Burkina Faso - Niena Dionkele Rice Development Project

Document of The World Bank FOR OFFICIAL USE ONLY Report No. 13226 PERFORMANCE AUDIT REPORT BURKINA FASO NIENA DIONKELE RICE DEVELOPMENT PROJECT (CREDIT 1013-BUR) JUNE 28, 1994 Operations Evaluation Department This document has a restricted 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 authorization. CURRENCY EQUIVALENTS Currency Unit: CFAF Appraisal year average - US$1.00 = CFAF 210 Completion year average - US$1.00 = CFAF 303 ABBREVIATIONS AND ACRONYMS CNCA - Caisse National de Credit Agricole CERCI - Centre d'Etudes et de Recherche sur les Culture Irrigues CRPA - Centre Regional de Production Agro-pastorale FAO/CP - Food and Agriculture OrganizationlWorld Bank Cooperative Program GTV - Gestion de Terroir Villageois IDA - International Development Association IRAT - Institut de Recherche d'Agriculture Tropicale MAE - Ministere d'Agriculture et Elevage ONBI - Office National de Barrages et d'Irrigation ORD - Organization Regionale de Developpement PAR - Performance Audit Report PNGTV - Programme Nationale de Gestion des Terroirs Villageoises PPF - Project Preparation Facility SRMS - Secondary Road Maintenance Service UNDP - United Nations Development Program WARDA - West African Rice Development Authority FISCAL YEAR January I - December 31 FOR OFFICIAL USE ONLY THE WORLD BANK Washington, D.C. 20433 U.S.A. Office of Director-General Operations Evaluation June 28, 1994 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Performance Audit Report on Burkina Faso Niena Dionkele Rice Development Project (Credit 1013-BUR) Attached is the Performance Audit Report on Burkina Faso - Third Rural Development Fund Project (Credit 1013-BUR) prepared by the Operations Evaluation Department. The objective of the project was to determine whether rice could be produced under the conditions of Southwest Burkina Faso, using an approach which provided only limited water control. At completion it was concluded that the approach was feasible but that the results did not justify an expansion into a second phase, as originally envisaged. However, the audit determined that the infrastructure and the system established by the project have since been abandoned and that the project was unsatisfactory and unsustainable. During the process of project preparation the proposals became increasingly modest. The final design was based on rainfed rice production with minimal water control and use of a polder to prevent flooding. However, aspects of this design (e.g. plot layout and polder size) were technically unsound as they implicitly required a higher level of water control than was provided. This deficiency contributed significantly to the problems experienced by the farmers and the eventual abandonment of the scheme. These problems became apparent during implementation, but were never effectively tackled, as appropriate decisions were not taken. In the event, Bank and Borrower priorities shifted elsewhere and adequate provision was not made to remedy the deficiencies identified earlier. The results of the project indicate the potential value of a formal mid-term review in a pilot project of this type to minimize the likelihood of the type of 'drift' which occurred, and also of a formal agreement on follow up actions at completion. On the positive side the work of the technical assistance team did help highlight the importance of the general land management issues in this region and played a significant early role in the development of the approach now being pursued under the IDA supported Environmental Management Project (Cr. 2229-BUR). Robert Picciotto by H. Eberhard K6pp Attachment This document has a restricted 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 authorization.  FOR OFFICIAL USE ONLY PERFORMANCE AUDIT REPORT BURKINA FASO NIENA DIONKELE RICE DEVELOPMENT PROJECT (CREDIT 1013-BUR) TABLE OF CONTENTS Page No. PREFACE.... i BASIC DATA SHEET ................................. iii EVALUATION SUMMARY .............................. v 1. INTRODUCTION ........................... 1 Background ................................ 1 2. PROJECT DESIGN .......................... 2 The Project Area ............................. 2 Identification ............................... 2 Feasibility Study ............................. 3 Preparation ................................ 4 Pre-appraisal. ............................... 5 A ppraisal................................... 6 3. IMPLEMENTATION ......................... 7 The Perimeters .............................. 7 Rice Production............................... 9 Technical Assistance............................ 10 Rice Industry Support Group ...................... 10 Institutional Support ........................... 12 Project Conditionality .......................... 12 Second Project Preparation ...... ....................... 13 Preparing for Handover ......................... 14 This report was prepared by John English (Task Manager), with administrative support provided by Constance Frye. This document has a restricted 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 authorization. Page No. 4. POST-IMPLEMENTATION ..................... 15 Management of the Perimeters ..................... 15 Village Land Management ....................... 16 5. FINDINGS AND LESSONS ..................... 18 Technical Issues ............................. 19 Resolving the Technical Issues .................... 21 The Organizational Problem ...................... 21 Post-Project Support ........................... 22 Lessons .................................... 23 MAPS - IBRD Nos. 14301R, 14613R -1- PERFORMANCE AUDIT REPORT BURKINA FASO NIENA DIONKELE RICE DEVELOPMENT PROJECT (CREDIT 1013-BUR) PREFACE This is the Performance Audit Report (PAR) on the Niena Dionkele Rice Development Project for which Credit 1013-BUR in the amount of US$6.5 million was approved 29 March, 1980. The original Closing Date, 31 December 1984, was extended by two years to 31 December 1986. A sum of US$2.83 million was cancelled at Closing. The PAR is based on the Project Completion Report prepared by the Africa Regional Office, on the Staff Appraisal Report, the President's Report and legal documents, on study of the project files, and on discussion with Bank staff. An OED mission visited Burkina Faso in September/October 1993. The mission discussed the effectiveness of the project with Government officials from the Ministries of Agriculture and Finance, as well as local authorities, private citizens and staff of the ongoing Environmental Management Project (Credit 2229-BUR). The mission also held discussions with officials of Norwegian Aid. The PCR provides a frank account of project experience, including performance of the Bank and Borrower, and the participants in the project. However, the infrastructure and system put in place have collapsed and project performance is now judged to be unsatisfactory. The audit primarily addresses the reasons for this deterioration. The draft PAR was sent to the Borrower for comments but none were received.  - Ill - PERFORMANCE AUDIT REPORT BURKINA FASO NIENA DIONKELE RICE DEVELOPMENT PROJECT (CREDIT 1013-BUR) BASIC DATA SHEET Key Project Data Item Appraisal Actual Actual as % of Estimate Appraisal Estimate Total Project Costs (US$ million) 7.12 3.9 54 Credit Amount (US$ million) 6.5 3.67 55 Disbursed (US$ million) 6.5 3.67 55 Cancelled (US$ million) 0 2.83 - Outstanding (as of 6/89) (US$ million) - 3.67 Economic Rate of Return (%) 9 negative - Cumulative Estimated and Actual Disbursements (in US$ m) FY80 FY81 FY82 FY83 FY84 FY85 FY86 FY87 FY88 Appraisal Estimate 0.1 1.3 3.2 4.8 6.0 6.5 - - - Actual 0.0 0.42 1.18 1.57 2.30 2.74 3.23 3.45 3.67 Actual as % of Appraisal 0 32.3 36.8 32.7 38.3 42.2 49.7 53.1 56.5 Date Final Disbursement July 16, 1987 Project Dates Original Revised Actual Appraisal 05/79 Negotiations 02/15/80 Board Approval 04/29/80 Credit Signing 06/26/80 Credit Effectiveness 12/05/80 Credit Closing 12/31/84 12/31/86 - iv - Staff Inputs (Staff weeks) FY75 FY76 FY77 FY78 FY79 FY80 FY81 FY82 FY83 FY84 FY85 FY86 FY87 FY88 Total Preappraisal 36.1 4.3 - 45.8 50.9 - - - - - - - - - 137.1 Appraisal 0.1 - - 8.3 39.4 - - - - - - - - 47.8 Negotiation - - - - - 4.2 - - - - - - - - 4.2 Supervision - - - - - 12.0 19.2 15.2 8.0 5.7 7.3 7.5 7.0 14.5 96.4 Other 0.5 - - 0.4 10.2 0.8 - - 0.8 - 1.1 1.4 6.8 13.1 35.1 Total 36.7 4.3 0.5 46.2 69.4 55.6 19.2 15.2 8.8 5.7 8.4 8.9 13.8 27.6 320.6 Mission Data Date No. of Staff Days Performance Month/Year Persons in Field Specializationa/ Ratingh/ Trend r/ Problemsd/ Appraisal 05/79 2 14 Ag(1),B(1) - - - Supervision 1 08/80 2 11 B(1),C(1) 1 1 - Supervision 2 03/81 3 11 Ag(1),B(1),C(1) 1 2 0 Supervision 3 11/81 2 12 Ag(1),C(1) 3 2 T,O Supervision 4 06/82 2 6 Ag(1),D(1) 3 2 T,M Supervision 5 12/82 1 5 Ag(1) 3 2 F,O Supervision 6 05/83 1 4 Ag(1) 3 2 F,M,T Supervision 7 11/83 2 5 Ag(1),D(1) 2 2 F Supervision 8 07/84 1 4 Ag(1) 2 2 F,P Supervision 9 02/85 1 4 Ag(1) 2 2 F,P Supervision 10 09/85 1 4 Ag(1) 2 2 P Supervision 11 06/86 3 5 Ag(2),C(1) 2 2 M Supervision 12 10/86 1 5 Ag(1) 2 2 M Supervision 13 05/87 1 4 Ag(1) 2 2 M Other Project Data Borrower Burkina Faso Executing Agency Ministry of Agriculture (formerly Rural Development) Fiscal Year of Borrower January 1 - December 31 Follow-on Project Name none Currency (abbreviation) CFAF Appraisal year average US$1.00 = CFAF 210 Completion year average US$1.00 = CFAF 303 a/ Ag = Agriculturalist; B = Economist; C = Engineer; D = Operations Officer. b/ 1 = problem free or minor problems; 2 = moderate problems; 3 = major problems. cl 1 = improving; 2 = stationary; 3 = deteriorating. d/ F = financial; M = managerial; T = technical; P = political; 0 = other. PERFORMANCE AUDIT REPORT BURKINA FASO NIENA DIONKELE RICE DEVELOPMENT PROJECT (CREDIT 1013-BUR) EVALUATION SUMMARY Background Project Design 1. Burkina Faso lies in West Africa in the 3. Preparation of the project was a long and Sahelian Zone to the south of the Sahara Desert. tortuous process and the primary objective of the The climate ranges from Sahelian in the north, project was never satisfactorily identified. As with a rainfall of 600 mm, to sub-humid in the the preparation/ appraisal proceeded, there was southwest, with rainfall of about 1,200 mm per continual change in the perceived primary focus year. The project had its genesis in the four of the project. The first note on record is from year drought which hit the Sahelian countries in 1973, by a reconnaissance mission which sug- the early 1970s. As a result of its severe gested the construction of a 300 ha polder to impact, ways were sought to make agriculture determine if satisfactory levels of rice production less susceptible to drought. These included could be obtain through protection from flood, fostering agricultural development in the with limited water control. A feasibility study, relatively sparsely settled southwest of the undertaken in 1975, examined three alternatives country. Particular attention was given to the for rice production. These ranged from potential for irrigation. Interest in rice irrigation of just over 2,000 ha for rice, to production was also stimulated by the fact that, irrigation of about 300 ha and partial water although it was not a traditional grain, control over about 1,800 ha. The preparation consumption was increasing, particularly in the mission in 1977 confirmed that irrigation and urban areas. Also, in the late 1970s, about half full water control was not economically viable in of the supply was imported, a particular concern this area and recommended a multi-component, given the limited export potential of the country. area development project. This would have had It was considered that there was considerable rice production (with limited water control) on potential for rice production in the plains and 1,400 ha as its principal component. valleys of the southwest. 4. Concerns in the Bank about the viability 2. The project area is a watershed of about of the approach led to the sending of a pre- 900 km2, situated about 100 km to the NW of appraisal mission, which considered a smaller Bobo-Dioulasso, in the southwest of the country. operation of the same type. Strong reservations Annual rainfall averages just over 1,000 mm but were expressed as to whether even this was not is highly variable, both within and between based on overoptimistic assumptions. Finally it seasons. A flood plain of about 7,000 ha has was agreed that a phased approach was desirable formed because of a restricted channel as the and the appraisal mission proposed a much river draining the watershed runs through low reduced, pilot operation. This confined its hills, and was considered to be a promising site attention to rice and was based on the construc- for rice production. tion of a dike to provide flood protection to an - vi - area of 400 ha where rainfed rice would be mert project, and it was agreed that the continu- grown. Given the sectoral issues that had been ation of the related work in the Niena-Dionkele raised during preparation on policy for the rice area would become a component of this larger sub-sector, it was decided to include a support project, then programmed for FY1988, and in component, including an adviser on policy. This the meantime these activities should be continued also enabled the project to be presented as a by the existing Niena Dionkele project team, "Rice Development Project". under funding to be arranged. Project activities related to rice production were handed over in Implementation Experience 1987 to the Hauts-Bassins Regional Development Organization (ORD). 5. The technical problems inherent in this approach to rice production under conditions of Post Project Developments highly variable rainfall soon became apparent, when the first season was dry. Implementation 8. These transfers took a few months to was also hampered by severe delays in getting effect and problems began to arise almost technical assistance personnel in the field. Two immediately. The PCR, while recognizing many years of better rainfall led to improved results, of the problems encountered during the and initial appraisal targets were met. However, implementation of the project, was modestly accessibility to the single polder was difficult for optimistic: "the project showed that it was most of the villages in the project area, which possible to develop bottomlands for rainfed rice were more than 5 km away from it. As a result, cultivation". It did note requirements for the polder was never fully utilized. Because of production in the project perimeter to be initial delays and the appreciation of the dollar, sustainable: availability of inputs, adequate disbursements lagged and the project was markets and research for disease control. These extended by two years. were not met. Government had announced that ORDs would be abolished and replaced by 6. The original conception underlying the organizations (CRPAs) with a narrower phased approach to development in the area had extension, rather than a development, function. been that a substantial increase in rice The 1987 revolution held up this process and production in the plain would be the main disrupted programs. In 1988 the area suffered emphasis of a second phase of the project. heavy flooding and much of the rice crop was However, it also became clear that the prime lost. Farmers could not repay loans taken for concern of the local population was the land fertilizers and sprays and village revolving funds pressure resulting from in-migration from other collapsed. Because of their insolvency new regions of the country which had been loans were not feasible. Increased incidence of encouraged by the reduction of the scourge of disease and marketing problems reduced returns, river blindness. At the same time this issue was and the protected area was gradually abandoned. becoming a major national preoccupation of Government, and of the Bank and other aid 9. Support for rice production might have agencies. In late 1985 the project produced an been given through the special funding which outline proposal for a second phase with a had been arranged to support the pilot land radical shift in emphasis and an avowedly management effort. However, this effort was experimental approach, giving top priority to immediately thrown in turmoil by political addressing the land issue. changes. The latter might not have been predictable, but it is clear that the Bank's main 7. By this time work had begun on the focus was on preparing an Environmental preparation of a separate national land manage- Management project. In so far as Niena - vii - Dionkele area was a good example of the 13. On the technical side there were two key problems which such a project would aim to questions: address, it was of some interest, but in practice it was "relegated to the back burner". The type (a) whether, with flood protection only, rice of interim funding arrangements made inevitably could be grown as a rainfed crop under place heavy demands on staff time in relation to the conditions of the Niena Dionkele the resources involved. As a result staff spent area; time on this project but with minimal impact in the field. (b) whether such production was compatible with the existing local farming system. 10. With hindsight, it might have been better to have extended the original project by two 14. The approach adopted was flawed by years, to guarantee support, or to have three basic technical failures: terminated the land management activities at the end of the project period. Continuing a dialogue (a) design of the dike. This did not enable with the villagers over the clarification of land any control over water inside the polder. rights, in the context of an anticipated broader However, the extent to which the polder development effort, which then failed to would drain naturally was not known, materialize was counterproductive. There was and the arrangements for the control of no counterweight to the tensions which were the weir which controlled water levels in bound to arise in such an exercise. the plain was not settled. There was also no detailed plan for the execution of Findings and Lessons the initial works; 11. The fundamental problem was a failure (b) lack of a recognition that such a design to clarify the basic objectives of the project and, would require fundamental change in the in particular, when it was agreed that a pilot management of the cropped area, operation was needed, to establish exactly what compared to irrigation. Each farmer it was that the project was attempting to was to be given a single, half hectare determine, and devising an appropriate plot. But, given the highly variable organizational structure for it. The project tried rainfall conditions of Niena Dionkele, to meet too many objectives, ranging from and the variation inside the polder, the testing a new approach to rice production in likelihood that a single plot would have small flood plains (a very local issue) to good growing conditions was poor. It strengthening the capacity of the central would have been preferable to have ministries to address sectoral issues in rice allocated at least two or three smaller production. plots to each farmer; 12. One area where the project did have a (c) inadequate participation of local farmers positive outcome, however, was in the work of in the design so that the likelihood of a the technical assistance team in developing ideas successful operation could be maxi- which became incorporated into the evolving mized. Because of lack of water control national approach to 'village land management'. and rainfall variability, preplanning of This work provided a valuable field level case in field work was difficult. However, the early stages of this evolution. many of the farmers lived far from the perimeter. A smaller perimeter, which could have been taken up by farmers - viii - from one, adjacent village, would have Lessons been better. 18. The difficulties of this project and its 15. There was also a failure to make a clear ultimate failure illustrate the importance in pilot decision on the performance of the project at operations of laying out clearly the objectives of completion. It became apparent during imple- the 'investigation' which will be undertaken and mentation that the polder, as designed, would the steps by which they would be assessed. not work. However, proposals to investigate Where a project is clearly experimental, it is possible steps to improve its performance were preferable that this be the central theme. Such an not pursued, but no clear decision that the effort approach would enable the project to have been had failed was made, nor were adequate arrange- clearly presented as a pilot, on a smaller, more ments made for continued support of project appropriate scale. The pre-appraisal had operations. Unfortunately this meant that the indicated that, if the approach proved feasible, issues related to management of the polder, the type of production being tested would be which were to have been investigated as a part economically viable. of the preparation for the second phase, never were resolved and little more is known now than 19. For such projects clear arrangements for a decade ago. review of the investigations are also advisable. In this case a mid-term review would have been 16. This has been doubly unfortunate since valuable in ensuring that all parties have an the effect has been to discourage the Bank from opportunity to assess the situation and agree on involvement in any significant further activities steps to be taken. By the third year of related to rainfed paddy rice production. implementation of this project, problems with its However, under the climatic conditions of the operation were clear and implicitly accepted, but moister regions of Burkina Faso, significant not formally agreed. areas become waterlogged in average or wetter years and large amounts of surface water have to 20. Similarly the outcome underlines the be managed. As a result, rice production will importance of a formal conclusion to the project undoubtedly have a role to play in the intensifi- and agreement on next steps. In this case the cation of agriculture in such areas, as it is one of effective conclusion was that the pilot had been the few crops which can generate a significant a partial success, but no provision was made to output from waterlogged or shallow flood condi- remedy the deficiencies identified earlier. This tions in the tropics. lack of clarity effectively undermined also those activities on which more specific measures were 17. The organizational structure established taken, and left everyone concerned little the was an attempt to straddle the local and national wiser about the basic issues the project was to aspects of the project, but predictably did not address, rainfed rice production with limited work. Project management reported to the water control. regional ORD but had a national Coordinating Committee. Neither relationship worked well. 21. Small investigative operations are staff The national, sectoral element of the project had intensive. Even when these are funded from no impact. A small pilot operation focussing on other sources, as with the follow-up land the technical issues could have been based within management activities, the Bank must make staff the ORD and would have assured better support, provision for support work. If this cannot be and made day to day management easier. justified, then the basis for seeking third party support should be called into question. -1- PERFORMANCE AUDIT REPORT BURKINA FASO NIENA DIONKELE RICE DEVELOPMENT PROJECT (CREDIT 1013-BUR) 1. INTRODUCTION Background 1.1 Burkina Faso lies in West Africa in the Sahelian Zone to the south of the Sahara Desert. At the time this project was appraised in 1979 its population was about 6 million, but has since grown to almost 9 million. The country has an area of about 275,000 km2. There are pronounced regional differences in climate, soils, and population density. The climate ranges from Sahelian in the north, with a rainfall of 600 mm, to sub-humid in the southwest, with rainfall of about 1,200 mm per year. The low and erratic rainfall in the north results in generally unreliable conditions for crop production and greater reliance on livestock for sustenance. For historic reasons, the Mossi plateau, in the center north is the most densely populated region in the country. For this reason and because of its erratic rainfall and relatively poor soils, agriculture is based on cropping, but is primarily subsistence oriented. In contrast in the west and southwest, soil and rainfall conditions are more favorable, population densities are lower, access to external markets is easier, and production is more market oriented. 1.2 Burkina Faso is bordered to the south by Cote d'Ivoire and Ghana. Over the past 50 years both of these countries have developed relatively substantial agricultural exports, particularly cocoa and coffee, but also other humid zone products such as rubber and palm oil. Poor access to international markets, and the agricultural limitations noted above, have limited similar development of cash crop agriculture in Burkina Faso, and the sector is overwhelmingly subsistence oriented. The country has few mineral resources so agriculture remains the basic source of livelihood for the overwhelming majority of the population, and supplies about 90 percent of recorded exports. 1.3 The Niena Dionkele area is in southwest of the country. The project had its genesis in the four year drought which hit the Sahelian countries in the early 1970s. As a result of its severe impact, ways were sought to make agriculture less drought susceptible. These included fostering agricultural development in the relatively sparsely settled southwest of the country. Particular attention was given to the potential for irrigation, especially in the southwest. Interest in rice production was also stimulated by the fact that, although it was not a traditional grain, consumption was increasing particularly in the urban areas. Also, in the late 1970s about half of the supply was imported, which was a particular concern given the limited export potential of the country. It was considered that there was considerable potential for potential for rice production in the plains and valley bottoms of the southwest. "Swampland and irrigated land has received little development effort to date, but has a large potential, which the Government is now anxious to exploit, both from the point of view of meeting internal rice demand and to develop areas which present reduced risk of crop failure during the droughts which periodically afflict the country." (Preappraisal Report para. 1.09) -2- 1.4 The primary objective of the project was to promote rice production. However, the area developed by the project for this purpose is now uncultivated, although it was in production in 1987 when the project closed, and the Project Completion Report was cautiously optimistic as to its future. The PCR provides an overview of the principal problems faced in implementation and of the outcome at completion. In order to provide a picture of why the effort finally failed, this report reviews the chronological sequence of events during project preparation, implementation and its aftermath more deeply than is normal in an audit. This illustrates the changing perceptions and priorities of those involved, since these appear to have been the major reason for the unsatisfactory outcome. 2. PROJECT DESIGN The Project Area 2.1 The project area is a watershed of about 900 km2, situated about 100 km to the NW of Bobo-Dioulasso in the southwest of Burkina Faso (see map). Annual rainfall averages just over 1,000 mm but is highly variable, both within and between seasons. The watershed is drained by the Dougo River, which flows NW into Mali and eventually into the Niger. A flood plain of about 7,000 ha has been formed because of a restricted channel as the river runs through low hills. More than 40 years ago a road was built, which crossed the river just upstream of this section close to the village of Karamassasso. This structure had the effect of further restricting flow in periods of heavy flow, and causing increased floods in the plain. 2.2 In the mid-fifties this river crossing had been reconstructed to include a sluice gate, to try and provide some control over water levels in the plain, with the objective of encouraging production of rice, possibly on a semi-mechanized basis. When closed, this gate maintained the water level at 331 meters above sea level. Rice production never took off, apparently because the control of water levels was never really effective, because the restriction on stream flow created by the structure caused excessive flooding in the plain. Total paddy land was said to have extended to about 1,500 ha in the early sixties. Given subsequent events, this level sounds implausible but, in any event, interest declined quickly as a result of poor sluice gate control, drainage problems, lack of interest on the part of the authorities, and marketing difficulties. Villagers surrounding the inundated plain, however, continued to grow small areas of rice for their own consumption. 2.3 Population in the sub-region has been relatively low. The appraisal report estimated that in the 11 villages surrounding the flood plain itself there were, in the late 1970's, about 1,500 households (i.e., 8,000 people). These villages utilized an area of about 40,000-50,000 ha. Slopes are generally modest, soils are loamy to clayey/sandy and have good water retention, and are only slightly acid. With the moderate rainfall farmers produce maize, cotton, millet and sorghum. Small areas near the villages have also been used for rice production. During the dry season, the flood plain has been used to graze cattle. On this basis there has been significant market-oriented production for some time. Identification 2.4 The first intimation of the possibility of a Bank interest in a project in the Niena-Dionkele area is contained in a brief note prepared in 1973. This reported on a visit to the area made as a adjunct -3- to a more detailed assessment of another possible rice scheme in SW Burkina Faso. This visit was undertaken by a Bank staff member together with a member of the Republic of China technical assistance team then developing the Kou Rice Scheme, located between the Niena Dionkele area and Bobo- Dioulasso. 2.5 A possible approach was suggested by the 1973 report, based on the construction of a polder to protect against the most severe floods. This was considered necessary because of the effect of the naturally restricted outlet in causing floods is accentuated by the Karamassasso crossing. A 300 ha pilot was suggested and a sketch design provided. This envisaged a polder with an encircling drain to evacuate flood water. At the upper end of the polder a small diversion structure was proposed to allow additional water to be diverted into the area if required. The polder would have a central drain, with a pump at the low point to allow removal of excess water. This would provide some, although obviously limited, control over water levels in the polder. As essentially a run-of-the-river scheme it could suffer from lack of water in dry years, but should result in reduced flooding in wet years. The very preliminary cost estimate for a 300 ha polder of this type was CFAF 43.2 million (US$226,000), or CFAF 172,800 ($735) per ha. Feasibility Study 2.6 In 1974 a more extensive pre-identification mission was undertaken to review the potential of the Niena-Dionkele and two other flood plains in the same general area. This considered that a project might be feasible which developed about 1,000 ha under irrigation and another 1,000 ha for a single wet season rice crop, together with upland crop development. Because of the limited local population, this project would also cater for settlement of about 1,400 families. The report recommended that a detailed study be undertaken to prepare a 'Plan Directeur' for the whole watershed, together with a more detailed feasibility study of a rice oriented project. 2.7 This study was prepared in 1975/76. It judged that the most suitable areas for rice production were those just above the sluice gate level of 331 m.a.s.l., but that the major problem would be the inundations following heavy rainfall. It identified three alternative approaches to rice production. These all proposed to develop just over 2100 ha for rice production. The area under full water control ranged from 300 to 1500 ha, with partial water control over the remainder. In all cases rice development would be associated with the development of rainfed agriculture on farms belonging to the beneficiaries. 2.8 Capital costs and economic rates of return for these options were estimated to range from US$9.4 million and an ERR of 6% for the most intensive case to US$3.2 million and an ERR of 11%, for the least intensive option. 2.9 The low rate of return for the most completely irrigated option reflected the fact that there are no good storage sites in the area to permit much more than the provision of supplemental water for a single, wet season rice crop. (This is typical for Burkina Faso as a whole). Some intensification is likely in such a situation, eg through use of supplemental water for other crops, (as a second or single crop), but this would not have significantly improved the ERR. The Bank's conclusion was that the only really feasible approach was the least intensive, eliminating the 315 ha of full irrigation as this was estimated to cost twice as much per ton of rice produced as the bottomland rice production. -4- Preparation 2.10 Following discussions with Government, FAO was requested to undertake preparation of a project proposal on the basis of the least intensive of the alternatives identified by the feasibility study. This reported in 1978 and proposed a project aimed at increasing cereal and cotton production by: - development of bottomlands to permit better drainage and provide a protection channel for the rice fields; - preparation of 1,400 ha of paddy land divided into 1.75 ha plots per family; - adoption of a four year rotation in the uplands for food crops and cotton, alternating with a grass fallow for livestock fodder; - provision of short- and medium-term credit to farmers for purchase of inputs and draft animals; - strengthening of extension services; - development of infrastructure, including a rice mill. 2.11 This proposal was based on works to improve the drainage and flow of the two principal rivers running through the plain, the Dougo and the Kuo, to reduce flooding in the plain and to facilitate the drainage of excess water in connection with the improved operation of the existing sluice. This sluice would also serve to maintain a high water table at the end of the wet season to support rice production. Protection of the rice growing area was proposed by means of drains and dikes to divert runoff water downstream. 2.12 It was estimated that 870 households would participate in the project, 80-90 percent of those in the villages adjacent to the plain, and that each would cultivate 1.75 ha of paddy within the area developed by the project, as well as improving their upland production. 2.13 The principal concerns raised in the review of this report related to: the likely response of the local population and the consequent availability of labor for crop production of the intensity proposed; the untested nature of the method proposed to control water levels and the related uncertain behavior of the water table; the organizational arrangements for construction of the necessary river and drainage structures; and marketing of additional cash crops, especially rice. One reviewer noted that the proposal was essentially for an integrated agricultural project in which the water control and road works accounted for about 50 percent of costs. He felt that, given the uncertainties on water control, it might be advisable to phase the effort, starting with an emphasis on the upland crop production, with a limited test of the approach to paddy. 2.14 Because of the extent of these concerns it was decided to mount a six-person pre-appraisal mission. This was to include a sociologist in order to investigate the concerns over the issue of likely participation of the local households and also of the use by transhumant herders of the flood plain in the dry season. Changes in economic opportunities, including migration of adult males to work, were reducing the commonality of interest and coherence of village groups. To complicate the picture further, -5- migrants, primarily Mossi from the north-center of the country, had been moving into the area. If villages became majority Mossi, they sometimes were in a position to supplant the traditional leadership. As a result the local Bolon and Senoufo, who had historically been antagonistic to the Mossi, were fearful of actions which might encourage further Mossi immigration to the area. In part these problems had already been factored into the proposed operation as it was scaled back to eliminate provision for settlement of families from outside the area. However, this overall situation of social tension and flux clearly suggested that there would be difficulties in developing local organizations necessary to support the type of development envisaged. Pre-appraisal 2.15 Reflecting the concerns expressed on the proposal of the preparation report, the pre- appraisal initially investigated a development of about 1,000 ha of bottomland for rice production, together with 3,000 ha of upland cropping (cotton, sorghum, millet, groundnuts, cowpeas and sesame). It was anticipated that 1,250 of the 1,370 households estimated to be in the 10 villages now included in the project area would participate. On this basis the estimated cost per family was US$3,600 (which was relatively high) and the estimated rate of return was only 8.7 percent. Considerable concerns were expressed in review about the firmness of this rate of return, based as it was on an assumption that the local rice yields of 1 ton/ha would be doubled and that project interventions would also lead to increases in yields of the dryland crops of between 48 percent and 112 percent. Given the high cost per family, primarily for the works linked to rice, questions were raised as to whether one could not be more selective in the area chosen, to reduce costs, or whether another area altogether should not be considered. One without the particular topographic features of Niena Dionkele, which added to costs and risks. 2.16 As work proceeded concerns continued to be expressed and one staff member was moved to write the following memo: "1. I oppose going ahead with Niena Dionkele as presently presented. 2. 1 do not accept the 9 percent IRR because dryland crop yields are too high, the change to the proposed crop mix is too optimistic, the increase (84 percent) in total labor requirement is too much to expect, too high a percentage of farmers are expected to reach the intermediate range and only US$75,000 new expenditure is budgeted to go into extension and training. In my view, the rate of return is 6 percent or less. 3. Too few farmers (1,250) are covered and the cost per family (US$3,600) is too high for what it achieves (the net gain in cereals or cotton could be considerably less than predicted). 4. Annual recurrent costs per family to be met by Government are high. The recurrent cost at full development of services directly beneficial to farmer and project objectives amounts to US$117 per annum and subsidies to US$45. The mission proposes a water charge of $136 per ha on 1,000 ha, but so far, the Kou Valley has only been able to charge US$27; if the farmer is charged, it amounts to a payment of US$91 per family. 5. In the anxiety to have a project, the salient objective of the project seems to be overlooked. We are trying to grow rice by swamp draina. If the costs could be reduced and/or the yield increased, we -6- have the sort of rice project that the country should develop. Now that we know what we are looking for, let us find the area and/or the technology to do it." 2.17 Revised Proposal as Appraised. In response to these continued concerns a further proposal was prepared. This restricted bottomland rice development to a single 400 ha polder which would be developed over three years, and be divided into 1/2 ha plots to provide for 800 families. It was envisaged that, depending on the outcome of this initial phase, a second stage of five years would extend the rice area to almost 1,500 ha. A major justification for this approach was that initial costs would be reduced as, with the use of a single 400 ha polder, it was considered that "it's development would not require digging expensive channels and constructing long dikes for flood protection. Only interception drains with small protecting dikes would be required" (Pre-Appraisal Report Annex 2, p. 3). It was also noted that at the site envisaged, it would also be possible to observe the "influence of the capillary water supply from the water table created by the Karamassasso Lake" (Annex 2, p. 3). Concerns, expressed mostly forcefully in the memorandum quoted above, were eventually assuaged by a clear position by technical staff that the proposed approach would work, and by presenting the project as a pilot. 2.18 Organization and Management. During the preparation process there had been considerable discussion over whether the operation should be free standing, or a component of a larger operation. The latter option loomed larger as the proposal was downsized and especially as support was already being given to the ORD Hauts-Bassins, which was the rural development agency which covered the area. The pre-appraisal finally proposed that the operation be implemented by a semi-autonomous self-accounting Project Management Unit under the ORD Hauts-Bassins with a Burkinabe Project Manager, responsible to the Director of the ORD. Project headquarters would be in the village of Ndorola. Appraisal 2.19 Given the extent of the effort at the pre-appraisal stage only a two person appraisal mission was mounted to review the revised proposal with Government and to resolve the remaining issues. These related to marketing (prices finance, purchasing organization and imports), rice processing and the Kou Valley rice mill, cost recovery and subsidies, Government's contribution, civil works construction capability, and monitoring and evaluation. 2.20 The most important changes which were agreed from the proposals of the pre-appraisal mission were, increased support for development of a rice industry policy, inclusion of a career advancement program for senior staff, support for field trials and an increase in the amount of credit available to farmers for the purchase of work oxen and equipment. Of these the most significant was the clarification and the expansion of the elements aimed at development of the rice sector. This finally comprised input of four technical assistance personnel and supporting funding. These were a Rice Industry Adviser to the Ministry of Rural Development, a Senior Evaluation Officer, a manager for the Kou Valley Rice Mill, and a Cooperatives Specialist. This was a somewhat eclectic group and to give its activity some coherence in operation as well as in description, it was decided that they should be located in Bobo-Dioulasso. The Rice Industry Support Adviser was to act as leader of the group and specifically to: prepare a rice industry policy study; assist the Ministry to co-ordinate ongoing activities (such as marketing policies) which have a bearing on the success of the pilot project: and implement a career advancement program for rice industry specialists. -7- 2.21 To ensure market outlet for the scheme it was proposed to provide financial assistance for the rehabilitation of the Kou rice mill with an expatriate manager for 3 years as part of the TA team. It was also agreed that the Centre d'Etudes et de Recherche sur les Cultures Irriguees (CERCI; a UNDP/FAO financed and managed research station located near Bobo-Dioulasso) should establish a small out-station at the project site to conduct site specific research. 2.22 As modified it was estimated that the ERR for the first phase was 6.5 percent, with 11 percent for the overall two phase project. The principal risks were considered to be poor farmer response, cost overruns, erratic rainfall and lower than projected yields. However, it was estimated that a two year delay in the benefits, or a one year delay plus 10 percent cost overrun, would only reduce the rate of return to the 8 percent level. This risk was considered acceptable. 2.23 The mission urged the Government to appoint a Burkinabe Project Manager as soon as possible in order to facilitate project start-up. A Project Preparation Facility advance of US$120,000 was also made available to Government to finance initial activities, including the operating expenses of the project manager, building construction, and a study on the rehabilitation of the Kou Valley rice mill. 2.24 No significant issues are recorded as being raised during the project review process. Bank staff interviewed for this audit indicated that, with the changes made at appraisal, management was satisfied that the project was worth pursuing, although risky. As approved by the Board, the project cost was estimated to be US$7.1 million, of which $6.5 million would be covered by an IDA credit. 3. IMPLEMENTATION The Perimeters 3.1 The Bank was extremely anxious that implementation should get off to a quick start (as noted in para. 2.23). This applied particularly to the construction of the perimeter, where there was concern that cropping should be started without a long delay. Prior to appraisal there had been discussion of the alternatives of force account or use of contractors for this construction. At appraisal the Government Agency came in with a cost estimate twice that of the mission. It was, therefore, agreed to tender privately and to do so before Board presentation, in April 1980. Tender documents were drawn up, reviewed by a Bank consultant, contractors qualified, and bids solicited before negotiations. Tenders had also been invited for related land development work, where force account appeared cheaper. 3.2 The three bids received were within the range of the appraisal estimate. However, the two low bidders then withdrew, saying they would only accept if the related land development work was included. After some pressure they agreed to drop this requirement. However, a series of delays then ensued. The contract could not be formally awarded until the project became effective. The contract was still held up in bureaucratic channels when a supervision mission visited in February, 1981. The contractor was finally persuaded to start work in advance of receipt of the contract in order to allow for preparation of at least part of the land within the polder to enable cropping during the 1981 season, starting in late June. 3.3 In the event, this was achieved and about 150 ha of rice was planted by about 300 farmers, a not inconsiderable achievement in the circumstances. Unfortunately the success did not extend to the results of the crop. Rainfall for the season was some 20 percent below normal for the season, with August (a critical month) being even drier. The result was that much of the area within the polder had insufficient water and no worthwhile crop was obtained on nearly 50 percent of the area planted. The yield on the harvested area was an estimated 1.2 tons per ha., or only just below the appraisal estimate for the first year. This yield, in a below average rainfall year, was about average for the area under normal conditions. 3.4 However, this experience had quickly highlighted some of the problems which were to result from the approach to water control which had been adopted. The supervision mission in September 1981 noted that "the project works consisted of protecting the rice fields against flooding (from the run- offs from the foothills as well as from the lake formed by the Karamassasso weir). However, during dry years it becomes most important to use the scarce local run-offs to supplement the scarce local rainfall. In order to divert as much run-off as possible from the foothills directly to the side ditches of the tracks and from there to the paddy fields, simple structures are proposed. These would consist of low earth dikes provisionally constructed across the collectors; inlet pipes fitted with gates would permit water control."' It was agreed that ONBI would prepare preliminary designs for such improvements. 3.5 The mission also raised the question of the operation of the Karamassasso sluice gate, which was under the nominal control of the Prefect. Government was requested to vest the Project manager with the full operational control of the gate in order to improve water control for rice production. This was a bone of contention throughout the project and reflects the competing pressures for use of land in the area. Apparently it was only for a very short time towards the end of the project that operational control was achieved, and it is not clear that much was achieved by it. 3.6 The designs appear not to have been prepared, but the Ministry of Rural Development did prepare a report on the project for the arrival of the next supervision mission. This report raised a number of issues related to the design which had been agreed for the project, in particular that the lack of effective control over water meant that the level of expenditure on works had been excessive. It recommended that a number of alternative solutions be investigated; pumping water into the perimeter from the lake formed by the sluice gate to supplement rainfall; raising the gates by 0.5 m to ensure water for this purpose; small barrages upstream to provide supplementary water by gravity; and that a second phase should comprise several small perimeters, close to each village. 3.7 The supervision mission did not include an engineer and no decision appears to have been taken on what to do on the perimeter, but the mission did note that increasing the level of the gates would run the risk of cutting off at least one of the villages for a significant part of the rainy season, the existing dike might have to be raised and the benefits were uncertain. 3.8 The following mission, in October 1982, discussed detailed terms of reference for a feasibility study for the second phase of the project. This included study of the potential for improving the conditions for rice production through improvements in operation of the sluice and of improvements to the river channel below it. This was to include estimates, based on existing hydrological data, of the 1. It should be noted that this was what was essentially proposed by the very first, 1973 reconnaissance mission, whose ideas seem to have been lost throughout the subsequent work. -9- periods and extent of flooding produced by different plans of operation of the sluice, and an analysis of a number of alternative development options for the second phase. 3.9 A preliminary selection of consultants was made in August 1983, but then was bogged down following the revolution. In the meantime the evolving situation of the project changed the perception of the immediate priorities. After considerable delay the technical assistance personnel began to arrive in 1983; the revolution of 1983, which brought the Sankara Regime to power, created great uncertainty about Government structures and priorities; and rice yields in the project polder improved markedly in the 1982 and 1983 seasons. As the perspective of project management broadened, it became apparent that the top priority for the local villagers was the issue related to use of the uplands and the pressure from immigrants. As a result, the project team began to evolve its own ideas about the second phase and, in mid 1984, it was tentatively agreed that the feasibility study would not be proceeded with and that the second phase proposal would be prepared by the project team, using other inputs as they saw fit. Their proposals will be discussed later. Rice Production 3.10 As noted, major problems were encountered in the first cropping season in the polder, 1981. Despite considerable efforts by project management, it was not possible to prepare all the target area of 300 ha for cropping. In the event 150 ha was planted. However, low rainfall and late planting resulted in no crop being obtained on about 70 ha. The average yield was only about 600 kg/ha, although on the harvested area it was about 1100 kg, close to the projected yield. The crop loss was aggravated by the fact that, on the advice of CERCI and WARDA, based on information on palatability and the expected availability of moisture in an average year, the project decided to provide seed of a long duration (135 day) variety, IR 1529, to the 300 participating farmers. This variety had been successfully used in the nearby Kou valley irrigated rice scheme. A variety with a shorter growing period (eg. of 110 days) would have been less adversely affected. IR 1529 is also susceptible to 'blast' (pyriculariosis), and to scalding, if the soil surface is not permanently flooded. 3.11 As a result of this experience, it was apparent that, even with the limited slope within the polder area, significantly different conditions were likely to be faced within it. It was, therefore, decided that for the 1982 cropping season two varieties would be recommended; IR 1529 for the lower half and the shorter duration rainfed variety IRAT 144 for the upper, and less well watered, half of the polder. It was also decided, because of the difficulties of the first season, to limit the cropped area to 155 ha. In the event rainfall was about average and some flooding occurred, but the average yield obtained increased dramatically to 2,200 kg/ha. On the strength of this result the area planted in 1983 was sharply higher at over 300 ha, good rains were obtained and a yield of 1,800 kg/ha was achieved. Production figures for the perimeter are summarized in the table below: Crop Year Area Planted (ha) Yield (kg/ha Target Actual Target Actual 1981/82 300 155 1,125 600 1982/83 300 155 1,400 2,200 1983/84 400 314 1,750 1,800 1984/85 400 333 2,100 2,000 1985/86 400 325 2,100 1,700 1986/87 400 195 2,200 2,300 - 10 - 3.12 The perimeter developed was close to the village of Ndorola and the project offices. Two other villages were within 5 km of it, but the other four villages covered by the project are 10 km or more away. It was very difficult for farmers in these villages to manage plots within the perimeter. To provide some support for them two steps were taken. Villagers had been cultivating rice in small plots adjacent to their villages. The project supplied fertilizer on credit to farmers producing in the perimeter. This facility was gradually extended to production outside the main perimeter and by 1986 this covered 150 hectares. Secondly, two small areas, totalling about 60 ha were opened up in 1984 and 1985 near to the villages of Niena and Dionkele. These were deep ploughed and small bunds constructed by the farmers, so that development costs were low. Yields appear to have been equivalent to those in the main perimeter in wet years, but slightly lower in dry seasons. However, these areas have since been abandoned, for reasons to be discussed below, and little trace now exists of them. Technical Assistance 3.13 The project contained significant input of technical assistance, especially for the Rice Industry Support Group. Steps were taken by the Bank to start recruitment immediately the project was signed in April 1980, but numerous delays occurred in actually getting individuals hired and in the field. The full team was not in place until mid-1983. Approximate dates of service are shown below: Position Dates in Post Deputy Project Manager 08/83 - 12/86' Senior Evaluation Officer 05/83 - 08/86 Rice Industry Adviser 04/83 - 12/85 Manager Kou Valley Rice Mill 07/81 - 04/84 Cooperatives Adviser 01/82 - 06/85 Continued for six months under separate contract after end of ND Project. b/ Part-time (max. of 2 days per week). Supplied by Netherlands assistance to Kou Valley Project. 3.14 The delays in getting the full team in the field had a major impact on the project, and it should be remembered that the original objective had been to start the second phase in 1985. The delays in arrival of TA staff clearly put an excessive load on the Burkinabe Project Director who had to cope with all the procurement problems associated with project start up (for equipment as well as the works described above) as well as with the technical problems which have been described. One can only imagine the overall effect on morale of project staff as the delays continued. It is small wonder that the project was in rather a 'reactive' mode in the first three years. The results which were achieved must be considered a tribute to the efforts of those who were on the ground. It was not until early 1984 that the full team began to get the measure of the situation and to decide that a different emphasis was required for the second phase from that envisaged at appraisal, and to question the emphasis of the TOR for the consultant study which had been prepared in late 1982 (paras. 3.8-3.10). Rice Industry Support Group 3.15 Kou Valley Rice Mill. Fortunately the manager of the Kou Valley rice mill was the one expatriate who was recruited quickly. However, his ability to improve its operation was hampered by the fact that much of its Taiwanese equipment was obsolete and spare parts were difficult to obtain. Also, - 11 - poor initial layout of the plant meant that it was not possible to set up the additional equipment needed to improve the quality of rice produced. As a result, it was finally decided that major work on the facility was not justified. 3.16 The mill had two other drawbacks from the point of view of the project. Since the Kou scheme produced over 3,000 tons of paddy per year, the 300 - 400 tons supplied by the Niena-Dionkele project was minor and the needs of the Kou Valley farmers took precedence. This meant that the needs of the Niena-Dionkele farmers had to wait until there was spare capacity. This complicated marketing efforts. In addition, agreement was never reached on setting up autonomous financial management for the mill. For these reasons the project decided not to extend the contract of the mill manager and ceased to provide funds for the operation and maintenance, when he departed in 1984. 3.17 Rice Industry Adviser. This position, and the notion of the industry support group, were inserted into the project only at appraisal, and appear to have been largely aimed at giving the overall package more coherence and balance and, in effect, to enable it to be presented as a rice development project. As a result, one is left with the impression that the elements of what the team leader was to do had not been completely thought through. It had been envisaged that this adviser would act as a team leader and for that reason was located at Bobo-Dioulasso, rather than at the Ministry of Rural Development in Ouagadougou. At the same time he was to advise on all matters of production, marketing and rice processing in Burkina Faso. All this might have been possible if an exceptional individual could have been found and established a rapid rapport with the relevant ministry staff. 3.18 The failure to incorporate the adviser into the structure of the ministry and his remoteness from those in a position to make decisions prevented him from having the desired impact. Furthermore, it is not clear that Bank staff gave this element of the program the attention commensurate with its presentation in the appraisal report. The problems outlined above of getting the project started without the TA staff and of the development and management of the perimeter took the prime attention of the Bank staff. As the TA staff came fully on board in 1983 the significance of the non-rice issues in the project area was becoming apparent and the prime focus of attention was shifted to the nature of a follow- up project. 3.19 In the event the adviser was unable to provide any real leadership. The Rice Industry Policy Study never got past the stage of a draft report on "Issues Related to Rice and Rice Cultivation", and was submitted too late for the Government to take advantage of it. The fact, reported by the PCR, that the Secretary General of the MAE, who had been in post for nearly a year, was unaware of the presence of the adviser until the day his report was submitted, is another indication that the Bank had given little attention to this aspect of the project, even though it clearly was primarily the result of a Bank rather than a Government initiative. It also indicates how unlikely it was that the intended reinforcement of the ministry would occur without strong Bank support. 3.20 Cooperatives Specialist. It was initially envisaged that the project would recruit a cooperative specialist to work both with the project and the Kou Valley project, where a cooperative society was a key element in rice marketing. However, considerable difficulties were experienced in obtaining a suitable candidate. In the meantime, a cooperative specialist was engaged for the Kou Valley project,as part of Dutch technical assistance, and it was arranged that he assist Niena Dionkele on a one day per week basis. Under the project a cooperative was to be established to assist particularly in input supply and rice marketing. However, this encountered a number of problems. Because of the distances - 12 - they had to travel to the perimeter, farmers from villages other than Ndorola and Teoule (which were close to it) began to lose interest in the project. Most of the farmers already belonged to functional village groups, but the divergence of interest between villages and inter ethnic differences in priorities (para. 2.26) also made it difficult to achieve the consensus necessary for establishing an area wide cooperative. Operating on such a limited, part-time basis, the adviser was never able to achieve enough continuity of contact and support to get a cooperative off the ground. This was to have unfortunate consequences after the end of the project. Institutional Support 3.21 The project was bureaucratically in a somewhat anomalous position since it was in a district in which the ORD Hauts-Bassins was responsible for agricultural development activity. The project as prepared by FAO was envisaged, in effect, as an integrated agricultural development project within a small area. However, after appraisal the focus of the project was narrowed to rice and it was agreed that services related to upland crops would be provided through the programs of the ORD. The project was to be implemented by a semi-autonomous self-accounting Project Management Unit. This unit was to be under the ORD Hauts-Bassins, with the manager being responsible to the ORD Director. A project Executive Committee was established at the national level. At the regional level the ORD Director was to coordinate departments and organizations directly concerned with the project. 3.22 This arrangement did not work well. The Executive Committee was large and unwieldy and seldom met. The need for the project management to report through the ORD Director created an extra layer of bureaucracy which contributed to delays, particularly in areas such as the initial contracts for perimeter works. This was further exacerbated after the Sankarist revolution in 1983 which ushered in a period in which public servants came under great suspicion. This greatly hampered initiative and willingness to make waves at lower levels in the system. As an example, the supervision report in mid 1983 notes that the mission had to step in and get the various parties together in order to permit action to be taken to avoid activities planned for the 1983 season slipping by a year. When in 1985 the Ministry of Rural Development was split and a new Ministry of Agriculture and Livestock took over responsibility for the ORDs, these problems were increased. Project Conditionality 3.23 Although there were some continuing issues related to conditionality, these were not fundamental to the operation of the project. The major areas were problems arose were: (a) Counterpart Funding. Given the general financial difficulties of Government this was a continuing problem, and counterpart funds were often slow in being transferred. After approval the rise in value of the dollar and SDR relative to the CFA Franc meant that the local currency value of the credit increased. In January 1985 IDA agreed to increase disbursement percentage for local staff salaries and operating costs from 88 percent to 94 percent. (b) Management Committee. As noted the Management Committee, although established, did not meet twice yearly as required (paras. 3.23-24). - 13 - (c) Water Charges. The agreement provided that the Government would institute a general system of water charges to cover at least recurrent costs of operation of water control projects. A proposal was prepared and submitted to the Council of Ministers just before the 1983 revolution. It was not pursued thereafter. Specifically for the project, there was to be a charge for use of the perimeter of 8,000 CFA ($35.4) per ha in the first year, rising to 20,000 ($88.5) in the third year. Farmers paid the 8,000 CFA in the first year, but the increases were never implemented. Farmers in the Kou Valley irrigation scheme paid only 3,000 CFA/ha ($13.3) and, given the problems with the perimeter, higher charges than Kou Valley were not tenable. 4,000 CFA per ha ($17.7) were collected in Year 2, and the project began a study to determine reasonable charges, but there is no record of action having been taken. This issue reflects the inability of the project, after the 1983 revolution, to address sectoral issues as envisaged. (d) Road Construction. Road construction was to be carried out by the Secondary Road Maintenance Service (SRMS). Work was delayed when it was apparent that the main 14 km route to be constructed, would require substantially more work than estimated at appraisal (para. 3.29). Following the revolution the SRMS was abolished and the work was not undertaken. Second Project Preparation 3.24 Detailed terms of reference for preparation studies for the second phase were prepared in late 1982 (para. 3.10). Because of limited response and bureaucratic delays a contract had still not been awarded by late 1983. By this time the TA team had finally been fully assembled, and the project had time and resources to give more attention to priorities for follow up. Meanwhile, it had become apparent that rice production was not the top priority for most local farmers and that there was increasing concern over the immigration, particularly from the Mossi plateau, and the resulting competition for land. Villagers felt that they had little control over this process. Tensions within some of the villages were increasing. The project had taken some imaginative steps to try and address the problems. In some cases the situation had deteriorated to the point where the relevant groups were unwilling even to meet one another. The project tried to overcome this impasse by video-taping meetings with the different parties and then replaying the film to the villagers so that they could see the different points of view. This helped but was clearly not a solution to the overall problem. 3.25 At this time central Government began to evolve ideas on ways to provide some security in law for local communities over their own land ("Terroir"). Based on field survey work undertaken by the Senior Evaluation Adviser and a Burkinabe anthropologist, the project began to evolve its own ideas on how the problem might be tackled in the project area and a draft report on "Amenagement des Terroirs Villageois" was prepared in late 1984. 3.26 The original conception underlying the phased approach to development in the area had been that a substantial increase in rice production in the plain would be the main emphasis of the second phase of the project. The problems of management of the polder had raised some doubt about this and the proposed preparation studies were to study alternative management of water resources in the plain (para. 3.10). The original feasibility study proposals of 1976 had envisaged the settlement of migrants to meet labor requirements of up to 3,500 ha of rice. The FAO preparation report had skirted this issue, - 14 - but it was now apparent that attempting to implement such an approach would not be acceptable to the local population. 3.27 In late 1985 the project produced an outline proposal for the second phase which would mean a radical shift in emphasis. This had an avowedly experimental approach, and proposed giving top priority to addressing the land issue. It would first require each village to define the area it claimed (its 'Terroir'). Then the village would have to decide for itself, through the village development committee or other appropriate body, how that area would be used. If there were different ethnic groups within the community, there would have to be agreement on whether they would have rights over specific parts of the village area. The resulting plan for use of the village area would imply certain development actions and the role of the project would be to work with the community to implement those activities which could be justified. 3.28 The project would also include steps to upgrade local infrastructure, especially local roads. The problems of road access, had been a significant factor in reducing farmer interest in the polder. The first phase had included the upgrading of a road across the center of the area but this had not been built. Questions arose about the original design, as it ran across an area flooded in the wet season. This issue was further complicated when the possibility of raising the sluice gate was raised (para. 3.6), as this would have increased the seasonal flood. These questions remain to be resolved. Preparing for Handover 3.29 The project completion date had been extended twice from December 1984 to December 1986. As of the end of August 1985, only US$2.74 million of the US$6.5 million credit had been disbursed. There had been some shortfalls in expenditure on TA and also on farmer credit, but the principal reason for the low level of disbursement had been the rise in the value of the dollar after project approval. It was estimated that, by the end of 1986, US$2.5 million would remain undisbursed. 3.30 By the end of 1985 the Bank had become convinced of the seriousness of the land related problems and of the desirability of tackling them in some fashion along the lines of the approach outlined in the project's draft on "Amenagement des Terroirs Villageois" (para. 3.26). Work had begun on the preparation of a national land management project, and it had been agreed that the continuation of Niena- Dionkele would become a component of this larger project, then programmed for FY1988. The Bank was concerned that the closing of the ongoing project before the land management project was effective would lead to a halt of project activities and a loss of the momentum built up by the project and the loss of an effective team. A Bank supervision mission, together with the project team, prepared an outline program for a two year program of work in the project area, to be undertaken in the context of the evolving land management project. This was essentially along the lines of the earlier project team proposal, and was estimated to cost about US$500,000. In April 1986, the Bank proposed that the closing date of the Niena Dionkele project be extended by a further two years from its then closing date of December 1986, to December 1988. This would ensure continuity into the time at which the proposed project was expected to be effective. 3.31 However, following further discussions, in August it was decided not to propose a further extension. There were three main reasons for this. "Firstly, the important work on the Village Land Management Project being undertaken by the project team is becoming more and more removed from the initial aim of the Niena-Dionkele project, which was the conversion of a swampy area to productive - 15 - rice fields. Any request for an additional, third, extension of the project would consequently be difficult to justify on the grounds of requiring extra time to complete the initial goals. Secondly, the only justification for extending the project would be to finance the creation of new rice perimeters; this would require more time than a one year extension would allow and could, therefore, not be completed before the final closing date. Thirdly, it is difficult to justify "freezing" scarce IDA funds if there are alternative sources of financing that could be used for financing the preparation of the Land Management Project." At that time alternative sources of funding (UNDP, Norwegian Government Funds, or a Bank PPF) were being explored and appeared likely to be available. As Government also appeared to prefer to wrap up the Niena Dionkele project, it was agreed that it would be closed on December 31, 1986, and that the project activities related to rice production should be handed over to the ORD Hauts-Bassins, and that the activities related to 'Gestion des Terroirs' should be continued by the existing team (including the TA Deputy Project Manager if his contract could be extended) under funding to be arranged. 4. POST-IMPLEMENTATION Management of the Perimeters 4.1 However, these transfers of responsibility did not take place smoothly. A supervision mission in March 1987 determined that no real progress had been made in either case. It was agreed that a program would be drawn up detailing the activities to be undertaken by the ORD to support rice production in the 1987/88 season. This was done but it is now not clear to what extent this program was actually implemented by the ORD. It appears that the 1987 harvest in the areas supported by the project was reasonable, although actual figures are unavailable. The draft PCR was prepared by a mission in early 1988, and so reflects the position as the handover was becoming effective and before the events noted below. Problems quickly arose, however, as a series of external events occurred, which had a serious impact on the continued activities of the project. 4.2 In the first half of 1987 there was a sharp deterioration in the political situation in Burkina Faso. There had been periods since the 1983 coup when political pressures had caused problems for public servants and others but these became particularly acute at this time. This situation ended with a coup in September and the ousting of the Sankara regime. 4.3 Changes in the institutions supporting rural development were also underway. It had already been decided to abolish the ORDs and replaced by "Regional Centers for Agro-Pastoral Promotion" (CRPAs). The latter did not have a broad rural development mandate, but were to focus more on extension. What was particularly harmful to the project was that the decision had already been made in early 1987 but was not actually implemented for a year. The result was that the general uncertainty noted above, was made even more acute for the staff in the ORDs. Another problem was that the CRPAs did not have the degree of financial autonomy enjoyed by the ORDs. This resulted in a cutback in field activities and, in particular, the ability of the organization to maintain any infrastructure. The CRPAs also did not have independent funds to undertake activities to support cooperatives or supply inputs. 4.4 This became crucial after the 1988 cropping year, when there were serious floods in the Niena Dionkele area and a large part of the rice crop was lost. As a result many farmers were unable (or unwilling) to repay their input credits and the Village Groups fell into default. The CRPA did not - 16 - have funds to make further advances, and credit was now only available for groups from the Agricultural Credit Bank (CNCA). This, however, had (and has) strict conditions for lending, including the solvency of the borrowing group. The Village Groups could not demonstrate this and, therefore, could not borrow. This posed a severe problem as rice in the area was susceptible to 'blast', particularly when the soil surface was not flooded. Spraying was necessary to avoid serious crop loss and, when farmers were unable to spray in dry years, yields fell dramatically. 4.5 Marketing difficulties also increased. After the project ceased support, the technical difficulties of keeping the old Kou Valley mill operational increased. Internal problems within the Kou Valley Cooperative also increased and it is reported to have collapsed around 1990 and the rice mill closed. Marketing is still carried out through a parastatal supermarket, but with the reduced production from the Niena-Dionkele area and the financial difficulties of the Village Groups, individuals are more on their own, and marketing costs have risen. At the same time the relative prices for cotton and other grains are reported to have risen in the area. As noted (para. 3.12), cropping within the perimeters had begun to decline before the end of the project and, in the face of these new difficulties, it was soon abandoned altogether. 4.6 For a brief period at the end of the formal project implementation, the Project had control of the key required for the operation of the Karamassasso sluice gate. As noted (para. 3.5), this gate and its role in the management of water levels in the plain were a matter of contention throughout the project life. However, as rice production in the perimeter declined the ability to control the water level became moot. The audit mission was informed that the key is now held by a group involved in fishery in the lake. The local prefect obviously did not know of the earlier issues related to the water levels but, of course, felt that any problems could be resolved. Unfortunately, despite the fact that the operation of the gate has been a subject of contention since the first discussion of the project, there is still no systematic data available on the effect on the water table and rice production of different operating strategies for the gate. Village Land Management 4.7 Difficulties also soon surfaced on the village land management side of the continuation of the project's operations. It had been agreed that a Coordination Unit would be established in the office of the Presidency to guide and oversee the pilot operations in the field, and the Government of Norway agreed to provide funding. However, by March, although a special account had been opened, this unit had not formally been established. Although field work with villages had been continued in the Niena Dionkele area, field staff were beginning to become concerned about potential problems implied by the lack of formal support from the center. 4.8 This situation was clearly beginning to be impacted by the deteriorating political situation (para. 4.2). In fact the expatriate member of the project team went so far as to send a hand written note to the Bank Task Manager in May which noted: "The seminar on agricultural land reform has been put off for the third time, indefinitely... Because of this delay we have had to restart our sensitization campaign without the support of the authorities." 4.9 Meanwhile the uncertain status of the Coordination Unit was continuing to delay decisions. The Unit was finally established formally in August 1987 and placed in the 'Department du Monde Rural' of the Presidency. The coup of September immediately created a new set of uncertainties - 17 - over the program. A supervision mission arrived in Burkina Faso on November 1Ith to be informed that the department had been dissolved on November 9th and that it was planned to reallocate responsibility for the coordination of village land management program to the Ministry of Planning. This new arrangement was not finally regularized until April 1988. 4.10 The Niena Dionkele team had from the early part of the year been engaged on collecting baseline data on the region. In October they began an extensive survey to determine the distribution of land under the control of the different villages in the region and to determine in each village the system of land rights, and the correspondence between village and territory in order to determine the territory claimed by each village and the acceptance by other villages of the borders claimed. This survey indicated that there were five villages where conflicts existed. 4.11 The November supervision mission emphasized the importance of reaching a rapid agreement with those villages in which there were not land conflicts and, once this had been achieved, to find out the existing rights of land use and the relative importance of Mossi and Fulbe in them. The team should also work in an intensive way with those villages where there were conflicts in order to reach an agreement on boundaries etc. as soon as possible. The mission was also concerned that the project be able to combine the land management work with development initiatives. "The intervention of the State must not be reduced to research on actual social and economic conditions." 4.12 In practice, however, it was difficult for the field team to heed this advice. The reorganizations at the center meant that little support was forthcoming from that quarter. Then, as a further element in governmental reorganization following the coup, the ORDs were disbanded and replaced by CRPAs (para. 4.3). While the ORDs might have had access to funding for such work, the CRPAs did not and, in any event, in their first few months of existence were trying to sort out their functions, personnel etc.. Once the Coordinating Unit was formally up and running in the Ministry of Planning, it is not clear that the situation improved much for the field teams. The program was broadly supported by the donor community and increasing numbers were expressing interest in supporting the program. Missions arrived with increasing frequency, as the unit was still trying to reorient itself to its new situation. 4.13 The original plan, which provided for funding for the Niena-Dionkele extension, was based on the assumption that a project would be appraised before the end of 1988 and approved by mid 1989, so that it would be possible to continue support to ND under retro-active financing or some similar means, when the two year tranche of funding terminated at the end of 1988. The turmoil of 1987 had made that timetable no longer feasible, but the Bank and other prospective financiers were keen to move preparation along as fast as possible. This further restricted the ability of the Coordinating Unit to devote time to working with individual pilot operations in the field. 4.14 A supervision mission in mid 1988 noted that 'Comites de Gestion' were being established in each of the villages in the project area. In addition to the traditional ethnic groups, through their leadership in land matters, these were including representatives of the Mossi and Fulbe inhabitants, and other relevant village groups, although concern was expressed that this was leading to groups being large and potentially unwieldy. Five of the committees had more than 10 members. 4.15 The consultant also noted a further problem which was beginning to appear. "In discussions with the population, a more global approach to economic problems was proposed by both - 18 - extension agents and villagers, resulting in concrete demands by the local population. In light of this, the project is now planning small projects in the different villages, which will focus on the creation of one secondary road; water pumps; primary health centers; buildings for the storage of agricultural inputs; and mills. I agree with this approach, although it could lead to persistent demands which will not be met in the short term. The project is not a bank for rural investment projects and this should be made clear to the villagers. Expectations which cannot be satisfied usually provoke discreditability and frustration. The main aim of the project is to convince the local population that proper management of existing local resources is in itself a 'kind of developmental strategy'. This idea is perhaps difficult to communicate but a 'populist' strategy can generate a series of uncontrolled pressures." 4.16 Unfortunately, these fears were realized. The expatriate member of the team left in mid 1988 and the effort limped along, being funded through a special trust fund arrangement. Quite why events took the turn they did is not entirely clear. Those remaining from the team on the ground reported to the audit that they did not receive support and approval from the center. Others interviewed implied that they could not understand why operations ground to a halt, but felt that the field team lacked initiative. Some stated that funds ran out, others believed funds were available, but were not utilized. What is clear is that, while the pilot effort in the Niena Dionkele is well reported in the early stages of the preparation of the Environmental Management Project (EMP), after early 1989 it fades from view. There now appears to be no institutional memory in the headquarters unit of the EMP of what was happening in Niena Dionkele at the time. 4.17 A mission in October 1989 recommended that responsibility for activities be transferred to the local CRPA. However, as far as could be determined by the Audit mission, the CRPA had limited resources for any 'Gestion de Terroir' work, and support from the center diminished as the emphasis on project preparation and appraisal increased. Frustration set in among the villagers and, given the potential problems among the social groups, the underlying tensions began to surface. In the absence of any follow through in terms of concrete action, agreement within the local 'Comites de Gestion' became extremely difficult to achieve. In consequence the input to the effort was wound down. In a number of cases, preliminary village land zonation had been agreed, but whether these proposals would still be viable is not yet clear. 4.18 The central unit of the Environmental Management Project (Cr. 2229-BUR), which is supporting the PNGTV has attempted in the latter half of 1993 to reestablish contact with the villages. There remains considerable tension and frustration at the village level and it will clearly take a sustained effort to rebuild the confidence which existed at the time of closure of the Niena Dionkele project. The major problems of the area, linked to agriculture, are the lack of effective demand for any increased production, which is linked to the access problems still facing a number of the villages. The new project is hampered in that it can only provide direct support to activities directly related to land management and not to local infrastructure. In the absence of any formal framework for support of local infrastructural development, this problem will remain difficult to address. 5. FINDINGS AND LESSONS 5.1 At completion, the PCR, while recognizing many of the problems outlined in the earlier sections, was modestly optimistic; "The project showed that it was possible to develop bottomlands for - 19 - rainfed rice cultivation." Farmers were cultivating rice in the project area before the project started and continue to do so on a small scale. However, the perimeter developed by the project is now unutilized and, therefore, the approach promoted by the project did not work. The PCR did note requirements for production in the project perimeter to be sustainable: availability of inputs, adequate markets and research for disease control. These were not met, and little now remains of the project's effort. 5.2 The audit did not attempt to re-estimate the rate of return from the project. Since there has been no production for the past three seasons, and no reason to expect that it would be resumed without significant further inputs, the ERR is clearly below zero. The audit did not believe that an attempt to obtain a precise estimate was justified. 5.3 While the immediate objectives of the project were not achieved, it should be noted that, during the second half of its implementation, the project did provide a valuable input to the evolving discussion of how to tackle the increasing land management problem. The initial proposals for the second phase, produced by the technical assistance team, did provide one of the few attempts at that time to investigate this problem in depth and prepare proposals for tackling it at the village level. The 'Programme National de Gestion des Terroirs Villageois', which subsequently evolved, incorporated many of the ideas generated in Niena Dionkele. However, it should be noted that this result was dependent on the existence of non-project links. If the Bank had not been becoming interested in the issue at the time and encouraged the team to pursue it, it is likely that a limited amount of work would have been done, and then left on the shelf at completion. 5.4 The Audit has considered it necessary to lay out rather comprehensively the history of the project since so much of why it failed after completion results from the earlier events. It is the judgement of the audit that the long term failure of this project effort relates both to problems in project design and in implementation. 5.5 The fundamental problem was a failure to clarify the basic objectives of the project and, in particular when it was agreed that a pilot operation was needed, to establish exactly what it was that the project was attempting to determine, to design an operation of appropriate scale and devise an appropriate organizational structure for it. The objectives of the project were too disparate; to try out a new approach to rice production in small flood plains (a very local issue), and to strengthen the capacity of the central ministries to address sectoral issues in rice production. Technical Issues 5.6 On the technical side there were two key questions: (a) whether, with flood protection only, rice could be grown as a rainfed crop under the conditions of the Niena Dionkele area; (b) whether such production was compatible with the existing local farming system. 5.7 Chapter 2 of this report outlined the twists and turns which occurred during the preparation of this project. At each stage there was a shift in the basic objectives of the project. Although at appraisal the rice production was clearly stated to be rainfed, this was not always clear through the preparation process. The feasibility study looked at the possibility of irrigation as one of the - 20 - options, but this was obviously not economically viable. The proposal was progressively scaled back to the least costly, involving just a protection dike. However, use of the protection dike was decided at a late stage with little investigation, and two other key elements of project design (plot layout and the scale of the perimeter) suggest that the implications of purely rainfed production were not fully taken on board by the engineers or agronomists. 5.8 The approach adopted was flawed by three basic technical failures: (a) the design of the polder; (b) lack of a recognition that a design which did not enable any control over water inside the polder would require fundamental change in the management of the cropped area, compared to irrigation; and (c) inadequate participation of local farmers in the design so that the likelihood of a successful operation could be maximized. 5.9 The dike. The dike was designed as a simple flood protection structure and there are no structures to permit water to be brought into, or evacuated from, the cropped area enclosed within the dike. While considerable feasibility and preparation work had been carried out, there were still basic gaps in knowledge. The arrangements for control of the sluice gate below the project site, which controlled water levels in the plain were never settled and, linked to that, the extent to which the polder would drain naturally without drainage structures was not known. There was also no detailed plan for the execution of the initial works. Technically the project was not ready for appraisal. 5.10 Plot Layout. Each participant had a single plot of a half hectare. It was assumed that field bunds between these plots would provide adequate water control to provide for a single rice crop. It was not recognized that, under the highly variable rainfall conditions of Niena Dionkele, this was highly unlikely. Even though slope was minimal, in dry seasons part of the area had inadequate water, while in a wet season part was flooded too deeply for the recommended varieties. The result was that the chance of a single plot having good growing conditions was poor. Under rainfed conditions of this type, farmers would normally respond by having several smaller plots at different levels, so that at least some would have adequate water whatever the conditions. 5.11 Size of the Perimeter. There was insufficient attention paid to the existing farming system, and of the management implications for the farmers of an additional rainfed rice plot. While project was to investigate the compatibility of this type of rice production with the existing systems, and appraisal had raised the question of labor availability, design was not modified to take this into account. Many of the farmers lived far from the perimeter and it was difficult for them to keep a close eye on conditions on their plot. Most major operations; cultivation, weeding or harvesting on a half hectare plot would take more than a day. If the area had been irrigated, farmers would have known in advance when their plots would be ready for cultivation etc., and would have been able to plan their trips well in advance. Under rainfed conditions the appropriate time for the different operations was difficult to predict. As a result it was more difficult than it need have been for farmers to manage their plots well and achieve good yields. It would have been preferable to have constructed smaller perimeters close to the villages. - 21 - 5.12 The overall viability of the project as presented was based on the completion of the larger two-phase sequence of operations. This approach tended to induce a larger first phase than would have been needed to test the approach to rice production. A purely experimental approach, based, for example, on an initial 100 hectare polder, would have been much more flexible. This might have been justified on the expected viability of the larger scale operation if the technology proved successful. Resolving the Technical Issues 5.13 The design was shown to be unsatisfactory after the first year of cropping when, because of lack of rainfall, half the crop was lost. Supervision did follow up on this issue and investigations to determine appropriate measures were agreed as part of the preparation for the second phase. However, as the monitoring and evaluation work got fully underway, the difficulties for local farmers in using the polder became apparent. It was soon agreed to change the focus of the second phase to upland management problems, implicitly accepting that the use of the polder was incompatible with the priorities of the local systems. The project also began to provide support for rice growing (through credit and inputs) for farmers cropping outside the polder. However, no formal decision was taken to accept that the polder was not effective, or to pursue investigations on possible steps to improve its performance. 5.14 Similarly, at completion no definitive decision was taken as to whether the pilot had failed. The PCR implicitly accepts the failure of the effort when it notes that "the solution of a polder without water control will have to be rethought." Loose arrangements were made for continued management of the polder (paras. 4.1-4.6), but no provision was made for the deficiencies identified earlier to be remedied after the end of the project period. There is still no clear evidence on which to base a judgement as to what might be done with the polder. 5.15 The Niena-Dionkele area was chosen for a pilot project because it is one of a number of smallish flood plains with similar conditions. As population grows and development proceeds, land use in these areas will be intensified. At least in average or wetter years significant areas become waterlogged and large amounts of surface water have to be managed. Given the amount of water to be managed, rice will almost certainly have a role to play as an economically viable land use under these conditions. However, full water control will not be feasible in the foreseeable future and some form of rainfed paddy production will be necessary. 5.16 Unfortunately, since the project failed to follow up on the issues raised early in the project period, little was in fact learned about improving rainfed rice production in this region, the possible role for polders, or (in this specific situation) the impact of alternative management strategies for the weir on upstream cropping. Moreover, as a result of this unsatisfactory experience the Bank has been discouraged from any further significant involvement in rainfed paddy rice production and others have not filled this gap. The Organizational Problem 5.17 The attempt to create a small project which operated at both the very local and national levels led to an organizational structure which was almost destined to fail. The prime objective was to conduct a pilot operation in rice production. This required detailed attention of the manager at the local level, with a relatively small project team. In this capacity it reported to the local ORD, which was appropriate. However, at the same time the project had a Rice Industry Support Group, including an - 22 - adviser to the Ministry of Rural Development, to focus on regional marketing and national policy issues. The latter in particular involved activity which was entirely outside the remit of the ORD. 5.18 The solution of creating an Executive Committee at the national level to oversee the project as a whole did not work. The committee was large and unwieldy and seldom met. It is difficult to envisage a wide range of national level organizations taking a great interest in a small rainfed rice trial which was the day to day responsibility of the ORD. The other elements of the project also were unlikely to require extensive central oversight. The result was to create a frustration rather than a support for project management, and a source of potential problems for the ORD. 5.19 Furthermore, the Rice Industry Adviser was located in Bobo Dioulasso, 250 km from the ministry he was supposed to be advising. Even without the turmoil which led to several ministerial changes in the latter half of the project period, it would have been difficult for him to develop satisfactory working relationships with those he was supposed to advise. 5.20 A small pilot project focussing only on the field problem of the potential for rice production in plains subject to irregular flood, under the general control of the ORD would have been preferable. This would have freed project management to concentrate on the field work and would have resulted in a better relationship with, and support from, the ORD. Post-Project Support 5.21 Support for rice production might have been given through the special funding which had been arranged to support the pilot land management effort. However, this effort was immediately thrown in turmoil by political changes. The latter might not have been predictable, but it is clear that the Bank's main focus was on preparing an Environmental Management project. In so far as Niena Dionkele area was a good example of the problems which such a project would aim to address, it was of some interest, but in practice it was "relegated to the back burner". The type of interim funding arrangements made inevitably place heavy demands on staff time in relation to the resources involved. As a result staff spent time on this project but with minimal impact in the field. 5.22 With hindsight, it might have been better to have extended the original project by two years, to guarantee support, or to have terminated the land management activities at the end of the project period. Continuing a dialogue with the villagers over the clarification of land rights, in the context of an anticipated broader development effort, which then failed to materialize was counterproductive. There was no counterweight to the tensions which were bound to arise in such an exercise. 5.23 Experience with village level teams in the ongoing Environmental Management project suggests that a similar problem is arising. While the effort is intended to respond to the expressed needs of the villages undertaking land management programs, the project itself only has funds for financing land management activities. If villages wish to undertake other small projects, the project has to try and assist them in obtaining resources from elsewhere. In the absence of any coherent framework for planning and funding small rural projects, this becomes a very ad hoc affair, which complicates project relationships. This issue is more specifically relevant to the Third Rural Development Fund Project and is discussed in detail in the audit of that project. - 23 - 5.24 It should not be concluded that the project failed because of the land related problems, i.e., because immigration increased local concerns over maintaining rights over land and so undermined the effort to manage the polder. Its failure resulted from an inability to agree on the prime goals of the activity and to stick with them until they were resolved one way or the other. This, in essence, was the problem from the very beginning, and "every puff of wind" resulted in a change in direction. Lessons 5.25 The difficulties of this project and its ultimate failure illustrate the importance in pilot operations of laying out clearly the objectives of the 'investigation' which will be undertaken and the steps by which they would be assessed. Where a project is clearly experimental, it is preferable that this be the central theme. Such an approach would enable the project to have been clearly presented as a pilot, on a smaller, more appropriate scale. The pre-appraisal had indicated that, if the approach proved feasible, the type of production being tested would be economically viable. 5.26 For such projects clear arrangements for review of the investigations are advisable. In this case a mid-term review would have been valuable in ensuring that all parties had an opportunity to assess the situation and agree on steps to be taken. By the third year of implementation of this project problems with its operation were clear and implicitly accepted, but not formally agreed. 5.27 Similarly, the outcome underlines the importance of a formal conclusion to the project and agreement and commitments on follow-up. In this case the effective conclusion was that the pilot had been a partial success, but no provision was made to remedy the deficiencies identified earlier. This lack of clarity effectively undermined also those activities on which more specific measures were taken, and left everyone concerned little the wiser about the basic issue the project was to address - rainfed rice production with limited water control. 5.28 Small investigative operations are staff intensive. Even when these are funded from other sources, as with the follow-up land management activities, the Bank must make staff provision for support work. If this cannot be justified then the basis for seeking third party support should be called into question.  tt z--- ---- .02 5 O * Ä - - 1-020 3k\ loo cr * - - 0o c -c- 0.0 = 5; 9°~ 0r C D7 0--.* 5 .® s oo - o eo O 25or N S-- -. 0• 0 n - a. r o .~ o - fl ,.°0 0o C 0 - 3 ~ 00,0 o o * < * n -. .- Z002, 'o 0 0// z - -D z 0 ~z2  BURKINA FASO NIENA DIONKELE RICE DEVELOPMENT PROJECT-, PROJET DE DEVELOPPEMENT AGRICOLE DE NIENA DIONKELE PROJECT AREA / ZONE DU PROJET Dl Gr MAJOR ROADS / ROUTES IMPORTANTS RAILWAY / CHEMIN DE FER -- RIVER S / R IVIE RES Ouah,guya - - - INTERNATIONAL BOUNDARIES I FRONTIERES A LI -- O U A G A D O U G O U K e aF i q DAFonkee 02 ýou BotIDmo~ B E N l N F araka-Ba y Bant" G H A N A 17 -"1BURINA FASO A FFR 1 O U E .° y . ,dg "- , - N s . °h, °f "y Ke LOME TE R$ C O T E D'Y1 V0O RE