Senegal - Debi - Lampsar Irrigation Project

- Dociment of The World Bank FOR OFFICIAL USE ONLY Report No. 7149 PROJECT COMPLETION REPORT SENEGAL DEBI-LANPSAR IRRIGATION PROJECT (CREDIT 775-SE) March 2, 1988 Africa Regional Office Thi docmnent hs a restiuted dsbaton and May be used by recipens only In the pedrorman of their offichd duties. Its contents may not otherwise be didosed without Wodd Bank adthoiu  CURRENCY EQUIVALENTS Currency Unit - CFA Franc (CFAF) US$1.00 - CFAF 378 CFAF 1.00 a US$,00263 WEIGHTS AND MEASURES Metric System ABBREVIATIONS BNDS Banque Nationale de Developp_ment du Senegal CCCE Caisse Centrale do Cooporation Economiqt.e (France) Aid Agency CPG Comite Paritaire do Gestion ERR Economic Rate of Return FAC Fonds d'Aide et do Coopiration (France) Aid Agency FAOICP Food and Agriculture Organization/World Bank Cooperative Program ICB International Competitive Bidding IDA International Development Association IRR laternal Rate of Return LCB Local Competitive Bidding lOFE Ministry of Finance and Economic Affairs MOFA Ministry of Foreign Affairs muV Manufacturing Unit Index OED Operations Evaluation Department ONCAD Office National do Coop;ration et d'Assistance pour le Developpement PCR Projoct Completion Report PIVs Small village irrigation perimeters PMU Project Management Unit PPAM Project Performance Audit Memorandum PPAR Prcject Performance Audit Report SAED Socilte Nationale d'Amenagement et d'Exploitation des Terres du Delta, the Implementing Agency SAR Staff Appraisal Report SUMA Section d'Utilisation du Materiel Agricole FISCAL YEAR OF BORROWER Government of Senegal t July 1 to June 31  MOR OCAL VUI ONLY THE WORLD SANK Washiton, D.C. 20433 U.S.A. Olikd O4 00".Gsui opw*ku IwAlDM March 2, 1988 MEMORANDUN TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECTt Project Completion Report on Senegal Debi-Lgamsar Irritation Proiect (Credit 775-SE) Attached, for information, is a copy of the report entitled 'Project Completion Report s Senegal Debi-Lampsar Irrigation Project (Credit 775-SE)" prepared by the Africa Region. No further evaluation of this project by the Operations Evaluation Department has been made. Attachment Tb doumet ha a nicdisdbut and my be ud by Ucpit a* In f ptifoan of their olhcbl dutbz Its oonXts may not othowie be disclod witbout World ank aoutnfio.  FOR OFFICIA UR ONLY PROJECT COMPLETION REPORT SEMEGAL DEBI-LAMPSAR IRRIGATION PROJECT (Credit 775-SE) TABLE OF CONTENTS Pa. No. Preface . ....... . $ Basic Data Shoet . . ... ii Evaluation Suary .......... .. ... v PROJECT COMPLETION REPORT iT Introduction ........... ... ... . . 1 II. Implement-t$en .... . ... .. .. 3 III. Agricultural Impact ................................ 8 IV. Economic Rate of Return Analysis ................ 17 V. Institutional Performance and Development 29 VI. Special Issues ...... ... .'.. . 26 Annexes Maps: IBRD Nos. 12863, 12864 I . '- , - - wf vs bvrs ts' nr thseperfoanncm  PROJECT COMPLITIOW UPRT SENEGAL DEBI-LLMPSAR IRRIGATION PROJECT (Credit 7?5-SE) MRFACE This is the Project Completion Report (PCR) of the Debi-Lampsar Irrigation Projoct in Senegal, for wilch Credit 775-SE, in the amount of US$20 million was approved on March 7, 1978. The original Cridit Closing Date wee June 30, 1983, but after s-veral extensions, the eventual Closing Date was Octobor 29, 1986. The project was the twelfth Bank Credit In the agricultural sector and the second for Irrigation development in Senegal. It Is expected to be followed by the Irrigation IV Project, which as of this writing waits Board approval. The PCR was prepared by the West Africa Region following a mission to Senegal from February 9 to 20, 1987, and is based on the Lampear Project Completion Report prepared by the Implementing agency (SAED), a review of the Staff Appraisal Report, SAR No. 1685a-SE, and the President's Report, No. P-2192a-SE, both dated Februray 21, 1978, and the Credit Agreement of March 17, 1978; relevant Bank files, including correspondence with the Borrower, internal Bank memoranda, minutes of the Board's discussions, Bank supervision reports, project progress reports prepared by the Borrower; as well as interview with past and current officials of the Bank and in Senegal who have been associated with the project. The roport vas not subjected to an audit by OED, but was sent to Government for couments on January 5, 1988; however, no comments were received.  - ii - SENEGAL DEBI-LAMPSAR IRRIGATION PROJECT (Credit 775-SE) PROJECT COMPLETION REPORT Basic Data Sheet Key Project Data Appraisal Actual or Actual as % of Estimate Estimated Appraisal Estimate Actual Total Costs (US$ Millions) 35.0 36.7 105 Credit Amount (US$ Millions) 20.0 19.4 97 Cofinancing 15.0 17.3 Date of Board Approval - 03/07/78 Date Signing - 03/17/78 Date Effectiveness - 07/17/78 Date Physical Components Completed 10/80 12/81 Clos.ng Date 06/30/83 10/29/86 Economic Rate of Return (%) 10 -4 Institutional Performance fair Agricultural Performance good Number of Main Beneficiaries .1 (family) 1100 2627 239 Staff Input (weeks) FY77 FY78 FY79 FY80 FY81 FY82 FY83 FY84 FYs FY86 TOTAL Identification/ Preparation Pre-appr'l 40.5 - - - - - 40.5 Appraisal 87.8 - - - - 87.8 Negotiation 4.8 - - - - 4.8 Supervision 2.9 22.3 37.1 33.1 3.8 9.0 8.3 2.8 0.2 119.5 TOTAL 90.7 22.3 37.1 33.1 3.8 9.0 8.3 2.8 0.2 252.6 Follow-On Project Name Irrigation Technical Assistance Project Credit Number 1632-SE Credit Amount (US$ Million) 4.9 D&te of Board Approval 10/29/85  - iii - Mission Data Date No. of Mandays Specializations Performance Trend Types of Mission (%o/yr) Persons In field Represented 2/ Ratins 3/ 4/ Froblems 5/ Identification - - - Preparation 1/ - - - Pre-appraisal 08/76 6 10 a,b,e,f,g - - - Appraisal 02/77 8 18 *,b,c,d,e - - - Supervision 1 06/78 1 5 a I - M 2 11/78 1 2 a 2 2 M,0 3 03/79 2 4 a 2 1 T 4 07/79 3 12 b,c,d 2 2 F,M,O 5 03/80 2 13 c,d 2 2 F,M,O 6 07/80 3 15 b,c,t 2 2 PFM,O 7 11/80 3 11 b,c,d 2 1 F,.,O 8 07/81 1 9 d 1 1 F,O () 02/82 4 17 1 1 9 10/82 2 8 c,d 1 2 0 10 04/83 1 2 d 2 2 0 11 11/83 1 6 4 1 2 0 () 03/84 1 1 1 2 0 12 07/84 2 30 d,e 1 2 0 1/ Carried out under Credit 5-18-SE. T/ Specializations: a - Irrigation Engineer, b - Agriculturist. c - Financial Analyst, d - Rural Engineer, e - Agro-economist, f - Drainage Engineer, g - Soil Specialist. 3/ Performance Rating: 1 - Problem free or minor problems, 2 - Moderate problems, 3 - Major Problems. 4/ Trend: 1 - Improving, 2 - Stationary, 3 - Deteriorating. S/ Types of Problems: F - Financial, M - Managerial, T - Technical, P - Political, 0-other. 0 indicates supervision missions that were not included in the official numbering sequence.  - iv - Exchange Rates Name of Currency CFA Year: Appraisal Year Average 245 Intervening Year Average 327 Completion Year Average 378 Cumulative Estimated and Actual Disbursements FY78 FY79 FY80 FY81 FY82 FY83 FY84 FY8s FY86 Appraisal Estimate (US$ Million) 1.0 6.8 13.8 18.4 19.5 20.0 - - - Actual Disbursement (US$ Million) 0.0 3.9 8.5 12.5 14.8 16.5 1;.1 18.7 19.4 Actual as Z of Estimate 0 57 62 68 76 83 86 94 97 Date of Final Disbursement October 29, 1986  -v - SENEGAL DEBI-LAMPSAR IRRIGATION PROJECT (Credit 775-SE) PROJECT COMPLETION REPORT Evaluation Summary Introduction 1. This was the second irrigation project the Bank financed in Senegal and like the first project was located in the Senegal River Delta. At the time of project preparation, some 10.000 ha with full water control were already in operation in the Delta, with an additional 240,000 ha thought appropriate for development once the planned Diama and Manantali dams were completed. Government wan emphasizing development of this potentially rich resource, not only to substitute domestic vice production for growing imports, but also to generate economic growth in this poor region. Experience with existing perimeters during the 1972/73 drought led to the conclusion that partial water control irrigation designs, though relatively cheap, could not assure production. Hence the design of the project's irrigatioc systems, developed with the assiseance of a Technical Assistance Credit, sought to balance low investment cost, effective water control, and simple management appropriate for farmers inexperienced in irrigated rice production. The project was identified and prepared within the Bank's West Africa Agricultural Projects Division, with appraisal completed by February, 1978, and Board approval in March, 1978. ObJectives 2. The project's objectives were to develop irrigation on 1100 ha at Debi and 2200 ha at Lampsar, to benefit some 1100 farmers and their families. The project was designed to (a) level land, establish earthworks, and install and equip pump stations; (b) assist farmers to grow and increase yields of paddy and tomatoes, initially during one crop season per year; (c) conduct applied farming research appropriate to development of small-farm irrigation in the delta; (d) prepare a development plan for irrigation resources in the Senegal River Basin in view of dams then in the planning stage; and (e) strengthen the implementing agency (SAED) to give it the operational and planning capabilities. to oversee irrigation development throughout the Senegal River Basin. Total project costs were estimated at US$35 million, with the IDA credit of US$20 million going primarily to development of Lampsar and preparation of the development plan, a US$5.1 million Kuwait Fund credit enabling development of Debi, a FAC grant of US$2.4 million covering technical assistance, and Government's contribution of US$7 million covering a variety of local costs. Implementation Experience 3. Project construction began in January, 1979, after a delay of almost a year caused by management changes at the implementing agency and difficulties in finalizing the works supervision contract. While most of  - vi - the origlally foreseen project components were completed well before the intended final closing date of June 30. 1983, this was extended several times to permit disbursement on added components including creation of small village perimeters and rehabilitation of the Savoigne perimeter and canal. Final disbursement and the eventual closing date occured on October 29, 1986. Although total disbursement of the Bank's US$ 20 million credit only reached US$19.4 million, total project costs were US$36.7 million rather than the appraised eost of US$35 million. 4. The originally designated net area for irrigation could not be fully developed because of excessively saline and sandy soils encountered, and because of a denser network of canals and dikes that resulted from chifting from 3 ha to about 1 ha farm allocations. Over 20X (750 ha) of the appraised area was thus lost, although this was partially compensated for by development of other perimeters that were added to the project. Production of tomatoes was abandoned after difficulties with the first crop showed that the project soils were too heavy and saline for remunerative yields. Results S. Rice production has surpassed appraisal expectations both in yields per hectare (4.8 ton/ha vs. 3.5 ton/ha) and for total annual project production (13,015 tons vs. 10,150 tons). The reduced average farm size that was implemented still provided positive financial results, partly because of subsidized water costs and producer price supports. The economic result -- an IRR of -42 -- is not satisfactory, however, and is below the appraisal estimate because no to-matoes could be grown, the CFA value of paddy fell sharply, labor costs were underestimated, a project cost overrun was incurred, and the estimated life span of the investment was reduced (paras 4.01-4.03). Sustainability 6. Numerous uncertainties exist over future yields, crop rotations, and institutional arrangements at the project's perimeters. Ongoing disengagement of SAED from provision of production inputs and tractor services makes future production dependent on the development of private market sources of supply, and during a transition period, yields may fall. The refocussaing of SAED on planning and implementing irrigation development in the Senegal River Valley continues the long and difficult matching of capacity and mandate. 7. Full operation of the Diama and Manantali dams, expected beginning 1989, will enable double cropping, though socioeconomic constraints are expected to keep cropping intensity below what is technically feasible, at least initially. 8. Adequate maintenance of the irrigation network continues to pose difficulties because of insufficient operating funds, inadequate attention, and inappropriate and limited equipment. These factors are contributing to deterioration of the irrigation network and civil works that risks shortening the investment's usable life.  - vii - Findinas and Lessons Lessons to be learned from the project include: (a) The economics of rice irrigation projects in the delta are poor. Higher valued crops need to be added to the crop rotation In order to justify the Investment costs. Moreover, multi-crop rotations can be designed to reduce peak water demand, which would enable reducing the scale and cost of the irrigation infrastructure. (b) Achieving the benefits of participatory farmer associations may require experimentation before appropriate institutional arrangements are found. At Debi-Lampsar, a range of groupings exist, from the small village groups to the perimeter-wide Comit6s Paritaires. This latter has only existed since 1983. and is proving to be a good forum between SAED and farmers for resolution of important production and management issues. (c) There are indications that water-borne illnesses are increasing as a consequence of the project, a situation likely to worsen once water is present for double cropping. Resources are currently inadequate for monitoring or mounting effective programs to limJt these diseases. (d) Better soil testing and mapping is needed to avoid the situation where land in a perimeter has to remain undeveloped because of excessive salinity or sandiness. (e) Options for building flexidility into project implementation need exploring. Designs and contracts should enable changing the scale of complementary investments when such components as irrigated area are subject to revision. At Debi-Lampsar, the originally designed pump capacity was installed despite a significant reduction in need.  SENEGAL DEBI-LAMPSAR IRRIGATION PROJECT (Credit 775-SE) PROJECT COYPLETION REPORT I. INTRODUCTION Context 1.01 An enduring priority of Government agricultural policy in Senegal is development of water resources to reduce dependGncy on risky rainfed cereals production and the substantial costs of imported rice. The untapped potential of the Senegal River has been, even before the creation Oc the Organisation pour la Mise en Valeur du fleuve Senegal (OMVS) by Senegal, Mali, and Mauritania in 1972, the primary focus of the Government's irrigation investment strategy. Regulation of the river's flow through planned construction of Manantali Dam (upstream) and Diama iam (delta) would create the potential _for irrigation of about 240,000 ha in Senegal. In the mid 1970's, Senegal, as part of OMVS, began to seek international assistance in support of the tremendous institutional, technical and financial effort needed to meet the target of 5w0Q ha additional irrigated land per year. The government agency charged with coordinating this effort was the Socift6 Nationale d'Am6nagement et d'Exploitation des Terres du Delta (SAED), originally created in 1965. 1.02 The Bank Group's first investment in development of the river's irrigation potential began in 1973 with the Senegal River Polders Project (Credit 350-SE). This IDA credit of US$ 4.5 million foresaw development of 3200 ha at Dagana in the upstream portion of the river's delta to permit double cropping, as well as improvement of irrigation to allow single cropping on 1780 ha at Debi and Lampsar. Implementation delays caused by SAED's cumbersome administrative and financial procedures, difficulties obtaining counterpart funds, as well as exchange rate changes led to Dagana's costs exceeding total budget provisions. Nevertheless, costs per hectare were still reasonable by West African standards. Meanwhile, droughts in the early 1970's exposed the risks of partial water control perimeters, such as was originally designed for Debi and Lampsar. Further, plans for Diama Dam made double cropping at these locations a feasibility, though requiring a different technical design. In reformulating Credit 350-SE, the Government requested financing to establish detailed technical designs for complete water control at Debi and Lampsar. Credit 5-18-SE was signed in 1975 providing US$ 1.0 million towards deveiopment of these engineering and cost estimate stuties. Hence the Debi-Lampsar Project (Credit 775-SE) was only the second Bank Group project in irrigation in Senegal. All other Bank loans (one) and credits (10), totalling US$ 52.3 million as of 1978 for the agricultural sector, were for non-irrigation projects.  -2- Objectives and Components 1.03 The concept of large-scale irrigation projects in the Senegal River Basin addressed Government objectives of a) developing water resources to reduce the risks of droughts; b) produce rice to offset the trend of growing imports; and c) foster economic growth in the river basin, both to alleviate poverty of the current residents, and to encourage voluntary migration from the densely populated groundnut basin. 1.04 The present project was designed to fulfill these objectives through upgrading the perimeters at Debi and Lampear from submersion irrigation to complete water control. At appraisal, improvements were to encompass 1100 ha at Debi and 2220 ha at Lampsar. For Debi, this entailed construction of a pumping station drawing directly from the Senegal River, an irrigation network of primary, secondary, and drainage canals, as well as land levelling. The Lampsar site consisted of 11 sub-perimeters strung along the Lampsar marigot, an offshoot of the Senegal River in the delta. Construction of dikes on both banks of the marigot for flood protection, 14 pump stations (one for each sub-perimeter and three drainage system stations), the canal system, and land levelling were the main irrigation components. In addition, both perimeters were provided with an internal dirt road network, and building infrastructure consisting of staff housing, a perimeter office, a workshop, and central and field warehouses. Both perimeters were also equipped with agricultural machinery, trucks and light vehicules. 1.05 The direct beneficiaries were to be some 1100 farmers and their families, each to receive a parcel of 3 ha. They would benefit from increased income resulting from production of paddy and tomatoes (for canning at a nearby factory). SAED was to provide substantial assistance in these new endeavors by supplying mechanical land preparation services, crop inputs, marketing outlets, and extension services. 1.06 The project also sought to develop SAED's capacity to fulfill its role as the agency responsible for Senegal's planning and execution of irrigation development in the River Basin. Five technical assistants were provided to reinforce planning, financial, and applied farming research capabilities within SAED. Funds were also earmarked for consultants to develop a master plan of irrigation development in the River Basin for Senegal, as well as detailed technical studies of perimeters appropriate for subsequent financing. Financins Plan 1.07 Total project costs were estimated at USS 35 million (see Annex 1). A Kuwait Fund loan was to cover the complete costs of the Debi component, or US$ 5.1 million. The French (FAC) grant was designated for the five expatriate technical assistants to SAED, at a cost of US$ 2.4 million. The IDA credit of US$ 20 million was divided among the development of Lampsar (US$ 12.5 million), technical assistance to applied research and irrigation development planning (US$ 3.5 million), refinancing the engineering credit S-18-SE (US$ 1 million), and price and physical contingencies (US$ 3 million). Government counterpart funds were  - 3 - envisioned to total US$ 7.0 million. Hence the IDA loan was the largest block of financing, contributing 572 of the total resources, and 801 of the foreign exchange costs of Lampsar perimeter, the planning studies, and applied research. Pre-I!mlementation Issues and Processlng 1.08 Appraisal of the project raised six principal issues, two of which -- cost recovery and organizational aspects - proved to be troublesome during implementation. Charges to farmers by SAED for water and agricultural service costs were uniform throughout the delta at appraisal, with reasonable recovery rates, i.e. 86S in 1976, but well below cost recovery. It was recommended that the charges for water be raised to recover operating and maintenance costs, plus as much of c capital costs as would still maintain adequate farmer income. 1.09 Proposals for reorganization of SAED pre-dated the project, but further steps were proposed to make operational the delegation of authority to project units and greater financial autonomy for SAED. As a state corporation, SAED had to comply with complex financial and administrative procedures, and had very little authority in budgetting and expenditure. To remedy the delays and deterioration of SAED's financial situation, it was proposed that a decree redefine SAED's relation to the State. Short-term financing by the State was also arranged as a step to financial viability for SAED. 1.10 Other issues were resolved by a) dropping from the project the Diagambal perimeter for which USAID had arranged financing; b) filling a financing gap through Kuwait Fund participation; c) evaluating project benefits conservatively on the assumption that Diama Dam would not soon be operational so as to permit double cropping; and d) arranging an agreement between OMVS and Government on water rights, as would become relevant upon Diama's completion. 1.11 Negotiations for financing the project took place in January 1978 with only minor difficulties. The water rights resolution requested by the Bank was adopted in time for the negotiations. However, the decree pertaining to the reorganization of SAED could not be ratified before negotiation because of procedures necessary for amendment of State law. Further, participation of the Kuwait Fund could not be made final as their commitment awaited results of their own project evaluation. To hasten this process between Government and the Kuwait Fund, the Credit Agreement was amended to permit suspension of IDA disbursements for the project if the Kuwait Funds finance was still unavailable by end-December, 1978. 1.12 Presentation before the Board occurred in tarch 1978, and did not occasion any difficulties. Credit signing took place the same month, and the credit became effective in July 1978. II. IMPLEMENTATION 2.01 Irrigation works of the project were completed by 1984,  -4- consisting of 3050 net hectares as compared to the appraisal target of 2980 net hectares. Despite the realized area surpassing the target, however, a number of implementation delays and project revisions occurred. Start-UJp 2.02 Start-up of construction work at Debi and Lampsar faced separate problems. Those at Lampsar stemmed largely from transitional difficulties as SAED went through a change of General Director in early 1978. Delays were caused by the slow pace of preparing the major contract documents, and disagreements between the World Bank and SAED, which wanted to select a local firm as supervising contractor. These difficulties were resolved, and earthmoving and civil works began in January 1979, a year later than foreseen in the Staff Appraisal Report (SAR). 2.03 For Debi delays resulted from several financing problems. The Kuwait Fund loan agreement was not signed until February 1979, two months after the deadline permitting suspension of disbursements from the IDA loan. Then the legal documents, which had been signed in Kuwait. were lost in transit between Senegal's Ministry of Foreign Affairs and the Ministry of Finance and Economic Affairs. It was not until February 1980 that the documents were replaced and the Kuwait Fund loan finally made effective. Meanwhile, experience gained during the work on Lampsar led the World Bank to request a further delay on Debi to permit additional mapping to avoid saline and sandy soils. Revisions 2.04 Several changes were made at Lampsar in the area to be developed. First, because of 151 devaluation of the dollar between the loan signing and project start-up, as well as local price increases, there was a concern about cost overruns. It was decided to forego development of the Ndiougue sub-perimeter, where sandier soils would have led to particularly high earthmoving costs. Further area was lost due to a denser irrigation network required by SAED's decision to establish 1.5 ha rather than 3.0 ha farms (para 3.15), and because some of the original area had to be used instead as a clay earth source for canal embankments. Lampsar's net area was in this initiUl phase reduced from 2200 to 1670 ha. As a cheaper alternative to the Ndioungoe sub-perimeter, SAED proposed rehabilitation of the 300 ha Savoigne perimeter adjacent to Lampsar. This was accepted by the World Bank and completed in 1982. Finally, the strong recovery of the dollar to over 150% of its SAR value in 1983 eased the local financing constraint, and it was agreed in early 1984 to develop a portion of the original Ndioungue sub-perimeter -- 140 ha, plus about 200 ha of small perimeters (PIVs). The decision to develop the PIVa was in response to villagers' requests as drought conditions made rainfed cultivation of millet impossible. 2.05 At Debi, the results of the additional soil testing led to a reduction in area to be developed from 1100 ha to 740 ha. 2.06 Apart from these changes in area under development, there was also a significant change in the basic farm system. As mentioned, the farm  - 5 - size to be distributed was halved to a target of about 1.5 ha per family although in practice the average is closer to 1.0 ha. Further, tomatoes were dropped from the crop cycle, leaving paddy as the only production of the perimeters. The reasons for these production changes are discussed in Section III. 2.07 At the request of the borrower, a revision was made in financing arrangements under the Credit Agreement whereby credit disbursements as a proportion of total costs were permitted up to 100Z for all categories. The original limits were 802 of irrigation and building costs, 752 of operational costs (excluding salaries), 85% of technical assistance, and 752 of total vehicule and equipment costs. The revision was only relevant for technical assistance as there was no financial contribution by the Government to this category; all other categories did receive government counterpart funds (see Annex II). Implementation Schedule 2.08 After the initial start-up delays at Lampsar and Debi, the major contracts for land levelling, earthmoving and civil works were completed in the time allotted. Work went particularly quickly at Debi since the same contractors as for Lampsar were engaged, and the necessary equipment was already on location (see Table 1). 2.09 Difficulties with equipment contracts restricted the first agricultural campaign in 1980/81 to only five of Lampsar's sub-perimeters. One reason for the delay on equipment delivery was that contract awards were still decided by the Ministry of Rural Development. SAED's technical evaluations of bidding documents were on occasion not adequately considered, and resolving this issue postponed several contract awards. As a consequence, delayed delivery of pumps, agricultural machinery and siphon tubing did not give sufficient time for equipment installation and preparation of all the sub-perimeters of Lampsar until 1981. Procedurally, the contract-granting process wst improved with SAED's change in legal status in 1982, as it was given authority to make contract awards. Procurement 2.10 On the whole, Bank procurement procedures were respected and were ir accordance with Project Agreement specifications. These were that (i) international competitive bidding (ICB) apply to all earthworks, civil works and irrigation equipment contracts surpassing US$ 100,000; (ii) local competitive bidding (LCB) be used for agricultural and other equipment contracts between US$ 15,000 and US$ 100,000, but not surpassing US$ 1 million total; and (iii) direct shopping be allowed for materials contracts under US$ 15,000, but not surpassing US$ 0.5 million in total. All bid documents and contract awards exceeding US$ 100,000 required prior World Bank approval, as did negotiated contracts for consultants.  SENEGAL DEBI-LAMPSAR IRRIGATION PROJECT (Credit 775-SE) PROJECT CO)MPLETION REPORT TABLE 1: Project Targets and Realization -- Area and Timetable Appraisal Realised Perimeter Area First Harvest Area First Harvest (hectaree (year) (hectare) (year) Debi 1100 1979 740 1982 Lampsar 2200 1810 Bifeche 200 1979 152 1980 Lampsar 160 1979 111 1980 N'Delle 260 1979 158 1980 N'Diaye 160 1979 125 1980 N'Gomene 200 1979 200 1980 * Tilene 130 1980 105 1981 : Pont Gendarme 200 1980 170 1981 N'Diougou 170 1980 140 1984 Bodiene 320 1980 278 1981 Polo 280 1980 227 1981 N'Gao 140 1980 144 1981 Savoigne - - 300 1983 Small Village Perimeters - 200 1984  -7- 2.11 An exception was made in the construction of buildings which was to be awarded through ICB but was finally awarded through LCB. This was agreed to by the World Bank at the request of the borrover in view of international contractor disinterest, and economies of cost and time in being able to construct the Lampsar buildings in conjunction with the adjacent N'Diayi training center, which had been contracted under LCB. 2.12 Another exception was granted on the US$ I million total limit on contracts to be awarded under LCB. With the addition of small village perimeters, Savoigue rehabilitation and Ndioungue development after the main work of the project was completed, the value of LCB contracts increased to about US$ 2.3 million. 2.13 Disbursements from the IDA credit, broken down by type of contract, are shown in Annex III. The fourteen ICB contracts accounted for 722 of credit disbursements. There were 35 LCB contracts with an average value well below the minimum requiring World Bank review. Despite the large number of LCB contracts, they only comprised 122 of total disbursements. Supervision of these contracts by SAED would have been simplified by grouping them into larger lots. costs 2.14 Total project costs were US$ 36.7 million compared with US$ 35.0 million at appraisal, for an overrun of 5% (see Annex I). Project costs were influenced at first by a depreciating dollar, but benefitted beginning in 1981 from a strengthening dollar which decreased the dollar cost of the CPA denominated contracts. Since about half of project costs, such as irrigation development costs at Debi, and much of the technical assistance, came after the dollar strengthened, exchange rate movements are not the primary factor in the modest cost overrun. If the exchange rate at appraisal (CFA 245/US$) had held throughout the project, there would have been an 11 overrun in dollar costs, as there was in CFA terms. Appraisal provisions for physical contingencies were 152 for all works, equipment, agricultural inputs and operational costs, 202 for applied research, and 0X for other studies. The weighted average price contingency applied to total project costs was 232. Because of the delay in project start-up, this price contingency proved inadequate. 2.15 The largest cost overrun was for irrigation investments at Lampsar, where expenditure was almost US$ 5 million greater than, or 151% over, the appraised costs including price and physical contingencies. Costs of earthmoving were higher than anticipated, and components to rehabilitate Savoigne and develop village perimeters were added. There were also substantial overruns for vehicules and equipment, by 29% of appraised costs at Debi, and by over 1202 at Lampsar. Finally, there were disturbing increases in salary and operational costs. Salaries at Lampsar were 317% of appraisal estimates, mostly because of a rapid increase in staff size. The salary and operational cost increase was partly the result of SAED decentralisation not totally foreseen at appraisal, but considered desirable. Staff increases were excessive however, and an implementation of cutbacks has begun. Including all project costs, per hectare costs of  - 8 - irrigation development were US$ 8,500 at Debi and US$ 10,000 at Lampsar, compared with appraisal estimates of US$ 6,800 and USS 8,400. Disbursements 2.16 A total of US$ 19.4 million was disbursed out of the US$ 20.0 million World Bank credit. At the original closing date of June 1983 only US$ 16.5 million had been disbursed, but the effective closing date was extended several times to permit further withdrawals for work on Savoigne, Ndioungue and the small village perimeters. The loan balance of US$ 0.6 million was cancelled in October 1986. Projected and actual disbursements by category are shown in Annex I. A comparison of the actual disbursement profile and that projected at appraisal is shown in the Basic Data Sheet. Contractor Performance 2.17 Technical designs, physical works, provision of equipment, supervision and the audit that were done by contractors were in general carried out satisfactorily. Several difficulties that did arise were solved without major project delays. At Lampsar, land levelling was inadequate on 500 ha and had to be redone by the contractor. Also, at three perimeters, the contractor constructed canals with walls that were too steep causing problems with erosion, apparently because the proper equipment was not yet on location. Remedial grading was eventually carried out on these canals by SAED. The same contractor later executed irrigation earthworks at Debi with no problems. Delivery of pumps to Lampsar also experienced problems when the pumps were diverted by the Government before installation to emergency pumping of drinking water for Dakar. SAED subsequently refused delivery of the used pumps, until an agreement was negotiated whereby the contractor refurbished the pumps and extended the warranty from one year to three agricultural seasons. III. AGRICULTURAL IMPACT Summary of Incremental Output 3.01 Incremental agricultural production resulting from the project has surpassed appraisal expectations for paddy, but has not been realized for tomatoes. Additional paddy production has reached 11,655 tons per year, or 33% better than originally foreseen (see Table 2). This is because realized yields averaging 4.5 T/ha at Lampsar and 5.5 T/ha at Debi are significantly higher than the conservative appraisal estimates of 3.5 T/ha. Production of tomatoes for canning has been abandonned, and the absence of this high value crop from the rotation severely affects the financial and economic benefits of the project. Before the project, a smaller area (depending on rainfall) averaged only 0.8 T/ha, the lower yields being the consequence of negligable water control and no chemical input use. Cropping Pattern, Land Use, and Yields 3.02 Attempts to grow tomatoes at Lampsar were abandonned for several reasons. Appraisal crop rotations foresaw tomatoes entering every third  -9- year on to paddy fields. With the heavy clay soils of the area, this turned out to require difficult and expensive mechanical preparations of the land, since parcels levelled for paddy flooding had to be converted to sloped ridges for tomatoes' row irrigation. If done improperly, ridging did not allow for an even distribution of water, and yield reduction or crop loss resulted. An additional problem was salinity of the soil, exacerbated by the groundwater level being close to the surface, and salt levels building up through surface evaporation. Although this problem was thoroughly anticipated and salinity levels determined to be acceptable prior to appraisal, the problem was accentuated by ridging difficulties, which resulted in inadequate maintenance of fresh water delivery to the tomatoes' root zone. 3.03 This problem appears to be particular to Lampsar, and has been avoided in other delta irrigation perimeters through proper selection of soils on which to cultivate tomatoes. Irrigation perimeters are generally developped from natural basins, which have lighter soils on the upper slopes of the terrain. These soils are much easier to switch from ridged to flood irrigation, and hence are more appropriate for polyculture rotations. But at Lampsar, such light soils are rarer, or were not developped. 3.04 Realized area of harvested paddy in 1986/87 was officially 2733 ha, or 94% of the appraised total. Difficulties with land development encountered during project implementation accounted for the area reduction at Debi, but some of the lost area was compensated for by the Savoigne Perimeter rehabilitation and small village perimaters at Lampsar (para 2.04). Temporary problems with debt repayment at Lampsar and late availability of irrigation water at Savoigne also idled about 300 ha in 1986/87. 3.05 Paddy continues to be cultivated only once per year with the exception of a small second crop of 200 ha first attempted in 1986. A second crop is currently limited by uncertainty over the availability of non-saline irrigation water, a situation which will persist at least until the Manantali Dam on an upstream tributary of the Senegal River is fully operational, anticipated in 1989. The Diama Dam near the mouth of the Senegal River, designed to prevent the upstream movement of salt water as the fresh water flow declines in the dry season, is not at present capable of guaranteeing this function. Lacking a dike on the Mauritanian side of the river, as well as an adequate flow of fresh water in the dry season, Diama operators are not able to maintain a high enough water level behind the dam for the design of the watergates. These ha-ve to be opened when the downstream water level rises too high, thus letting salt water pass into the upper reaches of the delta. Diama's operators feel assured that this problem will be resolved once Manantali Dam permits regulation of an adequate, year-round flow of fresh water, even without the Mauritanian dike. This will ensure sufficient water for irrigation of additional crops per year, potentially two rice crops plus a fast maturing vegetable or legume. However, SAED personnel feel that initially a cropping intensity greater than 160% will be hard to achieve, particularly at Lampsar.  - 10 - SENAL DEIl-LAHPSAR !RRXGATION PROJECT (Credit 775-SE) PROJECT COMPLETION REPOR TABLE 2: lacremental Aricultura1 Production Appraisal Actual 1/ Without With Project Incremntal With Incremental Project at MaturitV Benefits Project Benefits PADDY Lampsar Area (ha) 1,200 1,800 2,000 Yield (T/ha) 0.8 3.5 4.5 Production (T) 960 6,300 5,340 9,000 8,040 Debi Area (ha) 500 1,100 730 Yield (T/ha) 0.8 3.5 5S.5 Production (T) 400 3,850 2,570 4,01S 3,615 Total Paddy Area (ha) 1,700 2,900 2,730 Yield (TSha) 0.8 3.S 4.8 Production (T) 1,360 10,130 8,790 13,015 11,65S TOMATOES Area (ha) 3400 0 0 Yield (T/ha) 30 0 0 Production (T) - 10,200 0 0 / 1986/87.  I - 11 - Reasons given are proximity to St. Louis, which offers alternative activities to LaApsar residents, who are in effect part-time farmers. Simple habituation to year-round cultivation is also thought to be at least a transitional problem. 3.06 At Lampear, the total area available has not been cultivated beginning in 1985 with the implementation of a new SAED policy for farmers not repaying their prcduction credits. SAED has begun to refuse additional production credits until farmers repay these debts, and at Lampsar, the land thus idled -- about 10X of the total - has not been redistributed or otherwise cultivated. The problem has not arisen at Debi, where debt repaynent of nearly 1002 is the norm. In fact, the situation appears to be particular to Lampsar; at other perimeters where the SAID policy is being applied to non-repayment of debt, the affected land is distributed among existing or new cultivators. But at Lampsar, village organization is such that farmers are so far unwilling to lay claim to the idle land of their neighbors. 3.07 Effectiv- area benefitting from irrigation is greater than official area at Lampsar because of water "pirating'. This consists of unauthorized use of perimeter irrigation water on adjacent parcels by single cultivators or informal village groups. The area involved at Lampsar is thought to have reached 200 ha, or a 102 addition to official area. Rice is usually cultivated. Such water pilfering is quite destructive to the irrigation earthworks, with syphons eroding primary canal embankments, PCU tubing being extended from secondary canals under roads and over drainage ditches, and even rudimentary canals connecting directly to the primary canal. Control of this situation iF to date limited to SAED intervening to impose water charges where "pirates" can be identified; actual suppression of the activity, especially during the recent drought years when rainfed agriculture was foresaken, would have created frictions with villages that' SAED preferred to avoid. Resolution of the problem will have to come through self-policing by cultivators, and may be aided by recent changes in perimeter institutions (para 5.11). 3.08 Yield performance has far outpaced appraisal expectations for paddy. These were projected to increase from 1.8 T/ha initially to 3.5 T/ha over four years. Initial yields already exceeded 3.5 T/ha, and through 1986 averaged 4.5 T/ha at Lampsar and 5.5 T/ha at Debi. The higher level of yields does not have any apparent explanation; total fertilizer use is in fact a bit iower than appraisal recommendations, although there is a higher proportion of urea being applied. 3.09 Yields at Debi are quite stable, but at Lampsar are more variable, and fell to 3.9 T/ha in 1986. The reason appears to be Lampsar's greater susceptibility to the late arrival of flood waters. There is an additional delay at Lampsar, since once the flood waters push back the salt water intrusion, the Lampsar marigot must still be filled before irrigation of the sub-perimeters becomes possible. Hence the date of planting at Lampsar is often later than at Debi, and this is related to lower yields. In fact, at the Savoigne perimeter, which depends on gravity flow for the primary canal from the Lampsar marigot, the arrival of water was so late in 1986 that it was decided to completely forego planting in the main season.  - 12 - 3.10 The reason for higher yields at Debi than at Lampsar is not entirely clear. Physical inputs for the two perimeters are almost identical. One suggestion is that Laspsar farmers, being quite close to St. Louis, have more varied activities and do not perform their crop operations in as timely or complete a fashion as at Debi. Agronomic practices, levels of input use, and soil quality are quite similar between the two areas, and in the absence of a clear technical basis for the yield difference, managerial explanations carry some weight. 3.11 A more perplexing yield difference exists between the above results, based on SAED crop cutting samples, and farmers' recall estimates, summarized in a 1986 SAED survey at Lampsar. Farmer estimates averaged 3.2 T/ha for 1985, well below the crop cutting estimate of 4.5 T/ha. SAED explains the difference as resulting from farmers not estimating some production as part of their declared total. They may not include quantities consumed by the family during the prolonged harvest and threshing period, in-kind payments to hired labor, gifts to parents, and religious tithes. If all these deductions do in fact occur, and are not reported as part of production, then the 4.5 T/ha SAED yield estimated does appear valid. 3.12 Yield trends in the future are uncertain because of competing influences. There will likely be a negative short-term impact of SAED disengagement from input provision. Private or cooperative tractor services and market supply of fertilizer and other chemicals are still inadequate to provide sufficient supply at peak-demand periods. Over a longer period, the ordinary practice of dry-soil tillage without subsequent puddling will lead to poor land levelling and lower yields. Farmer Participation and Benefits 3.13 A total of 2627 farm units were recorded by SAED as of 1986, encompassing 1945 units at Lampsar and 682 un'-ts at DEbi (see Table 3). Recipients were typically male heads of households, with average family size estimated at appraisal of eight members. Direct beneficiaries of the project thus total at least 21,000, and may be higher taking into account a recent SAED survey at Lampsar which showed average family size of almost thirteen members. 3.14 Distribution of land was facilitated by participation of permanent village residents in SAED's flood irrigation activities that pre-dated the current project. There were thus existing farmer groups that served, as a basis for determining eligible farmers. Legal ownership of the land was vested in the State by a 1966 law, and SAED was designated to select recipients of user rights. 3.15 Through a reduction of the land allocation per farm from the appraisal provision of 3 ha, the project has spread benefits across a greater number of recipients. A request for revision of the proposed farm model by SAED was motivated by several concerns. Demand was greater than land available using 3 ha farms, and there was no simple way to select among claimants. Productivity in existing SAED projects was observed to be higher on smaller farms, so reducing farm size within the limits of  - 13 - maintaining adequate financial incentives was also seen as a way of reaching higher total production. Finally, there was a concern that simple allocation of 3 ha tracts did not adequately calibrate farm size to variable family labor resources. A rough eotimate of 0,25 ha per active adult was determined by SAED and apparently applied to give bigger families more area. Recorded land allocations average one hectare per farm unit, which corresponds with the four active adults per lamily estimated at appraisal. 3.16 It is doubtful that this average 1.0 ha farm is meaningful as a representative farm in the project area. Recorded average farm sizes by sub-perimeter within Lampsar vary by a factor of over two-and-a-half, from a minimum of 0.7 ha at Keur S. Sow to a maximum of 1.8 ha at Ndioungue. The actual range of farm size as operated in practice may be even larger. While there are no figures on farm size distribution within the project, it is probable that informal social mechanisms have led to some effective reallocation of decision control over land, and that is not reported to SAED. Inflated claims of resident active adults per family, the existence of patron-client family relations, and the consequences of informal credit relations are only several channels by which decision control over land may have become more diverse than the simple one hectare farm model implies. This diversity has important consequences for financial benefits to participating farmers. 3.17 Financial returns to farmers estimated at appraisal are compared to results estimated for 1986 in Table 4. Only returns for rice production are compared since the appraisal farm model with tomato production is not practiced. Appraisal financial returns are adjusted to current values by appplying 1986 prices to appraisal quantities where this is possible. Mechanical services and water charges are treated differently as these were scheduled at appraisal to increase during project implementation, and these fee changes are used for the appraisal model's 1986 costs. 3.18 Actual net production value per hectare exceeds appraisal expectations by 20Z. Higher yields are being realized without significant increases in production inputs, although there is unforeseen substitution of hired for family labor during harvesting and threshing. Actual production value net of costs arrives at 3309 CFA per manday of family labor, exceeding the appraisal of 2594 CFA per manday, and far surpassing peak hired wages of about 1000 CFA per day. 3.19 The cash flow of rice production presents some problems once family consumption of own rice production and the range of farm size at Laupsar are taken into account. Annual rice consumption of an eight member family is perhaps 180 kg/year of rice (paddy * 0.65) per person, or 2.2 tons of paddy per year total. If families consume just this amount, then on an average 1 ha farm, marketings would be 532 of production. After paying for production costs (usually paid in-kind) the remaining cash income is 66,000 CFA ($190), or 857 CFA per manday ($2.45) of family labor. But for smaller farms within the range found at Lampsar, the cash value of production remaining after household consumption appears insufficient to cover production costs. For the 0.7 ha average farm of Keur S. Sow, household food needs absorb two-thirds of paddy production, and the  - 14 - SENRAL DEUI-LAPSAR IRRIGATION PROJECT (Credit 775-SE) PROJECT COaPTm ON REPOWr TABLE 3s Beneficiaries and Averame Farm Sin Perimeter Village Section Number of Average Area Average Family Average Area Per Farms Per Fare Site d/ Farm Adult d/ (ha) (ha) tAMPSAR- Laupsar 101 1.25 14.1 0.13 Bifeche 44 0.96 8.2 0.18 Savoigne 74 0.92 14.6 0.12 Ndelle 218 1.22 14.9 0.13 Keur S. Soa 301 0.69 1 9.0 0.15 Tilen. 181 1.39 12.1 0.18 Dtagambale 317 0.93 15.9 0.10 1 Ndiaye 155 1.13 12.7 0.17 Boundoum Peul 378 0.97 10.6 0.17 Ndioungue 95 1.81 h 13.4 0.19 h Roubantine 81 0.96 9.8 0.1S Sub-Total 1945 1.05 12.9 DEBI b- Debt 393 1.12 n.a. n.a. Tiguet 289 1.00 n.a. n.a. Sub-Total 682 1.07 n.a. n.a. 50IA S/ 2627 1.06 n.a. n.a. a1 1985-86. bf 1986-87. e/ Using average family size at appraisal of 8 members, there are about 21,000 beneficiaries. Using 1986 survey results from Laupsar, average family size Is 12.9 members, giving about 34,000 beneficiaries. d/ Family site and compositton data are from a 1986 survey conducted by SAED. See para _ for discussion of these data.  - 15 - residual marketings fall short of input costs by almost 10,000 CFA. Such households evidently compensate by falling behind in debt payments, by paying for production costs of paddy from other sources of income, or by consuming less rice from own production (see para 6.06). 3.20 While most farms at Lampsar and Debi do not currently face this cash flow problem, they are not far removed from its consequences. First, there is little scope for accommodating ongoing population growth in the project villages through sub-division of existing allocations. Second, although irrigation water fees are well below costs for its provision, raising them would fall particularly heavily on small farmers, even while larger farmers could feasibly absorb the higher rate. 3.21 Farmers have also benefitted from the opportunity to improve their farm management skills, in addition to the incremental financial rewards. The techniques of intensive irrigation production of paddy were new to the farmers at the project's outset, and were learned over a relatively short period of time. Timing and application of chemical fertilizers, use of herbicides and pesticides along with maintenance of spraying equipment, proper timing of seeding and cultural operations, and management of water resources are all skills that farmers have begun to master. These fundamentals of more sophisticated, productive, and flexible farming are transferable to other crop activities of farmers. 3.22 Not all the potential or desirable development of farm management skills has occurred, however. The substantial direct involvement of SAED in production and marketing has isolated farmers from a range of decisions. How to transfer these additional responsibilities to farmers is now a pressing problem as SAED seeks to disengage from these activities. Mechanized land preparation and input supply has in the past been provided by SAED, but farmers are in the future going to have to seek these inputs from private suppliers or perhaps through cooperative organization of tractor services. SAED will continue to provide a market outlet for paddy at an official producer price, though finance difficulties often delay or limit the purchasing campaign. But the high official price is usually sufficient to keep farmers from making sales on the parallel market. In short, farmers need to actively seek out and develop market relations for input supply and paddy sales was limited until recently. Whether they can acquire these economic management zkills apace with SAED's scheduled disengagement is a key unknown for the immediate future of the project. Other Non-Quantifiable Project Effects 3.23 Health. The overall health impact of the project is difficult to assess because improved food availability and access to potable water have coincided with an increase in water-borne or related diseases. On the positive side, 79Z of a Lampsar sample perceive greater food availability, accompanied by a switch from a millet-bssed to a rice-based diet. Debi's population has benefitted greatly from the construction of Diama Dam, which for the first time provides them with non-saline water during the dry season. But 452 of Lampsar respondants felt that family health was worse  - 16 - SENEGAL DE8I-LAMPSAR IRRIGATION PROJECT (Credit 775-SE) PROJECT COMPLETION REPORS TABS 4: Financial Returns to Paddy Production (CFA) Financial Returns per Hectare Finantial Returns per Exploitation Without Appraisal Actual Appraisal Actual Pro1ect Forecast ( tha) 1 1 1 3 0.7 1.8 Paddy Production (tons) 0.8 3.5 4.5 10.5 3.15 8.1 Value, Paddy 68,000 297,500 382,500 892,500 267,750 688,500 Value, Ray 3,400 14,900 19,100 13,388 34,400 Gross Producticti alwu 71,400 312,400 401,600 937,200 281,138 722,900 Production Inpt.ts 13,000 42,000 45,400 31,780 81,700 Mechanical Servires - 21,400 21,400 14,980 38,SOO Water Charges - 35,000 41,000 28,700 73,800 Hired Labor - 39,000 27,300 70,200 Financial Costs 13,000 98,400 146,800 295,200 102,760 264,200 Net Production Value 58,400 214,000 254,800 642,000 178,378 458,700 Family Labor (days) 50 82.5 77 247.5 S4 139 Net Production Value per Handay 1,168 2,594 3,309 2,594 3,303 3,300 Marketings (%)- 53 33 73 Harketings Value 212,800 92,863 500,200 Net Cash Incoe 66,000 - (-9,897) 236,000 Net Cash Income per Manday - - 857 - (-183) 1,638 1/ Leapsar V Remainder from total production after subtracting autoconsumption (180 kg rice * 1/.65 paddy conversion * 8 people - 2215 kg/yearlfamily). Marketin.p inctude in-kind payments for production inputs, including hired labor.  - 17 - than before, with 761 of all families remarking an increase in maladies, and 991 observing a greater problem with mosquitoes. This health problem is related to the longer period of standing water in irrigation systems and marigots, and perhaps to greater water contact of the project's population through the convenience of the irrigation networks cement structures as bathing and laundry spots. The environment for mosquitoes and water-borne disease vectors will become even more propitious after 1989 when irrigation water for year-round cropping will become available. Preventive health measures through prophylactics, canal maintenance, and mosquito control are minimal, as is SAED's collaboration with health services. Definition of institutianal responsibility and allocation of resources do not appear adequate to deal with the worsening health situation. 3.24 Family Stability. A Government objective of irrigation development in the delta is to provide economic opportunities for the local population and reduce migration to urban centers. This appears to have occurred on one level, as Lampsar respondants observe fewer family member departures in 922 of the cases, and 94Z more arrivals. Two-thirds of sampled families also claimed a reduced number of social problems wlthin the family. But the control of household heads over family labor is thought by some SAED observers to be diminishing, *in Pirt explaining the recourse to hired labor for harvest activities. Increased opportunities for cash income are apparently contributing to the financial independence of wives and sons, and loosening the ties of the extended family as a work unit. This limited evidence suggests that the local economy may be broadening but that labor arrangements for farm production are becoming more flexible and complex. Hence a more stable economy than existed previously under rainfed conditions is not necessarily translating into greater stability of family structure. 3.25 Job Creation. zecause the project has established a more labor intensive crop activity than existed previously, there has been an increase in productive labor. Much of this increase is accounted for by migrant laborers, who arrive from the Groundnut Basin for paddy harvesting and threshing. These laborers are not currently permanent migrants to the area, as there is only call for their services for at most two months per year. With the development -of year-round irrigation, this situation may change, as within the delta there will be two rice harvests and a tomato harvest for which labor is also hired. More permanent residence may be established by farm laborers under these conditions, with beneficial effects on the local economy. IV. ECONOMIC RATE OF RETURN ANALYSIS 4.01 The rec&lculated rate of return on the project is -42 compared to 102 at appraisal. The rate of return was recalculated taking into account costs of all realized investments and variable production inputs, plus costs after 1986 consisting of periodic replacement of medium term investments, production inputs, operations and maintenance and SAED project salaries stated in 1986 CFA. Costs associated with technical assistance, applied research, and general studies are excluded. Benefits resulting from paddy production are also in 1986 terms, with future prices of paddy, chemical inputs, as well as exchange rates calculated using World Bank  - 18 - forecasts (see Annex IV for details on the assumptions used for the economic analysis). 4.02 The rate of return calculated at appraisal and in this report is based on incremental net benefits, that is project net benefits greater than those realized from without-project agricultural activities. The recalculated rate of return retains the appraisal assumption that earlier agricultural production on the project site consisted of paddy on 2,900 ha, with an average yield of 0.8 TIha. The low average yields are due to uncertain soil moisture under flood recession agrici'lture, and the absence of purchased inputs. 4.03 The decrease in the rate of return is the result of (i) the inability to cultivate a tomato crop which at appraisal was to account for nearly half of net incremental benefits at Lampsar, and a third overall; (ii) beginning in 1986, sharply reduced CFA value of paddy output, resulting from depressed world market dollar prices, combined with a declining CFA/US$ exchange ratio. In fact, because of this output value decline, the project's net incremental benefit stream turns negative from 1986 to 1992; (iii) inclusion of periodic, post-project costs for medium term investments -- pumps, agricultural and maintenance machinery - without which production for the estimated 25 year life span of the earthworks and civil works will not be sustainable. Such periodic costs were not included in the appraisal calculation; (iv) a project cost overrun of 11% in CPA terms; (v) higher labor costs because of higher shadow wages in constant terms. Labor costs per hectare are evaluated at the market costs of hiring labor for harvesting (18,000 CFA/ha) and threshing (350 CFA/80 kg sack), plus a subsistence wage of 150 CFA/day for family labor that is used for the remaining 77 days of cultural operations. The opportunity cost of labor per hectare of paddy, calculated on the basis of a 4.5 ton yield, is thus 50,000 CPA; and (vi) shortening of the project life from 50 to 25 years. Fifty years seems excessive in view of the rapid deterioration of past irrigation investments in the delta, and SAED's own depreciation schedule which gives irrigation works a 20-year life span. Appraisal sensitivity calculations estimated that a 10% fall in benefits or a 102 increase in project costs would each cause a 12 decline in the internal rate of return. Extrapolation of these relations and shortening of the project life span explain much of the realised decline in economic profitability of the project. 4.04 Because there is considerable uncertainty over future effects of institutional, paddy price and exchange rate changes on the potential stream of project benefits, sensitivity analysis was used to determine factors which would produce a positive internal rate of return. An optimistic crop scenario still envisions only paddy production, but projects a cropping intensity of 200% from 1989 on, rather than the base case 160%. The resulting IRR only improves to -2%, because at the low paddy prices through 1993, the additional area merely adds production which occurs at a net economic loss. An alternative scenario substitutes a more stable, medium-term exchange rate profile for the base case's "#overshooting" assumption. The latter, which has the CFA/US$ rate strengthening from 349 in 1986 to 254 in 1989, is replaced by a profile which strengthens to 300 in 1987, returns to 325 through 1991, then  - 19 - averages 367 for the remainder of the project. With this alternative exchange rate scenario, the IRR rises to -3%. Finally, the project life was extended from 25 to 50 years, resulting in the IRR improving to 2%. Benefits in distant years have a fairly major impact on the IRR in this case, because its low value means such benefits are discounted less. 4.05 Note that in these IRR calculations the Manufacturing Unit Index (MUV) deflator has been used to adjust project investment costs to get constant 1986 CFA values. This is consistent with treatment of paddy output valuation, but as a practical matter, it is unclear whether using a domestic GDP deflator for project costs would not be preferable. The latter, because of high Senegalese inflation, increases investment costs in 1986 terms, and results in a base case IRR of -7%, compared to -4% using the MUV. Summing these three scenarios, for which the changes in IRR are two, one and six points respectively, or nine points in total, yields an IRR of 5%. V. INSTITUTIONAL PERFORMANCE AND DEVELOPMENT Organization and Administration 5.01 The implementing agency for the Debi-Lampsar Project was the Societe Nationale d'Am6nagement et d'Exploitation des Terres du Delta du Flewve Senegal (SAED). Just prior to project preparation, SAErD had been reorganized to adapt to its increasing responsibilities throughout the Senegal River Basin. This initial reorganization, which grouped perimeters into "project units" of about 5000 ha each, and created new divisions of Perimeter Coordination, Finance and Accounting, and Administration and Personnel, left several issues unaddressed that were important to the new project. While these were recognized in the appraisal report, the proposed solutions were overtaken by new difficulties and further organizational changes beginning in 1981 that altered the fundamental legal, procedural and structural characteristics of SAED. Many of the changes addressed institutional issues more general than the specific organizational needs of the Debi-Lampsar Project, and hence will not be treated in depth in this report. 5.02 SAED was created as a state corporation, and as such had to comply with complicated financial and administrative procedures. Most serious among these was the extremely tight control exercised by the Government over bidding procedures and expenditure. Accounts were kept by a central agency in Dakar for all public corporations, with almost all contracts and expenditures requiring complex clearance by this central agency. Prior experience was that this system was fraught with delays, inefficiency and deterioration of SAED's financial situation. To avoid these problems during project implementation, the credit opening was made contingent upon signing of a financing agreement by the Government assuring regular budget transfers to SAED's account, and a decree easing the requirements on prior approval of expenditures. 5.03 These limited procedural reforms were superseded by the recommendations of a joint CCCE/FAC/IDA review of SAED, which reported in March 1979. These were that (i) SAED's financial autonomy be extended  - 20 - through a change in legal status; (ii) SAED's activities be focussed on paddy production and maintenance of irrigation works, while abandoning execution of irrigation works and studies implementation; (iii) SAED's central services be organized under three new divisions -- Administration and Finance, Production, and Engineering; (iv) project perimeters be given greater autonomy throughl staffing with higher level managers, and suppression of the Coordination of Perimeters Division; (v) reduction of SAED's staff be implemented. This reorganizational approach was accepted at an interministerial meeting. With financing from FAC and the Public Sector Technical Assistance Project, a three member team of expatriate consultants was in place by March 1980 to work in conjunction with SAED's General Manager to carry out the reorganization. By July 1981, the new organizational structure was in operation, and SAED's change in legal status to a National Society, with greater budget autonomy, formalized by Law N' 81-57. Since, the mutual financial and operational responsibilities of the State and SAED have been defined over three year periods in a contractual Letter of Mission. Staff and Training Issues 5.04 Within this general and evolving attempt to improve SAED's institutional health, the project foresaw specific areas where strong management was necessary for proper project implementation. Funds for five technical assistants were provided, one for improving financial management procedures at SAED headquarters, a project director and mechanic for the Lampsar perimeter, and two applied research specialists. All five were expatriates. The project director for Debi was selected from within SAED. The strong pressure from the World Bank to include expatriate technical assistants in the project apparently was the source of some friction during project preparation. This did not cause any specific problems, but there was a notable absence in the project documents of a strategy for training SAED staff to eventually replace the expatriates. The World Bank was unwilling to finance overseas training for a SAED staff member who was to assume responsibility as director of Lampsar perimeter. Bilateral funds for this were eventually arranged on an ad hoc basis. This donor was involved from the beginning in project financing, and the World Bank could have shown more sensitivity to the expatriate issue by encouraging incorporation of a SAED staff training component in the project. As it was, any such training was left undefined and dependent upon an, as of then, incomplete training project under separate donor financing. 5.05 Greater perimeter autonomy was planned for Debi and Lampsar to give the project directors sufficient authority and flexibility to ensure timely completion of the investments and beginning of agricultural activities. Decentralisation was to be achieved through separate project management teams with technical and administrative independence, and partial financial autonomy via establishment of separate project accounts. In the initial recruitment stage, Lampsar was staffed with two extension officers, seven extension agents, and a procurement officer in addition to the expatriate project director and mechanic. Despite some difficulties in establishing operational procedures between the project and SAED headquarters, these were largely resolved by 1983, when decentralization into Autonomous Perimeters was institutionalized throughout the delta.  - 21 - 5.06 One consequence of devolving greater responsibility to the perimeter level was a rapid growth in project staff. Lampsar personnel expanded to 48 employees, consisting of 11 administrative staff, 17 extension workers, 9 tractor drivers, and 11 workshop employees. With SAED's disengagement from production services, the number of personnel will be reduced, although a specific calendar has not been establlshed. Production-oriented training for farmers and SAED personnel was not a specific component of the project since other donors were financing a separate but complementary training project for SAED. This included construction of a training center at N'Diaye, within Lampsar's area. The center became operational about when Lampsar investments were completed, and has since provided valuable training to operational staff of SAED. 5.07 Pump operators were recruited from project villages and given practical training. Thirteen pumpists were selected from village sections (para 5.10) at Lampsar by means of a simple examination. They were then given two months of practical training at already wuctioninug pump statiorns in the Delta, plus two months of mechanical training at the workshop in Ross-Bethio, and one month of course work. Tractor drivers were villagers who had previous experience as drivers for SAED elsewhere in the Delta. An additional six drivers who were trained for the project were villagers who had previously worked for SAED in other capacities. Farmer Organizations 5.08 Since the beginning of the project there has been a significant evolution of farmer organizations which collaborate with SAED to manage the perimeters. There are now three levels of organization, with farmers represented at each level. Along with the development of these organizations, there has been a fundamental shift in SAED's approach from firm direction to collaboration. 5.09 Farmer groups have existed since the project origins, and consist of about forty farmers who cultivate adjacent plots provisioned by the same secondary canal. These groups have a simple administrative structure including a President, Treasurer, and pump operator. The positions are elective. SAED has assisted these groups through literacy and administrative tra