Document of The World Bank FOR OFFICIAL USE ONLY Report No. 12628 PERFORMANCE AUDIT REPORT SOMALIA NORTHWEST REGION AGRICULTURAL DEVELOPMENT PROJECT (CREDIT 635-SO) NORTHWEST REGION AGRICULTURAL DEVELOPMENT PHASE II PROJECT (CREDIT 1538-SO) BAY REGION AGRICULTURAL DEVELOPENT PROJECT (CREDIT 972-SO) DECEMBER 30, 1993 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. FOR OFFICIAL USE ONLY CURRENCY EQUIVALENTS Name of currency (abbreviation) - Somalia shilling (So.Sh.) Exchange Rates US$1,00 = .iy- NWI NWII Appraisal 6.23 6.23 26.26 Intervening years average - - 39.5 Completion year average 491.0 35.64 491.0 ABBREVIATIONS ADS - Agricultural Development Service (of World Bank) AfDF - African Development Fund AFMET - Agricultural Extension and Farm Management Training Project BRDCC - Bay Region Development Coordination Committee CGDP - Comprehensive Groundwater Development Program CP - Food and Agriculture Organization/World Bank Cooperative Program CPS - Central Projects Staff (of the World Bank) DCA - Development Credit Agreement EEC - European Economic Community ERR - Economic Rate of Return FC - Financial Controller GDP - Gross Domestic Product GOS - Government of Somalia HQ - Headquarters ICRISAT - International Crops Research Institute for Semi-Arid Tropics IDA - International Development Association IDRC - International Development Research Center IFAD - International Fund for Agricultural Development MOA - Ministry of Agriculture M&E - Monitoring and Evaluation MPW - Ministry of Public Works MTR - Mid-term Review NMEF - National Monitoring and Evaluation Facility NWI - Northwest Region Agricultural Development Project I NWII - Northwest Region Agricultural Development Project II PADU - Pilot Agricultural Development Units PCR - Project Completion Report PD - Project Director PM - Project Manager PMU - Project Management Unit PPF - Project Preparation Facility This document has a restricted distribution and may be used by recipients only in the performance of their oncial duties. Its contents may not otherwise be disclosed without World Bank authorization. ABBREVIATIONS (continued) RMEA - World Bank Regional Mission in Eastern Africa SAR - Staff Appraisal Report SDA - Settlement Development Agency SOGREAH - Soci6td Grenobloise D'6tudes et D'application Hydrauliques TA - Technical Assistance TM - Technical Manager USAID - United States Agency for International Development WDA - Water Development Agency GLOSSARY Bund - A bank or dike constructed of earth for soil and water conservation Lowlands - relating to the Northwest projects, this is a confusing term in the project documents and studies, being used in two senses: (a) on the gently undulating plateau forming the project area, those slightly lower areas of level to gently sloping land with deep soils which comprise almost all of the land with agricultural potential, or (b) those parts of (a) where land slopes of more than 3 percent (sometimes 2 percent) allow water harvesting from uncropped land to increase water available to crops nearby. The term is emphasized in the text with quotation marks to alert the reader to its special meaning in this Somalia context. Water harvesting - the use of rainfall surface runoff from slopes and runoff from ephemeral streams to increase moisture available for plant growth. FOR OFFICIAL USE ONLY THE WORLD BANK Washington, D.C. 20433 U.S.A. Office of Director-General Operations Evaluation December 30, 1993 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Performance Audit Report on Somalia Northwest Region Agricultural Development Project (Credit 635-SO) Northwest Region Agricultural Development Phase II Project (Credit 1538-SO) Bay Region Agricultural Development Project (Credit 972-SO) Attached is the Performance Audit Report on Somalia - Northwest Region Agricultural Development Project (Credit 635-SO), Northwest Region Agricultural Development Phase I Project (Credit 1538-SO) and Bay Region Agricultural Development Project (Credit 972- SO) prepared by the Operations Evaluation Department. These projects to assist rainfed farming were less successful, before insecurity interrupted operations, than ambitious designs projected. Even so, the Bay Region project was rated satisfactory overall (reversing the PCR rating) largely because the project design was flexible such that at the planned review the infeasible production components were dropped in favor of investments in infrastructure and technology development. Achievements were then on target. Infrastructure investments combined with non-project policy changes contributed to production growth. Both Northwest projects were unsatisfactory (reversing the PCR judgment on the first one) despite substantial physical implementation, largely because the economic rate of return was low -- benefits to the main investment (soil and water conservation) were much less than projected. A trial program for semi-mechanized cropping was supported by the projects and initially showed promise, but then developed technical and cost recovery problems by the time it had become an incomplete Bank-assisted project. The audit emphasizes the need for detailed knowledge of human and natural resources during project design, the importance of progress monitoring and periodic re-evaluation of objectives and potential impact (to assess the need for strategy changes), and the need to consider financial and social aspects when attempting to adapt introduced technology to local conditions. It is hoped that this audit will contribute to resumption of operations in Somalia when conditions permit. 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 SOMALIA NORTHWEST REGION AGRICULTURAL DEVELOPMENT PROJECT (CREDIT 635-SO) NORTHWEST REGION AGRICULTURAL DEVELOPMENT PHASE II PROJECT (CREDIT 1538-SO) BAY REGION AGRICULTURAL DEVELOPMENT PROJECT (CREDIT 972-SO) TABLE OF CONTENTS Page No. PREFA C E .............................................. i BASIC DATA SHEETS ........................................ iii EVALUATION SUMMARY ..................................... xiii 1. INTRODUCTION .................................... 1 O bjectives .......................................... 2 2. NORTHWEST REGION PROJECTS ........................ 4 A. Northwest Agricultural Development Project .................. 4 Design/Preparation .................................. 4 Implementation .................................... 8 B. Northwest Agricultural Development Phase II Project ............. 12 Design/Preparation .................................. 12 Implementation .................................... 15 3. BAY REGION AGRICULTURAL DEVELOPMENT PROJECT ...... 18 Design/Preparation .................................. 18 Implementation .................................... 20 4. SEMI-MECHANIZED RAINFED FARMING OPERATION ......... 26 Background and Design ............................... 26 Implementation .................................... 28 This document has a restricted distribution and may be used by recipients only in the performance of their ofcial duties. Its contents may not otherwise be disclosed without World Bank authorization. TABLE OF CONTENTS (Continued) 5. PROJECT OUTCOMES ................................ 32 Northwest Region Agricultural Development Project and Northwest Region Agricultural Development Phase H Project . 32 Bay Region Agricultural Development Project .................... 34 Semi-mechanized Rainfed Farming Operation .................... 36 6. BANK PERFORMANCE ................................ 37 7. CONCLUSIONS, LESSONS AND FUTURE ACTIONS ........... 38 Conclusions: The Two Northwest Projects ...................... 39 Conclusions: The Bay Region Project ........................ 40 Conclusions: The Semi-mechanized Rainfed Farming Operation . . 40 Lessons Learned ............. 41 Suggestions for Future Activities ............................ 41 Working Papers (available on request from OED) 1. The Impact of Bunding in the Northwest 2. Brief Account of the Semi-mechanized Rainfed Farming Pilot Project - Outcome and Technology Issues Maps - IBRD No. 25134 - Bay Region ADP, Northwest Region ADP, and Agricultural Settlement Areas IBRD No. 25135 - Northwest Region Agricultural Development Project - Phase II Project Area IBRD No. 25136 - Bay Region Agricultural Development Project - Physiography and Project Infrastructure Development List of Boxes Page No. 1 - Comparative Technical Aspects of Farming in the Northwest and Bay Regions ...... ................................ . 3 2 - Effect of Rainfall on Crop Yields in the Northwest Project Area .............. 5 3 - Bunding in Northwest Somalia ....................................... 8 4 - Technical Assistance Management ................................... 10 5 - Research and Extension ...... ................................... 11 6 - Bunding Impact on Sorghum Yields .................................. 13 7 - Use of Feasibility Study and PMU Data During Phase l Appraisal ...... .................................. 14 8 - Technical Problems - Bunding and Erosion Control ..................... ..17 9 - Sorghum Yield Survey Findings (1987)................................. 18 10 - Changes of Bay Region Project Components at the Mid- Term Review ...... ......................................... 23 11 - PMU Summary of Mid-Term Review Outcome ...................... . 24 12 - Post MTR Research Record........................................ 25 13 - Similarity and Affinity Between the Bay Region and Drought Rehabilitation Projects ..... ............................ 27 14 - Technical and Institutional Details of the Semi-mechanized Rainfed Farming System .......................... ..29 15 - Audit Comments on the Semi-mechanized Rainfed Operation ......................................... 31 PERFORMANCE AUDIT REPORT SOMALIA NORTHWEST REGION AGRICULTURAL DEVELOPMENT PROJECT (Credit 635-SO) NORTHWEST REGION AGRICULTURAL DEVELOPMENT PHASE II PROJECT (Credit 1538-SO) BAY REGION AGRICULTURAL DEVELOPMENT PROJECT (Credit 972-SO) PREFACE This is a Performance Audit Report (PAR) on the Somalia Northwest Region Agricultural Development Project, Northwest Region Agricultural Development Phase II Project, and Bay Region Agricultural Development Project, involving IDA Credits 635-SO, 1538-SO and 972-SO in the amounts of US$10 million, US$10.6 million, and US$12 million to the Government of Somalia, with the objectives of improving rainfed farming. The Credits were approved on June 3, 1976, January 8, 1985 and December 27, 1979, respectively, and became effective on February 9, 1977, July 10, 1985 and September 12, 1980. Credits 635-SO and 972-SO were closed on June 30, 1985, and March 31, 1989, respectively, after both had been extended for two years. Due to the civil war in the country, Credit 1538-SO was suspended in March 1991, three months before the official Closing Date of June 30, 1991. SDR 5.7 million remains undisbursed. The last disbursement was on June 11, 1992. The credit account cannot be closed because the special account cannot be reconciled. The final disbursements for the two other Credits were on January 14, 1986, and December 29, 1989 and outstanding balances of US$0.05 million for Credit 635, and US$0.09 million for Credit 972 were cancelled. The PAR is based on Project Completion Reports (PCRs) dated November 2, 1992, May 28, 1993, and May 14, 1993, prepared by the Africa Regional Office, the Staff Appraisal and the President's Reports, the credit documents, a study of project files, and on discussions with IDA staff. Because of the prevailing security situation in Somalia, it was not possible to visit the country. However, the audit has taken advantage of the availability of a former staff member of the Bank who has visited the project areas and is familiar with the technical issues. The PCRs give a comprehensive account of the projects' evolution, implementation and results. The PAR elaborates on project concepts and designs, the implementation experiences, and, in particular, on technical aspects of resource conservation and rainfed crop production in Somalia. This desk audit should also contribute to resolving technology issues arising from these projects, in anticipation of the eventual resumption of operations in Somalia. In consequence, this audit is more detailed than usual and the Table of Contents lists two working papers (available from OED) on bunding and the further progress of the Semi-mechanized Rainfed Agriculture Project (Credit 1774-SO). The audit should also be of interest to those working on other semi-arid areas of Africa. The Task Manager for this report was Julian Blackwood, OEDD 1, and the consultant was Gert Stern. - Iii - PERFORMANCE AUDIT REPORT SOMALIA NORTHWEF REGION AGRICULTURAL DEVELOPMENT PROJECT (CREDIT 635-SO) BASIC DATA SHEET KEY PROJECT DATA Appraisal Actual or Actual as % of Estimate Current Estimate Appraisal Estimate Project Cost (US$ million) 13 14.1 101 Credit Amount (US$ million) 10.0 9.95 99 Canceled (US$ million) -- 0.05 -- Economic Rate of Return (%) 11 n.a. KEY PROJECT DATES Other Project Data Original Revised Actual Identification (FAO/IDA) - - 11/72 Preparation (FAO/IDA) - - 10/73 Appraisal - - 05/74 Negotiations - - 01/14-16/76 Date of Board Approval - - 06/03/76 Date of Credit Agreement - - 07/07/76 Date of Effectiveness 10/05/76 12/06/76 02/08/77 01/06/77 02/04/77 Date physical components completed Proportion then completed (%) Closing Date 06/30/83 06/30/84 06/30/85 CUMULATIVE ESTIMATED AND ACTUAL DISBURSEMENTS (US$ million) FY 78 79 80 81 82 83 84 85 86 Appraisal Estimate (US$ million) 0.7 3.3 6.0 7.5 9.0 10.0 - -- -- Actual (US$ million) 0.6 2.7 3.6 6.0 7.1 7.9 9.5 9.9 9.9 Actual as % of Estimate 86 82 60 80 79 79 95 99 99 Date of final disbursement: January 14, 1986 - Iv - STAFF INPUTS (staff weeks) FY 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 Total Identification/ Preparation 2.9 18.9 21.8 Appraisal 27.3 43.0 4.6 74.9 Negotiations 11.2 11.2 Supervision .3 13.5 7.9 13.0 14.2 12.7 14.6 7.3 7.0 2.2 92.7 PCR 0.4 8.1 5.4 13.9 Total 2.9 46.2 43.0 16.1 13.5 7.9 13.0 14.2 12.7 14.6 7.3 7.0 2.2 0.4 8.1 5.4 214.5 MISSION DATA Date No. of Specializ. Performance Types of (mo/yr.) Persons Represented b\ Rating c\ Problems d\ Preparation 09/73 5 PA,A,E,LS,EC Appraisal 04/74 4 EC,A,E,FA Supervision I 08/76 1 AE 1 MT II 03/77 1 A 1 - III 06/77 2 FA,LS - MT IV 03/78 1 A 2 MTO V 11/78 2 E,LS 2 MO VI 06/79 2 A,CS 2 MO VII 01/80 2 A,E 2 MTO VIII 03/80 1 A 2 MT IX 10/80 1 A 2 FTO X 03/81 1 A 2 TO XI 10/81 2 AE,A 2 TO XII 04/82 2 AE,A 1 TPF XIII 02/83 2 AE,A 1 PO XIV 09/83 3 FA,AE,A 1 0 XV 03/84 1 A 1 0 XVI/e 09/84 1 A 1 OT XVII 06/85 1 FA 1 OT OTHER PROJECT DATA Borrower : Government of Somalia Implementing agency Northwest Region Agricultural Development Project Implementing Unit (under Ministry of Agriculture) Fiscal year of borrower January I - December 31 Follow-on project: Name : Northwest Region Agricultural Development Project II Credit Number 1538-SO Amount (US$ million) 10.6 Approval Date January 8, 1985 /a Cannot be calculated with reasonable confidence due to the lack of reliable data. /b AE = Agricultural Economist; FA = Financial Analyst; LS = Livestock Specialist; A = Agriculturist; CS = Conservation Specialist; E = Engineer; EC = Economist; PA = Project Analyst. /c 1 = problem-free or minor problems; 2 = moderate problems. /d M = Management, T = Technical, 0 = Other, F = Financial, P = Political /e From late 1984 to mid-1985, there was a serious cholera outbreak in the Northwestern part of Somalia including the project area, which prevented Bank supervision visits. Concluding supervisions were combined with Phase II project. The above does not include project implementation reviews held in Somalia by Government and IDA on an annual basis. - Vii - PERFORMANCE AUDIT REPORT SOMALIA NORTHWEST REGION AGRICULTURAL DEVELOPMENT PROJECT - PHASE II (CREDIT 1538-SO) BASIC DATA SHEET KEY PROJECT DATA Appraisal Actual or Actual as % of Estimate Current Estimate Appraisal Estimate Project Cost (US$ million) 25.6 n.a. Credit Amount (US$ million) 10.6 6.3 59 Canceled (US$ million) La - - Cofinancing - IFAD (US$ million) 7.0 n.a. - EEC (US$ million) 5.4 n.a. Economic Rate of Return (%) 16 n.a. KEY PROJECT DATES Other Project Data Original Revised Actual Identification 05/83 - 05/83 Preparation 09/83 11/83 12/83 Appraisal 12/83 03/84 01/85 Negotiations 10/84 11/84 01/85 Date of Board Approval 02/85 - 01/85 Date of Credit Signature - - 02/10/85 Date of Effectiveness 04/85 07/85 07/10/85 Date of Credit Closing 06/30/91 - 06/30/91 Credit Completion 12/31/90 - 06/30/91 CUMULATIVE ESTIMATED AND ACTUAL DISBURSEMENTS (US$ million) FY85 FY86 FY87 FY88 FY89 FY90 FY91 Appraisal Estimate (US$ million) - Annual -- 1.0 3.4 2.2 1.7 1.55 0.75 - Cumulative -- 1.0 4.4 6.6 8.3 9.85 10.6 Actual (US$ million) - Annual 0.3 1.8 1.9 1.8 0.2 0.2 0.1 - Cumulative 0.3 2.1 4.0 5.8 6.0 6.2 63 Actual as % of Appraisal Estimate 3 27 38 54 56 58 59 Date of final disbursement: June 11, 1992 - Viii - STAFF INPUTS (staff weeks) 3 1984 1985 1986 1987 1988 1989 1990 1991 Preappraisal 9.2 35.9 Appraisal 20.3 10.4 Negotiations 3.7 Supervision 2.8 19.5 15.3 11.9 4.4 3.1 0.5 PCR 7.0 Total 9.2 56.2 16.9 19.5 15.3 11.9 4.4 3.1 7.5 MISSION DATA Date No. of Specializ. Performance Types of (mo/yr.) Persons Represented /b Rating /c Problems /d Preparation 10/83 1 FA Appraisal 03/84 4 FA,LS,Ag,LS Supervision I 05/85 1 FA * II 02/86 2 FA,C * III 03/86 4 FA,F,ES,C 1 F, M, I IV 10/86 3 FA,F,C 1 F, M, I V 01/87 1 FA * VI 06/87 1 FA 2 M, I VII 12/87 1 Eco 2 M, I VIII 12/88 1 Eco * IX 12/90 1 FA 3 I, L, M OTHER PROJECT DATA Borrower : Government of Somalia Implementing agency Project Management Unit Fiscal year of borrower January 1 - December 31 Follow-on project: None /a Credit account not yet closed (see Preface). /b Ag = Agronomist, C = Consultant, Eco = Economist, ES = Extension Specialist, F = Forester, FA = Financial Analyst, LS = Livestock Specialist. /c = problem-free or minor problems; 2 = moderate problems; 3 = major problems. /d F = Financial, I = Impact, L = Legal, M = Management. * No Form 590 found on file. - ix - PERFORMANCE AUDIT REPORT SOMALIA BAY REGION AGRICULTURAL DEVELOPMENT PROJECT (CREDIT 972-SO) BASIC DATA SHEET KEY PROJECT DATA Appraisal Actual or Actual as % of Estimate Current Estimate Appraisal Estimate Project Cost (US$ million) 45.0 51.0 113 Credit Amount (US$ million) 12.0 11.91 99 Canceled (US$ million) 0.9 Economic Rate of Return(%) 20 5 KEY PROJECT DATES Other Project Data Original Revised Actual Identification (Project Brief) - - 11-12/76 Preparation - - 06/77 Preappraisal - - 12/77 Appraisal 07/77 - 06-07/78 Post-appraisal - - 12/78 Negotiations - - 11/6-9/79 Date of Board Approval - - 12/27/79 Date of Credit Signature - - 03/26/80 Date of Effectiveness - - 09/12/80 Project Completion 03/86 09/87 09/88 Closing Date 03/31/87 03/31/88 03/31/89 CUMULATIVE ESTIMATED AND ACTUAL DISBURSEMENTS (US$ million) FY 81 82 83 84 85 86 87 88 89 90 91 Appraisal Estimate (US$ m.) 0.4 1.74 5.46 7.86 9.54 11.22 12 - - - - Actual (US$ m.) 0 0.13 1.04 1.77 2.54 4.26 7.01 8.96 10.20 11.53 11.91 Actual as % of Est. 0 7.5 19 23 27 38 58 75 85 96 99 Date of final disbursement: December 29, 1989 STAFF INPUTS (staff weeks) FY 77 78 79 80 81 82 83 84 85 86 87 88 89 Total Preappraisal 83.2 22.4 105.6 Appraisal 69.2 8.3 77.5 Negotiation 2.7 2.7 Supervision 0.9 18.9 12.8 26.5 37.3 26.4 18.2 14.3 11.1 6.2 172.6 Total 83.2 22.4 69.2 11.9 18.9 12.8 26.5 37.3 26.4 18.2 14.3 11.1 6.2 358.4 MISSION DATA Date No. of Specializ. Performance Types of (mo/Vr) Persons Represented a\ Rating b\ Problems e\ Identification 11/76 3 AE,A,GS Preappraisal 12/77 1 E Appraisal 06/78 6 AE,H,FA,LS,CE,E Supervision I 07/80 3 AE,A,E 1 M II 03/81 4 AE,A,E,MES I M III 11/81 2 AE,A 2 M,F,O IV 06/82 1 AE 2 M,F,O V 12/82 2 AE,A 3 M,F,T,O VI 06/83 2 E,E VII 08/83 3 AE,A,CE 3 M,F,T,O VIII 04/84 5 AE,A,IS,FA 3 M,F,T IX 12/84 1 E X 02/85 3 AE,A,CE 3 M,T,F XI 06/85 2 AE,IE 3 M,T XII 02/86 1 AG 3 M,T,O XIII 10/86 1 A 2 M,I XIV 04/87 1 A 2 M,I XV 12/87 3 A,A,E 2 M,I XVI 05/88 2 E,H 2 M,I XVII 01/89 2 E,A 2 M,I OTHER PROJECT DATA Borrower : Government of Somalia Implementing agency Project Management Unit Fiscal year of borrower January I - December 31 Follow-on project: : None - XI - /a A = Agronomist; AE = Agricultural Economist; AG = Agriculturalist; CE = Civil Engineer; E = Economist; FA = Financial Analyst; GS = Ground Water Specialist; H = Hydrologist; = IE = Irrigation Engineer; LS = Livestock Specialist; MES = Monitoring and Evaluation Specialist. /b I = no significant problems; 2 = moderate problems; 3 = major problems. IC F = Financial; M = Managerial; T = Technical; I = Impact; 0 = Other. - Xiii - PERFORMANCE AUDIT REPORT SOMALIA NORTHWEST REGION AGRICULTURAL DEVELOPMENT PROJECT (CR. 635-SO) NORTHWEST REGION AGRICULTURAL DEVELOPMENT PHASE II PROJECT (CR. 1538-SO) BAY REGION AGRICULTURAL DEVELOPMENT PROJECT (CR. 972-SO) EVALUATION SUMMARY Introduction 1. The audit reviews three separate efforts by 4. The objectives and components of the the Bank to assist rainfed farming by technology second Northwest ADP approved in January 1985 investments. These operations were in the three were the same as before, with the addition of pilot main semi-arid areas of Somalia where rainfed programs for inputs procurement and distribution, crop production is possible. There were two and for catchment protection. Cofinancing was Northwest Agricultural Development Projects, and provided by the European Economic Community in the south the Bay Region Agricultural Develop- and the International Fund for Agricultural Devel- ment Project (ADP) and by semi-mechanized opment. Project design followed consultants' farming trials in the Shebelli valley. The projects recommendations, with production impact and supported the Government's policy to achieve benefit projections being based on data provided grain self-sufficiency and to foster settlement of by project staff. nomads. 5. The Bay Region ADP was approved in 2. Rainfed crop production in Somalia is December 1979. Main components were road and generally marginal and hazardous, because of low water supply construction (60 percent of project and erratic rainfall. The Northwest projects were cost), agricultural research and animal health based on existing local water harvesting and services, and establishment of pilot agricultural moisture conservation methods, while the semi- development units (PADUs). Project organization mechanized trials tested Australian/US technology. was complex, with five financiers, five imple- In contrast, for the Bay Project there was no local menting agencies and 20 technical assistance staff or ready-made foreign technology, but project from six sources. designers felt that crop production could be increased rapidly ' y adaptive research. 6. The semi-mechanized rainfed farming operation was financed under the Second North- 3. The first Northwest ADP was approved in west ADP and by the Bay Region project. The May 1976. The main components were bund trial program grew out of, and was initially construction by heavy machinery, domestic and funded from, an earlier unsuccessful settlement livestock water supplies, minor irrigation for project (Drought Rehabilitation Program Project, horticulture, and modest research and extension PAR #6011). New technology was to be intro- support. Bund design dated from British colonial duced and settlers were to become tenants receiv- times in the 1950s and generated relatively low ing services from the scheme rather than labor on rates of return. large state-run farms. - XIV - Implementation Experienceesults 7. Implementation of Northwest I and II was 10. All targets for Northwest I were attained very successful, despite serious security problems except that for minor irrigation. Northwest 1I also at times, mainly because of strong project man- made good progress with additional bunds, water agement and good Bank supervision. However, supplies and minor irrigation, but catchment both project and Bank staff paid inadequate atten- conservation and agricultural inputs components tion until NWII was underway to technical details did not progress beyond the planning stage. The and to production impact. Eventually Northwest modest research and extension programs had little II was brought to a halt by security problems in impact. Benefits from minor irrigation and water July 1988. Construction had progressed well, supplies must have been substantial but were not although technical defects in bunds were not quantified. There was a strong sense of farmers' corrected and some project vehicles had been ownership of project works, including bunds. diverted to military use. However, benefits from bunding are not as sub- stantial as originally assumed. 8. The first three years of the Bay Region project were plagued by the reluctance of road and 11. The Bay Region project provided a basic water agencies to second staff to the PMU, by all weather road network, drinking water supplies unsatisfactory consultants, and by delayed arrival and veterinary services. Vigorous and effective of research staff. Construction of roads and water dryland research held promise of medium term supplies proved more costly and slower than benefits for dryland farming nationally. Grain expected. A mid-term review in 1984 changed production and the area cultivated doubled, but the emphasis from agricultural production to estimating project-induced benefits proved diffi- research, including dropping the PADU concept cult. Project consultants concluded that 35 per- as impractical, and reducing the road program. cent of increased production was due to the The veterinary and water supplies programs project while the government put that figure at 60 received additional funding. Subsequent imple- percent. mentation was satisfactory except that there was no monitoring and evaluation. Delays caused 12. The area cultivated under the semi-mecha- mainly by fuel shortages necessitated a project nized farming operation did not expand materially extension of two years. A Phase II feasibility after 1983. The agricultural research program study was prepared in 1988 during the final (which was needed to solve agronomic problems) project months. never got going, while doubts about project viability, in particular the ability of farmers to pay 9. The semi-mechanized farming trials operating costs, were not resolved. started in January 1981. In the first year produc- tion aspects of the troubled settlement project Sustainab"M were transformed by strong technical assistance into what was viewed as a technology showpiece. 13. Had there not been conflict in Somalia, The operation's reputation then peaked in 1985, what were the chances of sustainability? In the and stagnated. Agronomic problems which had Northwest investments were mostly on-farm with appeared early on were never solved (weed con- quite a long life for the structures, but with trol, low crop yields - except for maize, and crop diminishing effectiveness. The PMU's permanent failures in drought years). In the last four years staff would have reverted to their parent services. the pilot operation was expanded into a new The small research/extension units constituted a project on its own. This project never took-off minimal service to the area with relatively little because a new set of consultants lacked dryland burden to the Government. farming expertise, the Bank was unable to super- vise during the last two years and security deterio- 14. The Bay Region project represents a rated until all operations were brought to a halt in different picture, with substantial but utilitarian December 1990. investments in public works (water supplies and - XV - roads) and in services, including a research insti- 18. The Bay Region project was handicapped tute, which would all need maintenance. Some of by design errors (because data was lacking) and these, notably veterinary and water supply servic- great complexity, but recovered well after the es could, in time, have become self-financing. redesign. Complexity would have been reduced But roads and research sustainability would de- and infrastructure implementation improved by pend mainly on Government finance, since a having two projects, with infrastructure staff in reasonably effective staff who could maintain the Bank and in Somalia managing road and water these services had been put in place. components. No explanation was found for the Bank not planning to support a follow-up project 15. The future of the semi-mechanized farm- as originally intended. ing operation was doubtful because of technical problems, lack of cost recovery and the uncertain 19. The semi-mechanized farming operation availability of expertise to manage the scheme and showed that technology can be transiitted from to resolve remaining problems. one continent to similar ecology in another, but that good research back-up is essential to deal Findings and Lessons with technical problems as they emerge. The operation also showed the dramatic impact that an 16. In the current country circumstances, outstanding consultant can have on a project rating these three projects' overall performance (supporting "sole source" technical assistance seems academic. But at the PCR stage Northwest procurement). While financial viability remains in I was rated satisfactory while the other two pro- doubt, there is enough evidence to hope that jects were rated unsatisfactory (the semi-mecha- skilled research might solve technical problems nized farming operation does not qualify for a and additional training and further hands-on separate rating, as a component). The audit technical assistance could produce effective local confirms the unsatisfactory rating of Northwest II management that could result in a sustainable and finds that the first project must also be re- operation. garded as unsatisfactory, despite physical results, since benefits to the main component (bunding) 20. Project reviews at appraisal were thorough are very limited. The audit would also reverse for the first Northwest project, less penetrating for the rating of the Bay Region project to satisfac- the second and inadequate for the Bay Region tory, despite a poor original design and low eco- project, where design errors should have been nomic rate of return, given the significant results challenged. Supervision was quantitatively ade- discussed in section V. quate for all three projects but for the semi-mech- anized operation there was a five-year supervision 17. The Northwest projects brought consider- gap from 1982-87 and in 1989 and 1990, with no able benefits through water supplies and irriga- record found of formal supervisions. The first tion, but bunding benefits were much less than Northwest project supervisions lacked technical originally expected. The audit believes that with expertise and failed to resolve technical design and minor works many project bunds could become evaluation shortcomings, but these problems were effective, but that some have been constructed in both dealt with during the second project. Bay unsuitable areas. Even so, bunds are generally Region project supervisions varied from censori- popular as farmers believe they improve security ous to collaborative with the latter receiving a of tenure. Outstanding implementation achieve- better response. ments by the PMU may have masked shortcom- ings in technical aspects and in monitoring and 21. Two good tools emerged from the Bay impact evaluation during implementation. Conse- Region project: the mid-term review, now virtu- quently, while there may have been data justifying ally mandatory, and annual donor conferences, bunding at appraisal of the first project, negative which are invaluable to keep cofinanced projects evaluation results by consultants were ignored by coordinated and on track. the Phase II appraisal and, for much of the time, by the Bank's supervision. 22. The main lessons learned are: - Xvi - (a) The audit confirms that making intensity and foreign exchange improvements to long established scarcity issues which are likely to traditional agricultural production affect replicability and sustain- systems, especially in resource ability prospects. The reverse of poor low rainfall areas, must start the conclusion above that the from a detailed understanding of success of the early activities was the existing natural and human due to exceptional "hands-on" resources, of the current produc- consultant input (which could not tion systems (livestock and be reproduced by the second crops), and with the full participa- group of consultants) is whether tion of the affected populations. the technology is too skill- intensive for general adoption in (b) Constant monitoring of progress Somalia. and periodic re-evaluation of objectives and potential impact (d) Some other important points are necessary to ensure that origi- include: that cofinancing should nal concepts are still sound and to be arranged early on in the pro- point out any needed changes. cessing cycle to avoid long delays which almost called for reapprais- (c) On the semi-mechanized farming al in two of the projects; innova- activities, questions remain about tive project designs need especial- whether it is prudent to pursue the ly skilled and perceptive review at resolution of technological prob- appraisal to identify design errors; lems associated with the introduc- and, technology that has yet to be tion of a foreign technology (to adopted by farmers is suitable for produce staple crops in a poor inclusion in a testing program, country) without reference to but should not be used as a basis financial, cost recovery, skill for projecting benefits. PERFORMANCE AUDIT REPORT SOMALIA NORTHWEST REGION AGRICULTURAL DEVELOPMENT PROJECT (Cr. 635-SO) NORTHWEST REGION AGRICULTURAL DEVELOPMENT PHASE II PROJECT (Cr. 1538-SO) BAY REGION AGRICULTURAL DEVELOPMENT PROJECT (Cr. 972-SO) 1. INTRODUCTION 1.1 This is the audit of the Northwest Agricultural Development Project, its follow-up project Northwest Agricultural Development Project Phase II and the Bay Region Agricultural Development Project. These projects represent the Bank's assistance to Somalia to improve rainfed farming in the few areas of that country where such farming is possible. The last two projects also provided bridging support for the continuation of semi-mechanized rainfed farming trials, until a separate project for that purpose could be developed. These trials had been initiated under the earlier Drought Rehabilitation Program Project (Credit 623-SO). In addition to making a contribution to reducing rural poverty, these operations supported the government's strategy for food self-sufficiency and settlement of nomads. 1.2 Most of Somalia is a semidesert, receiving less than 300 mm of rain per year, in two seasons. About 60 percent of the 6 million plus population are nomads. Livestock production by nomadic pastoralism is the predominant occupation in this harsh environment and livestock and livestock products were 80 percent of export earnings. Another major source of foreign exchange has been remittances from Somalis working abroad. The opportunity for employment in the Gulf was a formidable rival to the projects under audit, which suffered from skilled staff shortages throughout their implementation periods. 1.3 Only just over one percent of Somalia, about 700,000 ha, is cultivated. Of this, about 300,000 ha is irrigated or seasonally flooded land adjoining the Shebelli and Juba Rivers. The remaining 400,000 ha of cultivation is mostly in two patches which have micro-climates with slightly higher rainfall than elsewhere: about 300,000 ha in the Bay Region (north of Mogadishu) and just under 100,000 ha in a compact belt in the North Western region, which constitutes the project area there (Map IBRD 25134). Much more land is potentially cultivatable, but the main constraint is rainfall or irrigation water. The average annual rainfall in the Bay Region is 400-600 mm in two distinct seasons1/ and in the Northwest is 400-500 mm, also in two seasons, but with a shorter dry interval than in the southern areas.2/ Although this rainfall allows crop production, the moisture available to crops is very marginal and variations in rainfall amounts and timing make traditional cropping very risky. Crop yields, without irrigation, water conservation or harvesting, are low and each season is a gamble as to whether crops will mature or not. In these difficult conditions Somali farmers have developed over thousands of years their own technologies to make quite good use of the local ecological potential. For example, to survive frequent crop failures use is made of ancient and highly effective methods of storing grain in pits which 1/ The Gu rains and in April-June and the Der rains in October-December. 2/ The Gu rains in April-June and the Karan rains in July-September. -2- preserve sorghum in edible condition for up to ten years. The famine during the recent conflict was probably due not so much to lack of cultivation or crop failure, but to seizure of farmers' stored grain. 1.4 Experience elsewhere suggests that introducing lasting improvements to such long- established traditional agricultural systems is a risky undertaking which must start with thorough knowledge of the present situation. The audit returns to this point below. Objectives 1.5 The audit applauds the Bank's agriculture sector development strategy for Somalia in the 1970s and 1980s. In the face of the severe conditions described above and the resultant difficulty of identifying viable investment opportunities, the Bank tried to ensure that all important aspects of Somali agriculture received attention by filling the gaps left by assistance from other donors. Four IDA projects,J/ complemented by the Northern Rangeland Project (supported by the Kuwait Fund), dealt with the all important livestock industry, while two projects supported the improvement and expansion of irrigation.l/ The Drought Rehabilitation Program Project had responded to an emergency by helping to settle destitute nomads. Two extension/research projects I/ and the two Northwest Region Projects had primary focus on arable farming, while the Bay Region Project dealt with both crops and livestock. In all cases the projects' main beneficiaries were the poorest rural inhabitants. In the Northwest the government was especially interested in investing in rainfed farming to promote economic development, and thereby assuage the demand for autonomy. 1.6 The main objective of the first Northwest ADP was to increase agricultural production through soil and water conservation. For the Bay Region project the main objective was to increase crop and livestock production by better technology, increased use of inputs and development of infrastructure. Both projects shared secondary objectives of assisting settlement of nomads and preparing the basis for further investment. The aim of the Northwest Phase II project remained improved agricultural production, but with more emphasis on institution-building. The semi-mechanized rainfed farming trials (supported by these projects) were primarily to develop viable production systems for the two areas already settled under the Drought Rehabilitation Program Project. A secondary, more expansive hope was that the soil moisture conservation technology might have wider application in the rainfed areas. 1.7 There are similarities between the Bay and Northwest regions. Both are inhabited by people who are first and foremost pastoralists. Both sets of people have modified their nomadic customs to some extent, to establish villages and develop arable farming, primarily sorghum production. In both cases, stover for stock feed is as important to cultivators as grain. In both areas, lack of water supplies and roads are severe handicaps to farming. Rainfall is marginal for crop production in both areas, but otherwise there are differences between the two areas and their farming systems. In the Northwest, the gap between the two rainy seasons is only one month. By using photoperiod sensitive sorghum that does not flower until September, farmers can make use of the rainfall of both seasons to grow one sorghum 3/ Trans Juba Livestock Project (Credit 462-SO) Central Rangelands Development Project (Credit 906-SO) Central Rangeland Research and Development Project (Credit 1975-SO) Livestock Health Services Project (Credit 1647-SO). 4/ Farashane Irrigation Project (Credit 2063-SO) Bardera Technical Assistance Project (Credit 1865-SO) I/ Agricultural Extension and Farm Management Training Project (Credit 905-SO) Second Agricultural Extension Project (Credit 1794-SO) -3- crop. This usually improves yields and reduces the danger of crop failure. However, when the dry gap is excessive, as happened in 1987, the crop is damaged and yields are reduced. In the Bay Region there is usually a three months' gap between the two cropping seasons ("gu" and "der") and farmers attempt to grow two sorghum crops per year, both of which are more risky than the long-season single crop in the Northwest. Farmers have the option to resow sorghum during the der season or to ratoon the first sorghum crop if it has grown well during the gu season. Neither option is ideal. The der sown crop failure rate is high, since this is the less reliable of the two rainy seasons. Ratooned sorghum suffers from more serious pest and disease attack than the sown crop, and these pests and diseases are then carried over to damage the next gu crop. Other similarities and differences between the two regions are described in Box 1. Bar I - COMPARATIVE TECHNICAL ASPECTS OF FARMING IN THE NORTHWEST AND BAY REGIONS Soils, land preparation and sowing are different in the two regions. In the Northwest a permeable clay loam and a sandy clay loam are the main soil types used for arable farming. Farmers broadcast seed into the stubble of the previous crop and plough it in. Formerly ploughing was by oxen, but by the time the NW I project started about 80 percent of ploughing was by tractor. In the Bay Region heavy clay soils (vertisols), which are very difficult to work during the rains, are used. Farmers do not plough, but dig holes and sow several seeds in each hole. Usually the ground is bare and weed free at sowing time, having been grazed by livestock. Weeds are the main worry in both areas, but while in the Northwest oxen are commonly used for weed control, supplemented by hand weeding, in the Bay Region all weeding is by hand. A long standing effort to introduce ox cultivation into the Bay Region had not made any impact on farmers, and there was little demand for the services of the local branch of the Government tractor hire service, other than to "break" new land or to plough fields in which perennial weeds were difficult to control. In both areas, rainfall is marginal for crop production. However, the Northwest soils, being reasonably permeable, respond to water harvesting and to retaining water from rainfall on the field with physical barriers such as bunds. On the much less pervious clays of the Bay Region, such methods could damage crops by waterlogging. In both areas, either part or the whole family leave the village when water supplies dry up, and go in search of water and grazing for the livestock. Consequently, particularly in the Bay Region, labor is in short supply at the beginning of the rains, during the crop sowing season. Both regions host a large migrant livestock population in addition to the villagers' animals. In the Northwest the resultant stock overpopulation has denuded rocky slopes causing high runoff which erodes cultivated fields. In the Bay Region, the high stock population has resulted in thorn shrub thickets, but the much gentler slopes and pervious soils of the rangeland generates less runoff and relatively little erosion. There was one major difference between the two Regions of particular relevance to project design. In the Northwest the project aimed to expand a proven technology (or at least the appraisal mission and quite a lot of others believed it to be proven) that farmers liked, while in the Bay Region there was no improved technology. Whatever had been tried, for instance a multi-year effort to introduce animal traction, had had no impact on farmers' practices. 1.8 The next chapter outlines the two Northwest Agricultural Development projects, from design through implementation, while Chapter 3 deals with the Bay Region Agricultural Development Project. For the two Northwest projects the respective project completion reports give an accurate description of their features but have not analyzed conflicting reports on the benefits from the soil and water conservation works. Such analysis will be attempted in this audit. For the Bay Region project, which had several major components, the audit concentrates on technology aspects of the agricultural research and cropping development components. Chapter 4 traces the rapid rise of the semi-mechanized operation from the ashes of the Drought Rehabilitation Program Project to a possible Green Revolution for much of drought-prone Africa, to the eventual debate over its technical viability and financial sustainability. Chapter 5 discusses project outcomes, Chapter 6 is on the performance of the Bank, and the report ends with Chapter 7 on conclusions, lessons learned and suggestions for future Bank activities. 2. NORTHWEST REGION PROJECTS A. Northwest Agricultural Development Project Design/Preparation 2.1 Project designers often cannot find locally proved technology for semi-arid areas. But in this case there were proven technologies already in use, mainly bunding for soil and water conservation. The main benefits of the project were to come from bunds to retain and "harvest" rainfall. Box 2 discusses the effect of rainfall on crop yields in the project area. 2.2 Box 2 emphasizes that most years water harvesting and conservation could have a major impact on yields and, moreover, would be a prerequisite to the successful use of purchased agricultural inputs. 2.3 Soil and water conservation had been of major concern in the area since colonial times. The project area has a large livestock population. Residents are stock farmers first and cultivators second. In addition, nomadic stock move through the area. As a result, much of the range was denuded, resulting in high runoff, serious erosion on sloping arable land and gullying of livestock tracks. Agricultural Department staff introduced a bunding system, after trying alternative designs. The design selected for the project was the only bund that stood up to trampling by large herds that crossed fallow fields. During the 1950s bunds were constructed on 15,000 ha, using Department-owned tractors and ploughs to loosen the soil and oxen and scraper boards for shaping. Each team of department-owned oxen was matched by a farmers' team and no construction charges were levied.& With USAID assistance during 1963/64, bunds were constructed with bulldozers on a further 5,000 ha. 6/ Somali Land Protectorate, Department of Agriculture and Veterinary Services Annual Report - 1955. -5- Bar 2 - EFFECT OF RAINFALL ON CROP YIELDS IN THE NORTHWEST PROJECT AREA Annual rainfall for the three district headquarters in the project area is as follows: Years Rainfall mm Station Recorded Maximum Minimum Average Hargeisa 50 810 156 425 Gebile 16 548 331 445 Boorama 30 676 366 517 The project area stretches along the Ethiopian border from the regional capital Hargeisa to Boorama, about 120 km northwest with Gebile roughly in the center (Map: IBRD 25134). The preparation report for the first Northwest Project and the consultants' feasibility study for the second project both discuss the relationship between rainfall and crop yicld potential. The preparation report projected the following 100 year rainfall frequencies, using 37 years' rainfall data for Hargeisa and 3 years' for Boorama: Rainfall Frequency in 100 Years (Years) mm Hargelsa Boorama 151-250 5 - 251-350 30 12.5 351-450 30 31 451-550 25 22 551-650 7 22 More than 651 3 12.5 As the above figures show, rainfall increases and variability decreases towards the northwest. However, rain frequently comes in heavy storms and the total is not as important as the distribution. Thus, even a year with more than average annual rain may be a "drought" year for crop production, as happened in 1987, the year of the sorghum yield study. The water available to the crop was reduced because much of the rain from heavy storms was lost as run-off, except where it was kept on the fields by some means such as bunding. The feasibility study reports that even 300 mm of rainfall during the cropping season would not allow sorghum to develop without moisture stress (500 mm being the generally accepted amount required). Based on past data and discussions with farmers, the feasibility study and the 1987 Sorghum Yield Study estimated likely yields over a 10-year period as follows: Feasibility Study Sorthum Yield Study Years YIeld Years Yield In 10 Kg/Ha In 10 Kg/Ha 1 Zero 2 Zero 2 200 2 200 5 500 5 500 2 1,000 1 1,000 These projections imply "average yields" of 490 and 390 kg respectively. -6- 2.4 At identification and preparation, Bank and FAO/CP staff saw that the bunds constructed 10-15 years before were still intact, though in need of maintenance, and that they had prevented serious soil erosion. From discussions with local farmers and officials, the missions were convinced also that the bunds increased yields. The preparation report quoted results from two experimental sites7/ showing average yield increases of 108 percent for grain and 74 percent for stoverl/. The report also noted good results from water harvesting: "Various observations recorded over the last 20 years enable sorghum yields of both grain and stover to increase by: - 35 percent with only the rainfall on the interbund area retained; - 70 percent with additional catchment area equal to the size of the interbund area; - 100 percent with a catchment area twice the size of the interbund area."2/ No source for this data was quoted, and these conclusions stood unchallenged until 1983 when the feasibility study for NWII discussed the marked effect of different land slopes on bunding impact. This subject had not been raised at the preparation/appraisal stage of the first project. 2.5 Space is devoted to the design of the bunding component because doubts about the benefits of bunding are the main focus of the audit. There is no such debate about the other components, which were: (a) irrigated small gardens on fertile river terraces; (b) underground domestic water cisterns and ponds for livestock, using local designs; and (c) support for management, research, extension, technical assistance and training programs. Technical assistance included feasibility studies for a follow- up project and for "an integrated rural development project in the Bay Region." 2.6 The preparation report raised some concerns in the Bank, notably: (a) the low projected economic rate of return (ERR) of 8 percent; (b) project management capability; (c) whether bund construction was the best soil and water conservation option; and (d) issues relating to cost recovery, bund maintenance and farmer participation. The summary of the preparation report places these concerns in the context of the relatively poor resource endowment of Somalia: "The Northwest Region Agricultural Development Project contains many uncertain elements. However, considering the limited possibilities for 7/ Aburin Research Station - 1957 and 1958, Gebile - 1956 through 1960. 8/ Stalks or straw, used for fodder. Some sources suggest that farmers value their crops as much for stover as for grain, with stover being available to annuals even in low rainfall years when crops produce little grain. 2/ A similar increase of 100 percent from 2:1 catchment to crop area ratio was reported in the 1987 Sorghum Yield Study, from water harvesting trials at the local research station. -7- agricultural development in the country, it will probably be worth facing the risks and uncertainties of the project." 2.7 The appraisal mission terms of reference (TOR) stressed the need for paying particular attention to whether bunds were the most efficient approach to soil and water conservation. The visit to Somalia convinced the mission (an experienced soil conservation specialist was a member) of the applicability of bunding using the 1950s specifications. The mission was particularly alarmed by the "advanced stage of serious soil erosion" it saw and recommended adding a pilot gully control component. 2.8 An excerpt from the Staff Appraisal Report (SAR) in Box 3 discusses bunding alternatives. 2.9 As described in the PCR, the project was appraised by the Bank in May 1974, but was not presented to the Board until June 1976. There seemed to be great urgency to appraise immediately after preparation, but there was a long gap between field appraisal and Board presentation. This was evidently due to an unsuccessful search for cofinancing to meet a US$ 4.0 million financing gap. Eventually, the amount was found from IDA funds. 2.10 Returning to the appraisal process, a draft SAR was issued in December 1974 which resulted in Central Projects Staff (CPS) questioning in two memos whether the project should go forward on the grounds of the high cost per beneficiary--$5,640 per family, the resultant difficulty that farmers would have with cost recovery, and doubts about technical aspects. CPS suggested scrapping bunding in favor of technical assistance to develop alternative soil and water conservation methods. No response from the Region was found on file, but in November 1975 another CPS memo proposed replacing bulldozer technology with 1950s technology for bund construction (i.e., small tractor/plough units and oxen-pulled scraper boards). This sparked off quite a debate just as the project seemed to be moving towards negotiations. The Region pointed out: "that the proposed method would be slow, difficult to organize and supervise and that in any case, the ox population had declined since the 1950s, so that there would not be enough oxen available for a large-scale bunding exercise. In addition, if it were decided to change the technology, reappraisal would be needed, causing further serious delay to this already much delayed project." These arguments won the day. 2.11 From the above account it is apparent that the project did not have an easy ride through to approval, with soil and water conservation technology being a major issue. -8- Box 3 - BUNDING IN NORTHWEST SOMALIA *Soil and climatic conditions normally allow seasonal dry-land cultivation of food crops, and this now accounts for about 80 percent of the cropped area. Sorghum is by far the most important crop, although some maize and pulses are also grown. In years of low or poorly-distributed rainfall, crops often wither before reaching maturity except in the limited areas where field bunds have been constructed. These bunds utilize rainfall to the maximum extent and control erosion by limiting run-off. They consist of crescent shaped earth banks averaging about 120 m long and 75-85 cm high, I to 2 bunds per ha.' About 20,000 ha were bunded between 1954 and 1964, all of which are located in the Project area. About 90 percent of the bunds have been maintained in excellent condition. It is reported that no instance of complete crop failure has yet occurred on bunded land. At present bund construction is the only method used for controlling soil erosion. It has the added advantage of increasing moisture retention behind the bunds which facilitated crop production in areas which are devoid of water during most of the year. There is a significant difference in yields between bunded and non-bunded land. There is no technically viable alternative to bund construction which could provide a secure basis for crop production with a reasonable yield. Annual rainfed farm income is low, averaging about So.Sh. 640 (US$100), not including subsistence consumption valued at about So.Sh. 800 (US$128). Holdings of arable land average about 10 ha, and one third of this is traditionally fallow. Sorghum yields in the non-bunded area average about 3 quintals (of 50 kg) of grain per ha, and in the bunded areas about 6 quintals per ha. These low yields are attributable to limited rainfall, inadequate seed-bed preparation and weeding, broadcast sowing and poor seed. "b' The following are additional bund design features. The bunds are aligned along the contour but have .wings" at both ends that are turned uphill for 30 cm of the grade. These help to impound the run-off water to cover much of the land above the bund. But the impact of bunds on crop yields is highest when run-off from outside the bunded field can be directed onto it, either from land above the field or by means of small diversion ditches from the sides. This is "water harvesting". In these cases, and in any case during very heavy storms, surplus water flows round the ends of the wings of the bunds and a shallow drain directs the overflow to the land below the bund. The process is repeated, by each bund, down to the lowest edge of the field. b/ Appraisal of the Northwest Region Agricultural Development Project - Somalia, Report No. 912-SO, May 7, 1976. Implementation 2.12 Considering that the project area is about 1,600 km from Mogadishu, communications were poor, staff were reluctant to live and work in such a remote area, and that there were occasional security problems, it is quite an achievement that the project was implemented reasonably punctually and efficiently, with most appraisal targets being met. Any shortcomings were not the fault of project management and staff. The credit for this notable performance belongs largely to the outstanding project manager, who directed the project for seven years up to the appraisal of NWII. The record shows how one difficulty after another was overcome and how a bunding organization of almost 200 staff was built up. The staff were kept together and were reasonably well supplied with fuel and spares, whatever was happening in the area, including air attacks during one period. 2.13 Although construction started about six months later than shown in the SAR implementation schedule, the June 1979 supervision mission complimented the project staff on the quantity and quality of construction. Lack of aerial photographs (and the maps prepared from them), which had been thought to be essential for planning the works, proved no constraint. An experienced Somali soil conservation specialist had prepared a two-year work program. 2.14 Recruiting and retaining skilled staff in the face of much more profitable job opportunities abroad remained a problem throughout the project period. Shortages of fuel and lubricants was another difficulty that recurred frequently. Construction was helped by their being strong demand for bunding from farmers. Bunding was considered so important by some farmers that rather than wait their turn under the project, they paid in advance for the government tractor hire unit to do the work. 2.15 Progress of the other project components was mixed. The domestic water supply component was popular and made as good progress as periodical materials shortages (especially of cement) would permit. The pond construction program (for livestock), which was also carried out by the units constructing bunds, started slowly due to much debate among farmers over siting. Almost up to the time of the Closing Date, it seemed that only about half of the projected ponds would be constructed, but then there was increased activity and the target was achieved. The building program was much delayed by materials shortages and the lack of contractors in the Northwest. The contracts were ultimately completed but the quality of work under one of the two contracts was so poor that some of the buildings had to be demolished in Phase II. 2.16 The other soil conservation component was pilot gully plugging. It seems that the soil conservation specialist did not have much experience of this work and early attempts failed. Bank staff advised cancellation of further efforts, partly on the grounds of the lack of relevant skills of the project staff, but mainly because gully plugging would not address the main cause of serious erosion, namely heavy livestock concentrations denuding the range and creating track. The soundness of the decision to cease gully plugging is debatable. The later feasibility study for Phase II makes some practical proposals. 2.17 Consultants' services supporting the construction investments were an important part of the project. Performance was mixed and changes were made as described in Box 4. 2.18 Irrigation development was delayed by failure of the government's Water Development Agency to investigate groundwater availability. A contractor was selected instead but then to compound delays, the selected firm went into bankruptcy. The groundwater survey was eventually completed but there was time for only 20 of the projected 50 irrigated one-ha gardens to be developed before closing. The firm that eventually did the exploratory drilling did an excellent job. Similarly the consultants supporting horticultural and irrigation development (who were part of the group who prepared the Phase II Feasibility) also performed well. Of the two feasibility studies, the preparation of the Bay Region project was handed over to the FAO/CP, while a new firm was selected to prepare the NWII feasibility study which was completed in June 1983. 2.19 Lack of supporting technology to maximize benefits from bunding had been one of the criticisms of the project at appraisal. Project preparation staff had also been aware of this technology lack and had designed research and extension components, but neither component made an impact on farming methods during the two phases of the project (Box 5). - 10 - Bar 4 - TECHNICAL ASSISTANCE MANAGEMENT A firm to provide five long-term consultants was selected after international competition (their appointment was one of the effectiveness conditions). The first staff, especially the team leader, did not get on well with the PM. The first conflict was over location of the team. The PM wanted to split them up, proposing that they live in three widely scattered locations. This sparked a major debate (the Bank fortunately staying neutral, though sympathizing with the consultants--as does this audit), which ultimately brought the company's chairman out to Somalia. The PM won that contest and with some face saving changes to contract clauses, the chairman conceded all points. Next, the PM summarily sacked the team leader (for leaving the project area without permission). Then the Ogaden War in early 1978 led to the whole team leaving. (When Hargeisa was bombed, all expatriates were withdrawn. The PM agreed that the team could wait out the disturbance for four weeks in Mogadishu, but at the end of that time, he ordered them to return to the project area--although the security situation remained unsettled.' Consequently the firm invoked "force majeure* and cancelled the contract). In supervision records and the PCR the early withdrawal of the consultants is presented as a setback causing project delays, but the audit believes that it did not do much harm to the project. The PM then visited Pakistan and selected consultants for direct employment. On the whole they performed satisfactorily, especially the financial controller,' chief mechanic and horticulturist. However, the soil conservation specialist may not have had a firm grip on the technical aspects of bunding) or gully control but did well in staff training and getting the operation under way. Various agriculturists came and went and varied from ineffective to quite dynamic. Monitoring and evaluation remained weak and the PM may not have thought this subject a priority among his many concerns. It appears that the PM was more comfortable with directly recruited consultants than with commercial companies. The PM's relations with another firm, which carried out the Phase II feasibility studies and provided TA for irrigation, also became quite strained at one time). The British Embassy had advised UK citizens to stay away from the NW Region for the time being. Replaced after his departure by staff of the Bank's Agricultural Development Services. 2.20 The DCA required the Borrower to carry out studies that would facilitate the recovery of bunding costs from farmers, and by August 31, 1980, or a later date acceptable to IDA, submit proposals for cost recovery and then implement agreed measures to that effect. The original deadline was not maintained, but the PMU submitted proposals in July 1983. Projections were based on an incremental yield due to bunding of 0.24 tons sorghum/ha, which was the lowest of the various PMU observations of yield increases throughout the project area over three years. The proposals to levy So Shs 4.50 per linear meter constructed were approved by the government and IDA by the end of March 1984. The PMU then forwarded lists of farmers with bunded farms, and length of bund constructed in each case, to the Ministry of Finance to initiate collection. However, 1984 saw one of the worst droughts for decades and the government postponed collection of fees. As far as is known, no fees for Phase I bunding were collected. - 11 - Box 5 - RESEARCH AND EXTENSION The organization and management by the PMU of the research and extension activities was criticized from time to time both by supervision teams and others who felt that the two national services for research and extension should take over these services. But the distance from national headquarters would have precluded satisfactory integration with the national services, which had very little presence north of Mogadishu. There should be liaison of course, with headquarters and exchange staff visits, but as far as organization and management of agricultural services is concerned, the PMU remained the most satisfactory vehicle. Aburin Research Station, which was also to offer residential training to staff and farmers, was for most of the project period considered unsuitable for this purpose, due to lack of a domestic water supply. Development of another site was recommended by supervision missions, but a suitable site was not identified. (Lack of domestic water continued to be a serious handicap through both project phases, and it is surprising that neither the PMU nor supervision missions came up with a solution to this problem.) The Aburin station and research work on it were variously reported as badly neglected or well looked after and giving interesting results, usually depending on the presence or absence of a consultant agronomist in the PMU, and the interest displayed by him in research. However, it was always unlikely that this one-person research effort would identify "breakthrough technology" for the relatively harsh environment. Research program- were mostly relevant, but perhaps more attention should have been given to why farmers reverted from land preparation before the rains, to broadcasting the sorghum and maize seed into the previous crop's stubble, and then plowing the seed in. This practice was condemned as regressive and poor farming by all agriculturists connected with the project, including the consultants who prepared Phase II. The reasons for the universal preference for this type of husbandry by farmers, compared to what all the critics felt to be superior technology, was not established. It could have been that lack of time or funds or both made additional operations difficult, or farmers found them to be unprofitable. Extension services, particularly to support rainfed agriculture, did not develop as projected. Instead of building up to 23 field extension workers, the project never managed to attract more than nine such staff at one time, and these had very little transport. Reportedly, they did their best through demonstrations (in particular of spacing, weeding, and alternative crops), but these generated no change in farmers' cultivation methods. With hindsight, and had it been possible to provide adequate mobility, the extension staff could have served a more useful role by farm-to-farm visits, to check the adequacy of water retention and distribution in bunded fields. In contrast, the three persons horticultural advisory unit effective farm visits, possibly because horticulture was a new topic compared with trying to change long-established practices for cereal cropping. The Geed Deeble horticultural farm, which was to form part of the horticultural development effort, proved quite unsuitable for research and nursery purposes, again due to inadequate water supply. When efforts to improve that supply failed, it was decided to find a better site for the station, with costs to be funded by the Phase II project. Nevertheless, the station served as a demonstration orchard and supplier of small quantities of fruit trees. 2.21 Finally there was support for the PMU. Due to the leadership of the PM over seven years, this must be classed as an outstanding success. It was unfortunate that towards the end of his stewardship he was politically harassed and even jailed briefly. His replacement over the last 15 months of NWI and part of NWII implementation, also proved to be a capable administrator. He had familiarized himself with the work over several months as Deputy PM. With regard to training, the Unit initiated a large and successful local training program for the staff of the earth-moving units. Short-term job-related training abroad was also worthwhile, but of the eight candidates who visited the USA for longer-term fellowships, in the absence of bonding arrangements only three returned to Somalia. - 12 - 2.22 Bank supervision was intensive, and there were usually two supervisions per year spending 10-14 days in the field. There was also good continuity with two task managers from RMEA, Nairobi during the seven year implementation period. The second of these task managers visited the project ten times over a 4-1/2 year period. During the earlier supervisions an engineer and a soil conservation (construction) specialist had participated. If there is one criticism of the supervision effort, it is that similar expertise was not deployed once construction got well under way, when it would have been useful to check on how closely construction followed design and whether modifications were advisable. The missions made no project design changes, other than stopping the gully repair component. Suggestions for alternate bund construction methods, the need for overall catchment protection and various suggestions for improving cropping technology, which appeared in most supervision reports, were taken into account in the Phase II design. B. Northwest Agricultural Development Phase II Project Design/Preparation 2.23 The inclusion of funds in the Northwest I for a feasibility study for Phase II shows IDA commitment to longer-term assistance. The consultant circulated the inventory stage of the study (16 technical reports) and proposals for developing rainfed agriculture in the second half of 1982. Exploratory drilling delays held up completion of small-scale irrigation proposals until November 1983. Staff from the Bank's Nairobi office,IQ/ who were responsible for preparation, rated the feasibility study highly, particularly the detailed information on the region's resources and infrastructure. They generally accepted the project proposals except for some details, such as sorghum yields increasing from 870 kg/ha to 2,500 kg/ha, which were unverified by research or farmers' practices. 2.24 The NWII feasibility study's most interesting analysis was an assessment of the impact of bunding by comparing sorghum yields from bunded and unbunded fields, which, together with PMU results, is presented in Box 6. Bunds resulted in sorghum yield increases only in the so-called "lowlands" (see Glossary) where water harvesting is practiced. "Lowlands" occupy only 17 percent of the project area. 10/ Resident Mission in Eastern Africa (RMEA). - 13 - Box 6 - BUNDING IMPACT ON SORGHUM YIELDS "Lowland" (some slopes > 3 Slopes (2-3 percent Plateau (< 2 percent slope) percent) slope) Yield Tons/ba Bunded Nonbunded Bunded Nonburded Bunded Non-bunded Maximum 3.23 2.40 1.50 1.40 1.36 1.34 Minimum 0.48 0.51 0.92 1.04 0.34 0.51 Average 1.75 1.22 1 1.23 1.22 0.82 0.84 Index (cf. non- 208 145 146 145 97 100 bunded plateau) Index (cf. 100 64 70 70 39 40 bunded lowland) Source: Phase 11 teasibility study. Thus a yield increase of up a half was found for water harvesting by bunds in the "lowlands"", with no yield change for the slope or plateau areas. The "lowland" area occupied 17 percent of the project area. This result contrasts markedly with information provided by the Project Management Unit and reported by Bank supervision missions, which grew steadily more and more optimistic about project impact. This is illustrated by the following extract from the September/October 1983 supervision mission report on "Operating Results and Forecasts": "Incremental production for the Region for 1982 is estimated at about 6,350 tons of grain attributable to the project. This is about 159 percent of the appraisal target of 4,000 tons. This incremental increase is partially attributable to soil and moisture conservation efforts and partially to improved agronomic practices advocated by project extension workers. During the three crop years 1980-82, mean sorghum yields have risen from 13.4 qqlha to about 22 qq/ha, on unbunded/bunded farms - an increase of 66 percent. There are however year-to-year variations. Incremental yields on bunded areas rose by an average of 55 percent in 1980, 78 percent in 1981 and 66 percent in 1982. The 1983 season looks better than the previous 3 years but the crop has yet to be harvested. Over 30,000 family members are now benefitting from the project." Supervision missions had reported regularly on bunding impact, starting in early 1980 for the 1979 crop. Later in 1980, an economist joined the supervision missions to study cost recovery based on net incremental benefits from bunding. The March 1981 supervision mission suggested that "the high yields shown by project staff is probably due to faulty sampling methodology, through using 5m x 5m blocks close to the bunds rather than strips running at right angles to the bund." However, yield assessments in 1981, based on 100 randomly selected farms, appear to have been accepted with confidence. The question of benefits from bunding is discussed further in a working paper available from OED. &/ See definition in Glossary. 2.25 The TOR for the pre-appraisal mission in September 1983 stressed verification of achievements to date, with special reference to the impact of bunding on crop yields. The resultant revised Project Brief recommended "more of the same" as in the first project, with the addition of the import and distribution of production inputs and a pilot catchment protection scheme. Main risks were perceived as loss of management staff and cost recovery problems. There were some concerns at RMEA management level about bunding cost and benefits, but these were allayed by the preparation staff. - 14 - 2.26 Appraisal followed in March 1984. The mission included a headquarters agriculturist (part-time) to review agricultural aspects, including projected yield levels. The agriculturist expressed dismay at the way top-soil was stripped from part of the field to be used for constructing the bunds, drew attention to the importance of the water retention wings and water spreading ditches of the bunds, and the fact that despite feasibility study recommendations, bunding was still taking place on slopes of less than two percent. He cautioned against overoptimistic yield projections, especially assumptions based on unproven technology, and was concerned that the PMU's consultant was unable to produce "Bunded vs Unbunded" production figures for 1983. However, no doubts were recorded about the fundamental soundness of the first project, and by implication of the Phase II design, including satisfactory benefits from bunding. In retrospect, it is apparent that the appraisal mission was strongly influenced by the excellent record of the ongoing project, as elaborated further in Box 7. Box 7 - USE OF FEASIBILITY STUDY AND PMU DATA DURING PHASE II APPRAISAL While feasibility study recommendations were used to define the work to be done, there are no references in the SAR to comments in the feasibility study that some bunds constructed under the first project might not benefit the farmer, or that farmers like bunding because they believe that it confers security of tenure. The following SAR text is relevant: "The bunds constructed would be identical to those successfully constructed under the first project. ---------- . In addition, terracing, contour ridging, on-farm shelter windbreaks and agreed forestry activities would be introduced, following practices described in the soil and water conservation annex of the consultants' preparation report (Annex 6, Report No. 31-0297 R14, Volume 2). --------------. Actual on-farm interventions would depend on the degree of slope, rainfall regime, local experience and farmers' choice." The consultants had recommended contour hedging on slopes less than two percent, bunding on slopes above two percent, and terracing (a modified form of bunding) on "lowlands", whatever the slope, translating to about 20,000 ha of hedging and 13,000 ha of bunding and terracing. Farmers' choice represented a loophole to circumvent any of the above specifications. On the whole, farmers were keen on and demanded bunding and it is doubtful (and has not been documented) whether the alternative of contour hedging was acceptable to them. While accepting the feasibility study recommendations on how much bunding should be done and where, the SAR used PMU data to assess benefits: "Yields are based on sample field measurements of the unbunded and bunded lands in the three main areas of Hargeisa, Gebiley and Boorama over the four years 1980 through 1983. There is significant variation, not only between but within these areas. However, the average annual percentage increase from the measurements made within the three areas has been in the range of 55 to 78% between unbunded and bunded lands. No additional benefits, for either phase one or two project farms from investments in research and extension have been assumed." This paragraph ignores the field survey of the consultants, which had found in the 1981 season no yield increase on the plateau or on slopes, but considerable increase on low-lying fields that receive runoff from surrounding slopes (Box 6). The impact of bunding on yields was discussed by members of the appraisal team, and the PMU's survey methods were crude, but no one doubted that there were benefits from bunding and that the activities of the first project should be replicated. - 15 - 2.27 Although there was some debate on the pilot watershed protection component, the well- written SAR otherwise attracted few substantive comments.J1/ Late in the appraisal stage SDR 800,000 for the semi-mechanized pilot operations was added to support it until a separate project could be approved. This component, which was also supported by further bridging finance from the Bay Region Agricultural Development Project (Cr. 972-SO), is evaluated separately in this report. The second project had a much easier post-appraisal passage than its predecessor, with Board presentation being moved forward from February to January, 1985. The main risks perceived by the SAR were the quality of project management, difficulties of retaining staff, and security. Implementation 2.28 During the post-appraisal to effectiveness period, work on all aspects of the ongoing project continued, and good progress was made on the watershed protection part of the second project. A US$0.5 million Project Preparation Facility financed contracts for aerial photography and mapping, and for a detailed resources survey and design of the pilot catchment protection scheme. The handover from the outgoing PM to his equally capable successor proceeded smoothly. (The new PM's critique of the draft SAR, for instance, was impressive.) After a short delay the the project was declared effective on July 10, 1985, ten days after the first project's closure. (Such a smooth transition to a further phase of financing, without interrupting operations, is relatively unusual for this type of project.) 2.29 The work program continued to make good progress and in addition to bund and water points construction, mapping and aerial photography of the pilot watershed protection scheme were completed. The design consultants for that scheme were mobilized, a firm of consultants to replace the Phase I TA teams was selected and vehicle and equipment procurement made good progress. One new and important activity started: the collection of fees for bunding. 2.30 Supervision responsibility remained with RMEA in Nairobi. The first supervision (February 1986) was designed to check project progress and to explore the possibility of using labor- based construction methods - a recurrent theme that almost forced reappraisal of Phase I. Since then, labor-based construction had received attention in the Bank and this project seemed a natural candidate for such methods. A suggestion was drafted for trials on bund maintenance, for small-scale construction and for track improvement, but no record was found that these trials took place. 2.31 A strongly staffed mission in March 1986 concentrated on catchment protection and agronomy issues. 12/ This was the first technical focus on bunding by a supervision mission since early construction days, five years before. The mission was satisfied with construction of most of Phase I and Phase II works, except for evidence of inadequate works' supervision, some faulty construction (such as off-contour alignment causing waterlogging) and in particular, absence of, or faulty construction of bund retaining wings and water distribution ditches. (Similar comments had been made by another former agriculture officer, who had helped to develop bunding in British colonial times, when visiting at the PMU's invitation in 1983.) Increasing the number of construction supervisors and special attention to eliminating the faults observed were stressed by the supervision mission. While bunding is the preferred solution on slopes greater than two percent, the mission suggested trials of alternative measures ll/ World Bank, "Somalia, Northwest Region Agricultural Development Project - Phase II," Staff Appraisal Report No. 5292-SO (Washington, D.C., November 26, 1984). 12/ The mission included a consultant who was a retired Bank staff member who had helped to develop bunding in NW Somalia 30 years earlier. The Division Chief from HQ and the RMEA Section Chief joined the mission. - 16 - on lesser slopes. This was to form part of the agreed adaptive research program, which was also to try several cultural practices and variety screening to increase sorghum and maize yields. 2.32 Three other points concerned the mission: the lack of any attempt by farmers to maintain the bunds; the practice of multi-year fallowing of some bunded fields, and the deterioration of agricultural practices from pre-rain plowing and post-rain sowing, to broadcasting the seed into stubble and plowing in. Unaccountably, the mission did not investigate these practices, but recommended urgent studias of them. The records do not show any PMU follow-up to those recommendations. 2.33 One successful new feature of Phase II, recorded by this supervision mission, was the collection of a down payment for bunding (25 percent of the fee, which was to cover half of total bunding cost). This had reduced demand for bunding at first, but by the time of the supervision mission down payments were coming in almost faster than the PMU could process them. However, no mention was made of resuming collection of fees for earlier work under NWI, which had been in abeyance due to the severe 1984 drought. 2.34 The mission agreed with the PMU on measures to strengthen monitoring and evaluation, for which a small unit had been set up, and to introduce a new accounting system to cost operations as a part of determining fee levels. The mission graded the project "l" overall (i.e., problem free or minor problems), but reported that "--the 'Development Impact' is rated as "2" (moderate problems) because of the lack of a hitherto sound statistical base to measure increase in dryland farming grain production, even though other available information shows significant increases in this area." The report also stated, "Steps are being taken in 1986 to develop a sound statistical basis to measure benefits in grain production." 2.35 There were two more supervision missions over the next 15 months during which security in the area deteriorated and project performance suffered as a result. However, bunds and water supply construction continued at a reasonable rate. A new team provided by a consulting firm arrived and their presence was much more in evidence in the form of correspondence than with the individual consultants of the predecessor project. Eventually the typical situation occurred where the consultant corresponds with the Bank and others on behalf of the PMU. This would have been unheard of with the first PM. 2.36 Three events then caused particularly severe setbacks to the project: breakdown in security (that precluded visits by Bank staff in January 1987); the replacement of the PM by order of the President, and the departure of the long-serving and excellent ADS-employed financial controller. The new PM was judged by Bank supervisors to be unsuitable for the post, especially compared to the first two PMs, and it took several months to find a replacement financial controller. Hence management of the project deteriorated. The next supervision mission found little attempt to implement agreed recommendations and downgraded the project's overall rating to "2" (moderate problems) for the first time since April 1982. A further setback, and an ill-omen for the future, was the loss of seven project vehicles to military use. These were never recovered. During this period several technical aspects came under scrutiny and were debated as described in Box 8. 2.37 Some of the other components came under scrutiny. Stock water ponds were found to be silting up rapidly. A consultant from a UK hydrology institute was able to provide a design that decreased the problem and resulted in more satisfactory and less costly construction. The cost of domestic water storage cisterns kept escalating and there was a search for less costly construction methods, but there was no change during the project period. The small-scale irrigation component was almost a separate subproject, with a separate team of consultants and implementation supervisions by the EEC. Good progress was reported, up to the time when staff had to be withdrawn. There was much - 17 - Box 8 - TECHNICAL PROBLEMS - BUNDING AND EROSION CONTROL RMEA missions had to focus on bund design again because the project consultants had modified the design, despite strong opposition from farmers. IDA insisted on continuing with the original design while modifications were tested. As a follow-up, the June 1987 mission provided detailed bunding specifications, which included no bunding on land with less than one percent slope (except on fields, particularly on "low lands", where runoff can be diverted onto the field from adjoining areas). The Bank's consultant, who wrote the specifications, may not have had access to the earlier feasibility study survey which had found no yield differences on plateau land with less than two percent slope. A new technical problem then emerged: gully erosion by tunnelling, especially in lowland fields (the most productive of the project area). Evidently where there were pools, water would enter the soil through a "pot hole" and flush an unstable soil layer, thereby forming a tunnel into the stream system. After a while, the tunnel ceiling would collapse to start a new gully. A specialist from abroad suggested that the most effective remedy would be allowing the land to revert to bush, but realized that this is not practical. As a palliative, filling gullies with compacted earth was suggested, but for the longer term further study was recommended, but was not implemented before the untimely project end. discussion by supervision and project staff of research and extension, but as before there was no noticeable impact from either on the project or farming practices. In retrospect, it is reiterated that the extension staff could have checked on bunding quality and helped farmers make minor, but important improvements, for instance to enhance the water retention, harvesting and spreading functions of bunding systems. That might have been more productive than trying to change farmers' cultivation practices. 2.38 Repeated failure by the PMU and its consultants to evaluate and compare production on bunded and unbunded land in a statistically reliable way, caused the mission to insist on a short-term consultancy for that purpose. As a result a very impressive evaluation survey was conducted of one part of the project area, where it was possible to find enough bunded and unbunded land farmed by the same operators. One hundred samples were measured and all aspects of this survey were competently handled. The 1987 sorghum crop was measured and, to the surprise of project and Bank staff, there was no yield difference between bunded and unbunded land. Furthermore, yields were low, about 235 kg/ha, or only 190 kg/ha of sorghum when adjusted to take account of land occupied by bunds. A summary of additional findings is given in Box 9. 2.39 The last supervision mission took place in November/December 1987 and coincided with the interim results (substantially confirmed in the final report) of the yield survey; but made no reference to it. In fact, this supervision made little reference to any of the construction components of the project. The main stress was on water harvesting trials and demonstrations as part of agreed programs. A catchment protection plan for one of the pilot areas was agreed between the mission and PMU, but as far as is known implementation was never started. The failure of GOS to return project vehicles, which had been commandeered several months previously, continued to cause concern. Accounting defects in the absence of a financial controller were also causing concern, but despite all that, the project maintained its overall "2" rating. - 18 - Bar 9 - SORGHUM YIELD SURVEY FINDINGS (1987) While annual rainfall in 1987 was above average, a farmer survey had found that rain during the cropping period had been inadequate and poorly distributed. Other interesting results indicated that water storage improves yields; yields were positively correlated with bund height and that using topsoil to construct the bunds may depress yields. Confusingly, the only significant crop yield difference between bunded and unbunded fields was a yield increase on bunded land with slopes less than 1 percent. The survey hints, but does not confirm, that this result may be on "lowlands" receiving runoff, where the Phase II feasibility study had also found significant benefits from bunding. That survey also tested results from a water harvesting trial carried out by the project consultants at the Aburin Research Station. Bunds were spaced more widely and only 25 percent of the interbund area was cultivated, while 75 percent above the cultivated strip was used to generate runoff. Net yields averaged 1,870 kg/ha of cultivated strip, which amounts to 440 kg/ha of land, including cultivated and catchment land. (The audit notes, however, that if land is not a constraint, or runoff land cannot be cropped, then the conversion to a "whole area" basis is irrelevant. The issue is that a crop is grown with reasonable yields on an area on which labor and input costs have been incurred.) The survey noted that the bunds in the survey area had no water-retaining wings and that this may have diminished their effectiveness. The Aburin results, the higher yields in the water retention part of the bunds, and feasibility study findings indicate potentially good results from water retention and/or harvesting. Consequently, the follow-up work during the 1988 crop season was to extend the survey to three areas and pay particular attention to bund design. Hostilities precluded this second survey. 2.40 By June 1988 security problems had increased and the consultants were evacuated. There was a renewed hope in October when the PM visited Hargeisa and drew up plans to restart implementation based on Boorama, but that proved impossible. Because all project activities had ceased in NW Region, IDA agreed in April 1990 to reallocate $420,000 for the preparation of a National Environmental Action Plan. A Country Implementation Review mission in late November 1990 discussed closure of projects that could no longer be implemented, which resulted in termination of the Credit and cancellation of the unspent balance. Not long after that the government was ousted. 3. BAY REGION AGRICULTURAL DEVELOPMENT PROJECT Design/Preparation 3.1 This project was first proposed in 1973 by the same mission that identified the Northwest Region Agricultural Development Project. In fact, funding for consultant services to prepare the Bay Region Project was included in the Northwest Project, but in the event, the FAO/IBRD Cooperative Program prepared the project in 1976/77. Once agreed, preparation was rapid and appraisal was to follow immediately in July 1977. But the mission was aborted en route because of security problems in the project area. As so often happens, once the initial momentum was lost, processing began to lag. A new appraisal team was not fielded until a year later in June/July 1978, and (as happened with the NW I project), Board Presentation was delayed until late 1979 because of the need for cofinancing. The Credit was signed on March 26, 1980 almost three years after the departure of the original appraisal mission. 3.2 As approved the project was "intended, over a period of 6 years to increase crop and livestock production in the Bay Region" (Credit Agreement), and was the first phase of long term - 19 - development. It comprised three parts: (a) improvement of potable water supplies and road access to agricultural areas; (b) the development of an integrated, permanent and sustainable farming system, and (c) training of Somalis and technical assistance for building institutions in support of agricultural development. Four Pilot Agricultural Development Units (PADU) were to be established to raise output by integration of crop and livestock production. One late and important addition to the project proposals (to strengthen the implementation apparatus) was the establishment of the Bay Region Development Coordination Committee (BRDCC), consisting of the Regional Governor and the ministers of the five participating ministries. 3.3 Considering the lengthy period from initial identification through preparation to appraisal, with different staff involved, it is perhaps surprising that the initial enthusiasm about the areas' technology potential survived into implementation, when it emerged that there was little hard evidence to justify such optimism. These hopes had been kindled early on. The 1976 FAO/CP identification mission reported on the high potential of the area of "flat land with excellent limestone soil", where 10 percent of the population produces 75 percent of the country's grain crop. The mission claimed that with better varieties and higher plant population, yields could be doubled. The large livestock population could be improved by better water supplies, animal health services and production of dry weather forage. In consonance with these findings, the mission listed poor communications, lack of water supplies, inadequate animal health services and supplies and primitive farming methods as main constraints to greater productivity. The mission's groundwater specialist indicated a good chance of striking a shallow aquifer large enough for sizeable irrigation development, but emphasized that this would need checking by detailed survey, preferably before the preparation. (This did not prove possible and was not done until the project was in its third implementation year.) He was more certain about a sizeable deep aquifer underlying the limestone plateau, on which drilling anywhere should strike potable water in adequate amounts to meet human and animal requirements. 3.4 In the Bank, CPS questioned the information sources for the ID report and doubted that yields could be raised as predicted. CPS also asked why shortage of drinking water prevented the introduction of improved farming, and whether improved farming methods had been tried. This was the only substantive query about the validity of rainfed farming technology proposals by any headquarters staff, at any stage of the project up to almost the time of the mid-term review in early 1984. There were further CPS comments raising points that would need attention at preparation/appraisal, but the main concerns focussed on project complexity and Somalia's implementation capacity. CPS suggested after appraisal that incremental crop production benefits had been projected too far into the future and asked that the appraisal report specify which improvements were based on research results and which would need further testing. Apart from this comment, no one else questioned the technical assumptions on which the benefits were based, or discovered that none of these assumptions had been proved in the project area. 3.5 Criticism of the project design by the East Africa region's Infrastructure Division was more weighty. It pointed out that the location of some of the roads would not be known at the planning stage, yet detailed engineering and soil surveys for them are called for at that stage. The division also pointed out the staff and equipment shortage in the Ministry of Public Works (MPW) that would militate against force account construction by the Ministry. That warning proved prophetic. 3.6 Looking back, the complexity of this project stands out. A good description from an evaluation report prepared by the Project Implementation Unit of the National Planning Commission was: "This US$45 million project is complicated by five financiers (IDA, USAID, AfDF, IFAD and GOS) and five Ministries or Agencies (Ministry of Agriculture, Ministry of Livestock, Forest and Range, National Range Agency, Water Development Agency and Ministry of Public Works) trying to implement - 20 - the most ambitious regional development project undertaken in Somalia." In addition the project employed (or was to employ) 20 technical assistance staff from six different sources, and appointed yet another consulting firm to conduct the resources study. In retrospect, the project did well to achieve a respectable percentage of the implementation targets. Credit for that must go to project staff and to the task managers of the participating financing agencies. 3.7 The preceding paragraph leads to the question of whether there should have been a different project design. The project attempted to address the major constraints faced by the rural people of the Bay Region: lack of roads, lack of potable water and low crop and livestock productivity. An alternative would have been two separate projects - one each for infrastructure (roads and water) and agriculture (including livestock development). Probably the infrastructure parts of the project would have been more adequately prepared and supervised under the alternative arrangement. Implementation would have been smoother, since the implementing agencies would also be the project agencies, and would have been more willing to provide staff and equipment than proved the case. The multi-component, multi- sector type of project is now less common than in the 1970s and 80s, but if there is an overwhelming reason for development of different sectors concurrently, separate rather than combined projects for that purpose may be preferable. Implementation 3.8 Appointment of the Project Director (PD), the Technical Manager (TM), and Financial Controller (FC), the three key PMU staff, was a condition of credit effectiveness. Since GOS did not have funding for expatriate staff, it applied for a US$400,000 Project Preparation Facility (PPF), which was approved by the Bank shortly before the Credit was approved by the Board. This funding was to enable GOS to hire key PMU staff, procure vehicles/equipment needed for project start-up and to initiate the Bay Region Resources Study. 3.9 The first year of the project was devoted to building up the PMU, searching for and selecting consultants and negotiating with various participating agencies to second their regional staff to the PMU (i.e. preparing for project implementation). The ministries of Agriculture, Livestock Forest and Range, and National Range Agency did so quite readily, but the Ministry of Public Works (MPW), which was responsible for roadworks, and the Water Development Agency (WDA), charged with the water supplies construction, were reluctant to comply. The MPW eventually complied, but WDA never became part of the PMU. The compromise that emerged and which worked well was that all borehole drilling for exploratory and productive purposes would be done by contractors employed by WDA and funded by USAID under its Comprehensive Groundwater Development Program (CGDP). The PMU would equip these boreholes with pumps, tanks and troughs, and construct stock water ponds. 3.10 The third Bank supervision mission in January 1981 rated overall project status as 1/2 ("Problem free/minor problems, stable") - the last time it was so highly rated. The mission reiterated earlier fears about substantial cost overruns, estimated at US$8 million equivalent, but by this time consultants for construction, procurement and road construction (all to be financed by IDA and IFAD) had been selected, but those to be provided by USAID, notably for agricultural research, were not yet in sight. In fact they did not arrive until mid-1983. In the meantime, the research station and seed farm, which had been handed over to PMU management, was operating, albeit ineffectively, under recently recruited and therefore inexperienced staff. Also, at this time (January 1981), the prospects for water supply works were not at all clear. Procurement of farm, range and road equipment was getting underway, but in the absence of the specialist consultants it is not clear who provided guidance. Later there was criticism that some unsuitable equipment had been purchased. The most notable example was a large stonecrusher for road construction, which was ultimately sold and the proceeds used to pay for - 21 - other road equipment. Despite criticism, getting on with procurement avoided even greater implementation delays. An issue arose with regard to the firm selected to provide consultants' services for road construction. RMEA Nairobi pointed out that the firm had been disappointing in an earlier assignment in Somalia, but no further reference to this issue was found in the files and the appointment went ahead. In the event, the first group of staff from this firm proved unsatisfactory and were replaced in early 1983. 3.11 At this time, the PMU first expressed doubts about the Pilot Agricultural Development Units, in particular about the size of each unit. To provide space for PADU activities considerable bush clearing in the rangeland was foreseen (to permit cropping), but the PMU noted that funding for this purpose was not provided. The supervision mission counselled that establishment of PADUs be deferred until the resource studies had been completed and relevant technical assistance staff had arrived. 3.12 A joint HQ/RMEA supervision mission in October 1981 reported the slow progress, in particular that: - both WDA and MPW had not provided staff and equipment as agreed to the PMU for road and borehole construction, - USAID funded agricultural specialists had not arrived; - cost escalation was already projected at over US$12 million (or 27 percent); and - there was little liaison between the PMU and agricultural extension staff. On the plus side, USAID had appointed a veterinarian who quite soon got the veterinary program of stock vaccination and drug distribution underway. This component remained the "jewel in the crown" during the entire project period. 3.13 Thus, by 1982, infrastructure construction had started, the veterinary program was going well, and the PMU had been firmly established with all its technical assistance staff hired. This was more or less according to the appraisal schedule. However, two components were delayed: the construction of buildings, where the consultant took longer than expected on the design stage, and the recruitment of USAID-funded agricultural specialists. Building construction scheduled for completion during Year 3 of the project (1983) did not start until early 1985 and was completed during 1986. The consultants appear to have done most of the design work at the home office in Europe, which did not turn out to be a satisfactory arrangement. The agricultural team, which was to have been in place during the first project year (1981), finally arrived in mid-1983 and got the research program underway in the 1984 gu (April) season. Research assistance for sorghum improvement had been provided (in addition to the project) with funding from the International Development Research Center (IDRC of Canada). 3.14 Supervision responsibility was then transferred to RMEA, and this ushered in a period of rigorous and highly critical supervision reporting, which did the project's reputation permanent harm. Achievements, admittedly modest, were minimized, while shortcomings, admittedly plentiful, were emphasized. At the end of this supervision era, just after the mid-term review (MTR), the project was branded as a failure at both RMEA and Headquarters. A somewhat less severe and more helpful approach might have improved implementation, particularly since slow progress was either beyond the PMU's control, like slow appointment of USAID funded consultants, or stemmed from defective project design such as the PADU concept. - 22 - 3.15 The overall performance rating was dropped to 3/3 (major problems, deteriorating) following the supervision of December 1982. The project had been underway for almost three years and work on rainfed crops, a major focus of the project concept, had not started. Borehole drilling had got underway (but costs were triple SAR estimates) and early hostilities between the PMU and WDA had subsided. Road construction had also started, but initial results were disappointing in quantity and quality. The road construction rate was running at less than 20 percent of the SAR assumption and costs were up by 50 and 175 percent for two grades of road. The project consultants were judged unsatisfactory in planning and executing the contract. The resources survey had been completed and reported that: (a) there is no irrigation potential; and (b) the Pilot Agricultural Development Unit concept is not feasible because it would not be possible to retain such large areas (72,000 ha) for the exclusive use of the villagers of that locality, nor would it be feasible or desirable to direct their stock grazing regime in a particular way, when it was not known whether any particular way of range management was superior to current practices. 3.16 As a result of the new resource information and the slow rate and high cost of road and water supplies construction, the mission felt that project restructuring was inevitable. When this possibility was first raised in early 1982, Bank headquarters staff felt that appraisal had been carried out by a well qualified and competent team and that it was premature to challenge its proposals, but as more information became available HQ objections to restructuring faded. The files indicate that reappraisal with Board review, or even cancellation, were considered as possibilities. 3.17 The mid-term review took place in March/April 1984. As a preliminary, the PMU had hired a consultant team (the same firm that had carried out the resources survey) to draw up a three year work plan as stipulated in the DCA. Approval of this plan by IDA was a condition of disbursing any further IDA funds 3-1/2 years after the Effective Date. The plan was to have been drawn up by the project management, with the help of its consultants, but since the agricultural team was not yet in place and the water team was not part of PMU, the IDA task manager had urged the PMU to hire a separate consultant team. The IDA mid-term review team, however, appears to have relied more on its own judgment than on the consultants' work plan. 3.18 Although at the time of the review there were serious design problems, on the administrative side the situation was more encouraging. The PMU was firmly established and operating effectively, though heavily overstaffed, particularly with minor employees (130 out of the 500 PMU staff were watchmen, messengers, cooks and temporary staff). The long and short-term training programs were making good progress. The new Technical Manager of the PMU was considered more effective than his predecessor and the same seemed to hold true for the replacement consultant team supervising the roads program. In addition, due to the resources survey, there was accurate information available on land and water resources. A socio-economic survey was in progress, and some preliminary information gave both the project management and donors a better idea of the difficulties facing the Bay Region population and the measures used by them to deal with those difficulties. 3.19 The mid-term review was the turning point of the project and from then on implementation went reasonably well. The more modest targets for the roads program were achieved, albeit after two extensions of the closing date. The change in project components are detailed in Box 10. - 23 - Bar 10 - CHANGES OF BAY REGION PROJECT COMPONENTS AT THE MID-TERM REVIEW (a) the roads program was reduced from 560 km new roads and 250 km tracks to 300 km and 165 km respectively, with two instead of one road maintenance units to be established; (b) the water program for the limestone plateau remained at 60 boreholes, but on the basement complex, where no potable aquifer was found, 16 existing ponds were to be rehabilitated; (c) the veterinary program was not changed. A proposal for an animal disease survey, by the consultants who had proposed the revised work program, was turned down. This decision was later criticized by the PMU and the consultant who prepared the PCR; (d) the range improvement program (part of the PADU concept) was dropped on social and technical grounds. (Some large implements purchased for range improvement could not be reassigned or disposed of and may still be at the Seed Farm as a reminder of the abortive Pilot Agricultural Development Unit proposal); (e) The NMEF component was not altered; (f) The PMU support and staff training programs were not changed, other than to reduce the office and housing building program to fit the budget. (g) The agricultural intensification program was changed into a more comprehensive and intensive agricultural research program, which included a much stronger expatriate and Somali team of research staff (Somali scientists were returning after MSc training in the USA). A major off-station operational research effort, in collaboration with farmers, was added. (h) At the time that the new proposals were discussed by GOS, the Bank and donors, it was agreed to reallocate US$450,000 as bridging finance for the Semi-mechanized Rainfed Farming Pilot Operation, which was supplemented by SDR 800,000 bridging finance from the Northwest Agricultural Development Phase II Project. 3.20 The mid-term review proposals were accepted by cofinanciers and discussed in detail with GOS, which led to some amendments of details, but not of the general thrust of the new work program, which was adopted by all parties about mid-1984. The PMU, in one of its periodic reports, summarized the outcome of the mid-term review admirably as shown in Box 11. - 24 - Bax 11 - PMU SUMMARY OF MID-TERM REVIEW OUTCOME "Over the long term the original project objectives remain valid, except that range management and improvement is not considered feasible for technical and social reasons. The redesigned project would concentrate resources on the following: (a) Trying to find ways to reduce the risks of rainfed crop farming in the Bay Region by means of operationally oriented agricultural research and trials, including soil moisture conservation, working hand-in-hand with agricultural extension and plant protection services; (b) Consolidate the veterinary services already started in the region; (c) Provide reliable water supply where the population density is great and most vulnerable to frequent droughts; (d) Rebuild and maintain the economically indispensable roads of the Bay Region without adding significantly to the existing road network. The expected outcome upon project completion would be a minimum set of technical recommendations for rainfed crop improvements that can be widely and easily applied by farmers with the advice of extension agents, a low cost system of basic veterinary care, adequate water supply at strategic locations, and adequate maintenance of the improved roads network." 3.21 In general, the revised objectives and targets were achieved, except for those deriving from agricultural research, despite an intensive and well directed research effort over the three seasons between the mid-term review and project closure (see Box 12 below). A longer timespan was needed to find technology to improve crop and stock productivity in the Bay Region. 3.22 The main problem after the MTR was fuel shortages. From 1984 onwards this was a constant and recurrent theme in supervision reports, causing delays in road and water supplies construction; effecting the veterinary program (refrigerators storing vaccines ran out of kerosene and the vaccines were spoiled); and the agricultural trials programs on farmers' fields could not be supervised. (The two Northwest Projects also suffered from fuel shortages, but the impact appears to have been less severe than in the Bay Region.) Reliable fuel supplies are an obvious and important factor to take into account in project design in countries like Somalia which suffered from foreign exchange shortages or poorly developed fuel supply services. In this case no special arrangements were put in place to resolve the problem. 3.23 In early 1985 there was a change of task manager and with improving project performance the tone of supervision reports mellowed. The first sign of improvement was a change (in February 1985) from the 3/3 project status rating of the preceding two years to 3/2 (major problems but with an improving trend). By September 1986 the overall project rating was upgraded to 2 denoting moderate problems, and the project was no longer classified as a problem project. That rating remained the same until project closure in March 1985. - 25 - 3.24 The appraisal mission must be given its dues for pointing out the need for more intensive agricultural research than assumed by the project preparation team, but by the time of the mid-term review it became obvious that even broader and more intensive research than was projected at appraisal was required. The USAID funded TA team that arrived in mid-1985 consisted of an entomologist, a plant breeder and a soils agronomist to join the IDA financed dry land agronomist. By that time good collaboration with extension staff had been established and a program of trials on farmers' fields supervised by extension staff had been initiated. The main objectives were moisture conservation methods, sorghum improvement, weed and stem borer control (which seemed to be the agricultural problems that worried farmers most), and introduction of pulses and fodder crops. Successive supervision missions recorded complimentary reports about both the planning and execution of the research program and the increasingly important role of the Somali research staff. Three staff had returned after masters training in the USA as well as from shorter training abroad, most notably at ICRISAT. Some achievements of the post MTR research effort are recorded in Box 12. Bar 12 - POST MTR RESEARCH RECORD Earlier trial work had been handicapped by the small size of the station (only 12 ha); lack of laboratories, offices and equipment, and by the high variability of trial results which affected their reliability. Project works and purchases provided the necessary buildings and equipment, while amalgamating the nearby seed farm with the research station solved the station size problem and improved the management of both stations. The combined stations became known as the Bonka Research Complex. The most important event, however, was that the work of the soils agronomist found that station soils were severely phosphate deficient causing general stunting of plant growth and the high variability of trials. Phosphate application overcame that variability, but investigation of the applicability of this finding to farmers started only in 1988 and was not completed. The extent of the area affected by phosphate deficiency, the optimum phosphate application rates, the most appropriate fertilizers for farmer use and the economics of fertilizer use have yet to be established. Such work could lead to major improvements in productivity and farmers' incomes. Higher yields when phosphate is used may make use of other inputs worthwhile, such as stemborer control pesticides, and when rains are good or if reliable moisture conservation measures are found, other fertilizers. The stepped-up pace of research at Bonka, and successes like the soils work gave rise to the idea that Bonka should be regarded as the National Drylands Farming Research Center which was accepted by both GOS and the Bank before the end of the project. Resident researchers were supported by short-term specialist in plant pathology, forage production, sorghum physiology, statistical analysis of trials, farming systems surveys and animal traction. The transformation of Bonka from a moribund station to a vigorous and effective research center was an important project achievement. The change of most of the seed farm to research represented a further project modification. The appraisal concept was that superior varieties should be bulked at the farm and some foreign sorghum varieties that had looked promising in trials were bulked, but farmers found them inferior. However, when screening by the pulses research team identified a superior mung bean, the more modest bulking program proved useful and there was good uptake of the seed by farmers. 3.25 With regard to infrastructure components, the PMU developed a capacity for direct implementation of several construction programs, including equipping boreholes, construction of "Irish - 26 - bridges"./ for the new and improved roads, and a livestock ponds rehabilitation program. The ponds, in particular, were a success. A new design proposed by a short-term consultant was used and construction turned out to be so much below estimated cost that the original target of 18 ponds was expanded to 38, all of which filled up in the first rainy season following construction. The PMU also took over Irish bridge construction, because of unsatisfactory work by the road construction program. There are records of the number constructed but none to show whether work quality was better than that supervised by the road construction team. 3.26 In 1986 it became obvious that the road and water facilities construction would not be completed by March 31, 1987, the original Closing Date, and the Closing Date was extended by 12 months. Subsequently, when there was still some work to be completed, in particular the road construction and improvement program, additional short-term training needs had emerged and USAID had also extended its loan to September 1988, the Bank agreed to a second extension up to March 31, 1989. 3.27 During the last project year, the PMU hired consultants to prepare the Project Completion Report and the Bay Region Phase II Project. However, the files indicate that the Bank did not intend to finance a Phase II project, despite the project description in the SAR and in the DCA, which reads: "The Project ... forms the first phase of a long-term development, includes funds for the preparation of a follow up phase....". The Bank had, however, ensured continuation of the Bonka Research Complex by including funding in the Second Agricultural Extension Project (Credit 1794-SO). In the closing phases, the Project Director (PD) tried hard to persuade the Bank to continue to finance the expatriate financial controller and the expatriate chief mechanic of the road construction unit, because the Somali mechanical engineer had not returned from training in the USA. The Bank did not agree to this request. This refusal was typical of the more chilly climate that marked supervision when responsibility for it was returned to Headquarters early in 1988. The audit agrees with the PD that changes of task manager should be avoided, if possible. 4. SEMI-MECHANIZED RAINFED FARMING OPERATION Background and Design 4.1 The Semi-mechanized Rainfed Farming Operation arose from the mid-term review of the Drought Rehabilitation Program Project (Credit 623-SO). This project was the Bank's rapid response to a request for help from GOS, to settle destitute nomads who had lost their animals in the devastating 1974 drought. The US$10.8 million pilot scheme was prepared and appraised in two weeks by a two- person mission in July 1975 and was approved by the Board in April 1976. It was, over three years, to develop rainfed farming at three settlement areas, to produce about 7,000 tons cereal annually, from about 15,000 ha land. Credit 623-SO closed in September 1984, four years later than scheduled and has been reviewed in the Project Completion Reportl4/ which forms part of the Performance Audit Report.15/ 13/ Hard surface waterway allowing storm water to run over a road without damage, rather than a conventional bridge or culvert with water flowing under the road. 14/ Drought Rehabilitation Project (Credit 623-SO), Completion Report - September 20, 1985. 15/ Project Performance Audit Report, Somalia Drought Rehabilitation Project (Credit 623-SO), Report No. 6011 - December 26, 1985. - 27 - 4.2 In the 1979 crop season, after three years of project implementation, only 1,275 of the target of 15,000 ha of land had been cleared and estimated production totalled about 350 tons maize and 150 tons cowpeas, or about 7 percent of the target. This disappointing performance led to an invitation to an Australian firm of dryland farming specialists to participate in the mid-term review, which in turn suggested a pilot program that became known as the Semi-mechanized Rainfed Farming Operation. The PCR and PAR for the Drought Rehabilitation project evaluate the early years of the pilot operation. This audit expands on the findings of those reports, in particular on technical aspects. There were some links and affinities between the Bay Region and Drought Rehabilitation project areas as described in Box 13. Bar 13 - SIMILARITY AND AFFINITY BETWEEN THE BAY REGION AND DROUGHT REHABILITATION PROJECTS Both of these project designs lacked data about the resources of the project areas, both changed objectives from production to research, and both projects employed the same firm of consultants. One major difference was the reaction of Bank staff who reviewed the two projects after appraisal. While this audit has found that the processing reviews of the Bay Region Project had not adequately challenged the project's technical shortcomings, the converse occurred with regard to the Drought Rehabilitation Project. The Central Project Staff challenged the Region's project proposals so vigorously that a special meeting of the Loan Committee was called to consider whether the project should go ahead. The Loan Committee realized that the project was risky, but in view of the urgent need to address the drought victim problem, and the fact that the project formed part of a multi-donor financed drought rehabilitation program, the Committee decided to go ahead. The Drought Rehabilitation project PAR explains in some detail the unsatisfactory technical assistance provided for the project, which was probably one of the main causes of the poor implementation results. It was therefore ironic that the same firm was employed for one of the components of the Bay Region Project and there too performed unsatisfactorily, at least until the initial staff were replaced. 4.3 The pilot operation had two sites at the Kurtun Waarey and Sablale settlements in the Lower Shebelli Region of Somalia (Map IBRD 25134). These are situated along the Eastern side of the Shebelli River with headquarters about 140 and 220 km respectively southwest of Mogadishu. Soils are mostly black alluvial clays which are reasonably fertile, though there are some areas with high salt content that reduces crop yields. The soils had not been mapped when the project started. There were no rainfall records of the area but the nearest stations had averaged 400-600 mm per year mainly in two seasons - 70 percent in the April-July Gu season and 25 percent in the October/November Der season. As is usual in semi-arid areas, there are considerable year-to-year variations of rainfall totals and distribution. In this area, about 25 km inland from the sea, there are also light rains which fall between the Gu and Der rains, but the period from December through March is hot and dry. As in the two other project areas described in this audit, the low and erratic rainfall is the main constraint to crop production and was believed to be the main cause of the disappointing performance of the Drought Rehabilitation Program Project. 4.4 The area had virtually no resident population and before the settlements started there was no crop production. While there was a considerable wildlife population, there was no resident domestic livestock, due to infestation with tsetse fly carrying the trypanosomiasis pathogen. However, the area - 28 - served as an important dry weather grazing and stock watering reserve. Large numbers of stock from the Bay Region to the North and from other regions even further afield come to the Lower Shebelli grazing areas (that is why the veterinary services in the Bay Region are so important for treating animals against trypanosomiasis). 4.5 Bank staff supervising the Drought Rehabilitation project had been suggesting for some time, after it became obvious that crop production presented serious difficulties, a change of objectives from meeting land clearing and production targets, to production systems research on the already cleared area. The March 1979 supervision mission mentioned the possible applicability of Australian/USA dryland farming methods, including weed-free fallowing. SDA was reluctant to adopt such a change but agreed to a mid-term review. This took place in May 1980 and firmly recommended that resources be employed mostly for large scale trials and related technical assistance and staff training programs. SDA continued to resist this proposal and the Bank came close to suspending or even canceling the Credit.16/ As a consequence, SDA agreed to employ a dryland farming specialist and purchase the necessary implements to conduct dryland farming trials on up to 700 ha each at Kurtun Waarey and Sablale. In December 1980, the DCA was amended to reflect agreements reached. To plan and implement the trials program, SDA selected a specialist from the Australian firm which had participated in the mid-term review and which continued to provide technical assistance to the trials program. This firm also prepared a follow-up project some seven years later, but then SDA selected another firm to guide the follow-up project. Implementation 4.6 The Australian consultants arrived in early 1981 and quickly drew up an operations plan and then implemented it vigorously and efficiently. He was of outstanding caliber and the audit echoes the opinion expressed at the Sixth World Bank Agricultural Symposium in January 1986 (where he gave a presentation on the Somali trials) that "one must recognize the drive and skill with which the operation was planned, organized and managed. The technology could easily have failed if there had not been a highly skilled input of diplomacy, management and training." The irony at the end of the day is that with the deployment of similar skills and energy, the original project objectives with regard to land clearing and agricultural production might well have been attained, as planned, without recourse to moisture conserving technology. The strategy that was designed soon after arrival of the new specialist (Box 14) continued to be tested until the mounting hostilities brought the program to an end in December 1990. 16/ The DCA had been amended in July 1979 to introduce a new project component-the Project Implementation Unit, a national unit to help improve implementation of the entire agricultural portfolio. At the same time the Bank had agreed to extend the Project Closing Date by two years, to June 30, 1982. During the summer of 1980, the Bank was considering suspending disbursements on all categories, other than that to finance the PIU, or to cancel the Credit and finance the PIU from another project. - 29 - Bar 14 - TECHNICAL AND INSTITUTIONAL DETAILS OF THE SEMI-MECHANIZED RAINFED FARMING SYSTEM The basis of the system was the introduction of a four-crop rotation, combined with weed-free fallows for moisture conservation, in a strip farming system for efficient use of mechanical equipment. The rationale for the fallows system, tested particularly in semi-arid areas of Australia and the USA, was that cracking black clays can absorb and store a considerable amount of moisture if this is not withdrawn by weeds between crops. This stored moisture then augments rain in a subsequent cropping season. The design for Somalia was for a two-season (one year) fallow to precede each crop. Thus, only one-third of the land would be cropped in any one season, and the crop would benefit from the rainfall of three seasons. Maize and sunflower were chosen as the most suitable crops for the main (Gu) rains and cowpea and safflower for the much shorter and more erratic Der rains. Initialy, sorghum had been considered as the most suitable grain crop for this drought-prone environment, but after suffering massive bird damage it was replaced in the rotation by the less drought tolerant, but more bird-resistant maize. The design ensured that each crop would be grown on the same land only once in six years, mainly to avoid a pest and disease build up and the need to use pesticides. The standard field layout was known as the strip farming system. Each field was a 100 meter wide strip, up to five km long, to minimize tractor turning. Each cultivated strip was separated from the next one by a 33 meter wide strip of native vegetation to act as a wind-break and to help preserve the local flora and fauna. Not only was such a comprehensive and revolutionary program quickly planned, but a 400 ha area was sown in April 1981, mainly to test the suitability of various crops. The social engineering that ensued was equally comprehensive and dramatic. The existing settlement scheme was conceived during the days of USSR patronage of Somalia as a series of state farms. Supervision reports of the Drought Rehabilitation project consistently point to labor shortages for sowing and weeding crops, due to lack of interest of settlers in the farming operation, presumably due to low benefits and the fact that the original settlers were nomads, unused to growing crops. The new consultants convinced SDA to change the state farm system to one where the scheme management would provide key services to farmers on leased, individual holdings, much like the system pioneered by the Gezira Irrigation Scheme in Sudan after World War I and widely adopted elsewhere. The central management would plan the rotation, prepare the land, sow the crops and clean-weed the fallow, while farmers would weed, guard, harvest and market the crops, and pay for the services provided by SDA from the proceeds of the crops. It was believed that a farm family could cope with 3 ha of crop per season, which meant dividing the strips into 400 x 300 meter ' (12 ha) holdings; in each season 3 ha would be under crops, 6 ha would be clean weeded fallow and 3 ha would be bush strips. This system facilitated the use of large tractors and specialized equipment designed for working heavy clay soils. But it was discovered in the first year that effective mechanical weeding of crops on the wet clay soils was impossible and therefore hand-weeding (with strong incentives) was needed. This Gezira style of organization worked well, as long as farmers were left with a comfortable surplus at the end of the day. This last point did not become a serious problem until the start of the full project, six years later, when the question of the financial viability of the scheme with full cost recovery was belatedly taken up. Initially cost recovery efforts aimed only to establish the principle, rather than to recover all costs, since maintaining farmers' incentives was considered more important than the percentage of costs recovered. The settlers were happy with this system. Three arable strips of 100 X 300 meters plus 3 bush strips of 33-1/3 X 100 meters. 4.7 Probably the most valuable asset of the local consultant was his hands-on experience of doing all the tasks needed for successful dryland farming, including the operation, maintenance and repair of all implements and equipment (which had proved a problem under the parent settlement project). - 30 - Consequently he could personally train the local staff, and lead by example when necessary during peak cultivation times. This exceptional consultant input transformed the project and the tone of supervision reports changed from despair to euphoria, with the Drought Rehabilitation project status rating rising from 3 to 1. Even so, although results from the first 400 ha of cropping were encouraging, technical problems soon emerged that have never been completely solved. Most notable of these were the vigorous weed growth that made maintaining weed-free fallows more difficult than expected; the persistence of some shrubs on cleared new land that seemed to grow worse over time, rather than better; and the difficulty of maintaining seed viability (this ultimately led to recommendations to store seed in dry, air conditioned containers). 4.8 Despite emerging agronomic problems, SDA, MOA and even the Ministry of Finance, shared the Bank's enthusiasm for the new system. Consequently, in 1982 the strip system was extended to 900 ha at each site. Results in 1983 were particularly encouraging, especially in demonstrating the yield benefits of storing soil water by weedless fallows across two rainy seasons (long fallow). Yields were as follows (in kg/ha): Short Fallow Long fallow 17/ (one rainy season) (two rainy seasons) Sorghum 900 2,100 Safflower 470 740 Cowpeas - Kurtun Waarey 200 400 - Sablale zero 200 However, the Sablale failure for cowpeas after a short fallow, and the range of yield for Gu sunflower (270 kg at Sablale and 650 kg at Kurtun Waarey) should have perhaps tempered enthusiasm with the realization that conditions and results in that kind of environment are very variable. This message was brought home dramatically in 1984 when there was a serious drought in much of Eastern and Southern Africa. Crops failed in both seasons at Kurtun Waarey and during the Der at Sablale. In contrast, aptly demonstrating the erratic rainfall patterns of such areas, the Gu rains at Sablale in 1984 were quite favorable for crop production (461 mm fell in April, May and June resulting in a good maize crop of 1,700 kg after a short fallow and 2,500 kg after a long fallow). This Sablale experience earned an undeserved reputation as an example of the benefits of fallowing, although it was an island of good crops in the otherwise drought-stricken land, since it was also an island of favorable rain amid the otherwise general failure of the rains. Even double fallowing will not compensate for lack of rain at sowing or soon after. 4.9 Nevertheless, the reputation of the operation reached its zenith with the visit of the Bank's Vice President for Eastern and Southern Africa in 1985 and the presentation by the consultant of a paper at the Bank's Sixth Agricultural Symposium in January 1986.B/ The Drought Project PAR (dated .2/ This may not be a truly fair comparison because the two fallows may be the preceding Gu and Der, while the one fallow may be from Der only, with much less rain than the Gu. 18/ Development of Rainfed Agriculture under Arid Conditions-proceedings of the Sixth Agricultural Symposium. Ted J. Davis, Editor - Technical Session A - A Semi-mechanized System for Rainfed Agriculture in Semi-arid areas of East Africa. Andrew McPherson, Bingle Pty, Ltd. - 31 - December 1985),12/ contributed with generous praise of the operation (tempered with caution about experimental methods as quoted in Box 15). Bar 15 - AUDIT COMMENTS ON THE SEMI-MECHANIZED RAINFED OPERATION (from PAR #6011, Drought Rehabilitation Program Project) 35. "Yields have been impressive in the light of the area's semi-arid conditions. Maize yields have reached up to 3t/ha with an average of about 2t/ha. Cowpeas and safflower average about 6-800 kg/ha. Farmer/settler incomes have improved accordingly and interest in the scheme is strong. Due to insufficient land levelling and drainage of the irrigated settlement schemes, net return to dryland farmers are better than to irrigating settlers. 36. The consultants' work has also been beneficial to project organization. Due to the consultants' recommendation the two farm managers attended courses in Australia familiarizing themselves with the latest dryland farming techniques. To improve and secure proper maintenance of farm machinery and tractors a bonus system for the farm mechanics has been introduced. This system has not only prevented the quite common emigration trend which leads qualified and skilled Somalis to seek work in the Gulf states, but has also extended the life of project equipment -- some tractors are eight years old -- a striking difference to the scrapped machinery next door in the irrigation schemes. 37. There is of course scope for further improvements and refinements. Trials were undertaken in a rather crude fashion and while their layout may satisfy the curiosity of farm managers it may not convince scientists. There is no determination of medium-term priorities and most trials seem to be done on an ad hoc basis. Elimination of coincidental results through introduction of random location of trial plots has not been undertaken. The consultants are aware of this relatively minor problem ' and a trial/research program will be prepared. 38. Looking at the promising results of the introduced techniques several questions can be asked. Why has the Bank, in its many years of limited success with rainfed agricultural projects especially in Africa, never systematically explored the development possibilities of dryland farming systems in the semi-arid regions of Australia and the southwestern parts of the U.S.? Why did the Bank accept [the original project] consultants from a country that never had any experience in this sector? And finally, what is the Bank going to do to assure that the Somali experience will receive adequate attention within the Bank and in other countries with similar ecological zones?" Considered not so minor by this audit, nor by the CPS research specialist who had reviewed the PAR. 4.10 In September 1983 the Bank sent a strong mission to review rainfed farming systems, in particular the rainfed semi-mechanized system with a view to its possible application in the Bay Region. The mission included specialists on soils/dryland farming and on farm machinery. The mission confirmed the assumptions on water storage and the related crop yields projections. It stated that "the semi-mechanized farming system offers the only immediate prospects of gainfully using currently underutilized resources (land and labor) in much of the Lower Shebelli area," but that it would not be suitable for the Bay Region which was already being cultivated by agro-pastoralists. 19/ op. cit. - 32 - 4.11 The mission report apparently convinced the Bank that the technology was sound and that it was worth supporting further through a project. Interim finance was provided to maintain the trials for two years to further refine the technology while a project was processed. The firm managing the trials provided staff to prepare the project. 4.12 During the preparation period the principal of the firm, who had personally managed field operations up to then, handed over field responsibility to two new staff, but remained in Somalia in an advisory role. (The first supervision report of the later project was to note that the services provided to farmers had become less effective than during the early stages of the trials. Hence, the success of the early trials was apparently linked to an exceptional technical manager.) 4.13 The further history of this rainfed cropping initiative must await the PCR of that project and possibly an audit. In the meantime a brief account is given here to complete the picture. The new project was to treble the size of trial operations, conduct vigorous problem solving research and work out an appropriate institutional structure for the scheme. The project was approved in April 1987 but did not take off. This was partly because a new consulting firm which had little dry land farming expertise, was appointed, and because security continued to deteriorate (a consultant was killed and others were attacked). The operation came to a halt in late 1990. In these circumstances, the research program made little progress and land clearing and new settlement was slow. A further problem was that when fairly regular crop failures through drought were taken into account, average yields over time were much lower than projected from yields achieved in earlier good rainfall years. Settlers were discontented with the scheme management and started to make changes unilaterally, and agitated to be allowed to crop fallow land. The second Bank supervision mission cast doubt on the ability of settlers to both meet subsistence needs and repay any project costs. While two succeeding missions painted a less gloomy picture and recommended continuation of the pilot program, the doubts about viability were not resolved nor were the technical problems solved. 5. PROJECT OUTCOMES Northwest Region Agricultural Development Project and Northwest Region Agricultural Development Phase II Project 5.1 The physical implementation performance of these projects, particularly of the first phase, must have one of the best records of any Bank-funded project in Somalia. This was achieved in a remote corner of the country, by Somali staff with no previous experience of the Bank, and against a background of intermittent security problems. 5.2 The first project is on record for having attained all appraisal targets except for small- scale irrigation, through no fault of the project, and for gully control which had been halted at the suggestion of the Bank's task manager. The audit found that the bunding target in terms of length of bunds to be constructed may not have been met, and that the claim for area protected may also be excessive. With regards to length of bunds, the SAR (para. 4.02) projected 25,000 ha to be bunded with an average 300 m of bunds per hectare. That would have amounted to 7,500 km of bunding, while records submitted by the PMU for cost recovery purposes showed that a total of about 4,200 km of bunds had been constructed during the first project. With regard to area, the reporting system showed hectares bunded and number of farmers with project bunded land. The June 1987 supervision report (Attachment 2 - Key Factors) showed: - 33 - Project Farmers 5,275 Gross area protected 31,871 ha Net area bunded 16,643 ha Cropped bunded area 10,160 ha The gross area protected comprises fields which are only partly bunded. Typically the bunding is in the lower half of the field and the audit agrees with some observers who question whether the unbunded area is afforded any protection by the bunds below it. Consequently, the net area bunded would be a more accurate representation of the protected/treated area. That is not to denigrate the outstanding records of the PMU, but just to correct overestimates of actual achievements. 5.3 Gully control work was one of the key components, and had potential for supporting bunding work by preventing expansion of gullies into arable land. No evidence was found to support the PCR's conclusion that the early gully control efforts had proved prohibitively expensive, but a complete soil conservation plan for the area would have been desirable. However, the implementation of such a plan was nowhere in sight, although the idea was embodied in the NWII project as the Pilot Catchment Protection Scheme. In the meantime, useful gully plugging work by inexpensive means (such as were ultimately suggested for the Watershed Protection Pilot Scheme), could have been initiated. 5.4 Research and extension work had no impact on local farming practices. It was always unlikely that the one-person research effort would generate "breakthrough" technology in this difficult farming area. However, the water harvesting trial, designed by the project consultants, gave potentially useful results that could have led to bund design modifications and to improved crop yields. Adding to the many suggestions by supervision missions and feasibility study consultants for improving research and extension, in retrospect there were three potentially rewarding activities. First, for research, the challenge was to find out exactly why farmers were growing their crops the way they were. For example, an understanding of why farmers were broadcasting crops into stubble, rather than ploughing before the rainy season, could lead to adaptive research to identify superior practices acceptable to farmers. Second, extension staff could have played a useful role in visiting each and every bunded field to ascertain the condition of the bunds and whether work was needed to improve or rehabilitate them. Third, if the country had remained peaceful, the intensified dryland farming research at the Bonka station in the Bay Region could have produced useful results. 5.5 While domestic and stock water supplies also drew some criticism for poor management, both investments were much appreciated by the local people who had helped to locate the supplies and plan the system. 5.6 Bunds remained very I )pular with farmers, albeit allegedly more because they were perceived to confer security of tenure than for productivity benefits. As pointed out by the PCR, there have been conflicting reports about benefit levels and hostilities shut down the project before a follow-up crop yield survey could shed further light on the issue. However, the audit has reviewed available documents in detail and discussed the projects with staff involved in appraisal and supervision. Quantification of erosion control benefits has remained difficult. Undoubtedly some bunding has been done where there is little erosion hazard or potential for water harvesting. But there may be other bunds, of unknown number, that with relatively little extra work could improve conservation and water harvesting performance. 5.7 The appraisal mission of the first project estimated that the ERR would rise by an additional two percent from soil conservation benefits, as distinct from those from moisture conservation, and the consultants preparing the Phase II project assigned an additional three percent ERR for arresting - 34 - land degradation. These levels are in line with recent Bank economic work on catchment protection.n/ However, the audit agrees with the PCR finding that the economic rate of return arising from bunding cannot be re-estimated with any confidence. Additional benefits arose from investments in irrigated gardens and water supplies, but no estimates were found by the audit. 5.8 Sustainability might be considered a moot point given the fighting that afflicted the area. There is also evidence from project supervision, the 1987 sorghum yield study and the evaluation by USAID of its Northwest Region Bunding Project of 1963-66 2.1/, that farmers do not maintain the bunds and that ultimately the bunds deteriorate. Nevertheless, they are of such robust construction that they should mostly survive even periods of armed conflict. Similarly, the water supplies works should last a long time, but they are more susceptible to war damage and only time will tell what shape they are in. The irrigation works may not have survived. Bay Region Agricultural Development Project 5.9 As indicated in the Project Completion Report,22/ the reduced infrastructure targets set by the mid-term review were achieved. The project installed a reasonable all weather road system, potable water supplies were laid on for residents and their herds, a low cost but effective veterinary service was established (Map IBRD 25736) and an agricultural research system was started. While the objective of improving productivity from arable and rangelands may not have been attained, both the USAID evaluation of the project22/ and that by the consultant employed to prepare the PCRL4/ noted arable area expansion and considerable increase in sorghum production. While much of this may be due to decontrol of marketing and climatically favorable seasons, both sources independently ascribe some of the increase to project infrastructure development. Notes on individual components follow. 5.10 With the dropping of PADUs and associated range improvement trials, both the PMU and observers, notably the USAID Completion Report2/ and the consultant who helped to prepare the PCR, felt that range management and livestock husbandry research and extension had been neglected. Although the livestock subsector is by far the most important in most of Somalia, including the Bay Region, specific range interventions at that time were not justified. The Central Rangelands Project was feeling its way towards helping improve rangelands and the lives of nomads. Most findings from that project would either be directly applicable in the Bay Region or would point the way to the most effective interventions. 5.11 The PMU built up a formidable project organization and by 1988 supervised a staff of about 490, distributed as follows: 20/ Watershed Development in Asia - Strategies and Technologies, World Bank Technical Paper Number 127. l Agency for International Development, "Project Impact Evaluation Report No. 62 - A Soil and Water Conservation Project in Two Sites in Somalia: Seventeen Years Later,' August 1985. 22/ Project Completion Report - Somalia - Bay Region Agricultural Development Project (Credit 972-SO), Report No. 11874, May 14, 1993. 23/ Bay Region Agricultural Development Project (649-0113, Mogadishu, Somalia, Final Evaluation - May 1988, R. W. Duell and R. Fox). 24/ Bay Region Agricultural Development Project Phase I Project Preparation Study - Annex 1, Project Completion Report, Sir M. Macdonald and Partners, Ltd., December 1988. 25/ Already quoted. - 35 - PMU 175 Water Development 7 Road Unit 112 Agricultural Services 114 (does not include Extension) Veterinary Services 8 Total 492 The audit agrees with criticism by supervision missions that the PMU itself was heavily overstaffed. 5.12 As the PCR points out, the M&E section of the PMU did not function at all, although in the last project year the agricultural services carried out farm surveys. Apparently, the PMU received no help for its own M&E needs from the National M&E Facility, of which it was nominally the supervising agency. Even so, PMU reporting was good and punctual. 5.13 The PCR reports that the component to strengthen the National Monitoring and Evaluation Facility (NMEF) was successfully implemented, but the audit found no evidence of the work carried out by the Facility. As the Facility had little connection with the main project, which was complex and always required all the time available to supervision missions, NMEF received little attention from those missions, at least after the start-up period. 5.14 There was a major and well conceived and executed training program which sent 22 candidates for training at the masters level and over 30 short-term trainees abroad. Unfortunately six of the masters candidates did not return to Somalia, but this problem was solved for agricultural fellowships, by requiring in-country thesis work supervised and examined in Somalia. All participating project units benefitted, but the main beneficiary was the agricultural research unit with 16 masters candidates (of which five did not return) and about 20 short-term trainees. 5.15 The vast technical assistance component was of variable quality.M/ In at least three instances, the original selections proved weak and were replaced by more satisfactory candidates. There was a feeling by some observers that the project had been overloaded with consultants, but in most cases the TA provided specialists that Somalia did not have at the start of the project (for instance for roads and water supplies' construction, veterinary services and agricultural research). The relationship between PMU and consultants seems to have been good on the whole, and the PD tried hard to retain some key foreign staff beyond the Closing Date. The several resource and socio-economic studies were an additional and important asset resulting from the technical assistance program. They gave comprehensive information about the physical and human resources, and development constraints of the region. Not all these reports could be found in the Bank's records, but it is to be hoped they can be located in the future, when they may be needed, as most of the information will remain valid. 5.16 With regard to production, in the course of field work in 1988, to prepare the PCR, the consultants found that data of the National Food Early Warning System showed that the area sown to sorghum had increased from 193,000 ha in 1982 to 300,000 in 1987, while production rose from 84,000 to 166,000 tons. Lacking project evaluation data, the consultant organized a "rapid" field study to identify the factors responsible for the area and production increases. They concluded that at least a 35 percent share of the increase was due to project infrastructure. Assuming further that the project had not had an impact on yields, the consultants (using 1982 yields of 435 kg/ha) estimated that over 15,000 tons 26/ The audit was not able to track the total amount of TA used by the project, but believes it exceeded the 100 consultant years projected at appraisal. - 36 - of the incremental 80,000 tons of sorghum had been brought about by the project (about 35,000 tons was attributed to favorable weather and 30,000 tons to other causes, mainly market liberalization). The government, in reviewing the consultants' findings, felt that 60 percent of the area increase was due to project impact, which would translate to 30,000 tons of grain. About 11,000 tons of increased grain was projected at appraisal for ten years after project start. 5.17 On the basis of their project impact assumptions and projections of benefits from adoption of phosphate fertilizers by farmers (based on existing research results), the consultants calculated the ERR to be only two percent. Even adopting the government's assumptions, the ERR is only five percent (compared to 20 percent calculated at appraisal). While benefits, project costs and project operating costs were similar in the SAR and consultant's projections, there was a big difference in on-farm cost. These were quite low in the SAR and very high in the PCR. In fact, the latter projection seems to indicate that although the project might bring about a major increase in agricultural production by technical improvements, production costs, predominantly labor, would be so high as to make such efforts unprofitable. On-farm costs in the PCR are probably on the high side. 5.18 The audit contends, however, that both sets of projections are based on speculations about activities that either cannot be verified or about practices that had not yet been fully tested, let alone adopted by farmers, and are therefore too uncertain for meaningful ERR projections. 5.19 The Bank's divisional management had indicated concern over the financial sustainability of the project and it is therefore surprising that more emphasis was not put on GOS compliance with covenants to conduct studies to ascertain appropriate water charges, or to initiate similar studies for updating charges for veterinary services. Equally, more concern should have been shown to measures to improve the maintenance of roads and water supplies. It is both surprising and regrettable that the Bank did not continue to support the development of the Bay Region, by a follow-up project, as was the intention expressed in the SAR and in the DCA project description. The files show that while the Bank supported funding for the feasibility study of a follow-up project, there was not a place for it in the Lending Program. Instead Bank staff advised encouraging other donors to support the Phase II project. Although the subsequent civil strife frustrated such efforts in any case, this was an unsatisfactory weak ending to Bank participation in this area. Semi-Mechanized Rainfed Farming Operation 5.20 The semi-mechanized operation at first turned a failed project into a technology showpiece. There was widespread enthusiasm for the concept. The management system served as a model to convert moribund large farms into enterprises in which centralized management gave services to participating farmers. The role of the participating farmers in making management decisions had not been satisfactorily established, but farmers' pressure was bringing about some changes. 5.21 The operation had pioneered management training to ensure continuity after the departure of consultants. The special one-year courses in Australia emphasized practical aspects of scheme management including farm machinery operations, repair and maintenance. However, there were conflicting reports about the long-term effectiveness of the training. The basis for Somali management of mechanized agriculture had been established but needed some reinforcement. It was regrettable that the termination of the services of the Australian consultants spelled the end of the specialized training courses that the consultants had organized in Australia. 5.22 Despite early enthusiasm, up to the time that follow-up project operations ended the "long fallow-strip cultivation system" had not been established as a financially viable or technically proven - 37 - dryland farming system in Somalia. Technical problems that had emerged at the outset of the trials, in particular low cost methods for effective weed control on fallows had not been found and the most appropriate rotation was being debated. In fact, with increasing crop failures even in the main rainy season, in years with low or poorly distributed rain, the effectiveness of fallowing was being questioned. Similarly, the capacity of local staff to manage this type of scheme effectively without expatriate assistance had not been clearly established. Cost recovery seemed a distant hope since most of a family's grain was needed for subsistence, leaving little, if any for sale. 6. BANK PERFORMANCE 6.1 Given later doubts about benefits of the Northwest projects, a query remains on whether project processing reviews were adequate. The post appraisal review by CPS of the first Northwest Project was sharp and vigorous. Criticisms of bund construction methods, high costs and moderate returns caused much debate and came close to forcing reappraisal. The high cost of bunding arose from their massive dimensions and mechanized construction. Bund design has been justified by: (a) resistance to trampling by large livestock herds, giving the bunds reasonable longevity and (b) the 1987 yield survey finding that yields were positively correlated with bund height. CPS had also suggested reversion to the use of oxen and light tractors. This was impractical on several grounds, including that there was no longer a sufficient number of oxen in the project area for such work. By contrast, review of the second project's preparation and appraisal documents were almost a formality. There was no reason to doubt the excellently written appraisal report. In addition, it takes strong nerves to oppose processing of the clone of a project consistently graded "1" for the past two or three implementation years. 6.2 There was a lack of resource data and a lack of proven technology to improve either crop or livestock production in Bay Region. Resource and socio-economic surveys should have been carried out prior to project preparation, but were presumably ruled out so that the project would not be delayed. In fact it took three years from the time that the preparation report was published until the project was signed, and there would have been time for the needed surveys without affecting the ultimate project approval date. Lack of resource data led the preparation mission to propose settlement on uncultivated land with assumed dryland cropping potential, which proved wrong, and the appraisal to propose the pilot agricultural development units which the subsequent resource survey found to be superfluous. Both preparation and appraisal reports overestimated the quality and quantity of the potable aquifer and underestimated borehole costs. 6.3 Could these problems have been avoided? With hindsight, CPS should have insisted on major design changes that were ultimately introduced at the mid-term review, and should have queried that the project rate of return from crops was based on improved farming methods that had neither been tested nor accepted by any farmers in the area. Reviewers should have challenged the feasibility of the PADU proposal. 6.4 Supervision by the Bank showed interesting facets in each case. Supervision of the first Northwest Project was good as far as frequency, time in the field and continuity are concerned. There were two flaws. The first that no engineer or soil conservation specialist participated after bunding got underway. Consequently an important bund design change (or construction fault) was undetected until supervision of the Phase II project started. The change or defect (the omission of water retaining wings) is the reason why most Phase I bunds are ineffective for moisture conservation. The second flaw was inability to achieve reliable impact evaluation. The PMU conducted annual evaluations as designed by - 38 - their consultants, starting with the first year of bunding. Supervisions commented and advised on these results but evidently did not manage to improve the quality of the analysis. These shortcomings were put right during the second phase which saw multi-sector supervisions that brought technical and evaluation defects to management's attention. Despite these determined efforts there was evidently little improvement of technical aspects, but a highly informative impact survey was conducted as a result of supervision mission pressure. 6.5 Bay Region project supervision went through four phases. The first with HQ staff supervising gave the impression of a fairly friendly and helpful approach. The project was deemed to be making quite good progress. Supervision was then switched to RMEA and became less congenial, mentioning the few achievements but emphasizing the many blemishes. The project was labelled a "problem project". The audit feels that most of the problems encountered stemmed from design faults, including complexity. A helpful rather than critical attitude to project management might have paid better dividends. However, the task manager must be credited with an excellent job in modifying the project through the mid-term review. This led into the third supervision phase, when implementation improved. There was a friendlier note in supervision reports, with project management given credit for progress and with implementation status raising the rating out of the problem project grade. 6.6 Unfortunately, the fourth stage, after the Bank's 1987 reorganization when supervision reverted to HQ, ushered in again a more chilly period with reports resembling the pre-MTR versions. Not surprisingly the project manager has expressed regret at task manager changes. Supervision frequency and time allotted was quite satisfactory and the mix of expertise during supervisions seems to have been satisfactory. 6.7 While supervisions for the semi-mechanized operation were satisfactory and perceptive when they took place, there were two serious lapses. The first lapse covered the entire bridging finance period. No records of project supervision were found for the period from mid-1982 to mid-1987. It almost seems that between closing of the Drought Rehabilitation Program Project and appraisal of the Semi-mechanized Rainfed Agriculture Pilot Project, there was no specific task manager for the operation. However, a greater lapse overhangs this prolonged experience from the beginning - the prolonged failure to evaluate the economic and financial (that is, cost recovery) implications of the proposed system. It seems that staff in Somalia (including the early consultants), and Bank staff, were so impressed with the technological potential being explored that they ignored cost and return aspects for far too long. 7. CONCLUSIONS, LESSONS AND FUTURE ACTIONS 7.1 Given the recent history of Somalia and the current lack of a government, it may be considered academic to discuss the usual ratings of project performance. The sustainability section offers some comments, but ultimately the main concern should be to see what project assets can be used or rehabilitated when operations are again possible in the country. However, for the record, at the PCR stage overall performance of the first Northwest project was rated satisfactory, although an economic rate of return was not calculated. In part with the benefit of hindsight from the experience of the follow-up project, it now seems that the overall performance should be rated unsatisfactory, given that the very small part of the bunded area (about one-sixth) producing benefits implies a low or negative rate of return (even in the absence of more recent calamities). Similarly the second project repeated the unsatisfactory outcome of the first one. - 39 - 7.2 At the PCR stage performance of the Bay Region project was rated unsatisfactory, mainly because of a low economic rate of return. Although latest events must confirm the low rate of return, the audit upgrades the performance of the project to satisfactory. Admittedly the original project design was flawed technically and over-complex, but the two stage plan with a mid-term review proved its worth when the objectives were reduced to more realistic levels. The switch to concentrating on infrastructure and technology development could have led to greater benefits if later chaotic events had not intervened. Moreover, even though the ERR is low, project-induced production increases (as last measured) were up to appraisal expectations. The rest of this section reviews the findings for each project, presents the overall lessons and makes suggestions for future activities. Conclusions: The Two Northwest Projects 7.3 As to results, the two projects brought undoubted benefits to the Northwest Region by way of water supplies and minor irrigation components, but these are difficult to quantify. However, the question of benefits from bunding is more complex. As has been explored in section H, the PMU consistently claimed high benefits but two independent surveys found benefits from moisture conservation confined to water harvesting, that is where "lowlands" receive water from adjoining slopes. In addition, all slopes that have been bunded have been protected from soil erosion, but the water retention function of most of the bunds constructed under the first project is negligible because bund wings were omitted - hence crop yields are not improved. Even so, wherever there is runoff, either on bunded fields or in areas adjoining them, minor works would restore the effectiveness of such bunds with resultant production benefits. Lastly, some bunds, but it is not known how many, have been constructed in areas with little slope and on pervious soils. Since there is insignificant runoff in such areas, and therefore little erosion hazard, these bunds will not give any benefits. 7.4 The benefits from soil conservation alone are difficult to measure since although available estimates of economic rates of return are between two to three percent. If such low rates of return lead to rejection of soil conservation proposals, substantial social costs could arise later since areas at risk would become permanently unsuitable for farming in 20-30 years. This suggests that these ERR estimates fail to take into account a "without project" production decline to zero, or just that the cost of protection is high. 7.5 Another debate that touches on bunding benefits is the reason for the continued popularity of bunds with farmers. The feasibility study consultants reported that farmers believe that bunding improves security of tenure. This is reinforced by lack of maintenance by farmers, their reluctance to rehabilitate old bunds, and the demand for bund construction even on flat land. However, there are also reports of farmers recognizing the agricultural benefits of bunding, and a combination of both factors seems to be at work. 7.6 The lack of success of the small supporting agricultural research effort to generate technology improvements worried reviewers at various stages of the project. It is difficult to cater for the research needs of relatively small isolated agricultural areas. While it is unlikely that one professional researcher can have much impact on current technology, a costly multi-disciplinary research effort is not justifiable. With the development of the National Dryland Farming Research Station at Bonka, the chances of findings from there being useful in the Northwest enhances the value of any local research effort in adapting such findings. Such a combination seems the most likely solution. - 40 - Conclusions: The Bay Region Project 7.7 The audit concludes that the project was worthwhile and brought considerable benefits to this harsh, relatively waterless and drought-prone area. Improved communications and water supplies no doubt brought the most tangible benefits, while the research effort was working towards realization of future benefits. Increased crop production projected at appraisal, but not really anticipated during implementation, proved a bonus, even though it was attributed to project infrastructure rather than to improved agricultural technology. 7.8 The audit considers that the large technical assistance component was essential. The project could not have been implemented without the technical assistance component. Of the 20 or so foreign experts (not counting T.A. staff for the NMEF), only one, the Technical Manager, was an advisor. All others, including seven funded by USAID and working under WDA on groundwater exploration and drilling, were employed because of the scarcity of local professional staff. However, at the time of project closure, with much training having been completed, it was uncertain how much technical assistance would be needed by a possible repeater project. 7.9 The audit considers the decision of the Bank not to finance a second phase project to have been regrettable - unless the impending disaster had been anticipated and there were to be no more projects for the time being in Somalia. The situation in the Bay Region favored further support: the PMU functioning effectively and roads, water supply research and animal health units were well established. Conclusions: The Semi-mechanized Rainfed Farming Operation 7.10 The potential for this type of agriculture in Somalia remains uncertain. At the time this pilot operation started, both government and Bank staff felt that there were large areas of Somalia that could be brought under more productive use by mechanized agriculture, particularly by introducing soil moisture conservation techniques which are widely used in semi-arid areas of other countries. Later soil surveys for the Semi-mechanized Rainfed Agriculture Pilot Project showed large areas of unsuitable saline or alkaline soils. Only 10,000 ha of good soils was identified. Such a limited area would support relatively few farmers and could account for only a fraction of the country's production, such that the costs of the subsequent trials and pilot project may not have been justified. But this question can remain open for the evaluation of the pilot project, in due course. 7.11 The original trial was triggered by the failure of rainfed crop production in the Drought Rehabilitation Program Project. It was believed that the failure was mainly due to the marginal and erratic rain of the area and that moisture conservation was the key to successful crop production. The main question now raised by critics of the operation was whether or not it could have been made viable on a large scale. There was certainly room for technology improvement, in particular for management of the fallows and rotation choice. 7.12 Critics have also noted that the operation required considerable amounts of foreign exchange for purchasing and operating the equipment. However, cultivation of the heavy clay soils of the Lower Shebelle Region is feasible only by tractor and the number of tractors needed (about 30 for the six farm units that were to have been in full operation at the end of the pilot project) would have been a small part of the existing national tractor fleet of over 600. 7.13 Similarly critics have felt that management of the operation was so complex as to require permanent technical assistance. It is not possible to come to a firm conclusion on this point, but - 41 - information from supervision reports indicates that the special training of Somali managers in Australia was effective. Nevertheless, the audit believes that the hands-on technical assistance at the field level may have been terminated too soon and the technical assistance provided from project headquarters was not an adequate substitute that early in the life of the pilot operation. 7.14 The most difficult point remaining is whether the technology was viable financially. During the pilot phase the mechanical services and inputs were provided to farmers without charge, such that there was adequate farmer response. Plot size per family was set at a level which was within the labor resources of typical families, but without allowing for the added complication of crop failure years. The more successful rotations just supplied sufficient production for the family's subsistence requirements. There was not a significant surplus for sale. Hence, the technologically viable system demonstrated by the trials implied a substantial and continuing subsidy and no possibility of cost recovery. It was not clear that the government was willing or able to make such a heavy commitment to this segment of the population. Lessons Learned 7.15 The main lessons learned are: (a) The audit confirms that making improvements to long established traditional agricultural production systems, especially in resource poor low rainfall areas, must start from a detailed understanding of the existing natural and human resources, of the current production systems (livestock and crops), and with the full participation of the affected populations. (b) Constant monitoring of progress and periodic re-evaluation of objectives and potential impact are necessary to ensure that original concepts are still sound and to point out any needed changes. (c) On the semi-mechanized farming activities, questions remain about whether it is prudent to pursue the resolution of technological problems associated with the introduction of a foreign technology (to produce staple crops in a poor country) without reference to financial, cost recovery, skill intensity and foreign exchange scarcity issues which are likely to affect replicability and sustainability prospects. The reverse of the conclusion above that the success of the early activities was due to exceptional "hands-on" consultant input (which could not be reproduced by the second group of consultants) is whether the technology is too skill- intensive for general adoption in Somalia. (d) Some other important points include: that cofinancing should be arranged early on in the processing cycle to avoid long delays which almost called for reappraisal in two of the projects; innovative project designs need especially skilled and perceptive review at appraisal to identify design errors; and, technology that has yet to be adopted by farmers is suitable for inclusion in a testing program, but should not be used as a basis for projecting benefits. Suggestions for Future Activities 7.16 When the time comes to resume lending operations, revisiting both the Northwest and Bay project areas would be justified, in the first place to ascertain the status of farming and of project - 42 - investments in the area. In the Northwest, in those areas where bunds can increase production simple surveys of bunded areas may be desirable to identify minor works to improve effectiveness. The promising trials initiated by consultants in the last year to generate runoff from interbund land to enhance yields on a cultivated strip are worth continuing. New trials may be worthwhile to utilize the bunds themselves for more productive purposes, such as hay production to help alleviate the chronic dry season animal fodder shortage. 7.17 In the Bay Region the completed project had sufficiently encouraging results to warrant a reconnaissance to see what remains of the works and services established by the project, and what can be done to rehabilitate at least regional water supplies, roads and animal health services. 7.18 The semi-mechanized operation presents something of a dilemma. On the one hand, there is the undisbursed balance of US$9.8 million of the Semi-mechanized Rainfed Agricultural Pilot Project; on the other hand, there must be many more urgent rehabilitation tasks than reactivating the project. The project may yet bring significant benefits in the long run, but would assist only a few hundred people in the short-term. If, as the audit visualizes, the equipment and most infrastructure has been destroyed and staff and settlers scattered, rehabilitating the project would be difficult, slow and costly. The best alternative may be to ask whether the balance of funds can be used for higher priority agricultural rehabilitation work. 7.19 As a result of the operations reviewed in this report, several resource and other studies have been conducted which remain valid and useful. Unfortunately, not all of them, or all parts of them, could be found in the Bank's files or Archives. They may not be available either in Somalia. Losing this information would be unfortunate, but copies may still be available from the original sources. In preparation for resumed operations, the Bank should obtain at least one copy of each of the studies for storage by the information system. This should include all studies by Hunting Technical Services and by Louis Berger International on the Bay Region; socio-economic studies and Bonka Station Research reports by Wyoming University; the Bay Region Phase II Feasibility Study by Sir M. Macdonald and Partners, Limited; the Northwest II feasibility study by SOGREAH and the Northwest watershed protection studies by TAMS. There are also aerial survey maps and photographs used for these studies. For the South Shebelle Settlement Areas there are project feasibility studies for the Semi-mechanized Rainfed Agriculture Pilot Project by Bingle and other studies of that area by Hunting Technical Services. IBRD 2513A ErI REP.OFYEMEN DJIBOUTI DJIBOUTI * Loyada Condal /eI Maed t Adado Bosaso Matt Los Koreh Berbera E NORTHWEST REGION -oma Scusuan ADP I -nd ADP 11 ýeybbt Tag Walaae< Týg Buroo G.rd - geis Bender Bile Sololjeh 'o Duroks \las Anod SOMALIA ... F Boho wn BAY REGION ADP NORTHWEST REGION ADP, AND -- AGRICULTURAL SETTLEMENT AREAS PROJECT LOCATIONS BITUMINOUS SURFACED ROADS GRAVEL / EARTH SURFACED ROADS MAJOR PORTS Go oyo DOMESTIC AIRFIELDS INTERNATIONAL AIRPORTS RIVERS DISTRICT BOUNDARIES' REGION BOUNDARIES* - -- INTERNATIONAL BOUNDARIES -Dist,c and region bondorre s of 1984 Obb,o Fe, Fe, eleEl B r Hodd . hu Walid'd Gona -0 50 10C 1,50 200 250 300 - KlLOMETERS "arbe owa Adole Afgo' MOG D1iHU MOGADISHU A R E Go en Jn~ale ¯5¯ A U D l A R A B l A 5 -.Merca < jg 1ý ýk"giS/AREY AGRICULTURAL SETTLEMENT S U D A ' SA8BLALE AGRJCULTURAL SET TLEMENT Brova E1RITREA Z REP. OF YEMEN uJ Beles bo' Gogar, DAsou- Thi. rn.P h.. le p'.P.r-d by Th. Word Bon.I I TI P 7 Ja e iely for th. oneniene E T H P A Arara of rdr nd i. for th. Kisinaye inwrn.1 e of The World Bonk Group. The denominations used "SóMALIA ond th. boundori.. shown 0n thi. mp do not imply, on Uh D part of The World Bonk Grop, Mogodishu anyr dgent on th. lega KENYA ny terrioy or ooy Bedode ndorsment or acceptance of uch boundories TANZANIA NOVEMBER 1993 IBRD 2513l ERITREA 44. REP.OFYEMEN DJI BOUTI Bereda 12* DJIBOUTI *Loyada CardaI Adado Bosasoc /ZeilaLos Koreh Ma't Berbera Erigavo - lg I D0.a Scusciuban NORfTHWEST REGION -rama ADP and ADP I - ey Tug W.aloe*BroG ges . . - Bender Bell e. Amnob DOM L s \ Los Anod S OM A LIA Boso eW' BAY REGION ADP, NORTHWEST REGION ADP, AND AGRICULTURAL SETTLEMENT AREAS PROJECT LOCATIONS BITUMINOUS SURFACED ROADS GRAVEL / EARTH SURFACED ROADS MAJOR PORTS GaNoyo DOMESTIC AIRFIELDS INTERNATIONAL AIRPORTS RIVERS DISTRICT BOUNDARIES' O REGION BOUNDARIES' - - INTERNATIONAL BOUNDARIES -Ditct and regon boundores s of 1984 D,sa Mo,ebOb -Obb,o .Fer Fe, ele El BIjr ... Hoddur . ugk WoG dui Gonano 0 50 100 150 200 250 300 KLOMETERS od Afgo' MOGADISHU MOGADISHU A R E ---anaeS AUDI A R ABI A L. enmerca NhJ AREYAGRICULTURAL SETTLEMENT SCDMAN SABLALE AGRiCULTURAL SETTLEMENT S D A ~ITREA REP OF YEMEN uaBeles Thi, mp has b- p-pct-d IyT.Wrd Iah StflfH O Jaa e .cl,ively for th. connitn E T H 1 P A A,rr of-d-s and is for th (K,b-yeinrl .s o- Th. World Bank Group. The d-.,naions und ..dMAA Gnd th. bound.ri,ho..n on thi, map do no imvply on h. UGANDA part of Th. World B.nk Grop, Mogodishu ony jud mnt n th. legal K EN YA status Oo n territory ot any Baad do'somn,.n or occeptonc. of - such boundaries. TANZANIA NOVEMBER 1993 IBRD 2513 43-00* 4330' 4iO0 SAUDI ARABIA j --OMAN SOMALIAr \MA B'AY REGION AGRICULTURAL DEVELOPMENT PROJECT - PHYSIOGRAPHY AND PROJECT INFRASTRUCTURE DEVELOPMENT EXISTING ROADS: - CLASS I BOREHOLES E CLASS II HIGH YIELD LIVESTOCK DISPENSARIES - - - CLASS III MODERATE YIELD . 5. IMPROVEMENTS UNDER THE PROJECT (SEPT '86) LOW YIELD PROPOSED IMPROVEMENTS KENYA I . (SEPT '86, BELIEVED COMPLETED) / aVff LIMESTONE PLATEAU o- ANZAN I LIMESTONE DEPRESSION BAKOO*REGION 7 BASEMENT COMPLEX :BAY REGION \ - To CULTVABLE LAND IDENTIFIED PROBABLE CULTIVABLE LAND VO MAJOR TOWNS 0 OTHER TOWNS ESCARPMENT RIVER\\/ DISTRICT BOUNDARIES laI Dole KILOMETE S -O RGONBONARE iat -- N ERNA 10fLIOUN ARIE 5o 100 Bnk .. BA D - o 30044' N E E1 B., AkoW 7-' NOVEMBER 1993 IBRD 2511 430 43030 - 444 Gull of Aden BERBERA ~I -lBo n D rdu_r -10~~~~ _.tCd.L,r Bod, Y D dýbr> W ~., 1,,c].I HDrhale S OM A ll A ..1X-nba~eyn NORTHWEST REGION AGRICULTURAL ° . S DEVELOPMENT PROJECT-PHASEII .G - ' bl Dae Project Area GEB.Sbad Dhu PHASE IPROJECT AREA T -Xdhn Bandar.H,ns PHASE I PROJ1ECT AREAS xdh n'a HARGEI A Wanaa POTENTIAL IRRIGATION ASEAS 0 c] IRRIGATED FARMS (PHASE I) E T H I 0 P 1A HUMAN WATER POINTS AERIAL SURVEY COVERAGE / ASPHALT ROADS MAIN TRACKS RIVERS INTERNATIONAL BOUNDARIES OQ51 J1bley 2 2~ MOGADISHU 0 5 10 15 20 25 301 KILOMETERS .Bla ano OBal, K-, 430 4330 440 44030 OGIOER T 993
World Bank Group · Project Performance Assessment Report
Somalia - Bay and Northwest Regions Agricultural Development Projects
View original document
The full text is hosted by the publishing organisation. lawenc.com indexes the metadata and links to the official source.
Full text
Key facts
Organisation
World Bank Group
Document type
Project Performance Assessment Report
Country
Somalia
Source
World Bank