Mozambique - Agricultural Services Rehabilitation and Development Project

Document of The World Banak FOR OFFICIAL USE ONLY Report No: 22445 IMPLEMENTATION COMPLETION REPORT (IDA-23370) ONA CREDIT IN THE AMOUNT OF US$35 MILLION TO THE MOZAMBIQUE FOR A AGRICULTURAL SERVICES REHABILITATION AND DEVELOPMENT PROJECT June 12, 2001 Rural Development Operations Eastern and Southern Africa Africa Region This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.  CURRENCY EQUIVALENTS (Exchange Rate Effective ) Currency Unit = Mozambique Metical At appraisal Mt. 1,450 = US$ 1 US$ 1 = Mt. 16,600 At Completion FISCAL YEAR January 1 December 31 ABBREVIATIONS AND ACRONYMS ARDP Agricultural Rehabilitation and Development Project ASRDP Agricultural Services Rehabilitation and Development Project BPD People's Development Bank DINAGECA National Directorate for Geography and Cadaster DNER National Directorate of Rural Extension FM/FI Farmer-Managed/Farmer-Inplemented FSR Farming Systems Research INIA National Institute for Agricultural Research IPM Integrated Pest Management JVC Joint-Venture Company M&E Monitoring and Evaluation MADER Ministry of Agriculture and Rural Development PROAGRI Agricultural Sector Public Expenditure Program QAG Quality Assurance Group RRP Rural Rehabilitation Project T&V Training and Visit Extension Vice President: Callisto Madavo Country Manager/Director: Darius Mans Sector Manager/Director: Sushma Ganguly Task Team Leader/Task Manager: Miriam Brandao/David Nielson  FOR OFFICLAL USE ONLY MOZAMBIQUE AGRICULTURAL SERVICES REHABILITATION AND DEVELOPMENT PROJECT CONTENTS Page No. 1. Project Data 1 2. Principal Performance Ratings 1 3. Assessment of Development Objective and Design, and of Quality at Entry 2 4. Achievement of Objective and Outputs 5 5. Major Factors Affecting Implementation and Outcome 12 6. Sustainability 13 7. Bank and Borrower Performance 14 8. Lessons Learned 15 9. Partner Comments 16 10. Additional Information 17 Annexes Annex 1. Key Performance Indicators/Log Frame Matrix 17 Annex 2. Project Costs and Financing 17 Annex 3. Economic Costs and Benefits 18 Annex 4. Bank Inputs 19 Annex 5. Ratings for Achievement of Objectives/Outputs of Components 20 Annex 6. Ratings of Bank and Borrower Performance 21 Annex 7. List of Supporting Documnents 22 Annex 8. Republic of Mozambique Implementation Completion Report 27 MAP (S) Map of Mozambique 42 This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not be otherwise disclosed without World Bank authorization.   1. PROJECT DATA Report Date: March 20, 2001 Name: AGRIC. SERVICES REHABILITATION AND DEVELOPMENT Country/Department: MOZAMBIQUE/AFTRI L/C Number: Cr. 2337-MOZ Sector/Subsector: Agriculture/Services Region: Africa KEY DATES Activity Original Revised Appraisal June/July, 1992 Approval 11 February, 1992 Effectiveness 15 December, 1992 MTR December, 1997 No formal MTR, project restructured 1997 Closing December, 2000 June, 2000 Borrower Government of Mozamlbique Implementing Agency Ministry of Agriculture and Fisheries (now Ministry of Agriculture and Rural Development, - MADER) Staff Current At Appraisal Vice President C. Madavo E. Jaycox Country Manager D. Mans S. Denning Sector Manager S. Ganguly C. Helman Team Leader of ICR D. Nielson ICR Primary Author FAO/CP Team Leader of Appraisal - G. Keynan The project was appraised in July 1991, approved in February 1992 and signed on 30 March 1992. The original closing date was 31 December 2000. The actual closing date was earlier by six months (30 June 2000) because by then the Agricultural Sector Public Expenditure Program (PROAGRI) had become effective and was financing the same activities. This is a core accountability ICR. 2. PRINCIPAL PERFORMANCE RATINGS Outcome Unsatisfactory Sustainability Likely (under P'ROAGRI) Institutional Development Impact Modest Bank Performance Unsatisfactory Borrower Performance Unsatisfactory QAG ICR Quality at Entry Poor Poor Project at Risk at any time: No No  3. ASSESSMENT OF DEVELOPMENT OBJECTIVES, DESIGN AND QUALITY AT ENTRY 3.1 Original Objective: The main objective of the Agricultural Services Rehabilitation and Development Project (ASRDP), as stated in the SAR, was to increase production and returns from foodcrop and cotton cultivation in the family subsector through the rehabilitation and development of effective agricultural services and the strengthening of institutional capacity in the Provinces of Nampula and Cabo Delgado. It was estimated at appraisal that some 140,000 farm families would be reached directly by the supported agricultural services. This was expected to lead to increased yields and average annual increase in production on the order of 23,000 tons for seed cotton, 38,000 tons for maize and 7,600 tons for beans. The objective was clear and continues to be important to the country. The objective was consistent with the Bank's Country Assistance Strategy (CAS) and with the government's attempt to develop and disseminate improved farming practices aimed at increasing agricultural productivity and rural household incomes. Although food security today in Mozambique is less problematic than in 1992, yield levels of food and cash crops remain low, and the objective of reduced poverty and increased incomes in rural areas still drives the government's strategy and actions in rural areas throughout the country. 3.2 Revised Objective: The objectives were not revised during implementation. 3.3 Original Components: The project objective was to be achieved through the following components: * upgraded agricultural services such as extension and applied research; = working capital and investment credit for agricultural production and marketing; - rural water supply facilities; * land use and management services; and * institutional development including training of personnel of MADER. 3.4 Revised Components: The project was officially re-structured in 1997. This re-structuring was motivated largely by a desire to rationalise and consolidate the various World Bank instruments of support to the rural sector in Mozambique and was only to a lesser extent driven by concerns with ASRDP in and of itself. The restructuring involved changes to ASRDP, but also to the Rural Rehabilitation Project (RRP) and, in addition, included the cancellation of the Agricultural Rehabilitation and Development Project (ARDP). The restructuring aimed at focusing IDA support on the priority needs in the sector at the time and aligning implementation responsibilities with the country's new institutional framework. ASRDP and RRP were designed during the civil war largely as area- based operations focusing, by necessity, on safe and accessible rural areas. They included similar components (land management and rural water supply), but operated in different areas of the country. In practice they were only loosely co-ordinated, reflecting the lack of a coherent policy framework and the fragmented institutional set-up that existed at the time of preparation. Political 2  stability since the signing of the peace accord in October, 1992 permitted the Government to begin a shift from emergency operations toward the implementation of a longer-term rural development strategy. Slow credit disbursements under ASRDP prior to 1996/97 made it clear that expectations had been optimistic at appraisal regarding the speed at which it would be possible to re-establish effective research and extension institutions in Mozambique as well as the private sector's demand for credit and the ability/willingness of financial institutions to meet actual demand"'. The restructuring of IDA-financed operations in the agricultural sector included a redirection of available resources to activities which would become part of the Agricultural Sector Public Expenditure Program (PROAGRI), then under preparation, and a streamlining of the activities which were being financed under the ASRDP, the RRP and the ARDP. The restructuring involved: the transfer of agricultural services financed under ARDP (agricultural extension and cashew rehabilitation activities) to ASRDP; the transfer of Rural Water Supply Component from ASRDP to RRP; the-cancellation of the remaining balance of ARDP; the cancellation of the Credit Component under ASRDP; and the transfer to ASRDP of studies and surveys which had originally been included under the Land component of the RRP. The revised total IDA financing for ASRDP was US$ 25.7 million against US$ 35.0 million at appraisal. The restructured ASRDP was to have expanded geographic coverage (adding Gaza and Inhambane, the two provinces of ARDP) and was to focus primarily on agricultural extension, applied research, and land. Further, it was to go forward under a now-unified management. The ASRDP extended its coverage to Gaza and Inhambane Provinces in February and August 1999 respectively. The project was designed to have bi-annual and mid-term reviews. These should have provided opportunities for lessons learned (under ASRDP, but also in agricultural services programs in other countries) to be taken into account over the course of a flexible and evolving implementation of the Program. The technical reviews were not undertaken and consequently, opportunities to review and modify approaches may have been missed - this may have been particularly evident with regard to the government extension service. The mid-term review was also not undertaken per se, rather it was subsumed in the project's restructuring. Consequently, here too an opportunity was missed to analyze past performance, to address the issues impeding implementation and impact, and to re-direct the technical content of project's components. At the time of restructuring, the project's main development objective remained basically unchanged: "rebuilding national research and extension institutions able to help smallholders increase productivity and incomes". 3.5 Quality at Entry: From a theoretical point of view, the components were reasonably related to achieving the project objective. An exception to this might be the Rural Water Supply Component which is not directly related to the specified objectives of increasing the production and returns from food crops and cotton. At the time of project design the country was still engulfed in civil war and the basic infrastructure was dilapidated. Implementation capacity and analytical capability at the national, provincial, district and village levels were inadequate, making it difficult to achieve impact in a short time, and so the project implementation period was relatively long (eight years). The ICR mission supports the QAG supervision assessment which stated that: "The project design was far too complex for a country emerging out of civil war and with limited implementation capacity". The number of project components, the number of implementing Bank memorandum with the proposed amendments to the Development Credit Agreements of ASRDP and the Rural Rehabilitation Project (RRP) dated 7 May 1997. 3  agencies involved and their weak management and implementation capacity, and the range of policy and institutional improvements required for effective implementation, made the design of the program too complex and risky. Poor salaries of government staff, weak financial systems, and inadequate and tardy counterpart funding were problems prevalent at the time of project appraisal and continue to persist today. The underlying assumption that support for human resource development and training in management, and the formation of a relatively highly-paid project implementation unit, would generate adequate commitment to efficient and effective implementation of the project components was optimistic. With regard to the research and extension components, the project aimed at the establishment of a (modified) Training and Visit (T&V) extension system and a Farming Systems Research (FSR) program. This was following a UNDP/FAO project during the period 1981 to 1990 in which the extension services of the then state cotton companies had been supported in establishing a system similar to T&V in Nampula and Cabo Delgado Provinces. In retrospect, it should have been clearer to the preparation team that the investment and recurrent costs of these systems were high, and that this would place a high constraint on sustainability. Also, the achievement of the main objective of the FSR approach, to instil a farming systems perspective in on-station and on-farm researchers and extensionists, required a major attitudinal change within the research/extension system. A decentralisation of both research and extension beyond what was proposed under the project should probably have been explored as pre-requisites of achieving such changes. The project design also worked against decentralization of the extension services in two ways. First, because it placed most of the control over the project resources in the hands of the provincial extension supervisor. Two key players were not provided with either the means or authority over the project: the provincial and district agriculture directors. As a result, they had no incentive to become involved with the extension services and participate in the project's objectives and activities. Second, it created a layer of management (in the person of the component coordinator) between the provincial level and the project's implementation unit (PIU). The component coordinators (usually the director of each service) were the final arbiters over each component's activities and expenditures, and the PIU's role was to follow through on their decisions. For a number of reasons, activity plans and budgets requested by the provincial level were often altered at the central level (primarily to adjust to counterpart funding shortfalls) and not discussed with the provincial level. When adequate funds failed to arrive, the provincial level blamed the PIU which in turn blamed the component coordinators. This distrust between the central and provincial levels and the PIU was apparent throughout the life of the project. The proposed Unified Extension Service, with polyvalent Field Extension Workers, was consistent with the farming systems perspective underlying the agricultural research programs. However, it may have been unrealistic to expect that it would be feasible given the skill-levels prevalent among extensionists at the time. The amount of training required to turn commodity-focused extensionists, who had at most an elementary -level education, into polyvalent extensionists with a farming systems perspective, was underestimated in the project design. The QAG supervision assessment questions the soundness of project design on the grounds of its complexity. The ICR also rates the quality at entry unsatisfactory. The project objective was and remains consistent with government priorities and the Bank's CAS, but there were deficiencies in the design. When the extension system was established (with the project's support) the project staff, particularly at the field level, had limited formal training. It was thus unrealistic to adopt an extension approach which demanded more independent and creative work at that level (as required by the FSR approach). It is only now, eight years later, after MADER has gained some experience and there are better trained staff available, that there exists the basis to embark on more complex paradigms. 4  It was also unrealistic to have expected the project to have acted be able to accomplish attitudinal change in the institutions themselves. First, changes in attitude depended on MADER's political will and institutional arrangements, areas which are somewhat beyond the scope of a single project. It was through regular, hands-on supervision missions, which made every effort to interact with the involved staff, project beneficiaries and other stakeholders, that field evidence was taken as the basis for evaluations, mid-term corrections, and the incorporation of new initiatives. The adoption of this methodology by the extension managers was expected to lead to attitudinal and institutional change. Yet, it is obvious that supervision missions had to be limited in number and duration and could not serve as a substitute for changing the incentive structures within MADER's internal processes. Last, it was unreasonable to assume that aLn enabling environment (markets, inputs, financial price incentives, roads, etc.) for the adoption of the extended technology would be established during the life of the project. At the time of approval of the project, the participation of the cotton joint venture companies in managing extension on a pilot basis, and of the financial intermediaries in managing the credit component, had not been obtained, and consequently these activities were not ready for implementation. Despite efforts to establish participation agreements with companies/banks, no agreements were reached during the project. The availability of credit was correctly seen as crucial to enable the adoption of technologies using purchased inputs, and the failure of this component and its subsequent cancellation, while justified under the circumstances, removed a critical piece of the project's original logic. 4. ACHIEVEMENT OF OBJECTIVES AND OUTPUTS 4.1 Outcome/Achievement of Objectives: A key assumption underlying ASRDP was that support to the rehabilitation and development of effective agricultural research, extension and land use management would increase the efficiency and effectiveness of these services. This in turn would result in increased adoption of improved farming practices, which would lead to higher on-farm productivity and higher incomes for rural households. The assessment of outputs by components which follows will show that, by the end of the project, whilst progress was made in establishing the extension networks and FSR teams, efficiency and effectiveness, or value for money and impact, of the government agricultural services were not appreciably improved. The objective of strengthening the institutional capacity of MADER has been achieved to the extent that a system for management of extension services is now in place. Improvements in the delivery of research and extension services were certainly realized in the project districts. Training at various levels combined with the provision of transport facilities, operating budgets and travel allowances, contributed to some increase in adoption of improved farming practices. However, other aspects of capacity building have not been achieved. The project suffered from poor financial management, delays in disbursement of project funds, late procurement and delivery of goods, delayed and incomplete civil works, misappropriation of funds and a lack of an effective M&E system. The capacity to analyze and to react to farmers' constraints and opportunities remains limited - perhaps because little progress has been made towards decentralization. With regard to the expected increase in agricultural productivity and rural incomes, the ICR mission could not come to a firm conclusion. There is consistent evidence that in areas influenced by the project (i.e., Nampula, Cabo Delgado, and Gaza), farmers have widely adopted some of the varieties recommended by the research-extension system. These include maize 5  (Matuba and Manica varieties), cowpea (Variety IT] 8), and high-producing cassava and sweet-potato. But, since data on impact are not available, these cannot be quantified. Despite the specific activities in the project's design and assurances obtained at negotiation, and the continuous references to this problem (i.e. absence of meaningful M&E and impact assessment systems) in almost every aide-memoire and IDA supervision report, little effective action was 2 taken during the project to redress the situation . The QAG supervision assessment refers to "some degree of misplaced optimism in the rating of the project's development impact". Supervision missions, through conversations with farmers, suggested that extension had achieved impact in terms of increased productivity and production levels. However, such anecdotal evidence cannot suffice as objective substantiation of the positive impact of the project. Evidence obtained through talking to a relatively small proportion of relevant farmers in the presence of an often large group of government and donor representatives, can lead to inconclusive and misleading information. Since little reliable, objective, and systematic measurement of yield levels took place, development impact cannot be conclusively evaluated. As was expressed by the last supervision mission, although the growth rate of the agricultural sector has been extremely high since 1 992 (average of 14% per year) and marketed agricultural output from smallholders has increased significantly, there are strong indications that this growth is due to an expansion of the production factors of land and labor, and not to increases in productivity. Food security has improved, but it is far from clear that the per capita income of rural people has increased or that the objective to reduce poverty has been met-. 4.2 Output by Components: Extension This component supported the rehabilitation and development of a T&V-based, unified extension service in 12 districts - 6 in Nampula and 6 in Cabo Delgado Province. In the SAR it is estimated that, by the end of the eighth project year, the extension service would have reached directly about 140,000 farming families who, together, would be cultivating some 300,000 hectares (225,000 hectares of food crops and 75,000 hectares of cotton). The expectations placed on the extensionists were high. They were the key persons on whom the T&V system relied to achieve impact at farm level, and were assumed to take on a polyvalent role which implied that they had to master a farming systems perspective. They had to be prepared to provide advisory services in the production of food, cash crops, and livestock as well as in the management of farm enterprises. It was unrealistic to assume that, starting from such a low level of education and experience, the training provided by the project would be adequate to internalize such a farming systems perspective in such a short time. For resource-poor farmers there are inevitable compromises in crop and livestock management. The extension service was not (and still is not) equipped to provide reliable advice on how to handle these inevitable trade-offs in farm management. The extensionists increased contacts with farmers and undertook demonstrations, on- farm trials and field days. A beneficiary assessment conducted in 1997 indicated that in two selected districts in Nampula Province, 15-20% of the households were in direct contact with extensionists (the target in the SAR was set at 50%). In the beneficiary assessment, no adoption rates were established nor were crop samples taken to assess impact on productivity levels. According to this assessment, farmers complained that no new extension messages were forthcoming. This lack of new extension messages might have been due to insufficient 'on-the- shelf technology' that is relevant to smallholders and that has successfully passed the stage of 2 Supervision Mission Report, April 1999. Supervision Mission Report, February-March 2000. 6  Farmer-Managed/Farmer-Implemented (FMFI) on-farm trials. In the whole of Nampula Province a total of 137 extensionists now work regularly with some 23,000 families in 7 districts - roughly 4% of the farm households in the province4. About 17,600 families in 4 districts were reached through the extension system in Cabo Delgado in the season 1997/985. Thus, in the two provinces a total of 40,000 households in 11 districts are in direct contact with extensionists (target set in the SAR was 140,000 in 12 districts). The little data available at district and provincial levels suggest that over the project life no increases in productivity took place6. This is supported by the observation by the supervision mission in February 2000 that "extension and research activities continue to be carried out in routine fashion, without a strong emphasis on achieving impact at the level of the farmer"7. Included in plans for the extension component was a pilot activity to contract-out the management and delivery of extension services to a Joint-Venture Company (JVC). During the first project year a contract was to be signed with a JVC. Cotton extensionists in the pilot area would be trained to become polyvalent extension workers in a period of 2-4 years in order to generate alternative extension delivery mechanisms. Attempts to implement these pilot activities in alternative delivery mechanisms for extension continued throughout the project but without success. One possible explanation is that MADER and DNER resisted the proposals as they feared that the arrangements would diminish their own power and control over funds. Perhaps more importantly, both the public and the private sectors were experiencing a market economy for the first time and, given the context, neither was really prepared for the kind of "advanced" partnership which had been envisioned in the pilots. Although some informal, short-term collaboration took place, attempts to contract a cotton company to provide extension on food crops were unsuccessful. Commodity-focused, single-crop extension is not easily re-oriented to incorporate food crops, because of the overriding interest in profits from the specific commodity. Moreover, the conventional task of commodity 8 extensionists is to provide production inputs to farmers and recipe-type advice for their use . Extensionists previously specializing in cotton had to become generalists and internalize the farming systems perspective. The lesson here is that the establishing of targets in projects, particularly those of an institutional nature, deserves a proper assessment of local reality and cannot be done through a direct transfer from other parts of the world. Also, at the time, even in other African countries with longer experience in public provision of extension services there is not much experience in outsourcing. Thus, it was difficult to identify experiences to be visited when DNER was planning a Bank-recommended study tour. Various supervision missions recommended the (re-)creation of the position of Extension-Research Liaison Officers (ERLOs) at the provincial level. DNER did not view this as necessary and no ERLOs were created in either Nampula nor Cabo Delgado provinces. The total number of vehicles (59), motorcycles (88) and bicycles (620) procured and distributed to extension staff could only be established for the period 1992-1998 because the M&E system installed in the project implementation unit (PIU) and quarterly project progress reports were discontinued. Originally, the PIU maintained a database of basic input indicators as well as financial and procurement information. However, with the mainstreaming of the unit into the directorate of economics of MADER in 1999, the collection of these types of indicators was discontinued. 4 DNER, Quarterly Report April-June 2000. 5 Supervision Mission report, April-June 1998. 6 The reliability of the data on acreage cultivated and total production could not be established - these data are based on estimates provided by extensionists. 7 Supervision Mission Report, February-March 2000. 8 Supervision Mission Report, October-November 1998. 7  Under this component, the project had some achievements - namely the establishment and operation of basic extension program in the projects provinces, and some limited impact of the established program. However, the program has still achieved only limited coverage, even in the provinces where it started. In those areas where it has operated, impact appears to have been only marginal. The program has persisted in clinging to methodologies and institutional frameworks which were not performing well - and this despite the emergence of alternative methodologies and approaches even within other programs in Mozambique. Therefore, the extension component is rated as unsatisfactory. Research The research component was intended to provide for the establishment of modest research capacity in cotton breeding, food crops, IPM and FSR. In addition, basic seed production and longer-term agronomic studies on soil and water conservation, soil fertility and farm economics were to be initiated. INIA's FSR program was to focus on maintenance of soil fertility and the rational use of inputs. Multi-disciplinary FSR teams of agronomists, economists and sociologists were to be established, and training and technical assistance provided. The SAR indicated that: "The project placed emphasis on the farming systems approach, integrated pest management, environmental awareness and preservation, and farmers organization, as the basis for selecting the topics of [training] courses, workshops and seminars". The impact of this component of the project has been positive, but slow and uneven. Some cotton breeding work at Namialo has been done, as well as some seed multiplication (in the season 1999/2000 planting seed for only 140 ha was produced). Progress with the IPM program was slow, in part affected by the failure of the contractor to complete civil works at Namialo, including the proposed entomology laboratory, the equipment for which remains boxed and unused. The FSR work in Nampula did not get started until 1998 - although it is a positive sign that this work which was launched through long-term technical assistance is now being carried forward by two Mozambican agronomists who are staff members of MADER. Two more FSR teams (based in the Lichinga and Sussundenga agricultural stations) are currently operating in Mozambique as a result of the project's influence. Before the project, there were no university- trained staff based in the provinces. The FSR team in Nampula have embarked enthusiastically on some interesting research work, but the scope of their mandate area is huge and well-beyond their current capacity to deliver. In all areas of research, productivity has been severely hampered by the fact that requests for computers, FSR literature and funds for training of extensionists have repeatedly gone unmet. INIA has not yet succeeded in the establishment of a social science capacity (two economists occupy management positions), and effective FSR work cannot be implemented without holistic social science input. This lack of social scientists is explained by the very limited availability of these professionals in Mozambique (economists prefer urban assignments and a Sociology Faculty was only created at Eduardo Mondlane University (UEM) two years ago). Extension specialists, adult education experts, agro-economists and rural sociologists are needed in FSR work and in the strengthening of farmer organizations, but they have not been available locally and MADER has been unable to hire new staff (hiring restrictions apply in almost all public institutions). Thus, the strong emphasis in the SAR on FSR, IPM and soil fertility management and conservation has not been realized in visible ways in the project area. Research protocols of on-station and on-farm experiments, which if kept would reveal the thrust, and quality of the research being undertaken, are inaccessible (or do not exist?). It should be noted that, starting in 1996, the project has played a major role in triggering the beginnings of institutional reform (and subsequent decentralization) of the research system. While progress has been slow, it has been obvious for some time that no "quick-fixes" were possible. Respecting MADER's own pace has resulted in a much longer preparatory phase 8  than anticipated at appraisal. Yet, in terms of PROAGRI, the agricultural research component is considered to be one of the more advanced areas of the Ministry in terms of institutional reform and will be treated as a pilot by MADER's (recently-created) Reform Unit. At the time of project restructuring, the cashew research activities started under ARDP were transferred to ASRDP. These included the completion of civil works at research and cashew multiplication stations, continuation of work on selection for oidium resistance through vegetative propagation and the production of grafted plants for distribution. The research work has made progress in oidium resistance and in improving the grafting success rate, but the numbers of new plants for distribution still falls short of the 1.5 million p.a. target (it is estimated that around 650,000 trees were planted last year). Many of the civil works were not completed. In general, the research agenda supported under the project has not been significantly driven by clients, and no systematic mechanisms have been built into the system to ensure that research is oriented to farmers' needs. Much of the or-farm experimentation supported under the project was simply replication of on-station work and researchers failed to distinguish between different types of on-farm trials9. Most on-farm trials are research-driven and focus on variety testing (few FMFI on-farm trials have been implemented). Twenty-five years after the emergence of the first FSR programs in Eastern and Southern Afi-ica more client-orientation could be expected. Although communication problems (language) and the civil war undoubtedly contributed to the lack of knowledge on the FSR approach, by now FSR should have been institutionalized. The institutionalization and internalization of the FSR methodology (as well as other important issues such as gender sensitivity, results evaluation and accountability, public- private collaboration, etc.) could not rely solely on training, they should have been triggered by political commitment along with sufficient analytical capacity - both of which were clearly lacking in MADER. The ASRDP design and subsequent supervision clearly underestimated the difficulties inherent in attitudinal change. Overall, the outcome for the research component is rated as unsatisfactory. Credit The credit component was intended to provide seasonal working-capital credit to the family farm sector under group credit arrangements, and to rural traders on an individual basis (a revolving fund was to have been established). It was anticipated that under these arrangements, the number of farm families which would be able to gain access to credit would be in the range of 30-35,000. Roughly 1,000-1,200 credit groups were to be formed through which the individual farmers would work in seeking credit. Some 500 cornmercial farmers were also to get credit. Credit was to be offered through the banking system and under market conditions at unsubsidised rates. Short-term TA was to be provided under the component. The component was never implemented and, at the time of restructuring in 1997, it was cancelled. This wa sin part because at the time of project launch, in the early nineties the banking system was already in a process of being privatized and the People's Development Bank (BPD) was not willing/able to take decisions with regard to its involvement in rural credit. Moreover, the BPD did not have experience with credit supply to the family sector and agriculture was considered a risky enterprise. Smallholders could not provide collateral guarantees and the official registration of farmer associations was and still is difficult and time-consuming. The inclusion of the credit component in the project was based on unrealistic and over-optimistic assumptions concerning local capacity and willingness to participate. While the project was correct in recognising the need for working capital in order for farmers to adopt technologies requiring purchased inputs, the project made no progress in this area and the plans which it did make were unfulfilled. This component is rated as highly unsatisfactory. Supervision Mission, 1996. 9  Rural Water Supply The rural water supply component aimed at the rehabilitation and construction of rural water facilities (wells and boreholes) in the Province of Nampula. Various inventories, studies and plans were to have been undertaken and completed. Rural communities were to have been mobilised to participate in the operation and maintenance of rural water supply systems. It was anticipated that some 250,000 farm families (out of a total number of 280,000 in the targeted districts) would benefit from this component. TA was to have been be provided. Until the transfer of these activities to the RRP, some of the inventories, studies and plans were made and TA was provided. Five new water supply points were installed and 15 were rehabilitated. At the time of restructuring, this component was transferred to the RRP. Activities under the component are evaluated in the ICR for that project. Land Use and ManaQement The main activities under the land use and management component were to collect, analyze and present basic data on land use and land tenure using aerial photography, and then to conduct a broad boundary cadastral surveying exercise, covering the Provinces of Nampula and Cabo Delgado. MADER's cadastral mapping, titling and registration activities were to be strengthened through TA, training and by funding surveys and other activities related to land use and land management. The primary objective was to establish a technically sound and socio- economically acceptable basis for improved land use and land management. Another type of support provided by the component was directed towards the legal and regulatory work of the Land Commission. The land component was added, in part, to avoid conflicts between cotton producing JVCs, who occupy the best land, and food crop-growing farmer communities. Implementation of the component was marginally satisfactory. The weak institutional capacity of the three agencies involved in this component, i.e., the National Directorate for Geography and Cadaster (DINAGECA), National Institute for Agricultural Research (INIA) and National Directorate of Rural Extension (DNER) was a major constraint to successful implementation, and linkages between them remained poor. Equipment was provided and a considerable amount was spent on staff training, mainly abroad. Unfortunately, this training was focused on central-level staff and had little effect on improving the capacity at provincial level, where most of the titles have to be processed. In only one district (Monapo) has all community land been demarcated so far (this work started in 1998). Socio-economic studies have been implemented in Monapo and Montepuez districts. Soil and land use maps are made by INIA (funded by the project) but they are not effectively used. Formulation of soil conservation packages and land management practices is done by INIA, but it is not clear what has been the impact at farm level. An impact study on land titling policies, as proposed in the SAR, has not been implemented. In the family sector title deeds are not a necessity yet, but for farmers' associations they could become important to secure institutional credit in the future. Proiect Mana2ement The effectiveness of the project's management appears to declined in 1996, and management became much less effective after 1998. This coincided with a change in the leadership of MADER as a result of the emergence of PROAGRI (which was perceived to be of higher priority as MADER's main financing mechanism in the future). There was a slackening of control over the management and administrative routines of the project. The PIU (a MADER structure not under the administrative or disciplinary control of the Bank), reported to the directorate of economics of MADER, which also had a change in leadership. Project progress reports ceased to be produced in 1999, and there was less commitment to implementing project 10  activities. This was seen in the fact that many of the recommendations and agreed actions following supervision missions either were implemented with delays, or were never implemented at all. This in turn prompted closer, hands-on management by the PIU, which in turn caused antagonism between Government and project staff. In addition, PROAGRI represented a total departure from the "project" approach, and there was a perception that the principles and managernent of ASRDP established in the first half of the project could be modified so as to be more like ithose being put into place for the PROAGRI "program". This, however, coincided with increased instances of financial mismanagement and lack of technical progress of the project. Reports of "doubtful and non-justified" transactions were made initially in 1995 following an audit report. A review of the financial transactions for 1999 and 2000 revealed further unjustified expenditures'