Mauritania - Gorgol Irrigation Project

Document of The World Bank FOR OFFMCIAL USE ONLY Repow No.7566 PROJECT COMPLETION REPORT MAURITANIA GORGOL NOIR IRRIGATION PROJECT (CREDIT 1068-MAU) DECEMBER 29, 1988 Africa Regional OZfice Wsi document ha a resdrtd dItrbuton *md may be _ed by rme4pnts miy It dte performne d thir offiei dudes.Is cont nts my n oderwie be disclosd withot Wodd Bank auhorludon.  MAURITANIA GORGOL NOIR IRRIGATION PROJECT (CREDIT 1068-MAU) PROJECT COMPLETION REPORT Monetary Unit : Ouguiya (UM) US$ = 73 UM (1987 average) UM = US$0.014 (1987 average) List of Acronyms ADPAED Abu Dhabi Fund for Arab Economic Development GSA Commissariat a la s6curit6 alimentaire EDF European Development Fund FAC Fonds d'aide de cooperation (France) FND Fonds national de d6veloppement GOL Government of Libya GPC Groupement pre-cooperatif GPD Gorgol Project Directorate IDB Islamic Development Bank IFAD International Fund for Agricultural Development KfW Kreditanstalt far Wiederaufbau MDR Minist4re de developpement rural SDF Saudi Development Fund SONADER Societe nationale pour le developpement rural  FOR OFMCIUL Us ONLY THE WORLD BANK Washington, D.C. 20433 U.S.A. Olhte W Dwect.GeWt4a Operatom hwalitinsf December 29, 1988 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Pass-Through Audit Report on Mauritania Goraol Noir Irrigation Proiect (Credit 1068-MAU) Attached, for information, is a copy of a report entitled 'Pass-Through Audit Report on Mauritania, Gorgol Noir Irrigation Pro4ect (Credit 1068-MAU)" prepared by the Africa Regional Office. Full evaluation of this project has not been made by the Operations Evaluation Department. Attachment This document has a restricted distribution and may be used by recipients only in the performance of their oMcial duties. Its contents may not otherwise be disclosed without World Dank authorition.  FOR OFFIACIL USE ONLY MAURITANIA GORGOL NOIR IRRIGATION PROJECT (Cr. 1068-MAU) PROJECT COMPLETION REPORT Table of Contents PaRe Preface ..................................................... Basic Data Sheet ............................. *.* ............ Evaluation Summary ........................... v I. INTRODUCTION ...................................................1 II. PROJECT IDENTIFICATION, PREPARATION AND APPRAISAL .... 2 III. PROJECT IMPLEMENTATION ... ... . . ............. . . 6 IV. INSTITUTIONAL PERFORMANCE AND DEVELOPMENT ............ 12 V. AGRICULTURAL IMPACT AND ECONOMIC RE-EVALUATION .,.,,,. 17 A. Impact on Agricultural Production .........**,, , 17 B. Economic Re-evaluation ......................... 19 VI. BANK PERFORMANCE ...... .......... ................. 20 VII. CONCLUSIONS AND LESSONS LEARNED ..................... 21 Annex: Comments from the Saudi Fund for Development ...... 37 Maps: IBRD Nos. 13753R1 and 13754R1 This document has a ratricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authoriation.  - i - MAURITANIA GORGOL NOIR IRRIGATION PROJECT (Credit 1068-MAU) PROJECT COMPLETION REPORT PREFACE This is a Project Completion Report (PCR) for the Gorgol Noir Irrigation Project, for which Credit 1068-MAU was approved in September 1980 for the sum of US$15.0 million. The Credit was almost fully disbursed and was closed in March 1987. This PCR was prepared by the Agriculture Division of the Sahelian Department, Africa Region (AF5AG) and the IBRD/FAO Cooperative Programme, and was based on: a review of Staff Appraisal Report No. 2517a-MAU dated August 21, 1980; the Development Credit Agreement and Project Agreement, both dated November 21, 1980; the draft PCR prepared by SONADER's Gorgol Noir Irrigation Project Directorate and dated March 18, 1988; review of supervision reports; and visits to the project site and discussions with project staff as well as with SONADER Head Office staff in June 1987. No further evaluation of this project has been done by the Operations Evaluation Department (OED). The draft report was sent to the Borrower and cofinanciers for comment on October 7, 1988. Comments from the Saudi Fund for Development have been taken into account, or answered in the margin of the telex which appears as an annex in the report.  - ii - MAURITANIA GORGOL NOIR IRRIGATION PROJECT PROJECT COMPLETION REPORT Basic Data Sheet KRY PROJECT DATA Item Appraisal Revised Actual or Actual as % of Estimate Estimate l/Estimated Actual SAR Estimate Project Cost (US$ million) 93.2 93.2 93.2 (end 1988) 100 Overrun (Z) - - 0 Credit Amount (US$ million) 15.0 15.0 15.0 - Disbursed 15.0 15.0 14.8 (Dec. 1986) 99 - Cancelled ^ - 0.1 - Reimbursed - Achievement of Physical Targets Completion Date of Dam 12/82 8/83 6/83 Conveyance Canal (km) 3.5 3.5 3.5 100 Main Canals: P1 (km) 12.3 10.9 10.9 89 P2 (km) 21.1 15.1 15.1 72 Main Drains (km) 18.5 15.0 15.0 81 Main Roads (km) 40.0 40.0 40.0 100 Secondary Canals (km) 90.0 69.0 69.0 t/ 77 Secondary Drains (km) 105.5 70.0 70.0 66 Road Network (km) 149.0 85.0 7.0 5 1/ Although the project cost has remained unchanged, the targets were scaled down. 2/ Of which 60 km have only embankments constructed.  - iii - Tertiary Network and Agricultural Development 3,600.0 ha 2,000.0 ha 600.0 ha 17 Rice Mill 1.0 -- -- Economic Rate of Return Appraisals 7.4 2.7 Financial Performance: Poor Institutional Performance: Satisfactory Cumulative, Estimated and Actual Disbursements (UM Millions) 1981 1982 1983 1984 1985 1986 Estimated 4.7 8.8 12.2 13.2 14.2 15.0 Actual 3.1 7.0 8.6 11.4 11.6 14.8 Actual as 2 of Estimated 66 80 70 86 82 99 Country Exchange Rate Wmsu of Currency (abbreviation) Ouguiya (UM) Exchange Rates US$1 = UM Appraisal Year (1978) 46.2 1981 - Average 48.3 1982 - Average 52.2 1983 - Average 54.8 1984 - Average 63.8 1985 - Average 77.1 1986 - Average 70.0 1987 - Average 73.0 STAFF INPUTS L 2L 7JL 7 2JL 80 i JL __L _ M4 L -i 8 - InMa Pr.appralaaI 0.2 0.8 80.8 9.6 Z1.0 7.8 - - - - - - - - - 97.7 Appraisal - - - - 48.8 39.9 8.8 - - - - - - - 94.0 Ne"otiation - - 5.0 - - - - - - - 5.0 Super.vion - - - - - - - 11.0 18.6 84.0 12.2 13.9 1.S 6.2 11.2 116.8 OthMr 0 -4 0 .1 4_ _ 0.1 _ _ _ _ & _ 4.0 Tatal 0.2 0.8 80.8 9.6 29.4 87.0 40.0 21.9 18.7 34.0 12.2 18.9 9.8 8.2 11.2 317.1  - iv - Other Proiect Data First Mention in Files June 1974 Government's Application February 1978 Appraisal 9/26 - 10/6/78 Negotiations 7/30/80 Board Approval 9/16/80 Credit Agreement Date 11/21/80 Effectiveness Date 11/19/81 Closing Date 03/31/87 Borrower: Islamic Republic of Mauritania Fiscal Year of Borrower: January 1 - Dec. 31 Executing Agency: Societ4 nationale de developpement rural (SONADER) Follow-on Project Name: Irrigation Improvement Project Mission Data Item Month/Year No. of No. of Man- Date of days persons 1/ Weeks Report Re-identification 1976 Preparation Feb. 1978 Pre-appraisal May. 1978 Appraisal Sept. 1978 Total Supervision 1 Feb. 1981 6 le 1 03/17/81 Supervision II June 1981 10 2e, a 3 07/20/81 Supervision III Nov. 1981 9 3e, e, a 4 12/05/81 Supervision IV Nov. 1982 12 3e, f, s 5 01/17/83 Supervision V Apr. 1983 16 3e, a, 1 7 04/07/83 Supervision VI Nov. 1983 10 le 2 11/30/83 Supervision VII May 1984 12 2e, a 4 06/29/84 Supervision VIII Dec. 1984 10 2e, a 3 12/19/84 Supervision IX Apr. 1985 6 2e, a 2 05/31/85 Supervision X Dec. 1986 7 le 1 01/21/87 Completion June 1987 7 2/ 3e, a, b 3 11/03/87 Total 105 35 2/ 1/ e = engineer; a = agriculturist; f = financial analyst 2/ Mission combined with identification/preparation of second phase;  - v - PROJECT COMPLETION REPORT Mauritania Gorgol Noir Irrigation Project (Credit 1068-MAU) EVALUATION SUMARY 1. The Project was the third Bank Group operation in the irzigation subsector in Mauritania. It was designed to provide irrigation atvd land development for some 3,600 ha, with the objectives of reducing the country's dependence on food imports and increasing the incomes and living standards of some 4,600 families by stabilizing their agriculture. The Project, as planned at appraisal, included: (i) construction of a dam and water conveyance works; (ii) construction of the Gorgol perimeter. including irrightion canals, drains, land levelling, access roads, agricultural inpu:-' and vehicles, and community buildings and facilities (schools, dispen _..is, etc.); (iii) provision of soil preparation and threshing services, inputs, credit, and extension and marketing services; (iv) technical assistance and equipment and vehicles for execution and initial operation of the Project. 2. Due to lengthy negotiations, late award of contracts, and problems with some of the nine external financiers, project effectiveness was delayed by almost a year (from January 1981 to November 1981.) A number of serious problems surfaced in the early implementation phases (paras 25-27, 28 (a)). The most important of these was a major cost overrun of which it was estimated in mid-1982 that, were it allowed to continue, it would amount to 52Z by project's end. The cost overrun on the dam and conveyance works was not major and arose mainly from the aforementioned delay before project effectiveness. The more serious overrun involved the tertiary network and on-farm development, and arose from: the abandonment of the force account approach, which in any case had been severely under-costed; substantial increases in prices and quantities over appraisal estimates; and the inclusion of facilities, machinery and staff not originally foreseen. 3. Since co-financiers and Government ruled out additional financing to meet the increased costs, the Project underwent a substantial reformulation in 1983 (para 28 (b)). This involved a significant reduction in the area to be irrigated and major revisions in the plans for tertiary networks and on-farm works, in the settlement program and in agricultural development. The dam retained the same desiga, but the primary, secondary and tertiary irrigation and drainage networks and roads were scaled back to serve a net area of about 2,000 ha (reduced from the 3,600 ha proposed originally). The settlement program brought in 1,100 families (instead of 4,400 at appraisal), and under the agricultural development component, all farming operations avre made the responsibility of beneficiaries. No rice mill was constructe. and technical assistance was reduced.  - vi - 4. The different components performed differently because of the changes made in the project's content and scope (paras 29-34). The dam was completed according to original specifications, as were the main conveyance canals. The secondary irrigation canal network was only partly constructed. The tertiary drainage network and land levelling works were only carried out on an area of 600 ha, as against 2,000 during project reformulation. All of the planned feeder roads were constructed along the main canals. While settlement and agricultural development were slowed by delays in delivery of irrigation works, the second season of irrigated rice gave an average yield of 4.6 t/ha for wet season rice, a result that slightly exceeded appraisal estimates. Extension services and technical assistance performed adequately overall. 5. The first phase of the Project is clearly a failure in economic terms. The estimated ERR for the overall development program is 32. 6. The major project objectives, i.e., reducing the country's dependence on food imports and ensuring an adequate income to settler-farmers, are likely to be achieved. Also, valuable experience was gained in implementing this first major irrigation project in Mauritania, with probable benefit to future projects to be carried out under the 'post-dam" plan for development of the Senegal River Valley. 7. The major lessons learned are the following: (a) There are no short-cuts to having final designs based on accurate base data for estimating project costs and construction schedules; (b) a large number of donors is best avoided but, if this is unavoidable, firm commitments should be obtained before final negotiations; (c) the project would have benefitted from greater legal input, both in design and supervision; (d) irrigation projects in Mauritania should be monitored regularly with appropriate modifications made by a strong Bank supervision team, as was the case with this Project.  MAURITANIA GORGOL NOIR IRRIGATION PROJECT (CREDIT 1068-MAU) PROJECT COMPLETION REPORT I. INTRODUCTION 1. Mauritania's economic performance during the 19709 was characterized by a high investment rate (22 to 322 of GDP) combined with slow growth (annual GDP growth rates of 1.0 to 1.5Z at constant prices). The near stagnation of the economy was due to prolonged drought, misguided investment policies and the war over the former Spanish Sahara. This period was followed by the fall of the 18-year old government and the country's withdrawal from the war. 2. An IMF-supported stabilization program was launched in 1978 and continued in 1980, but the decline continued through the l9POs due to the drop in world prices for iron ore, a worsening of the drought in 1983/84 and the mounting debt service burden (which reached a projected 402 of export earnings on debts contracted up to the end of 1984). Over the decade ending in 1983, the Mauritanian economy grew by an average of 2.32, barely keeping pace with population growth. 3. The Government's response to this situation was to launch a comprehensive Economic and Financial Recovery Program for the period 1985/88 in collaboration with the Consultative Group of donors for Mauritania. This program involves a range of economic policy adjustments and institutional reform measures, including flexible exchange rates, encouragement of the private sector, reform of the banking, public and para-public sectors, and an investument program giving priority to rural development, mining, infrastructure development and public services. It is still too early to judge the program's effects, but prelimin&ry results appear encouraging. 4. The Government continues to place high priority on the development of irrigation in order to both diversify the economy and reduce its dependence on food imports. The unreliable rainfall does not allow rainfed agriculture to be considered as a viable alternative for reaching these goals. The country's irrigation potential is about 150,000 ha (140,000 along the Senegal River and 10,000 ha in the Gorgol Valley) of which only about 13,000 ha have been developed. The latter includes about 4,000 ha developed and irrigated by individual farmer with almost no Government help. 5. The Bank's association with the 4rrigation sub-sector dates back to June 1974 when, on the basls of a UNDP-financed feasibility study, a US$ 1.1 million IDA credit (S-16-MAU) was provided to finance preparation of the final designs of two dams and the first phase of the irrigation development of the Gorgol valley. In FY77, an IDA credit for US$ 3.5 million was made available for the SONADER Technical Assistance Project (Credit 694-MAU) which was intended to strengthen the project preparation  and implementation capacity of SONADER (Soci4td nationale pour le developpement rural), which took over these functions from the Rural Engineering Department of the Ministry of Rural Development (MDR). 6. Other Bank 'roup operations I.n Mauritania in the irrigation sector include: the Gorgol Irrigation Project presented herein (Credit 1068-MAU; US$ 15.0 million, 1981); the Second Technical Assistance to the Rural Sector (Credit 1414-MAU; USS 8.1 million, 1983); and the Small Scale Irrigation Project (1571-MAU; US$ 7.5 million; 1985). 7. The Gorgol Irrigation Project was designed to provide irrigation and land development for some 3,600 ha at a total cost of US$ 93.2 million with the objectives of reducing the country's dependence on food imports and stabilizing the agriculture and thus increasing incomes and living standards of some 4,600 farm families. As expected at appraisal, the Project was completed in September 1986 and the IDA credit closed in March 1987. 8. This Project Completion Report (PCR) is based on a review of the Staff Appraisr.l Report (SAR) 2517a-MAU dated August 21, 1980; the Development Credit Agreement tDCA) and Project Agreement (PA), both dated November 21, 1980; the draft PCR prepared by the Project Directorate of SONADER; and the project supervision reports. It was prepared following visite to the project site and discussions with project staff as well as SONADER head office staff in June 1987. II. PROJECT IDENTIFICATION, PREPARATION AND APPRAISAL Identification 9. Irrigation development of the Gorgol River Valley (see map) was always favored by the Government because (i) it would increase and stabilize Mauritanian cereal production and reduce dependence on food imports; (ii) it could be the nucleus of development in the Southeastern region which lacks other development possibilities; and (iii) it did not depend on international agreements or water control structures in other countries as was the case with development of the Senegal River. Project identification started in 1971 with studies financed by UNDP. On the basis of the resulting preparation report, IDA financed under the Credit S-16-MAU the preparation of final designs for two proposed dams (El Bir and Foum Gleita) and for the first phase of the irrigation development program. However, when this work was completed in April 1975, the Bank refused to appraise the Project for the following reasons: a) There were serious doubts about the technical suitability of the design of the El Bir dam, and further topographic and engineering studies were deemed necessary; b) On the basis of past experience, it was felt that the Mauritanian irrigation development agencies did not have the  managerial capacity needed to handle medium to large-scale irrigation projects; and c) While a preliminary evaluation of the whole 10,000 ha project indicated that it would be viable, the first phase involving the construction of both dams and the development of 3,800 ha was expected to yield a very low rate of return. 10. A revised first phase, limited to the construction of only one dam (Foum Gleita) and the downstream development of 3,600 ha, was then identified on the basis of a Bank desk study and a field visit by an FAO/CP mission. This alternative was accepted by the Government after extensive discussions with the Bank and FAO. Preparation 11. The additional investigations and feasibility studies required t, prepare this revised first phase were supported by Credit 694-MAU (US$ 3.5 million) which also aimed at strengthening the managerial capacity of the newly established SONADER. Preliminary feasibility studies for the irrigation scheme were completed by Italian consultants in February 1978 and final documents including technical specifications, tender documents, execution plans and agricultural and socio-economic studies, were submitted in September 1978. A special study of the design and specifications of the Foum Gleita dam by a panel of dam experts was completed in April 1979. Pre-Appraisal 12. The Project was pre-appraised in March 1978 and the pre-appraisal report was distributed in May 1978. The Project, as then conceived, had a total cost estimated at US$ 52 million net of taxes, with the cost per family of beneficiaries at USS 11,000 and an ERR around 7Z. 13. Th' low rate of return was attributed to: a) the early stage of development of the sector, with the resulting need for a long lead time to settle farmers, and the high management and training costs involved in building up the executing agency; and b) the exorbitant cost of civil works (which would affect any major investment in Mauritania), mainly due to the high cost of importing technical and managerial skills, the need to transport equipment and materials over long distances on poor roads, lack of local maintenance and support services for equipment, and the lack of competition among contractors. 14. The request to the Bank's Loan Committee for permission te appraise the Project was justified on the grounds that: a) the Bank had a long prior association with the Gorgol Project and had financed much of the preparation work;  - 4 - b) the objectives of the Bank's technical assistance to SONADER had been met and it was now a viable organization capable of implementing the Project; c) there was a shortage of investment opportunities in Mauritania, particularly in the Southern Region; d) on account of the ready availability of funds as grant aid and soft loans, an appreciable proportion of the Gorgol project costs would not give rise to finance charges. 15. The Region was authorized to appraise the Project (despite its low rate of return and high investment cost per family), with the provision that the Project should be reviewed at appraisal, in view of the special conditions prevailing in Mauritania. Appraisal 16. Appraisal took place in September/October 1978. On the basis of the report of the panel of dam experts, the pre-appraisal proposal for an earth-fill dam was replaced by a proposal for a concrete arch dam at Foum Gleita. The resulting revised designs and cost estimates were to be available only in April 1979, when project cost estimates would be firmed up. Slightly higher costs for the irrigation works were also indicated by the engineering studies which were made available at appraisal. The economic prices of project outputs (paddy, milled rice, sorghum and maize) were also revised upwards following appraisal. 17. The appraisal mission's revised estimate of project costs at that time was USS 55 million, allowing for increased physical contingencies on the dam and irrigation costs due to the changes in design. The mission's preliminary calculation of the project's economic rate of return remained at about 7Z. In order to partly bridge the financing gaps that were evident at this stage, it was proposed to raise IDA's contribution to US$ 15 million. 18. Subsequent updating of project costs on the basis of bids made by potential contractors for the dam and irrigation works brought the figure up considerably, to US$ 93.2 (including physical and price contingencies as well as US$ 0.5 million for refinancing the credit S-16 MAU mentioned above). After much debate regarding the acceptability of the project's low ERR, it was decided to consult the Bank's President, Mr. McNamara, before processing the Project further. 19. The President very reluctantly agreed to pursue processing of the Gorgol Irrigation Project on the grounds of: - lack of development alternatives for the people of the region;  - higher project costs, partly reflecting tied procurement required by some donors; and - lack of basic infrastructure in the area. 20. After further refinements, the yellow-cover version of the SAR issued in August, 1980, showed an ERR of 7.4Z. Project Description 21. The Project, as planned at appraisal, included the following componenits (see map): (i) Dam and Conveyance Works - Construction of a concrete arch dam in Foum Gleita with a maximum height of 37 m and length of 95 m to give a storage capacity of 500 million m3; and - construction of a headreach canal with a maximum head capacity of 10.73 m3/s and lpngth of 3.6 km along the left bank, subdividing into a left bank canal reaching the irrigated area 3.0 km further downstream and a right bank canal (with a canal bridge) reaching the irrigation area after 1.9 km. (ii) Gorgol Irrigated Perimeter - Construction of two main canals (26.5 km long), 37 secondary canals (61 km), 360 km of tertiary canals and 484 km of drains; - on-farm development, including land levelling and access roads; - construction of houses, offices, storehouses, a rice mill, social infrastructure (schools, mosques, dispensaries, etc.) and other facilities close to the perimeter; and - provision of agricultural equipment and vehicles. (iii) Agricultural Development Provision to farmers of soil preparation (ploughing) and threshing services, and inputs (seeds, pesticides, fertilizers) on credit, and extension and marketing services. (iv) Technical Assistance  - 6 - Provision of technical assistance and equipment and vehicles for execution and initial operation of the Project. III. PROJECT IMPLEMENTATION Credit Negotiations, Effectiveness and Start-Up 22. Negotiations for the credit took place on July 30, 1980 follJwing technical discussions conducted from July 3 - 10, 1980. The Bank's Board of Directors approved the credit on September 16, 1980 and the agreement was signed on November 21, 1980. Conditions for effectiveness included, in addition to the normal legal opinion and authorization and ratification by Governments the creation of the Gorgol Project Directorate (GDP), appointment of its director, opening of a special operating account and deposit of initial amounts by Government, proof of effectiveness of the agreements with IFAD, IDB, SDF and Libya, and proof of signature of agreements with ADFAED, FAC, EDF and KFW. 3/ 23. The Project was expected to become effective on January 21, 1981. By February 1981, compliance with conditions for which the Government was entirely responsible was well advanced and the only missing document was the legal opinion. Legal problems persisted, as there were discrepancies between the financing plans mentioned in the World Bank's Development Credit Agreement (DCA) and the plans of certain other donors, with respect to the financing of some components; this was delaying payment to the contractors and reimbursement of SONADER, which was pushed to the limit of its own capacity to finance on-going operations. 24. Finally, on November 19, 1981, the IDA Credit became effective, thus rescuing the Projec, from the brink of disaster. However, doubts persisted as to when the Libyan credit would be available and there was no confirmation about the disbursement of IDB and SDF funds. 25. Because of the lengthy negotiations and late award of contracts, the project completion date was put back by one year (as compared with the date envisaged in the SAR) and the implementation schedule was changed accordingly. Other minor problems that arose during the start-up of the Project were: - rejection of the site near the dam selected by the contractors for the work base camp, as SONADER wished to use the buildings later for its project management unit; 31 see list of acronyms.  - 7 - - failure to provide the benchmarks for the topographic survey in the period specified in the works contract, owing to delays in the appointment of the chief of the topography section; - inadequate detail in topographic maps (1:5,000 scale, on which bids were called) resulting in the need for new, more detailed, surveys before construction could start; and - discrepancies between the financial regulations applicable to the Project and those attached to the tender documents. 26. Sociological activities commenced satisfactorily and a protocol was drawn up between the Government and traditional land owners in the project area on the use of the land to be irrigated under the Project. Financing difficulties, however, continued to plague the project, with Libya still not honoring its commitment and IDB deciding not to finance its share of local expenditures. The Government also failed to make timely deposits in the project operating account. 27. The performance of the Gorgol Project Directorate (GPD) was affected by the problems of achieving a satisfactory disbursement routine with ten different financing sources. Senior GPD staff, including the Project Director, were slow to move out to the project site, and compliance with this project condition was obtained only after a threat by IDA to suspend disbursements. Proiect Revision 28. Other problems surfaced during the early phases of project implementation. The more important of these were: (a) The cost overrun. By mid 1982, it became obvious that there was a major cost-overrun problem (see para. on project costs). The overrun on the dam and primary and secondary works was not major and arose mainly from the long delay before project effectiveness. The major overrun was on the tertiary network and on-farm development, and resulted from: (i) abandonment of the force-account approach, which in any case had been severely under-costed during appraisal; (ii) substantial increases in quantities, which had been estimated on the basis of insufficiently detailed topographic work and preliminary designs; (iii) price increases well above appraisal estimates; and (iv) inclusion of facilities, machinery and staff not originally foreseen.  All the co-financiers, as well as the Government, ruled out any supplementary financing to meet the increased costs. There was thus no alternative except to re-design the Project, essentially by reducing the area to be equipped to 2,000 ha. (b) the redesign of the settlement and on-farm development program. The settlement and on-farm development proposals were re-designed in order to permit beneficiary participation in land development, and to reduce costs. The reformulation of the Project involved a significant reduction in the area to be irrigated (see maps) and major revisions in the plans for construction of tertiary networks and on-farm works; settlement program; and agricultural development. The revised Project included (i) a concrete arch-dam to store about 500 MCM of water; (ii) primary, secondary and tertiary irrigation and drainage networks and roads to serve a net area of about 2,000 ha (reduced from 3,600 ha proposed originally); (iii) settlement of 4,100 families vs. 4,400 at appraisal; and (iv) agricultural development in which all farming operations were to be entirely the responsibility of the beneficiaries; (v) the acquisition of rice hullers instead of a rice mill; and (vi) technical assistance. The Proaramme of Execution 29. Preliminary cost breakdowns revealed that substantial overruns were likely to occur under the different work headings. The reappraisal study identified the nature of these overruns and evaluated the sums involved. Table 1 in the annex shows the two appraisals. 30. It was therefore decided to scale down the Project, particularly with regard to the canal network. 31. Concerning the tertiary canals, the original plan had been to have the work done on force account by SONADER, given the experience SONADER was acquiring in developing the Bogue perimeter, another large-scale irrigation area. However, the base studies proved too inaccurate with respect to topographical detail (with maps at 1:5.000 scale) to permit the work to commence. SONADER, therefore, contracted out a further topographical study involving aerial photo coverage and mapping at a scale of 1:2.000. 32. Work on the dam was delayed by a cement shortage resulting from a transport quota dispute between Mauritania and Senegal. The contractor suffered a loss for which it demanded compensation. In order to resolve the disagreement between the two parties (the contractor and SONADER), IDA suggested that the Mauritanian Government appeal to an international mediator. Based on the report of a designated expert, an out-of-court settlement was reached between the Government and the contractor, whose loss was evaluated at FF 68 million.  33. Concerning final on-farm developments, the donors decided (May 1983, Paris) to implement only a first tranche of 600 ha so as to test out the proposed agricultural development program and assess farmer interest in the Project. International tenders issued in July 1983 produced only one bid. Negotiations with the contractor concerned were finalized at UM 170.6 million, t4king into account the work that could be done by farmers. Though not originally planned, supervision of the work was entrusted by extension of the contract to the consultant engineer responsible for supervising the work on the main and secondary canals. 34. The main delays appear to have occurred on the seconidary works (two years) and, to an even greater extent, the tertiary works (nearly four years). This is essentially due to a lack of detail in the basic studies, which had two major impacts on the Project. First, it was impossible to carry out the work without complementary studies. Secondly, and more important, project costs were heavily underestimated and hence the credit negotiation process was based on incorrect figures. Proiect Costs 35. Table 3 gives actual project costs up to December 31, 1986 as against appraisal estimates. The original cost estimate for the whole area (3,600 ha) to be developed during 1980-86 was UM 4,197 million (in 1980 prices). The actual cost during 1981-86 for constructing the dam and headworks and developing an area of only 600 ha (one sixth as big) was, however, around UH 4,000 million (in current prices). Expressed in 1987 prices, these costs were about UM 5,200 million. With the remaining areas under the second phase of the Project (1987/89 - 1,400 ha and 1989/93 - 1,600 ha), the total cost is estimated at about UH 8,000 million (US$ 110 million) at the 1987 rate of exchange). Costs per hectare would thus be about US$ 30,000, of which about US$ 22,000 for the irrigation infrastructure, almost half of it for the dam and headreach and main canals. Performance of Project Components 36. The performance of individual project components varied according to changes in project scale and content. The current level of progress is as follows: - The dam has been completed in accordance with the initial specifications and final acceptance has taken place. - The main canal network has been built in accordance with the  - 10 - revised project objectives. Its characteristics are as follows: Canal Area Conveyance Scheduled Length Built Head Discharge Covered Canal Length (ha) (km) (km) (km) (m3/s) P1 1.145 1.9 10.4 9.0 3.1 P2 2.746 3.1 18.0 12.0 7.3 * The P1 canal crosses the Gorgol via a 60 m span, 5 girder canal bridge. - The secondary canal network was only partially completed. Of the planned 36 canals, only three were completed built, 19 were simply banked and 13 were not even begun. In linear terms, this means 9 km completed, 60 km begun and 20.5 not begun. - The tertiary canal network, land development and terminal on-farm developments covered an area of only 600 ha, corresponding to sectors S9, S10 and S2 (see maps). Work on the remaining 1,400 ha of Phase I of the Gorgol Noir Project is underway and is scheduled to end by mid-1989. - The external drainage network channelling the run-off from surrounding watersheds was completed for almost the entire perimeter (15 out of 18.5 km). This network is needed to protect the main canals and the internal land development. Some of the cross-drainage works originally planned as siphons were changed to aqueducts or box culverts. - The secondary drainage work was completed throughout the entire Phase I area of 2,000 ha, i.e. 70 km. The 35.5 km of Phase II remain. The tertiary drainage network, however, has been built only on the 600 ha that are developed. - The main road network was fully built all along the existing main canals (40 km). The secondary road network, however, covers only the section developed for irrigation (7 km out of 149 km). - Settlement and agricultural development were delayed by slow delivery of irrigation facilities, and the decision by donors to assess the agricultural development on the first tranche of 600 ha before continuing with further downstream development. However, by the second season of irrigated rice cultivation, over 550 ha (discounting the first pilot or demonstration season in 1984 in which 46 ha were cultivated) wet season rice yielded an average of 4.6 t/ha, which was slightly above appraisal expectations.  - 11 - Disbursements and Financial Sources 37. The Project was financed by nine external financiers and the Government. Cost overruns and the redesign of the project necessitated changes in financing. After Libya's default in 1984 (for UM 280 million) other donors had to increase their shares. As of end 1986, the project cost was still within the sum fixed initially, but its scope was substantially reduced: Donor 11 Financing Plan 1980 Actual Financing (end 1986) ................... (UM million). IDA 675 807.35 FIDA 450 296.86 IDB 270 348.46 EDF 733.5 802.95 FAC 243 200.98 KFW 360 532.92 ADF 288 360.92 Libya 450 _ SDF 450 457.30 Government 274.5 153.83 TOTAL 4,194.0 3,957.57 38. Details of the financing plan by donor and sub-component at time of appraisal and the last quarter of 1986, respectively, are given in Tables 4 and 5. Table 6 compares the phasing of the disbursements as estimated at appraisal and in practice. As might be expected, the delays in credit effectiveness and implementation of the Project and changes in its components caused disbursements to start late and to deviate from the anticipated pattern. 39. Delays in disbursement seriously threatened the start-up of the Project. Subsequent delays were caused by difficulties in implementation and with Project revisions, rather than by delays in disbursements, although some minor hold-ups were caused by Government's failure to make deposits in the Project Operating Account. The default of the Libyan Government and the refusal of IDB to finance local costs despite its earlier commitment were, as described before, major causes of delay in the project's execution. 1/ See list of acronyms at the beginning of the report.  - 12 _ IV. INSTITUTIONAL PERFORMANCE AND DEVELOPMENT 40. Establishment of the Project Directorate at Foum Gleita was one of the conditions of granting the credit. The SONADER GPD 41. The Gorgol Projec.t Directorate (GDP) under SONADER is an independently managed and financially autonomous body. It was set up to monitor project implementation and operations (for organizational aspects, see section on agricultural development). Supervision of Works 42. For project execution, the GPO set up a work team, led by a Mauritanian civil engineer and including two supervising engineers and a topographical brigade. It is clear from the size and diversity of the works already carried out, that the current team has acquired good on-the-job training. In all probability, given the experience of the last tranche of Phase I (development of 1,400 ha), the Mauritanian counterparts will be able to assume full responsibility for most of the supervisory aspects of future work. Water Management 43. The Project has not achieved its objectives of water control and rational water management, the main reason being that the perimeter is still not fully developed and water resources therefore have not been a management constraint. Since the supply of water in the reservoir (300 million m3) greatly exceeds the needs of the 600 irrigated ha, the tendency in distribution has been to waste water. Although project management tried to rationalize distribution, this wasteful trend imposed itself due to the difficulty of regulating flow in the network and enforcing irrigation timetables. Some improvements did follow after two years of uneven efforts. 44. Water management is currently organized in accordance with user demands and not according to crop water requirements. Since the project designers opted for upstream water control, any water not used by farmers goes into the drainage network 2nd is lost to the perimeter. The analysis of mean monthly discharges at canal level for 1986 clearly shows discharge to be mismatched with seasonal crop requirements. The monthly volume hovers around a near-constant figure of 5 million m3/month. The average per hectare volume delivered during the 1986 season is 63,430 m3, i.e., more than twice the average crop requirement. Farmers cannot be blamed for this, considering their complete unfamiliarity with irrigated agriculture at the time they joined the perimeter. SONADER staff in charge of this water management are too few in number, insufficiently trained for this kind of activity, and virtually overwhelmed by the plethora of minor defects in the calibration of the irrigation works.  - 13 - 45. Attempts had been made to reduce the nighttime canal flows by half to save water, but nothing came of it due to the number of sluice-gate manipulations involved and the canal side degradation produced by extreme variations in canal water levels. Farmers will have to be taught to irrigate during at least part of the night so as to adjust flow volumes closer to crop water requirements. Maintenance 46. Despite the relative newness of the irrigation infrastructure. major maintenance problems arose as soon as it became operational. This cannot be blamed on the quality of the work done by the contractor. Despite its poor working conditions, this firm did a good professional job (particularly in the case of an earth-made network). The maintenance problems arose in some cases from the restrictive design standards (such as the canal side slopes) and in others from insufficiently detailed norms (such as the cross-drainage box culverts). 47. The organization of maintenance differs depending on the works involved: (a) At dam level, maintenance consists of checking the functioning of the installed electromechanical or hydraulic equipment. Dam stability is monitored by systematic readings of gauges set up on the main structure of the dam. The consultant engineer responsible for engineering supervision was recently assigned to interpret the first set of readings; *b) At the level of the headreach and primary canals, the maintenance service is in the process of providing full protection for the inner and outer sides of the canals. The inner sides cave in and deteriorate through wave action, water wakes and the hooves of animals drinking from the canals. Protection is novel, and consists of sinking schist slabs into shores hit by wave action, particularly in areas where streamflow velocity is high, such as on curves. The external sides and ridge are degraded by rainwater run-off, which causes deep gullying. Protection in this case consists of filling in the gullies with wet earth and covering the ridge and sides with crushed, tamped-down schist rock. (c) Scouring occurs at the iuncture of the earthworks with the concrete-covered areas, usually because the dissipation basin is not big enough and because these areas are used by people for washing or drawing water. Prote tion consists of extending the cement lining with gabions. 1I (d) A series of operations need to be carried out systematically on the perimeter after the rainy season: 1/ The Saudi Fund for Development adds that "erosion of canal bank by rainwater run-off and wind can be prevented by seeding of short growing variety of grass in the unwetted part of the canal. Erosion by drinking animals can be prevented by construction of special places for animal watering and taking stern actions against people watering their animals at places other than those especially conRtrulted plae.es."  - 14 - Weeding and cleaning out the main canals, particularly in sections where water velocity is very low; Cleaning out the cross-drainage works (culverts); Reinforcament of external canal sides and internal drain sides; Cleaning out secondary drains and sometimes specific tertiary drains; and Regrading roads. Extension Services 48. The extension services have been remarkably effective in training settlers with little prior knowledge of farming, or of modern techniques of rice cultivation over two or three crop seasons. The ratio of extension workers to farmers has been high (1 per 96 farmers at present). Despite this satisfactory performance, extension workers still lack basic training. The good results were obtained only through the high level of motivation of the extension officers in charge and through the communication of technical recommendations in small, easily remembered instalments under the "Training and Visit' system. Technical Assistance 49. Technical assistance envisaged at appraisal included 113 man- months for supervision of construction and 546 man-months for implementing the Project. Some adjustments were made in view of revisions of project plans, e.g., shortening of the contract of the Technical Director avd increasing consultancy services to cover supervision of tertiary networks formerly expected to be done by force account. 50. Performance of technical assistance personnel has ranged from excellent to mediocre depending on the individual, but on the whole the jobs got done. However, as might be expected from the number of expatriates involved, some resistance to further inputs of technical assistance has arisen from GDP and SONADER staff generally. This technical assistance "fatigue* would need to be taken into account in future project preparation exercises involving SONADER. Procurement 51. SONADER's past experience with several projects has led to the establishment of effective procurement procedures; no major difficulties were encountered.  -15- Training 52. No specific provision was made for training under this Project, but well designed on-the-job training was provided for extension agents, irrigators and farmers. | Accounting and Reporting 53. The GPD has maintained project accounts satisfactorily as required under the DCA. Reporting on physical and financial progress also appears to be adequate. Agricultural Inputs. Credit and Marketing 54. Agricultural inputs were distributed free of charge during the first major cultivation season (1985 wet season). It was only in 1986, however, that inputs (mainly fertilizer and seed) were given at full cost, but on interest-free credit. The credit was recovered in kind and, by the end of 1986, the rate of recovery stood at 76.7Z. All inputs and credit were provided to farmers by the GPD. 55. Fertilizers were purchased from private dealers who delivered * them to the project site, but often with considerable delays. The situation was saved only by the fact that actual demand had been overestimated in 1985 and left-over stocks could thus be utilized in 1986 * when deliveries were delayed. The same events were repeated in 1987. 56. Although treated seeds of good quality were available from the government seed farm at Kaddi, farmers tended to revert to seed from local sources or to use their own seed. 57. Market outlets for surplus cereals (mainly rice) have been less than satisfactory. Although the Commissariat de securite alimentaire (CSA) theoretically buys the grain at an attractive guaranteed price, it is not usually on the scene when the farmers want to sell (generally soon after harvest) and farmers are compelled to sell at much lower prices (e.g., UM 10 to 12 per kg for rice vs UM 18.5 kg paid by CSA). Nor has Lhe CSA adequate financial resources to buy in sufficient quantities to have a significant impact on prices offered by the private trade. On average, CSA has purchased 30-40X of the paddy produced in the project area.  - 16 - Prices 58. Prices of both agricultural products and inputs were considerably higher in 1987 than at the time of appraisal. The farmgate price for paddy is UM 18.5 per kg compared to UM 10 at appraisal, a 12Z annual increase. This was mainly due to Government's increasing the guaranteed prices of cereals and removing subsidies on fertilizers. The economic prices of cereals have, however, dropped slightly in comparison with those in the SAR, and economic prices of inputs have gone up (Table 7). The economic price of paddy is UM 11 per kg compared to UM 16.6 at appraisal, an annt.-l decrease of 82. Farmers' Settlement and Organizations 59. About 1,100 farmer families, or about 4,100 people, have been settled so far (on the 600 ha) in the project area. The establishment and development of farmers' organizations was not among the original project proposals contained in the SAR. However, on the GPD's own initiative, ten voluntarv farmers' organizations, or Groupements precooperatifs (GPCs), each comprising 50 to 100 members, were set up, thus covering about two-thirds of the families settled so far. 60. In fact, the most obvious way for SONADER and GPD to disengage themselves from activities such as input distribution and maintenance of tertiary canals and drains and to involve the beneficiaries in the construction of social infrastructure is to promote the take-over of these functions by GPCs. Considerable progress has been made in this direction. 61. Overall, the farmers' settlement program has been implemented satisfactorily. However, one of the drawbacks encountered is that people from the same village and ethnic group have not been settled in common irrigation turn-outs, whereas GPCs have been formed along village and ethnic lines. Thus, GPCs often cut across irrigation areas and this makes it difficult to organize canal and drain maintenance operations through them. Covenants 62. In general, the covenants contained in the Development Credit Agreement (DCA) and Project Agreement (PA), have been complied with by the Borrower. Some revisions in the legal covenants of the DCA and PA were proposed at the co-financiers meeting in Paris in 1983 and the legal documents were amended accordingly. A few temporary problems in Borrower compliance that occurred were: - failure to deposit the required sums of money into the project operating account in a timely manner (PA Section 4.01 b); in some instances SONADER's Head Office retained the funds without passing them on to the GPD;  - 17 - - delays in furnishing financial statements of project accounts and auditors' reports (PA Section 4.03 a, ii). However, these lapses did not cause any important delays in the actual implementation of the Project. V. AGRICULTURAL IMPACT AND ECONOMIC RE-EVALUATION A. Impact on Agricultural Production Yields 63. Table 8 compares annual yields projected in the SAR with those actually obtained. In the case of rice, the yields obtained in 1986 (admittedly over a limited area) already exceed appraisal predictions marginally in the wet season (4.7 t/ha actual vs 4.5 tlha predicted), but fall short in the dry season (2.8 t/ha actual vs 4.5 t/ha predicted). Yields of maize and sorghum have been highly variable on the small areas planted and are nowhere near appraisal expectations. 64. Although the average yields of paddy are very satisfactory, examination of yields obtained by individual farmers shows R sizeable proportion of them getting very low yields on the order of 1-2 tons/ha, indicating that there is potential for further improvement. 65. Areas and Cropping intensities. Areas cropped are also below appraisal estimates, mainly due to delays in project implementation. Even the revised targets have not been met as shown by the following tablet Area Cultivated (ha) 1983 1984 1985 1986 DS WS DS WS DS WS DS WS Appraisal a/ 400 400 736 1,200 1,076 2,000 1,432 2,800 Revised (1982) - - - 580 290 980 490 1,360 Actual - - - 46 110 505 320 550 a/ Includes areas originally meant to be cultivated by the Project Authority. Note: DS = Dry Season WS = Wet Season Most of the farmers cultivated rice. Sorghum and maize were grown on small areas, and vegetables, expected to contribute significantly to farm incomes, were not grown at all. This is mainly due to the absence of demand for vegetables in the project area and may also be due to the almost  - 18 - exclusive concentration of the extension services on rice. Cultivation intensities rose from 1.1 in 1985, the first full year of cultivation, to 1.6 in 1986. Crop Production 66. Table 9 shows crop production levels envisaged at appraisal and actually achieved up to the end of 1986. Production of paddy in 1986 was 222 of appraisal estimates. Once again, in the case of rice at least, this reflects mainly the delayed delivery of irrigated land to farmers. Paddy production actually rose from 130 tons in 1984 (over 46 ha) to 1,600 tons in 1985 and 3,150 tons in 1986. On a pro-rata basis, the production in 1986 represents, for 550 ha, more or less the rice production level at full development envisaged in the original appraisal estimate. 67. As would be expected from the small areas cultivated with maize and sorghum and the mediocre results obtained, the production of these crops is well below appraisal projections. Use of Farm Inputs 70. Since no records have been maintained on the use of farm inputs in the project area, an accurate assessment cannot be made. In the first year of cultivation fertilizer was giver. free of charge, while in the second year it was given on interest-free credit although it was agreed during negotiations that a 122 interest rate would be charged. The recovery was combined with the recovery of water charges. It is not clear whether the general resentment shown by farmers to the recovery applied only to water charges or whether they had also assumed that the fertilizer was free once again. Farm Level Impact 71. Table 10 shows representative summary farm budgets for the cultivation year 1986. When compared with appraisal projections (SAR, Annex 2, Table 6), farmers, especially rice farmers, appear to be already doing much better than was expected in terms of the net value of production. The main reason for this appears to be the considerably higher farm-gate cereal prices in 1986 as compared with those that prevailed at the time of appraisal. Since 1986 was only the second full year of cultivation, even better results could be expected at full development when farmers have improved their skills further and yields have correspondingly increased. Cost Recovery 72. It was envisaged at appraisal that Project O&M costs would be recovered by the imposition of water charges on farmers. The raising of water charges to include the recovery of capital costs was thought to be Inadvisable, given the already high rent recovery index. Accordingly, at  - 19 - negotiations it was agreed that SONADER would collect water charges amounting to UM12,000/halyear. 73. No water charges were in fact collected during the first full year of cultivation on account of the relatively low yields obtained by farmers. In the second year (1986), water charges were fixed at UM10,000/ha/year. B. Economic Re-evaluation 74. This economic re-evaluation has been undertaken, with appropriate projections and the analysis has been carried out in 1987 prices, with past expenditures converted to 1987 prices by using the GDP deflator. 75. Actual costs during 1981-86 were converted to economic prices by adjusting the local portion with the standard conversion factor (SCF) of .867. Actual shares of local costs during 1981-86 were not available and the values from the SAR were used. Investment costs during 1987-89 for 1,400 ha were estimated from the contract with the construction company. Projected costs for 1989-93 period for the last 1,600 ha were taken directly from the FAO/CP. No adjustments were made for taxes and duties since the Project is exempted from all identifiable fiscal obligations. Economic prices were estimated for main crops and inputs (Table 7), and a shadow price derived for labor of UM34.2/day. 76. Delays in project implementation influenced the flow of benefits which are substanLially behind schedule. In 1986, season and off-season paddy was cropped on only 792 ha (21% of original projections) and sorghum/maize on only 25 ha (62 of original projections). 77. Financial prices of crops and inputs have increased between appraisal time and the present (Table 7). Economic prices of crops, however, have been reduced by about one-third for rice and by about 20Z for maize. These movements and the lag in benefits flow have reduced project profitability. 78. In addition to the benefits, benefits from fishing and flood recession crop production were also taken into account. It was assumed that because of the Project, incremental production of sorghum and millet have developed around the dam reservoir. The lake has created conditions for flood-recession production. 79. Economic analysis on the above assumptions shows (Table 11) that the estimated ERR is 2.7Z as compared with the appraisal estimate of 7.4Z.  - 20 - VI. BANK PERFORMANCE Overall Performance 80. Although delays in implementation have plagued the project, Government's major objectives in launching the Project viz. (i) to reduce the country's dependence on food aid, and (ii) to provide a steady income to farmers to meet their basic needs are well on the way to achievement. As the first major irrigation scheme in Mauritania, certain intrinsic risks were foreseen by the Bank and efforts made to reduce them. Thus, through a long and arduous process of preparation, numerous adjustments were made to arrive at the technical solution finally chosen, and an implementing institution (SONADER) was built up. 81. The risks that appeared most prominent at appraisal were: - The high cost of, and lack of experience with, the execution of major works by force account andlor through contracts in Mauritania; these were counteracted by pre-selecting bidders, obtaining firm contract prices, allowing a margin for uncertainty in bid proposals and having strong technical supervision; and - Possible shortfalls in projected yields resulting from the difficulty of converting semi-nomadic rainfed farmers to settled, irrigation farmers practicing double cropping. 82. In fact, the major problems that hampered the execution of the Project as planned were (a) different interpretations of the legal documents by other donors and by the Government, (b) the use of inadequately detailed, preliminary data in preparing the designs for the tertiary systems resulting in high cost overruns and long delays from the additional time required to generate more detailed basic data, and (c) the difficulties of coordinating disbursements among ten different financing sources (one of whom pulled out). 83. Although, in retrospect, the Bank has to take some of the responsibility for items (a) and (b) above, a striking feature of project implementation were the strong coordination and rapport between the Bank and the executing agency which led to effective solutions being worked out as problems arose in the course of execution. 84. The Bank did an excellent job in identifying implementation problems at an early stage and taking resolute action to revise project targets to more realistic levels and to re-design the settlement and agricultural development components. The Bank was also able to coordinate the activities of the donors effectively and arrive, after some initial difficulties, at a satisfactory disbursement rate. Supervision  - 21 - 85. The Bank supervised the Project adequately, fielding nine supervision missions. However there were long intervals without supervision missions (1982, 1985-86), although supplementary missions were active during one of these periods. Mission composition was technically strong and good staff continuity was maintained (see Basic Data Sheet). 86. Corrective action was taken to revise the Project before matters came to a head, although no attempt seems to have been made to assess the impact of the revisions on the economic rate of return of the Project. VII. CONCLUSIONS AND LESSONS LEARNED 87. It is too early to make final judgments on the successes and failures of the Project as a whole. First, although the Bank's share of the loan has been disbursed, project investments are still continuing from funds provided by the other donors. Secondly, the relegation of part of the original project objectives to a second phase (presently being prepared) in which first phase investments are regarded as sunk cost, and thirdly, the yet incomplete state of the first phase, make it difficult to do more than make a provisional economic re-evaluation of the Project. 88. Nevertheless, despite an apparent shortfall in the ultimate Project ERR in comparison to original appraisal expectations, the major objectives of reducing the country's dependence on food imports and ensuring an adequate income to settler-farmers are certainly likely to be achieved. Furthermore, much valuable experience was gained in implementing what is really the first major irrigation project in Mauritania, with probable benefit to future projects to he taken up for implementation under the plan for the development of the Senegal River Valley. As discussed exhaustively in the course of pre-appraisal, Mauritania lacks investment opportunities, especially in the Southern Region in which the Project is sited, and the low economic rate of return has to be viewed in this light. 89. The lessons to be learned from the Project are as follows: - It is absolutely essential for irrigation projects in Mauritania to have final designs, based on accurate base data, on which project costs could be estimated and construction scheduled before pre-appraisal; - A large number of donor agencies is best avoided, but if unavoidable, firm commitments and financing schedules for each donor should be agreed to before final negotiations. Dependence on unreliable or marginal donors for financing key components of the Project should also be avoided. A project coordination unit specifically charged with monitoring disbursements should bring problems encountered to the notice of the relevant donors without delay and call a meeting of donors, if necessary; - The project would have benefitted from greater legal input, both  - 22 - in design and supervision; - Irrigation projects in Mauritania should be monitored regularly with appropriate modifications made as needed by a strong Bank supervision team, as was the case in the Gorgol Irrigation Project.  MAURITANIA Projet Dua Gorgol (Credit - 1068 MAIl) GROSS DOMESTIC PRODUCT BY SECTOR OF ORIGIN, 1973-83 In Constant 1982 Prices (million Ougulya) 1973 1974 1975 1976 1977 1978 1979 3980 1981 1902 1983 Pri ary 8ector , 10,472 10 43? 8,785 9 015 9 537 9,271 9,460 10,527 11,466 11t656 13,416 4 a Agrlculture, Forestry 1,157 1,136 1,0506 1.0 ON ff3 981 1.267 1,608 1,335 1,167 868 i iventock 7,695 8,064 7,005 7,258 7,6008 7,935 8,011 8,200 8.541 8.768 9,357 Flshing 1,620 1,237 890 752 856 355 162 719 1,590 1.761 3.207 Secondary Sector 7,665 8,843 7,989 8 929 7944 8,295 9,1?7 8,580 8,576 fl228 7.752 Mining 5,103 5,858 4,55 4,947 3,845 1-T6i 4,637 4,338 4,377 3,550 3,86i Fish processing 234 300 419 296 318 436 459 483 508 535 562 Other Industry and Handicrafts ' 873 1.031 1,019 1,071 1.207 1.165 1,321 1,284 1,297 1.255 1,288 Construction, Public Works 1,455 1,654 1,993 2,613 2,584 2,926 2,760 2,475 2.394 2,888 2,061 Tertiary Sector 10 595 It 493 13,100 14,264 14,823 15,126 15,259 16 740 16,529 15,116 15,916 Transport and Communications' 2,16 2,01 1,971 2,496 2,697 2,772 2,870 2,962 2,904 3,035 3,2ii Trade and Other Non-Covernment 5,145 5,815 6,670 6,609 6,730 7,930 7,771 8,618 8,176 6,687 6.636 Covernment Services 3,285 3,664 4,459 5,159 5,396 4,624 4,618 5,160 5,449 5,594 6,009 COP at Factor Cost 28,132 30,773 29,874 32,208 32,304 32,692 33,876 35,847 36.51! 35,200 37.102 Indirect taxes net of hubsidies 3,097 4,347 3,179 3,486 3,n08 2,449 2,823 2,416 3,121 3.631 4,299 CDP at Constant 1982 Rarket Prices , , 31,829 35,120 33,053 35,694 35,312 35,141 36,699 38,263 39,692 38.837 41,401 Annual Crouth Rate 10.3 -5.8 8.0 -1.0 -.4 4.4 1.5 3.7 -2.2 6.6 '-I Mmicat Direction de