,000 ha and again no targeting of small farmers was attempted. The SAR had no comments on the effectiveness of the PLL accomplished under the previous project and made no reference to the benefits of PLL. According to the PCR, the target area was exceeded by 62 percent overall, again more as the result of the extension of project life than because of high demand. In Balochistan, however, the topography of the land is such that leveling is more land grading than PLL, and the laser equipment was not used. 6. No PLL was included in the watercourse improvement component of the Command Water Management Project. In the third OFWM project, however, it was again included. 7. The lack of interest in PLL by small farmers in Sindh was investigated by the Sindh Development Studies Centre (SDSC).42 Their report did not attempt to verify the area claimed to have received PLL, but believed it was exaggerated. (The Supervisory Consultants were supposed to verify the area claimed but were too short-staffed to make more than a superficial check.)43 The report concluded that, while most farmers were aware of the benefits of good land leveling, and practiced it as a normal part of their land preparation activities with the resources available to them, the vast majority had never heard of the PLL program or seen demonstrations of its effects. The main difficulties of implementing PLL on small farms were: (i) lack of equipment (only about 10 percent of cultivators owned a tractor); (ii) lack of time between crops to complete the PLL operation; and (iii) the relatively high cost. In addition, in Sindh, a tenant would not be able to undertake this operation without the agreement of his landlord. 8. From what the evaluation mission saw in the field, PLL is a program which essentially only benefits large farmers and, while there is now supposed to be no subsidy, there is clearly a very high one: technical assistance from departmental staff is free and the hire rates for equipment are nominal and, at most, cover only running costs. Only large farmers have sufficient land to make proper use of laser equipment, and sufficient resources to invest in the considerable capital cost of leveling. The program has never been evaluated: it would be extremely difficult to separate the benefits from the leveling per se from the benefits from additional inputs, such as fertilizer, which level land would make it economic to use. There is good reason, however, to believe that the benefits from land leveling are short-lived. Most farmers do not know how to set a plow so that it will not upset the levels that have been so accurately produced with the laser equipment. Within two or three years the land will have 42. Sindh Development Studies Centre, The Causes of Very Slow Response of Small Farmers to the Precision Land Leveling, September, 1989. 43. It was suggested to the evaluation mission that, because of the verification problem, it was easy for the departments concerned to show that they had fulfilled their leveling targets and thereby avoided the problems that a report showing large shortfalls would create. The exactness with which annual targets were supposedly met gives credence to this view. 129 Annex E reverted to its previous undulating state. There thus seems to be no justification for this component in the OFWM program. 9. The survey carried out for this impact study provides some revealing evidence on the lack of impact of the PLL program. On the 240 watercourses surveyed in Punjab, not a single farmer reported any land leveled by PLL. On the 210 watercourses in Sindh, only 36 watercourses (17 percent) reported that there had been any PLL. For the most part, less than half a dozen farmers on these watercourses had had their land leveled. elf 131 Annex F Irrigation System Rehabilitation and Sustainability Results of a Survey Carried out by the Irrigation Systems Research Institute of Pakistan 1. The impact evaluation mission's visit to Pakistan came at a time of year when it was not possible to determine the condition of the canals which had been rehabilitated under either the ISR of CWMP projects, to see whether the improvements which had been made under the projects had been sustained or not. The mission accordingly arranged for the International Sedimentation Research Institute of Pakistan (ISRIP) to carry out an engineering inspection/survey, during the canal closure period at the beginning of 1995, of a small sample of the canals improved under the two projects. ISRIP, in its former identity as WAPDA's Alluvial Channels Observation Project (ACOP), had carried out the hydraulic monitoring/evaluation of the rehabilitated canal distribution sub-systems under ISRP-I, including the effect of the improved water deliveries on irrigated agriculture. 2. A total of 15 channels were selected for study in Punjab, Sindh and NWFP provinces, distributed as follows: Province Project Branch Distributary Minor Total a) Punjab ISRP 1 2 1 4 CWMP - 1 1 2 b) Sindh ISRP 1 2 1 4 CWMP - 1 1 2 c) NWFP ISRP 1 1 1 3 Total 3 7 5 15 3. This was out of some 600 chanels in all four provinces which had been rehabilitated under ISRP-I and CWMP. The amount of rehabilitation had varied considerably: in a few cases there was complete head to tail rehabilitation; in most cases only short reaches were improved. The selection for the sample concentrated on channels where major earth works had been undertaken. A list of the channels selected is given in Table F. I together with their design discharge. The canals varied in size from 1,723 cusecs down to 6 cusecs. The engineering inspection concentrated primarily on: a) the condition of the channels as visually determined during the closure period; b) the condition of the canal structures; 44. There was need to do a survey because there is no routine source of objective information in Pakistan on the condition of the irrigation system, such as would have permitted a comparison of rehabilitated channels with non- rehabilitated channels. Annex F 132 c) t1W condition of outlets from the canals; d) collection of engineering data, e.g. berm width, freeboard, siltation, seepage; e) the potential maintenance needs and adequacy of maintenance funding; f) major accidents subsequent to rehabilitation, especially breaches; g) pre- and post-rehabilitation discharge data. Channel Condition 4. Tables F.2 through F.5 summarize the results of the engineering inspection. Table F.2, Condition ofBerm, shows that for nine out of the 15 canals the condition is good over the full length of the canal. Out of the other six canals, where the condition is defective to some extent, four of them are not seriously affected: the length affected is 15 percent Or less of the total. In only two cases is more than 30 percent of the length of the berm inadequate cr non-cxistent, Jurrian Distributary (31 percent) and Doso Minor in Sindh (98 perceni). Thus ia only one case is the condition seriously dangerous. Table F.3, Height of Freeboard, clrzsifies the propcrtion of the canal length according to the observed height of the freeboard. As thi was observed during the closure period, the measurements are estimates from water level marks made during the inspection. For most of the canals, a freeboard less than 1.5 feet indicates that the canal is being run at above design levels, either because it has silted or to pass extra discharge.45 Only three of the canals have 75 percent or more of their length with freeboard over 1.5 feet. Seven of them have freeboards of 0.5 feet or nil. This would tend to indicate that all the three PIDs are in breach of the Development Credit Agreement for ISRP which required them to maintain the integrity of the minimum designed freeboard [DCA, section 3.05 (b)].46 5. Table F.4, Other Indicators of Canal Condition, provides data on five different indicators. (i) Seepage: this shows that only one out of the si:: canals in Punjab was affected and then only slightly; that five out of the six in Sindh were affected, two of them seriously; and that none of the three canals in NWFP was affected; (ii) Sediment heaps, being sediment piled on canal banks during the process of canal cleaning, indicates :hat the canal has recently been desilted; the table shows that only a small proportion of the Punjab canals have such sediment heaps while in Sindh five out of the six canals have them, in three cases covering the entire canal length; in NWFP two of the three canals have them, but only along a small part of the length, the existence of these heaps after cleaning is an indication of inadequate O&M funding: there are enough funds to remove the silt from the channel but not enough to clear it away; (iii) Indication of windblown sediment is only important on one canal, 1-R Pacca (an offtake of the Thal Branch 45. 1.5 feet is generally adopted as an average freeboard for distrubutaries and minors; for branch or bigger canals (i.e. iza e survey Alipur Branch, Nabisar Branch and Machai Branch), 2.5 feet or more is required for safe passing of the designed discharge. 46. This covenant was inserted into the DCA in order to avoid the PIDs from further increasing the quantity of water conveyed down the rehabilitated channels and to set a pattern where the canals were no longer operated in an unsafe manner. The project had allowed rehabilitation to the then maximum discharge which was frequently greater than the design discharge, and the Bank wished to prevent any further unauthorized increases. ISRIP point out that a comprehensive survey with physical measurements would be needed to verify if the covenant were being adhered to. See also para. 12 on Canal Discharge. 133 Annex F of the Muzaffargarh Canal); (iv) Weed growth, meaning growth that affects water flow in the canal-not merely bush growth on the canal banks-is important on the Doso Distributary and the Nabisar Minor, both offtakes from the Mithrao system in Sindh; and (v) Condition of inspection path, recorded as the proportion of the path which is non-motorable, or in very poor condition; five canals (two in Punjab and three in Sindh) have parts of the inspection path impassable, or with difficulty, to vehicles, thus making proper inspection at least more time- consuming and indicating inadequate maintenance. Structure Condition 6. The form which ISRIP used to record the results of their visual inspection covers, among other things, the general condition of hydraulic structures, whether there are silt heaps and scouring, the condition of gates, damage to joints and the general condition of wooden weirs. For the six Punjab canals there are only one or two minor structural repairs reported as being required and the condition of all the other items is recorded as good. The most significant shortcoming was that the remodeling of the intake structures on the 1-R Pacca Distributary, which had been recommended following a 1983 Behavioral Analysis of the canal, had not been carried out and the canal could thus not take its due share of water/sediment. Despite this, on the whole, the condition of the structures on the Punjab canals can be regarded as satisfactory. The picture for the Sindh canals is mixed: for the ISRP canals, the following defects were reported: Nabisar Branch: most of the regulator gates are in poor condition and there is scouring below most structures; Sanwari Distributary: three road bridges need repair and there is scouring below most structures; Nabisar Minor: head regulator needs repair as do two road bridges; scouring below most structures; 7. The two CWMP canals are in reasonable repair except for (i) the gates on one regulator on the Mithiana Distributary which is causing seepage and another gate at the Head Regulator is required for smooth regulation and (ii) on the Detha Minor there is erosion below the head regulator and the outlets at the tail regulator are all damaged; the railway bridge/syphon needs rehabilitation as it prevents water delivery to the tail reaches; 8. The three NWFP canals have their structures in generally good repair, though there are signs of scouring in places on the Machai Branch, and the head regulator on the Ismailia Distributary has a cracked plate. Outlet Condition 9. The condition of outlets is summarized in Table F.5. Only one canal in Punjab requires any outlet repairs; in aggregate this represents 3 percent of the outlets inspected in Punjab. In Sindh outlet repairs are needed on five out of the six canals inspected, though these are minimal on the two CWMP canals. On the four ISRP canals repairs are needed on between 39 and 95 percent of all outlets; on aggregate for these four canals, 89 percent of all outlets need repair. In NWFP only one outlet needed repair, less than one percent of the total inspected. (From the Annex F 134 mission's own inspection of 36 watercourses in Sindh, most of whose outlets needed repair, the main reason was that they had been damaged to admit more water.) Maintenance Funding 10. One of the important topics of the ISRIP survey was the assessment of the extent to which the canals rehabilitated under ISRP-I and CWMP were being properly maintained. The above evidence from the visual survey reports indicates that for some aspects the situation is satisfactory, but discussions that ISRIP surveyors had with PID staff indicate a general shortage of O&M funds. The inspection revealed, particularly in Sindh, that although the channels inspected could be operated satisfactorily, their condition had deteriorated beyond what would be expected so soon after rehabilitation. The effects of rehabilitation only last about five years, after which the maintenance requirements of rehabilitated channels are similar to those of non- rehabilitated channels. Given inadequate O&M funding PID staff endeavor to maintain the head and middle reaches of channels at the cost of neglecting the tails. In addition these discussions revealed that inadequate funding from O&M was the cause of some of the defects observed, or was holding up desirable improvements. For instance, funds were inadequate to carry out the needed strengthening of the Alipur Branch right bank in Punjab although the bank had become weak and the berm had disappeared; O&M funding was reportedly inadequate to properly maintain the Nabisar Branch, the Doso Distributary, the Nabisar Minor, and the Detha Minor in Sindh (i.e. all four ISRP canals in Sindh). The NWFP canals were being adequately maintained. Breaches 11. Breaches were reported on only two out of the 15 canals surveyed: the 3-R Distributary in Punjab, the result of constructing the canal through fill; and the Mithiani Distributary in Sindh, the result of an inadequate number of control gates at the tail. The general problem of overtopping, which prevailed before the projects, seems to have been greatly reduced. In the past breaches were mostly due to non-availability of freeboard resulting in over-topping. The encroachment of freeboard was the result of pushing more water than design down canals. It is unclear how far the pushing of more water than design was the result of increases in supply as a result of Mangla and Tarbela storage, and how far the result of increased demand from farmers. Canal Discharge 12. The available data collected on canal discharges is not sufficiently complete to make any aggregate observations about whether the surveyed canals are being run within the design limits or whether they are now carrying more water than pre-rehabilitation, or just carrying it now without so much risk of breaching or overtopping. The survey revealed that the relationship between gauge levels and discharges (rating curves) is not known for many of the channels and that the rating curves are not properly developed, maintained and updated by the PIDs. Indeed at some of the control points there are not even any gauges against which to measure discharge. Thus for many of the channels the amount of water being discharged is not known: channels are run at full supply level as long as water is available. Equity in Distribution 13. One of the objectives of the rehabilitation program was to improve water distribution to tail watercourses. It is impossible to tell from the survey data how far this has occurred. It is 135 Annex F possible, however, to say that the distribution is still unsatisfactory on many of the canals surveyed and that tail watercourses still suffer from inadequate supplies. In Punjab, four out of the six canals surveyed report water shortages at the tails. In Sindh, for five out of the six canals surveyed, water was reaching the tails as designed. For the sixth canal (Detha Minor), an abandoned railway bridge at Tando Adam is blocking the flow and the tail reaches rarely receive any water at all. In NWFP, there are tail end shortages on the Machai Branch, but water reaches the tails on the other two canals. Annex F 136 Table F.1: Canals Observed by ISRIP Canal Design Head Discharge (Cusecs) PUNJAB ISRP 1. Alipur Branch (Muzaffargarh System) 937 2. 1-R Pacca Distributary (Muzaffagarh System) 237 3. Jurrian Distributary (Lower Chenab Canal System) 49 4. Dabbi Shah Minor (ex Upper Maggi Distributary) 60 CWMP 5. 3-R Distributary (ex Pakpattan Canal) 244 6. Pacca Dalla Minor (ex Shahkot Distributary) 6 SINDH ISRP 7. Nabisar Branch (Mithrao Canal System) 483 8. Sanwari Distributary (" " " ) 63 9. Doso Distributary (" " " ) 108 10. Nabisar Minor (" " " ) 64 CWMP 11. Mithiani Distributary (ex Sehra Branch, Rohri Canal System) 189 12. Detha Minor (ex Mithiani Distributary) 23 NWFP ISRP 13. Machai Branch (Upper Swat Canal System) 1723 14. Ismailia Distributary (ex Machai Branch) 64 15. Swabi Minor (Upper Swat Canal System) 6 137 Annex F Table F.2: Condition of Berm (Percent of Length) Canal Good/ Inadequate None PUNJAB 1. Alipur Branch 92 4 4 2. 1-RPaccaDist. 86 8 6 3. Jurrian Dist. 69 8 23 4. Dabbi Shah Minor 100 - - 5. 3-R Dist. 100 - - 6. Pacca Dalla Minor 100 - - SINDH 7. Nabisar Branch 100 - - 8. Sanwari Dist. 100 - - 9. Doso Dist. - 48 52 10. Nabisar Minor 100 - - 11. Mithiani Dist. 85 - 15 12. Detha Minor 100 - - NWFP 13. Machai Branch 97 3 14. Ismailia Dist. 100 - 15. Swabi Minor 100 - Annex F 138 Table F.3: Height of Freeboard (feet) Percent of Canal Length Canal Number 2.0+ 1.5 1.0 0.5 Nil PUNJAB 1. Alipur Branch 25 75 - - - 2. 1-R Pacca Dist. 12 - - 4 84 3. Jurrian Dist. - - - 23 77 4. Dabbi Shah Minor - 100 - - - 5. 3-R Dist. 19 22 26 - 33 6. Pacca Dalla Minor - - 100 - - SINDH 7. Nabisar Branch - - - 100 - 8. Sanwari Dist. - - - 7 93 9. Doso Dist. - - - 100 10. Nabisar Minor - - - 10 90 11. Mithiani Dist. - 75 25 - - 12. Detha Minor 22 - 31 47 NWFP 13. Machai Branch 8 23 20 - 49a 14. Ismailia Dist. 10 - 15 75b 15. Swabi Minor - 10 70 10b10 a. 32 percent in vertical cut mostly through rock and 17 percent nil freeboard. b. Reported as 0.75 feet. 139 Annex F Table F.4: Other Indicators of Canal Condition % of Canal Length Affected Canal Number Sediment Indication Weed Condition of Heap Due of Wind Growth Inspection Path Seep- for Silt Blown (% none or age Clearance Sediment very poor) PUNJAB 1. Alipur Branch Nil * Nil Nil Nil 2. 1-R Pacca Dist. 15 Nil 66 Nil 5 3. Jurrian Dist. Nil Nil Nil Nil Nil 4. Dabbi Shah Minor Nil Nil Nil Nil 45 5. 3-R Dist. Nila 5 Nil Nil Nil 6. Pacca Dalla Min. Nil 10 Nil Nil Nil SINDH 7. Nabisar Branch 75 100 Nil Nil 35 8. Sanwari Dist. 20 100 Nil Nil Nil 9. Doso Dist. 94 22 Nil 50 22 10. Nabisar Minor 10 100 Nil 100 6 11. Mithiani Dist. 5b Nil Nil Nil Nil 12. Detha Minor Nil 8 Nil Nil Nil NWFP 13. Machai Branch Nil 10 Nil Nil Nil 14. Ismailia Dist. Nil 22 Nil Nil Nil 15. Swabi Minor Nil Nil Nil Nil Nil * Trace. a. Some sloughing of side lining. b. Some sloughing at tail regulator. Annex F 140 Table F.5: Condition of Outlets Canal Number Proportion of Outlets Needing Repair (%16) PUNJAB 1. Alipur Branch 2. 1-R Pacca Dist. - 3. Jurrian Dist. 16 4. Dabbi Shah Minor - 5. 3-R Dist. 6. Pacca Dalla Minor - SINDH 7. Nabisar Branch 39 8. Sanwari Dist. 95 9. Doso Dist. 84 10. Nabisar Minor 94 11. Mithiani Dist. 12 12. Detha Minor - NWFP 13. Machai Branch 14. Ismailia Dist. - 15. Swabi Minor 15 141 Annex G Results of the Impact Evaluation Farmer Survey The Survey Sample 1. Table G.1 shows the classification and distribution of the survey sample. Table G.1: Survey Cells in A Three-Way Stratification Project Crop zone Province/ Size of Sample Supply channel Watercourse Punjab Sindh ISRP OFWM Cotton/Wheat 15 15 Rice/Wheat X - ICWMP CWMP Cotton/Wheat 15 15 Rice/Wheat 15 - none OFWM Cotton/Wheat 15 15 Rice/Wheat 15 15 ISRP none Cotton/Wheat 15 15 Rice/Wheat X - none none Cotton/Wheat 15 15 Rice/Wheat 15 15 - denotes cells that do not exist in the survey population. X denotes population cells excluded from the sampling frame. 2. The first two columns relate to the project characteristics, whether the type of intervention was in the supply channel above the mogha (the mogha is the outlet through which water passes from the supply channel controlled by an irrigation department into the watercourse, controlled by farmers), or within the watercourse. The IER covers four projects- the two On-Farm Water Management projects where the intervention was entirely within the watercourse below the mogha, the Irrigation Systems Rehabilitation Project where the intervention was entirely in the supply channel above the mogha, and the Command Water Management Project where the interventions were both in the supply channel and on the watercourse, that is above and below the mogha. For the purposes of the survey, the two OFWM projects were treated as one program. This thus provides four combinations of supply channel/watercourse intervention: none/OFWM, ISRP/none, ISRP/OFWM and CWMP/CWMP. With a control group (none/none) this becomes five possible combinations for analytical purposes. In this report the terms just used are applied to the location of farmers interviewed, as to whether they are supplied by canals/watercourses which were upgraded under the various projects, or not (none). Annex G 142 Watercourse Profiles 3. There is no single typical watercourse in Pakistan. There are distinct differences between the "average" Sindhi and Punjabi watercourses, for example. In the survey sample, Punjabi watercourses were much longer (5,701 meters) and had far more land holdings than their Sindhi counterparts (3,011 meters), and individual Punjabi land holdings were smaller (Table G.2).47 It was easier for the survey consultants to assemble groups of farmers to interview in Punjab than in Sindh, and Punjabi interview groups were larger on average. Impacts Measured by The Survey Table G.2: Survey Watercourse Profile, by Province Number of Average Average Average Average sample watercourse watercourse number of number in watercourses length command land owners survey Province (meters) (acres) interview NWFP 15 1,848 185 47.8 5.8 Punjab 120 5,701 509 52.5 8.5 Sindh 105 3,011 456 19.0 6.8 Total 240 4,224 454 34.2 7.3 Source: Pakistan IER, Farmer Survey. 4. The areas of impact targeted by the survey are detailed in the appendix. The aspects of the survey covered in this annex are the following: * water supply and environment * water use and cropping patterns * input use * labor use * institutional development * women's issues * public health 47. It is unclear whether the difference in watercourse length between Punjab and Sindh is more than merely a difference in definition. The IER team discovered that in Punjab, for purposes of calculating the proportion of the total length of watercourse eligible for lining, the length of watercourse included not only the primary branches but also all of the subsiduary channels between plots. In Sindh, on the other hand, the "total" usually included the primary branches only. As a result, for watercourses of similar physical size and a set percentage to be lined, those in Punjab received more lining than those in Sindh. 143 Afinex G 5. The survey findings relating to these areas of impact are discussed in the rest of this annex. Water Supply 6. The primary objective of these projects was to get more water to the farmers' crops. Farmers were asked questions both about what had happened to the amount of water in the supply channel (i.e. the minor or distributary on which the watercourse was situated) and its reliability and also whether they had experienced an increase in the amount of water in the watercourse or in the amount of water delivered to the farm (i.e. what had happened downstream of the mogha). Given a fixed warabandi (i.e. the amount of time per week which each farmer is allowed to use the full supply in the watercourse), such increases could come from any of three sources: more water through the mogha, increased delivery efficiency in the watercourse (by reducing transmission "losses"), or increased reliance on tubewells. Mission interviews had established that farmers have a keen appreciation of changes in water flows (both above and below the mogha, as might be expected. 7. Changes in water levels in supply channels. Only 12 percent of the farmers, when interviewed in groups, thought that the level of water in the supply channel had increased. Ten percent thought it had decreased, while the rest thought there had been no change.48 Ironically, almost all those who thought there had been an increase were in the one group which had had no supply channel rehabilitation (none/OFWM). The other three project groups and the control were virtually unanimous that rehabilitation had not changed the water level in the supply channel, and indeed 40 percent in the control group thought the level had decreased. When questioned individually, about 30 percent of farmers were of the opinion that the water level had increased, but the pattern of responses by type of project was different. In the ISRP/none group only 16 percent of individual farmers thought that the supply level had increased, whereas in the other groups this proportion was 30-40 percent. The response was also different between provinces. In Punjab not one of the farmers in the ISRP/none and ISRP/OFWM groups, and only around ten percent in the other groups, thought levels had increased. In Sindh 56 percent of farmers thought there had been an increase. 8. Since ISRP was not designed to increase the flow in supply channels, merely to eliminate losses through breaches, even if in practice it was used to allow higher discharges to be safely carried, one would expect the survey to show that ISRP had at least provided farmers with greater reliability in their water supply, if not actual increases. In fact, the ISRP/none farmers groups reported reliability increases at a lower rate (18 percent) than even the control group. Both Punjab and Sindh farmers agreed on this point when questioned as groups-the individual questionnaire showed that 30 percent of Sindh farmers thought that it had increased reliability, but not so in Punjab. By contrast with this finding, watercourses that received improv6ments both above and below the mogha reported much higher rates (about 63 percent) of increased supply channel reliability than did the control or none/OFWM watercourses (around 27 percent). One explanation for the apparent failure of projects to increase reliability could be that the direct 48. Water level was taken as a proxy for discharge since farmers had no knowledge of discharges; to the extent that desilting was a part of rehabilitation, there could be an increase in discharge without any change in the supply level. Annex G 144 benefits of ISRP were too dispersed to register with farmers since breaches had been spread over a wide area.49 Another explanation is that canal renovation does nothing to reduce variations in supply that result from erratic flows above the distributary head. These, not breaches, are the primary source of unreliability in flow at the mogha. 9. Changes in water supply at the mogha. Some of the more outspoken farmers interviewed by the IER mission commented that the projects gave irrigation authorities a "ready excuse" to visit watercourses and adjust moghas--always an opportunity to extract payments from farmers. However, according to the survey, the presence of the projects was associated with changes in the discharge from the mogha on a relatively small number of watercourses. Sixteen percent of project farmer groups had their sanctioned mogha discharge changed (almost equally divided between increases and decreases), and 33 percent experienced changes in their actual discharge. No control farmer group reported sanctioned changes and only 9 percent experienced actual changes (thus confirming farmers' comments to the mission noted above). About two-thirds of the changes were increased supply.50 10. Projects where the watercourse had been renovated were more likely to have their mogha discharge increased than were the control or ISRP/none watercourses. And the likelihood of an increase in discharge was greater still in projects where both the supply channel and the watercourse had been improved. Despite this, the frequency of a decreased discharge was greatest on the CWMP watercourses. The below mogha projects were not intended to rebuild moghas, but either that is what was done, or these farmers have taken the opportunity to tamper with their outlet from the supply canal. Tampering, on the basis of the survey, does not appear to be widespread. Only 21 percent of project watercourses reported an increase in discharge at the mogha; and surprisingly, it was reportedly much more widespread in Punjab (36 percent) than in Sindh (four percent). 11. Changes in water supply to the farm. Although 21 percent of project watercourses reported an increase in the mogha discharge, 50 percent thought that the water supply in the watercourse had increased (the other 50 percent thought it had not increased). The difference would be from the combined effect of reduced conveyance losses in the watercourse and increased use of tubewell water. More farmers in Sindh (57 percent) thought there had been an increase than in Punjab (37 percent) largely because no farmers in the ISRP/ none group thought there had been an increase. Even one-third of the control watercourses thought the watercourse supply had increased. These figures are the consensus of interviews with farmer groups. When interviewed individually, a rather lower proportion, 40 percent, of project farmers, but about the same proportion of control farmers, thought the water supply in the watercourse had increased. 49. See Gleason, J.E. and Wolf, J.M Evaluation of the Impact of Canal Rehabilitation on Hydraulic and Agricultural Economics Indicators, Irrigation Systems Management Project Phase II, USAID, Islamabad, Pakistan, April 1993. 50. The monitoring results of both the USAID'pilot water management project and the Bank's OFWM-1 found that discharges in the supply channel, and hence through the mogha, decreased after watercourse renovation. 51. This is quite the opposite of the conclusions of the mission. During the field visits, the majority of watercourses seen by the mission had tampered moghas, particularly in Sindh. 145 Annex G 12. If watercourses reported increased water to the farm while reporting no increases in mogha discharge or in tubewell use, increased conveyance efficiency must have been the only source of the incremental water. The share of watercourses meeting this strict set of criteria was quite small (22 percent), nor was it significantly different from the control group's share for either the none/OFWM or the CWMP/CWMP group. Only for the ISRP/OFWM group was the difference significant. It is difficult to interpret the reasons for such a finding, particularly given the previous finding that ISRP did not increase water reliability in the absence of OFWM. Whatever the reason, the conclusion must be that, at best, only half the beneficiary farmers felt that they were getting more water as a result of the projects, and that only a proportion of this was the reslIt of reduced conveyance losses (the main project objective). 13. One unexpected result which the survey throws up, and which in the farmers' views has a major impact on the amount of water saved by renovation, is the great variability in the proportion of the watercourse which was lined. The amount permitted according to the project documents varied both between projects and between fresh and saline groundwater areas, usually up to a maximum of 30 percent. The survey found that the amount lined was as high as 59 percent; indeed the mission was told by OFWM staff in Sindh that they had been ordered to line 100 percent of some channels. The OFWM-I criteria were to line 15 percent in fresh groundwater areas and 30 percent in saline groundwater areas. The survey found that in Punjab there was no significant difference between the proportion lined in each type of area, it was 16 percent in fresh and 18 percent in saline. In Sindh there was a significant difference, the proportion lined being 22 percent in fresh and 32 percent in saline groundwater areas. The difference between provinces is the result of a definitional difference, that in Punjab the length of a watercourse includes the subsidiary channels between plots, whereas these are not included in Sindh: thus for any given amount of lining, the proportion will appear higher in Sindh. Two figures below show, first the percentage of watercourses receiving different lining shares, and second, the relationship between the proportion of lining and the reported percentage of watercourses receiving more water. The second figure would tend to suggest that beyond 25 percent lining, the additional amount of water decreases up to about 35-40 percent, and then begins to increase again (S-Sindh, P-Punjab, N-NWFP, SPN-all three provinces combined). Annex G 146 Figure 4: Distribution of WCs By Lining Share 60% 50% 40%.- 30% / 20% * *1 10% 0% 0-10% 10-20% 20-30% 30-40% >=40% Lining Share .-. SP Figure 5: Impact of IUning on Water to Farm 100% . 80% S60% N -- 20% 0%-1 0-10% 15-20% 25-30% 35-40% Lining Share .......S -.-.-P SPN 14. Tubewells. 41 percent of survey watercourses renovated under the project reported increased reliance on tubewell water, in other words much of the increase in water reaching the farm on project watercourses is associated with increased tubewell use. In addition, every control farmer group reported an increased reliance on tubewells. Since no farmer in the control group reported any lining of the watercourse, and only 6 percent thought their mogha discharge had increased, the bulk of the 33 percent of the control watercourses reporting increased water for the farm must be getting their additional water from their reported increase in the use of 147 Annex G tubewells.52 The pattern is nearly identical for the ISRP/none group except that no watercourse in this group reported a change in actual mogha discharge. 15. The aggregate response of greater reliance on tubewells masks an important provincial difference and an important difference related to groundwater quality. With its relative surplus of surface water, Sindhi farmers reported a decreased reliance on tubewells to supplement surface water.53 Farmers in Punjab, however, report a nearly universal increase in tubewell reliance where watercourses have not been renovated, and roughly three-fifths reported increased tubewell use after watercourse renovation. No increase in tubewell use after renovation indicates that in all probability watercourse renovation is substituting for tubewells as a source of additional water, particularly in Sindh-precisely the reaction which the Revised Action Program anticipated and wished to avoid by restricting renovation to saline groundwater areas. (However, the assumption that substitution occurs, and that this is undesirable, is an over simplification as discussed in the main text.) 16. The greater reliance on additional tubewell water in Punjab compared with Sindh is not unexpected, given the greater density of private tubewells in Punjab, the larger proportion of the area which is underlain by fresh groundwater and the relative scarcity of surface supplies. More detailed examination of the survey data, however, reveals some strange anomalies. The first anomaly is that the distribution of renovated watercourses by quality of groundwater bears no relation to the very different distribution of groundwater quality between provinces. The design of the OFWM projects, following the recommendations of the Revised Action Program, lvas to give priority to watercourses in areas underlain by saline groundwater, since these had greater need both for additional surface water and for reduction in seepage. The selection of watercourses for inclusion in the survey was done without reference to groundwater qualith' and so should provide an unbiased indication of the distribution of watercourses by groundwawr quality in the OFWM projects as a whole. Table G.3, shows this distribution in comparison with the estimated distribution, by area, of fresh and saline groundwater zones in the provinces as a whole. This leads to the conclusion that there was a substantial bias in the selection of watercourses for renovation in Sindh towards those with fresh groundwater. 52. The survey provides contradictory data on the control watercourses use of tubewells; on the one side there is a unanimous response that the number of tubewells and the hours tubewells are used have increased. On the other hand, only one-third report an increase in water supply to the farm. Where does the tubewell water go? 53. There were no valid responses to the question whether there was greater reliance on tubewells for farmer groups on Sindhi control watercourses, but to the question whether the working hours of tubewells had changed as a result of the project, only five percent of watercourses in the Sindh project groups reported an increase and 55 percent reported a decrease. Annex G 148 Table G.3: Watercourse Distribution by Groundwater Quality (percent) Groundwater Quality (by area) Groundwater Quality Reported in Survey (by number of watercourses) Fresh Saline Fresh Saline Punjab 79 21 36 64 Sindh 29 71 73 27 Average 59 41 47 53 17. It would appear from these figures that in Punjab watercourse selection did follow the project criteria to a great extent by favoring watercourses in areas underlain by saline groundwater.54 18. Not only does there appear to be a bias in Sindh towards the selection of watercourses with fresh groundwater, there also appears to be a bias in the selection of watercourses which already had a tubewell. Only 47 percent of the unrenovated watercourses (i.e. the control and ISRP/none groups) in the survey which had fresh groundwater had a tubewell, compared with 67 percent of the OFWM watercourses and 95 percent of the CWMP watercourses. This would tend to indicate that watercourse selection was biased towards influential farmers, since these would be most likely to have a tubewell in Sindh: a finding in accordance with the mission's own observations. It also suggests that such farmers valued increased surface flows to their farms, even though reduced seepage would increase their pumping costs as watertables fell. 19. The second anomaly is the surprising number of watercourses reportedly underlain by saline groundwater, but which had a tubewell. None of the watercourses in the survey underlain by saline groundwater which had not had the watercourse renovated (i.e. the ISRP/none and control groups) had a tubewell at the time the project commenced, while 48 percent of the OFWM watercourses and 77 percent of the CWMP watercourses underlain by saline groundwater had tubewells at that point. In response to the question whether the number of tubewells had increased during or after the project, none of those underlain by saline groundwater, and which had not had the watercourse renovated, reported an increase in tubewell numbers, while 40-50 percent of the renovated watercourses in saline areas reported an increase. There are two possible conclusions from these data: either that there was a tendency for the OFWM Directorate staff to record watercourses to be renovated as having saline groundwater, even when it was fresh, as this would avoid questions being asked about the reasons for the 54. All Punjab watercourses in the survey which were not renovated (i.e. ISPR/none and none/none) are excluded from this analysis because data on groundwater quality were not collected. The Sindh control watercourses are divided 72 percent with saline groundwater and 28 percent with fresh groundwater, which is close to the actual groundwater areas for Sindh. This reinforces the conclusion that watercourse selection for improvement was heavily biased towards fresh groundwater areas. Sindh authorities are invited to comment on this point. The mission has been unable to locate any analysis of watercourse improvement by groundwater type, although it should be possible to make one using OFWM Directorate records. 149 Annex G selection and it would allow them to line a higher proportion of the watercourse than if the groundwater was reported fresh; or that farmers are pumping water of higher salinity than is advisable on agronomic grounds. Either way, these watercourses were underlain by groundwater that was being used for irrigation and they were selected for renovation at the expense of those in areas where the salinity was too high for irrigation use, and hence where the need to control seepage was much greater. 20. Waterlogging and Salinity. The Revised Action Programme also recommended concentration of renovation initially on saline groundwater areas because water lost through seepage would not be recovered through pumping by private tubewells, while water lost in fresh groundwater areas could be recovered, so that watercourse renovation to prevent losses was less urgent, even undesirable if the aquifer is treated as a valuable over-season storage. The survey throws some light on project impacts on waterlogging and salinity, at least as perceived by farmers. The responses of the control groups showed up well the marked difference between Punjab and Sindh Provinces. In Sindh, every farmer group response on control watercourses reported increased waterlogging and a rising watertable. This bears out the frequently voiced contention that Sindh is getting more water than it can currently use efficiently, indeed that in some areas incremental supplies are harmful. In Punjab, every control farmer group reported no change in waterlogging, and two out of five reported that the watertable was dropping (in fresh groundwater zones this would indicate over-pumping). Almost every Sindhi control farmer group reported increased salinization, whereas not one of their Punjabi counterparts did so, and 60 percent of the Punjabi control farmer groups reported decreased saline areas. 21. In Sindh, on all projects except the ISRP/none, waterlogging and salinization does appear to have been reduced, but only in a limited way. On one quarter of the project watercourses farmer groups reported a decrease in areas waterlogged (an increase in the depth to the watertable) and almost one-third reported a decrease in saline areas. The projects had much greater impact on waterlogging in Punjab, where on average 80 percent of farmer groups reported a decrease. The projects had no reported impact on salinization in Punjab, and in neither province do the projects appear to have much impact on the respective provincial trends in watertable depth. There is no significant difference in the pattern of changes in Watertable depth between areas with saline or fresh groundwater. There is, however, in the extent of salinity. In Punjab, for both saline and fresh groundwater areas, over 80 percent of farmer groups reported that salinity either decreased or remained the same; in Sindh fresh groundwater areas 72 percent of groups reported similarly, but in saline groundwater areas in Sindh 93 percent of groups reported that salinity had increased, presumably because of the reduction in seepage-of comparatively fresh water. The finding was similar for both project and control watercourses even though the control watercourses were not renovated. It is difficult to explain why these control watercourses should report an increase in groundwater salinity since none of them had a tubewell and therefore had no first-hand knowledge of what was happening to groundwater salinity. In any case, in areas where the groundwater is already too saline to use for irrigation, an increase in salinity will not have any short-term impact on productivity. 22. Drinking water quality. An improvement in water quality was expected as a positive impact resulting from either increased water flow or more regular supplies. At the same time, though, a reduction in seepage of relatively fresh water in areas with saline groundwater could Annex G 150 have the opposite impact. Though the farmer survey asked six questions about changes in water quality, the responses followed conflicting patterns. For example, while the ISRP/OFWM group reported the highest rate of drinking and tubewell water turning "fresh" (68 and 83 percent), this group ranked second lowest only to ISRP/none in the percent of responses rating drinking or tubewell water "improved" (ten and eight percent). And while the none/OFWM and CWMP/CWMP groups had drinking and tubewell water turning fresh or saline in rates similar to the control, (except in Sindh saline groundwater areas where over 90 percent of farmers thought that the quality of tubewell water had deteriorated), these two groups had drinking water quality "improvement" rates significantly higher than the control (both had 32 compared to 15 percent). It may be that the positive responses to drinking water quality were driven by the improved clarity of watercourse supplies, to the extent that drinking water is drawn from surface supply, although as there was no difference between head and tail responses, there is slender evidence to support this. Water Use and Cropping Patterns 23. Increased water supply in the watercourse was expected to result in increased agricultural production. The survey asked a number of groups of questions about this aspect. It asked first whether the projects had increased the reliability of the water supply in the watercourse. The answers to this question have already been described above-namely that though supply channel reliability generally improved in both provinces, except for the ISRP/none watercourses, which reported a large number of cases of decreased reliability, watercourse renovation under the projects resulted in much more frequent improvements in reliability. It then asked, if the projects had increased reliability, whether this had increased their willingness to take risks, e.g. by changing cropping patterns to more water-intensive crops. The results of the survey indicate a strong willingness to take greater risks: 46 percent of project farmers reporting they would be willing to do so, compared with 12 percent of control farmers. The third question in this group asked whether the projects had affected the risks of farming that were outside the farmers' control. In Sindh, 55 percent of farmer groups , except the ISRP/none groups, thought they had increased these external risks. In Punjab only three percent thought they had and many (25-50 percent depending on the project) reported decreases. Unfortunately the survey did not probe further and inquire why they felt this way. 24. The second group of questions focused on what farmers had done with the extra water: whether it had been used to cultivate a larger area, to grow more water-intensive crops or to increase the delta on an unchanged area and crop mix (the least risky option). The survey found that neither the project nor control farmers were likely to increase the delta and to apply incremental water onto crops already receiving irrigation. They differed in what else they would do with it. Project farmers with extra water were much more likely to report spreading the water on more land (particularly in the rice-wheat zone), while control farmers were more likely to grow more water-intensive crops. This differs slightly from the IER field mission's observations that nearly all project farmers reported a shift into higher value crops, which were also more water-intensive, typically cotton, sugarcane or orchards. This may be the result of a bias in the mission's observations since the survey shows a tendency towards more water-intensive crops in the cotton/wheat zone. These survey findings on water use are confirmed by farmer group reports elsewhere in the survey that farmers shifted into fodder (primarily during kharif season), 151 Annex G vegetables, sugarcane (except NWFP), and "other kharif' season crops, largely by reducing fallow. These cropping shifts are likely to be the result of the combination of additional water and greater reliability of supply. Input use 25. Whatever changes took place i-nTput use were the result of the farmers' own initiatives. Although some provision was made in the projects for continued agricultural extension after project investments were completed, most project watercourse groups reported a decline in extension contact. 26. The survey asked farmer groups explicitly how changes in water supply affected their use of seeds, fertilizer, pesticides, ahd the amount of water per crop,. As regards seeds, the most interesting finding is that, for reasons which are unclear, farmers on project watercourses in the Sindh (unlike those on control watercourses) have shifted from using certified seeds to using non-certified seeds. They reported that their shift was due to changes in their water supply, irrespective of whether the water they received at the farm has gone up or not. In contrast, farmers on almost every control or project watercourse in Punjab with increases in farm water have increased their use of certified seeds. The difference between Punjabi and Sindhi demand for certified seed could be explained by quality differences in the certified seed available in the provinces."s The reasons for the difference between Sindhi project and control groups' demand cannot be explained in this way. 27. Given the strong trend in Pakistan toward greater use of pesticide and fertilizer, any project extension message on their use had little impact on project farmer groups' reported increase in fertilizer and pesticide use although 90 percent or more reported an increased use of fertilizers and pesticides, for reasons largely unrelated to water supply. There was no significant difference between project and control watercourses in this increase. But it is notable that all of the watercourses in the sample that reported an increase in water to the farm, but no increase in fertilizer or pesticide use, were in the Sindhi control group. This is weak evidence that the projects did have an impact on the agricultural message Sindhi farmers received. Labor Use 28. The labor required to direct, manage, and maintain watercourses was dramatically reduced after renovation. The farmer survey found that farmers on over two-thirds of renovated watercourses in both Sindh and Punjab reported decreased irrigation-related labor, while virtually no farmer in. either the control or ISRP/none groups reported any such decreases. On the contrary, on over half the control or ISRP/none watercourses in Sindh, farmers reported the labor required for watercourse maintenance increasing. And over half the ISRP/none watercourses reported an increase in the amount of labor required to guard and maintain the 55. See OED, Impact Evaluation Report No. 14761 -IN, June 30 1995. 56. A similar finding has been made in other studies: see Byrnes, Dec., 1992 and Planning and Development Department, Government of Sindh, April 1993, op. cit. Annex G 152 water. This is probably more a reflection of a general increase in surface flows in Sindh than of any project-related impact. 29. While labor for irrigation may have been decreased, overall labor use appears to have increased, probably a good indicator that production intensity was increasing as a result of the project. The majority of farmers reported increases in labor requirements for land preparation, harvesting and marketing. There is an important difference between provinces on the source of the extra labor: in Punjab, only about 20 percent of farmer groups reported an increase in the use of family labor, the rest reporting that it either remained the same or decreased; the use of hired labor, on the other hand, increased widely, 90 percent of the control and 77 percent of the project watercourses reporting an increase; in Sindh, by contrast, around 80 percent of farmer groups reported an increase in family labor, (presumably under the predominantly tenant farming system the use of hired labor is largely confined to landlords), while the use of hired labor was reported to have increased on just over half the control but only 17 percent of the renovated watercourses. Institutional Development Impacts 30. Watercourse User Associations (WUAs). The projects created large numbers of water user associations (WUAs) as part of the process of renovating watercourses, but did little to encourage their sustainability after project completion. In fact, because the renovation projects created so many temporary WUAs that became inactive immediately after receiving project subsidies, there is evidence that the projects' limited use of the organizing opportunity may have adversely affected the long-term potential of using WUAs as a vehicle for grassroots development. The field mission found credible evidence that the activities of social organizers under the CWMP project (financed under the USAID component) had increased the sustainability of WUA achievements. Unfortunately there is no support for this finding from the farmer survey. There were no differences between how farmer groups and individuals reported on WUA impacts other than a slightly higher latent Sindhi level of WUA activity. 31. WUA sustainability. There is an interesting difference in the sustainability of WUAs (or their non-project equivalent) formed voluntarily on control watercourses and compulsorily on project watercourses: none of the control group WUAs were reported dormant but 42 percent of project group WUAs were. The pre-existing non-project khal committees were more likely to have regular meetings than WUAs on project watercourses (31 versus 16 percent). The interest and membership in WUAs fall after project completion, perhaps because the demands for and justification of the organization were less after watercourse lining because most W-UAs do watercourse maintenance and little else, and labor requirements for maintenance were dramatically reduced under OFWM & CWMP. Even with this reduced labor input project watercourses were judged by those conducting the survey to be in better condition overall than control watercourses, with 50 percent of project watercourses being in good condition against only 20 percent of control. Maintenance, however, was of variable quality: one-third of both control and project watercourses were reported as in poor condition/needing major maintenance. 57. WUAs on project watercourses were not any more likely to have started new activities than those on control watercourses although 21 percent reported that they had undertaken some further renovation/improvement of the unlined portions after project completion. 153 Annex G 32. Grassroots leadership. The survey provides evidence that the WUAs created under the projects were as dominated by influential farmers as those either already existing or created on control watercourses, particularly in Sindh (see Figure 6). When asked where on the watercourse the important persons in the WUA lived, 54 percent of project fanner groups replied "at the head" and 25 percent "at the tail" compared to 19 percent and 69 percent respectively for control groups. This finding suggests that the benefits of investment are not evenly distributed along a watercourse. Even if water savings benefits are distributed all along the watercourse (which is doubtful given that most of the lining was in the head reaches), the projects clearly had the effect of concentrating local power and status toward the head reaches. Figure 6: Location of FO Leaders 100% 80% TAIL 'l 60% MIDDLE - .o 60. ". 40% ]HEAD 20% 0% CWM/ OFWM/ OFWM/ none/ none/ CWM ISRP none ISRP none Source: IER Farmer Survey Data 33. The IER field mission heard many agriculture officials claim that the representative and democratic structure of WUAs was a positive attribute of the watercourse projects. Given the above survey results and the findings of other Bank work, this claim appears unfounded.58 34. Farmer satisfaction with WUAs. The reduction in water disputes was the benefit of watercourse renovation projects mentioned most frequently to the field mission. Almost all farmers on Punjab and Sindh watercourses affected by projects below the mogha (i.e. OFWM and CWMP) reported disputes having decreased (81 to 88 percent) while only 2 percent of ISRP/none farmers reported decreased disputes. Even the control group had had some farmers (23 percent) reporting decreased disputes, but the much higher proportion on project 58. Meinzen-Dick, Reidinger and Manzardo, Participation in Irrigation. Environment Department Papers, Participation Series Paper No. 003, The World Bank, February 1995. Annex G 154 watercourses is strong evidence that watercourse lining greatly reduces arguments over water and suggesting that supply channel improvements, without watercourse renovation, have the perverse effect of increasing disputes. This reduction in disputes must also be the reason why so many farmers reported that labor requirements for guarding and maintaining the watercourse when they were actually irrigating had gone down. The answers were identical between provinces. This confirms that the reduction in disputes had a real economic impact beyond a more general change in social relations. The survey of individual farmers confirms this as a popular impact with 34 percent of OFWM and CWMP farmers with active WUAs responding that the WUA "decides disputes well" as their reason for satisfaction with their WUA, compared to 15 percent in the control group. However, the farmers with active WUAs in the none/OFWM and ISRP/OFWM groups differed from the CWMP farmers by reporting that the WUA "works for the benefit of farmers" as their number one reason for satisfaction with their WUA ( 40 compared to 24 percent). 35. Government agencies. In general, conflict may arise between government development agencies and intended beneficiaries when government changes policies, reallocates resources, or forces locals to bear the cost of externalities. But the design of these projects imposed little on farmers, and participation by local WUAs was voluntary. In the case of the watercourse projects, nothing was to be forced on farmers; they could elect to undertake subsidized investments and form a WUA or to not participate at all, and not have to change their existing arrangements for watercourse maintenance. And in the case of projects concerned only with upstream development above the mogha, the only farmer-related objective was the increased reliability of surface water supply. In this context, the survey results on farmer views of their relationships with government agencies are striking. 36. Individual project farmers in Punjab generally reported a worsening of their relationships with all three government agencies covered by the survey: the Departments of Irrigation and Agriculture and the OFWM Directorate. Sindhi farmers were much more likely to report no change in their relationships with government and to report improvements when reporting change. This is a reflection of the differences in land tenure and social systems between the two provinces. In Sindh most interaction with officials would be through politically influential zamindars with whose demands the local officials could not argue and who are usually able to influence water distribution arrangements, by one means or another, in their own favor. In Punjab, farmers usually had to obey whatever instructions were handed out by departmental officials. Figure 7 illustrates the diverging experiences of Punjabi and Sindhi farmers. 155 Annex G Figure 7: Farmer Relations with Gowrnment Agencies 100% ___________________________ 80% PID PAD OFWM S 60%.- a Sindh 40% Sndh Sindh 20 PunjabP Pu 20% - ...... -20% o -40% u -60% A -80% -100% o Improved * Worsened Source: IER Farmer Survey Data Notes: PID = Provincial Irrigation Department, PAD = Provincial Agriculture Department, OFWM = Provincial On-Farm Water Management Directorate. Since many farmers report "no change," single columns do not sum to 100 percent. 37. OFWM project farmers were most likely of all project farmers to report changes and exhibit inter-provincial differences in their relationships with government.59 When Punjabi farmers from all projects were pooled together, they were more likely to report a worsening in their relationship with their provincial Irrigation Department than with their Agriculture Department. By either action or inaction, the Punjab project agencies disappointed Punjabi farmers. Women's Issues 38. On 152 watercourses groups of women were interviewed with a separate questionnaire, the watercourses being selected at random from within the main sample. No women farmers were found to interview separately. 39. This survey found one surprising change in the workload of women, though it does not appear to be brought about by the project-the amount of time devoted to "other business" labor: 95 percent of control women's groups reported a change in this (and 72 percent were increases), but only 38 percent of project women's groups reported a change, of which 37 percent were increases. It would appear that the projects provided enough additional work to provide employment at home. This would be borne out by the survey results, which show a large propoirtion of project women reporting increased demand for agricultural labor (three quarters reporting increased demand for harvesting labor, about half for more labor in weeding, crop 59. The provincial OFWM Directorates were staffed by and reported to the provincial Agriculture Department. Annex G 156 processing and storage, and about one-third for looking after livestock), but for the fact that the control groups have even larger percentages reporting increases in these activities. This could be attributable to greater reliance on tubewell water in the control watercourses, but there is insufficient evidence to substantiate this with certainty. One shred of evidence is that more women on control watercourses report that the time to collect drinking water has decreased (27 percent) than do women on project watercourses (16 percent). 40. An engineering oversight was highlighted by the women's descriptions of the conditions for crossing watercourses and supply channels. Of the 50 Punjabi women's groups reporting a change in the conditions for crossing supply channels, 43 reported a worsening. Only 19 Sindhi women's groups reported a change, but 11 of them reported improvements. The Punjab Irrigation Department may have been less sensitive to local needs for bridges and other crossings than the Sindh Irrigation Department. Women on control and OFWM watercourses reported significantly less worsening of supply channel crossings than CWMP women, presumably because there were no changes to supply channels in the control and OFWM projects (15 and 30 versus 70 percent), but also suggesting that the CWMP may have ignored local concerns about bridges. The mission noticed cases in the field, when being shown CWMP lined supply channels, that farmers had built their own bridges to supplement the pukka bridges constructed by PID, which were too far apart to be convenient. Farmers reported that lined channels also prevented animals using them for drinking, and when they fell in they could not get out again. Only one-fifth of the women respondents reported that conditions for watering animals had improved. 41. Women in the OFWM and CWMP groups reported significantly higher improvements in watercourse crossing conditions than did women in the control and ISRP/none groups (88 and 100 versus ten and zero percent), suggesting another positive impact on the quality of life from watercourse improvements. But contrary to the finding for supply channel crossings, it was the Sindhi women on the ISRP/none watercourses who reported the worsening of watercourse crossing conditions. Again, this was almost certainly due to the fact that these projects did not renovate watercourses, but lined supply canals. Public Health Impacts 42. Malaria. The women's survey confirms earlier findings about rising watertables and improved drinking water quality, but also unveils other significant project impacts. While the sample as a whole reports increases in mosquito population and the spread of malaria, women on the OFWM and CWMP watercourses report significantly fewer such increases (40 and 34 versus 67 percent on control watercourses) and higher numbers of decreases (27 and 19 versus seven percent). This is not surprising given the dramatic reduction in watercourse right of way and increase in water velocity after watercourse improvements as well as the reduction in spillages and seepage which lessened the area of standing water (58 percent of project women report reduction in water ponding around the watercourse, but so do 67 percent on control watercourses). The women had a similar pattern to their responses about the population of houseflies (24 percent of project women reporting a decrease versus seven percent of control), which may be interpreted as an improvement in a more general indicator of community health. 157 Annex G 43. Splitting the women's group responses along provincial lines reveals that these improvements in public health were limited almost exclusively to Punjabi watercourses. Less than six percent of Sindhi women's groups reported improvements in the mosquito/malaria situation. The linkage of mosquito breeding to standing water suggests that Sindhi watercourses were getting more water than they use healthily-a finding consistent with the observation that Sindh was getting more surface water than it can handle. 44. The watercourse projects had a positive if limited impact on community hygiene through improvements in conditions for bathing babies and washing clothes and utensils. In Punjab, for instance, about 50 percent of women respondents in the OFWM and CWMP watercourses felt that they had benefited from improvements in baby bathing facilities, significantly higher than the control group or the ISRP none (13 percent and nil), although this is not surprising as neither of these groups provided such facilities. In Sindh, 75-100 percent of women felt that they had benefited on project watercourses (again with the exception of ISRP/none). Similarly high figures were reported for clothes washing facilities: 60 percent in Punjab and 75-100 percent in Sindh. Evidence from the field mission and other secondary sources confirms that women became fond of the pukka nuccas as washboards. Only between one-third and one half felt that conditions for washing utensils had improved. 159 Annex G Appendix 1 Survey Design Considerations Unit of Observation While the ISR and CWM projects improved the conditions of supply channels, the CWMP and OFWM projects addressed specifically the more local needs of watercourses. Each watercourse typically serves from 20 to 50 farmers and the impacts of the projects were expected to differ on farmers clustered either at the head or tail reaches of watercourses. So based on the JER field mission's visits to 85 watercourses and the input from Pakistani consultants who conducted the survey, the survey was designed to test hypotheses with three different units of observation. Sample Frame Project impacts were thought to have varied along three major dimensions: province, agro-climatic cropping zone, and type of project intervention. The sampling frame for the survey waF designed to provide cost and statistical efficiency by stratifying the population into cells along the three dimensions. Since Punjab and Sindh contain over 80 percent of the watercourses in Pakistan (see Table G 1.1), the survey analysis was restricted to these two provinces. A small number of watercourses in NWFP were sampled for descriptive purposes only. Table G1.2 illustrates the three-way stratification of the population and the design of the sample frame. Since group and individual interviews were carried out at each reach of each watercourse, watercourse reach serves effectively to stratify the observations along a fourth dimension. Annex G 160 Appendix 1 Table G1.1: Watercourse Distribution in Pakistan Punjab Sindh NWFP Baloch Total Watercourses in Pakistana 55,000 40,000 10,000 2,000 107,000 Watercourses renovated by all donorsb 14,219 5,646 3,354 1,611 24,830 Watercourses renovated under:cd OFWM I 2,440 396 279 125 3,240 OFWM II 3,262 890 517 488 5,157 CWMP 790 509 87 73 1,459 Total 6,492 1,795 883 686 9,856 Sources: a. Taken from the SAR for OFWM III. b. Based on 20 projects funded by 7 donors, as reported in the OFWM III SAR. Figures here incorporate numbers of watercourses reported in the OFWM 1, OFWM II, CWMP, and SCARP Transition Pilot Project PCRs. c. It is unclear to what extent individual watercourses have been improved under more than one project, so the actual numbers of renovated watercourses may be lower. d. Taken from respective PCRs. 161 Annex G Appendix 1 Table G1.2: Survey Cells in a Three-Way Stratification Project Crop zone Province Supply channel Watercourse Punjab Sindh ISRP OFWM Cotton/Wheat * * Rice/Wheat X - CWMP CWMP Cotton/Wheat * * Rice/Wheat * none OFWM Cotton/Wheat * * Rice/Wheat * * ISRP none Cotton/Wheat * * Rice/Wheat X - none none Cotton/Wheat * * Rice/Wheat * * - denotes cells that do not exist in the survey population. * denotes population cells included in the sampling frame. X denotes population cells excluded from the sampling frame. The Punjab-rice/wheat-ISRP cells were thought to contribute less to the analysis than additional observations in the remaining cells, so the two cells were dropped from the sampling frame. For the purposes of the two province/two crop zone hypotheses, the remaining cells in effect describe the survey population and the sample frame. Sample Size Nearly 10,000 (9%) of Pakistan's 107,000 watercourses were improved under three (OFWM I and II and CWMP) of the four projects covered by this IER. The survey sample covers the head and tail reaches of 143 of these watercourses, plus of 30 others affected by the fourth project (ISRP) and 67 unaffected watercourses (i.e. in the control group). On each reach of each watercourse, one group of men and three separate individual farmers were interviewed. On 152 watercourses, a group of women was interviewed with a separate questionnaire. The distribution of group and individual interviews for the 240 watercourses by province and cropping zone is in Table G1.3 below. Annex G 162 Appendix 1 Table G1.3: Watercourses Sampled by the Impact Evaluation Supply Water- Punjab Baloch- channel course Sindh NWFP istan Total In the Farm Surveya CWMP CWMP 30 15 - - 45 ISRP OFWM 15 15 - - 30 ISRP none 15 15 - - 30 none OFWM 30 30 8 - 68 none none 30 30 7 - 67 Total 120 105 15 - 240 By the IER mission 43 25 7 10 85 a. On each watercourse, two groups and six individuals were interviewed separately. The sample size was determined jointly by the number of cells and a minimum analytical subsan-ple size of 30 watercourses. For the purposes of most hypotheses, the analytical subsample is two or more cells, so that a cell size of 15 provided subsamples of 30 or more watercourses (and 60 or more group interviews and 180 or more farmer interviews). Thus the 15 sample frame cells in the above table required 225 watercourse observations, while another 15 watercourses in NWFP were sampled for descriptive purposes only. Sampling The relatively small number of supply channels affected by the projects lead to a purposive but representative selection of supply channels for the ISRP and CWMP project strata. All watercourses were selected randomly from within the first and last 20 percent of the watercourses on their given supply channels. Group interviews were open to all qualifying farmers at the respective watercourse reach, and both reaches were covered for each sample watercourse. Individual farmers at each reach were selected randomly, subject to no more than one of the three being a member of the decision-making committee of his village. For group interviews of women, watercourses were selected randomly from within cells and all qualifying women could participate. No women farmers were found to interview individually. To qualify 60. The minimum analytical subsample size (N) was based on calculations suggested by Casley and Kumar (1988), using the formula: N = K * K * R * (100- R) / (D * D); D = acceptable margin of error (10%), K = normal deviate for desired one-sided confidence (1.28 for 90%), and R = best guess of indicator (80%). An additional 4 observations were included to allow for discarded outliers. 163 Annex G Appendix 1 for either group or individual interviews, men and women must have been old enough to have been farming "before" the project, as defined at each watercourse. Other Design Issues Most information about project impact was collected through "before" and "now" questions, which provided a farmer-recall-dependent built in pre-project control. But as illustrated above, three different "control" groups were defined and sampled to provide a variety of without project subsamples Impact Indicators The survey attempted to collect farmers' perceptions of impacts directly. The indicators included in the survey are intended to reveal qualitative directions of change in selected areas of project impact without incurring the cost of a more comprehensive socio-economic investigation. This design strategy limited the IER's descriptive narrative of project beneficiaries, but it enabled a more focused investigation of project impacts. The areas of impact targeted by the survey were: Cropping patterns and yields Farm technology input use precision land leveling water use increased delta increased area on-farm labor requirements Stability farming risk water reliability timing of planting Environment watertable groundwater quality Community relations traditional versus formal farmer organizations nature of leadership, and participation joint activities (e.g. cleaning watercourse, purchasing tractor) disputes land tenure farmer relations with government agencies Women's perspectives same issues as above for female landowners Annex G 164 Appendix 1 participation in farmer organizations work pattern and load family health environment Data Collection A Pakistani firm collected the survey data through structured group and individual interviews. In each of the three covered provinces, the field team was comprised of a senior field coordinator and one or more interviewers, all of whom were conversant in the local languages. Field coordinators collected watercourse and supply channel lists from project supervising and implementing agencies to assemble the sampling frame. Care was taken to ensure that watercourse selection was independent of influence by agencies providing the lists. After sample selection, field coordinators returned to the respective agencies to collect basic watercourse data, creating watercourse profiles for use in the field. Field group interviews were conducted by an interviewer who obtained a consensus before recording answers onto pre-coded survey instruments. Female consultants were hired to interview the women's groups. The consultants completed the survey and created the subsequent database over an eight week period, and data cleaning iterations between OED and the consultants took place over another ten weeks. Lessons from the Survey Analysis of the farm survey data was delayed eight weeks due to pervasive data entry error. Fortunately, OED designed the survey with several built-in quality control checks-an investment which yielded high returns. Specifically, the survey included three yes/no questions each with follow up questions specific to the response. Simple tabulations of the follow-up responses aligned with the initial yes/no responses provided logical validity tests. Without built- in validity checks, it would have been difficult to ensure that proper attention was given to the quality of data entry. The Pakistani survey consultants had valuable prior experience with pre-testing draft survey questionnaires. This permitted an initial round of more open-ended farmer group interviews to elicit likely categories of responses to follow-up "why" questions. Pre-testing draft survey instruments: a) reduces the cost offield interviews since surveyors need only mark appropriate sections of pre-coded instruments; b) reduces data error by simplifying data recording and subsequent entry; and c) minimizes interpretation problems of vague responses like "other" or "none of the above." Pre-testing was carried out successfully on several open- ended farm survey questions, except the cropping intensity series.61 Datp Analysis Since most of the farm survey was qualitative, as quantitative recall answers were considered unreliable, most of the analyses were based on cross-tabulations, or contingency tables, between different dimensions of the strata illustrated above in Table Fl.2. The 61. The most dramatic reported change in cropping was in the "kharfother" crop. Pre-testing reported crops may have identified precisely what crops fell into this general category. 165 Annex G Appendix 1 conclusions drawn from the contingency tables were qualified by calculating a statistic that has a chi-square distribution.62 The statistic (hereafter, X2, or chi-square) was calculated from frequencies of observations in the cells of the cross-tabulations.63 62. The definition and use of the chi-square distribution is provided in standard undergraduate statistics texts. See for example, Mansfield, Edwin, Statistics for Business and Economics: Methods and Applications. W.W.Norton & Company, NY 1980. 63. chi - square = F- Qr-][) 2 ei X F I[-1 = (1, rc), for r rows and c columns, fi= actual frequency in cell i, ei= expected frequency in cell i, = nri*nci/N, nri= total frequency in cell i's row, nci= total frequency in cell i's column, and N = total subsample size of contingency table. Annex G 166 Appendix 1 When the chi-square-statistic exceeded the critical value for a given level of significance and the specific number of rows and columns, the respective rows and columns could be described as "statistically dependent."64 Since the dependence of the columns could vary by row, the statistics must be carefully interpreted, especially for large cross-tabulations. The most useful statistics were those based on pairwise comparisons, such as Sindh with Punjab, or watercourse projects with watercourse controls. The range of chi-square values varies with the dimensions and number of observations in a contingency table, so two variations of the chi-square were also useful. When comparing the statistical strength of relationships from contingency tables with identical dimensions but different numbers of observations, a mean chi-square (D) was calculated by dividing the chi- square by the number of observations in the contingency table. When comparing the relationships between tables with different dimensions, Cramer's V was used.65 In cases where the means of continuous variables are compared across the analytical subsamples, two statistics have been used to test hypotheses-the T-statistic for pairwise comparisons and the F-statistic for group comparisons. The use of both statistics are documented in elementary statistics textbooks. 64. Strictly speaking, such a test rejects a null hypothesis which states that the columns and rows of the contingency table are statistically independent. Critical values for the chi-square distribution are provided in the appendices of most statistics textbooks. 2 65. X= N C.ees \ m (r -),(c -1) r = number of rows in the contingency table, c = number of columns in the contingency table. Cramer's V ranges between 0 and I regardless of the dimensions of the contingency table. 167 Annex H National Irrigation Projects-Problems and Missed Opportunities All four projects reviewed in this report were national in scope, which made project implementation very difficult according to provincial authorities. This was to a considerable extent the result of dissatisfaction with the central coordination unit, and the provinces lack of autonomy, but the problem went deeper. Inter-provincial differences-hydrological, agricultural, social, economic and institutional-either have to be glossed over in the interests of a common project formula (for instance in the amount of canal lining), or require that special and complex considerations for each province be built into the project design. This was particularly noticeable in the case of Balochistan, most of which is not within the Indus Basin System and has different hydrological conditions (See Box H.1). Similarly, the degree to which irrigation system management has broken down (under the pressures of inadequate budgets, corrupt practices and the unrest resulting from the "Kalashnikov society" in some areas) varies greatly throughout the country. In such circumstances project designs must compromise and err towards the commonality, rather than fully meet the specific needs of each different situation. The national project coverage created another difficulty where projects were successive time slices of an on-going program (i.e. OFWM and ISRP), since the phasing of successor projects became very difficult because each province had a different rate of implementation. The province with the slowest rate of progress determined the overall speed. In practice, ways were found to ameliorate this so that provinces which had used up their share of allocated funds without having completed implementation were able to continue (e.g. by using funds from another project in the same province), but this is a source of great discontent in the provincial implementing agencies.66 Increasingly, the inter-provincial differences have created divergences in the way the projects were implemented. For instance, institutional differences: the different powers of the irrigation departments in Punjab and Sindh have led to different patterns of enforcement of canal regulations, and hence the enforcement, or not, of water allocations at the mogha. In Punjab, for instance, the Executive Engineer has powers to fine farmers for breach of regulations, whereas in Sindh, even a Superintending Engineer can only complain to the police, who are unable or unwilling to enforce canal regulations. Or again, social differences: the totally different land ownership patterns of Punjab and Sindh have resulted in the development of hugely different Water Users Associations (WUAs) in the two provinces, with many of the Sindh WUAs being dominated by a few large landowners, or even a single landowner, and with tenants barred from membership.67 As a result, the distribution of project benefits is much more equitable in Punjab or NWFP than in Sindh, where a large part of benefits may accrue to only one or two individuals on a watercourse. The difference in the rural power structure is well demonstrated by the level of cost recovery. At the end of 1993, for instance, both NWFP and Punjab had collected over 90 percent of the amount due under the cost recovery formulae of OFWM-I and II, whereas Sindh had collected just under 60 percent. 66. The national formula also lowers the overall rate of return to the project, since funds are allocated between provinces on a formula that is unrelated to the returns to investment: the benefit per unit of water saved in Sindh, for instance, is well below that in Punjab. 67. Some of the comments here about the large land holdings and semi-feudal system in Sindh also apply to the more recently developed irrigated areas of Punjab, particularly in the southwest of the province. Annex H 168 Box H.1: Balochistan's Involvement Balochistan was only a minor player in any of the four projects under evaluation. In total its share of project costs was only 5.9 percent, ranging from 8.5 percent on OFWM-II to 3.8 percent in OFWM-I." The types of investment financed under the Command Water Management and Irrigation Systems Rehabilitation projects were basically similar to those financed in the other three provinces. The investments financed under the two OFWM projects, however, were very different. The province does contain a small area irrigated from the Indus river system and some of the watercourses renovated were in this part, but much of the on-farm water management undertaken in the rest of the province was very different in character. The water source for the majority of the area covered by on-farm water management was groundwater, although in a few cases it was from karezes (long sloping tunnels carrying water from a mother well to an area suitable for cultivation), or from springs. The discharges from all these sources is small (often only 0.25 cubic feet or seven liters per second or less), with high conveyance losses when applied directly to fields. The USAID project had demonstrated that water storage tanks, with the capacity to store about eight to 12 hours of discharge, were a cost effective means of reducing losses by allowing discharge rates to be substantially increased particularly when combined with channel lining. There was strong demand for these tanks in Balochistan. The original target in OFWM-I was 37 tanks, later raised to 95, with 116 being constructed. In OFWM-II the target was only 60 tanks. By the end of the project 597 tanks had been constructed, almost ten times the target. Unlike with watercourse renovation, farmers were willing to build these tanks at their own expense, or enlarge those which the government was providing. Tank construction was 100 percent subsidized under both projects. A 25 percent charge was introduced for canal lining under the second project. The majority of the tanks and watercourses are owned by individual families. Because the demand for these improvements far outstripped the amount of finance available at the 100 percent subsidy rate, most of the families who participated in the projects are influential in some way or other, either politically or through relatives in government. They are not necessarily owners of large areas. Since it was a requirement for participation in the project that WUAs should be set up for each watercourse being improved, the records show that WUAs exist in the requisite numbers. They are, however, merely paper WUAs. The members are the immediate family and the decision making process has not changed. This is a typical example of the misguided application of uniform design for a project which covered four widely different provinces. The combination of a tank and a lined watercourse is highly profitable. Even a small area with the right climate and a reasonable water supply brings in a good income, particularly as most owners already had more land than they could cultivate with the available (unimproved) water supply. Most of the water is used to extend the area of high value fruits and vegetables, with the extreme case of it being sold as potable water in Quetta. Thus the rate of return on the farmers' outlays is very high. Even considering the full cost, the PCR estimate of an 18 percent ERR for OFWM-ll seems too low. The situation, however, is unsustainable. The groundwater over considerable areas is being overexploited, and watertable levels are falling fast. The project's tanks are not the main cause of this but, by making possible the more efficient, and therefore more profitable, use of groundwater, they increase the demand for more pumping, both from existing wells and also by the sinking of new ones. The heavy subsidy on all elements of the project civil works is exacerbating the situation, particularly as there is no equity in its distribution. (Under the third OFWM project, a 25 percent cost recovery rate for tanks has been introduced.) The equity situation is in fact worsening in some areas. The lower watertable deprives existing karezes of water, reducing the farmers who previously were shareholders to tenants of those who have caused the watertable to fall. a. In addition, and unlike any of the other three provinces, Balochistan has received Bank assistance under two minor irrigation projects covering only Balochistan (See report number 14709, Performance Audit Report, BMADP (Credit 1243-PAK), June 1995. 169 Annex H Discussion in Annex G on the Impact Evaluation Farmer Survey results points up some of the differences between the two main provinces with respect to the waterlogging and salinity issues. In Sindh, every farmer group response on control watercourses68 reported increased waterlogging and a rising watertable. This bears out the frequently voiced contention that Sindh is getting more water than it can currently use efficiently, indeed that in some areas incremental supplies are harmful. Conversely, in Punjab, every farmer group on control watercourses reported no change in waterlogging, and two out of five reported that the watertable was dropping (in fresh groundwater zones this would indicate over-pumping and perhaps reduced seepage following channel improvements). Almost every Sindhi farmer group on control watercourses reported increased salinization, whereas not one of their Punjabi counterparts did so, and 60 percent of the Punjabi control farmer groups reported decreased saline areas. This would suggest that quite different project strategies are required for the two provinces and that they are not readily bundled into one project, if optimal results and cost-effectiveness are important. The case for taking social differences into account in designing participatory interventions has been well stated in a recent Bank report: "In cases of very hierarchical social structures and inequitable distribution of assets (for example, Sindh in Pakistan) it may be unrealistic to expect fully equitable and democratic local organizations. Therefore, the Bank and government need to recognize their role in controlling vested interests and acting as advocates for the poor. The differential incentives of different categories of farmers ("head"/"tail", small/large) should be recognized and dealt with in program design (for instance, in defining water rights), along with the resulting problems of achieving collective action." (Participation in Irrigation, Environment Department, February, 1995, page 10). This suggests that a unified WUA strategy is not appropriate for both Sindh and Punjab. There was nothing in the Bank's procedures that made it necessary to make these operations national projects: there had been, and would be again, single-province projects (agricultural extension, seeds, livestock and drainage). On the other hand, it was much more convenient, for both the Bank and GOP, that they should be national projects. For the Bank, it was a question of balancing the number of annual lending operations (which was determined by Bank management) against a desired volume of lending both for the Bank as a whole and for the South Asia Region which includes Pakistan operations. For GOP, it was a question of maximizing the volume of Bank lending within a pre-determined number of projects. Four single-province projects, for say on-farm water management, would have reduced the annual flow of funds from the Bank to GOP, unless they had been much larger projects. This could then have given rise to inter-provincial allocation problems, if the federal government was being unfair. As it was the funds were allocated between provinces on the standard federal formula used for the Annual Development Plan.69 This formula also found favor at the political level in the provinces since the Bank was not seen to be favoring any particular province. The bundling of assistance for the four provinces together in these four projects may have been administratively convenient, and clearly facilitates the Bank's wholesaling of development assistance, but is likely to have reduced the overall impact of the assistance. 68. i.e. watercourses which were not improved under OFWM, nor received their supply from canals improved under ISRP or CWMP. 69. Ninety percent on the basis of population and ten percent divided equally between Balochistan and NWFP. Annex H 170 Operations more closely tailored to circumstances on the ground in each province, with appropriate timing, would have had better results. This problem has been recognized by operational staff at least since 1990 when a long paragraph in the Aide Memoire of a preparation mission for OFWMP III starts by observing that: "it is becoming increasingly difficult for the Bank to assist nation-wide projects," but concludes only that the issue should be studied and discussed at appraisal. In the event, OFWMP III and the most recent drainage project are national in scope. The actual distribution of project cost between provinces at the start of the project is shown in Table H. 1. This indicates that Punjab was expected to receive almost 57 percent of the total share, compared with only 21 percent for Sindh. At project completion, of the four projects 66 percent of the renovated watercourses were in Punjab. These data suggest that Punjab considerations would tend to dominate the project, despite the fact that the other three provinces are very different. Table H.1: Cost Distribution by Province Total original cost from SARs (incl. price contingencies and farmer contribution) Rs m OFWM-I ISRP-I OFWM-H CWMP Total Federal/Provincial Shares (Percent) Punjab 558 649 551 597 2,355 56.9 Sindh 80 395 158 237 870 21.0 NWFP 54 Ill 82 79 326 7.9 Balochistan 28 91 79 82 250 6.8 Federal 84 53 60 110 307 7.4 Total 804 1,299 930 1,105 4,138 100.0 171 Bibliography Ahmad, Masood and Kutcher, Gary P, Irrigation Planning with Environmental Considerations, A Case Study of Pakistan's Indus Basin; World Bank Technical Paper No. 166, April 1992. Bandaragoda, D. J. and Rehman, Saeed ur, Warabandi in Pakistan's Canal Irrigation Systems, IIMI Country Paper, Pakistan No. 7, 1995. Byrnes, Kerry J., Water Users Associations in World Bank-Assisted Irrigation Projects in Pakistan, Technical Paper No. 173, December 1992. Cernea, M. M (editor), Putting People First, 1985, page 33-"Ignoring Local Irrigation Organization." Chancellor, Felicity, Economics of Canal Lining-An Evaluation Method, in Proceedings of a Workshop on Canal Lining and Seepage, 18-21 October, 1993, International Waterlogging and Salinity Research Institute, Lahore, and HR Wallingford, U.K. Gleason, J.E. and Wolf, J.M. Evaluation of the Impact of Canal Rehabilitation on Hydraulic and Agricultural Economics Indicators, Irrigation Systems Management Project Phase II, USAID, Islamabad, Pakistan, April 1993. Government of Pakistan, Agricultural Statistics of Pakistan, 1987-88. Jones, W.I., The World Bank and Irrigation, A World Bank Operations Evaluation Study, OED, July 1995. Mansfield, Edwin, Statistics for Business and Economics: Methods and Applications. W.W.Norton & Company, NY, 1980. Meinzen-Dick, Reidinger and Manzardo, Participation in Irrigation. Environment Department Papers, Participation Series Paper No. 003, The World Bank, February 1995. Mohammad, Ghulam (1965), Private Tubewell Development and Cropping Patterns in West Pakistan, The Pakistan Development Review, Vol. V, No. 1 (Spring): pp. 1-53. Mott MacDonald International, Water Sector Investment Planning Study. January 1991 (for Federal Planning Cell, Lahore) Murray-Rust, D. Hammond, and Vander Velde, J., Changes in Hydraulic Performance and Comparative Costs of Lining and Desilting of Secondary Canal in Punjab, Pakistan, in Proceedings of a Workshop on Canal Lining and Seepage, 18-21 October, 1993, International Waterlogging and Salinity Research Institute, Lahore, and HR Wallingford, U.K. 172 National Engineering Services Pakistan (PVT) Limited, On-Farm Water Management III Project: Feasibility Study, Lahore, May 1990. Punjab Economic Research Institute, Planning and Development Department, GOPunjab: Evaluation of On-farm Management Programme in Punjab, December, 1985 and February 1993. Saleem, Mahamood Ali, Amin, Muhammad Sajid and Mazar-ul-Haq, February 1993, Evaluation of On-Farm Water Management Program in Punjab, Punjab Economic Research Institute, Publication No. 281. Sindh Development Studies Centre, The Causes of Very Slow Response ofSmall Farmers to the Precision Land Leveling, September 1989. USAID, The On-Farm Water Management Project in Pakistan, Project Impact Evaluation Report No. 35, June 1982 WAPDA, April 1994, First Time-Scaled Evaluation Report ofFarm Water Management Program, IDA Financed OFWM-I & II Projects, Watercourse Monitoring and Evaluation Directorate, Planning and Investigation Organization, Planning Division WAPDA, December 1985, Monitoring and Evaluation of On-Farm Water Management Program-I; Final Report, Watercourse Monitoring and Evaluation Directorate, Planning and Investigation Division, Planning Division. WAPDA, November 1977, An Initial View of Water and Associated Power Developments for RAP and Perspective Plan, (quoted on page 5 of the SAR of OFWM-II). Waryam, Ali Mohsin, "The Impact of Watercourse Improvement on Conveyance Efficiency". Paper prepared for presentation at The National Conference on On-Farm Water Management, May 29-30, 1994. Mona Reclamation Experimental Project, WAPDA, Bhalwal. Water and Power Development Authority, Master Planning and Review Division, Revised Action Programme for Irrigated Agriculture, April 1979. Water and Power Resources of West Pakistan: A Study in Sector Planning, Johns Hopkins Press, 1968. World Bank, Participation in Irrigation, Environment Department, February, 1995, page 10. World Bank, India-Punjab Irrigation and Drainage Project, Report No. 8056. World Bank, OED, PAR: Khairpur Tile Drainage and Irrigated Farming Development Project (Credit 648-PAK) World Bank, OED, PAR: Salinity Control and Reclamation Project (SCARP) Mardan (Credit 877-PAK) 173 World Bank, OED, Report No. 14709, Performance Audit Report, BMADP (Credit 1243-PAK), June 1995 World Bank, OED, SCARP Transition Pilot Project (Credit 693-PAK), Impact Evaluation Report, forthcoming. World Bank, OED, World Bank in Pakistan: Review of a Relationship (1960-1984), OED Report No. 6048, January 27, 1986 World Bank, Pakistan-A Strategy for Sustainable Agricultural Growth, November 3, 1994, Report No. 13092-PAK World Bank, Pakistan: Irrigation and Drainage: Issues and Options, Report No. 11 884-PAK, March 1994 World Bank, Pakistan: National Drainage Transition Program, Executive Project Summary, June 14, 1995. World Bank, SCARP Transition Pilot Project (Credit 1693-PAK), June 1995 (Report No. 14769) World Bank, The Agricultural Impact of the Kalabagh Dam, Report No. 6884-PAK, August, 1986 (Revised June, 1987) World Bank, Irrigation O&M and System Management in Southeast Asia-An OED Impact Study, Operations Evaluation Department, June, 1996. SRD Z지02 신‘ cC (D 0~ LI ccr z ii O. C cE