Mid-term evaluation

Burundi, a small, landlocked country in East Africa, had an estimated population of 3.8 million in 1976, with a growth rate of 2.2%. It is one of the most densely populated countries in Africa and one of the poorest countries in the World. About 95 % of the population live in the rural areas and rely on subsistence farming for their livelihoods; coffee is also grown and is the mainstay of foreign exchange earnings. Because of its diverse climate and ecology, Burundi's agricultural potential is very varied, ranging from tropical crops in the Imbo region (rice, cotton, sugar cane, groundnuts) to temperate fruit, vegetables and cereals in the higher areas. Despite these advantages, over-crowding has progressively reduced fallow periods and led to declining yields. Forest areas are also under pressure to satisfy the demand for domestic fuel, and increasing numbers of cattle are leading to over-grazing. The project area, the Ruzizi Valley, lies just 15 Km north of the capital, Bujumbura. and is a flat valley with deep alluvial soils deposited from the rivers which cross it and which subject it to periodical flooding. The climate is tropical and rainfall is around 857 mm per year. The land tenure system is quite mixed: in the cotton and rice areas the land belongs to the State and is allocated to farmers for specified purposes, outside these "development areas" the land belongs to the farmers and may be sold at will. In the project area there are approximately 3000 ha of "development areas"; these consist of farms of about 1.9 ha growing rice, groundnuts, maize, sweet potatoes, bananas, beans and a variety of other food crops - there are also a large number of fruit trees. In general the farmers in these areas use low intensity farming methods and rarely use fertilisers.

Project design and objectives

In general the agricultural sector has suffered from inadequate allocations (from the government) in terms of both production inputs and also services, and output prices have been kept low, reducing incentives for farmers. The main problems faced by the government in the late 1970s were under-nourishment and food scarcity at certain times of the year, and overcrowding in the densely populated hill areas. The movement of settlers into the Ruzizi valley occurred in the 1970s, due to developments in rice and cotton cultivation; the project intended to develop a part of the under-populated area in the valley for new settlement and increase production from other areas.

Target group

The population of the project area was about 14,000 people in 1977, and family sizes were about 4.5. Average farm sizes were 1.3 ha for food crops and fallow land and 0.8 ha for coffee in the coffee growing regions. The target group was defined as the whole farming population, which consisted of:

• farmers with 0.5 ha each, in the newly developed irrigated rice schemes, (1,091 total farmers in 1986; this number will increase to about 2,800 as the development and rehabilitation of irrigation schemes progresses);
• farmers with 4.0 ha each, organized in cotton production schemes (1,290 in 1986, "paysannats coton")
• independent smallholders (exploitants hors paysannat), including 2,723 farmers with 0.1-2.5 ha holdings in the Bujumbura Rice Scheme (most of whom will eventually become part of the first group) and 1,107 dryland farmers on scattered 1- 4 ha holdings.

Objectives and components

Project objectives were (i) to increase and diversify food crop production to meet subsistence needs, (ii) to improve the socio-economic conditions and the standard of living in the area, through water supply, electrification, and the construction of schools, health centres, and village depots; and (iii) to increase the cash income of target groups through the marketing of surpluses. Project interventions to meet these objectives, which were adjusted in 1983 on the basis of findings from socio-economic studies carried-out by the project, included:

• an agricultural extension programme to reach 4,980 smallholders cultivating irrigated rice (3, 420 ha), cotton and other rainfed crops (1,560 ha), providing support services such as supply of farm inputs, draught animals, agricultural credit, and marketing;
• infrastructure works for new and rehabilitated irrigation schemes, comprising training of rivers, drainage networks, primary canals, secondary and tertiary canal networks, and land leveling;
• improvement of socio-economic conditions by the construction of 5 schools and 3 health centres, the rehabilitation and improvement of 132 km of secondary roads, the provision of drinking water supplies, and housing improvement for 2 000 nearby settled families;
• assistance to the Project Management through the construction of offices, staff housing, electrification, and the construction of an administrative centre.

Total base project costs were estimated in 1979 at USD 29.0 mn, to which contingencies of USD 13.5 mn were added; this comprised the following components:

(i) Civil works	USD 5.15 mn (17.8%)
(ii) Resettlement	USD 3.1 mn (10.5%)
(iii) Social Infrastructure	USD 5.4 mn (18.4%)
(iv) Cooperative and Community Devel.	USD 0.05 mn (0.2%)
(v) Project Management	USD 9.6 mn (33%)
(vi) Credit and Input Supply	USD 2.8 mn (10%)
(vii) M & E	USD 0.8 mn (3%)
(viii) Studies, supervision	USD 2.1 mn (7.1%)

Project management, the largest single component, was concerned with the strengthening of the implementing agency, the Societe Regionale de Developpement de l'Imbo (SRDI), and included the development of an extension service, community development including cooperative groups, the distribution of farm inputs, establishment of an ox training centre, and training for extension and cooperative workers. To undertake these tasks SRDI received considerable expatriate TA, which not only supplied sustained support and training, but also undertook the various consultancy aspects of the project. Under social infrastructure, the second largest component, the project developed and rehabilitated access roads, and provided drinking water, electricity, schools, dispensaries and warehouses to the 15 new villages developed (and others which were already existing). The Civil works component was concerned with flood control measures, irrigation (rehabilitation of 1,500 ha and development of a further 620 ha) rehabilitation of various existing canals (including lining some canals to reduce water losses), land leveling to improve water control, and the establishment of a drainage network.

A small credit in kind facility was included in the project design to enable farmers to purchase the fertilisers, seeds, pesticides and small implements which were to be provided through SRDI. A number of detailed socio-economic and design studies were included in the project costs to allow further specification of the project facilities during the early stages of the project. M&E, which was undertaken by a Unit in the general management division of SRDI, i.e. covering all SRDI activities and projects, was equipped by the project for this task.

In the light of findings of technical and socio-economic studies carried-out by the project in 1983, project cost allocations were altered so that :

• provision for civil works was increased to USD 14.4 mn;
• provision for resettlement was reduced following spontaneous settlement along secondary roads;
• provision towards social infrastructure was increased by a larger financial contribution of the parallel European Development Fund (EDF) programme for health measures;
• technical assistance was reduced to USD 0.9 mn, after a review of existing local capacity.

The financing plan for the project was complicated by the number of donors and the different methods of financing. In summary the plan as proposed was:

• joint financing by IFAD (SDR 11.3 mn), ADF (UCF 8.0 mn), and OPEC (USD 2.0 mn) for civil works, water supply, electricity, training and community development, resettlement and agricultural inputs;
• a contribution by WFP for up to USD 0.85 mn in food improvement and compensation during rehabilitation works;
• a grant by EDF of USD 5.9 mn for social infrastructure and health components of the project, financed on a parallel basis.

Expected effects and assumptions

The project plans called for approximately 3,475 families to be resettled and a further 1,845 families were to be assisted to improve their productive capacity (this required the development of 2,400 ha of irrigated land and 3,550 ha of rainfed land). The value of incremental agricultural production was estimated at appraisal as USD 8.4 mn. at full development; this included a further 10,890 tons of rice and 476 tons of coffee, as well as more modest gains in groundnuts, maize, soya beans, bananas, and cotton. Beneficiaries' incomes were estimated to increase by about 65%. In general all food deficits of project beneficiaries would be covered, although animal protein requirements would still need to be purchased; surplus production from the project area would contribute to reducing food deficits throughout the region. Access to regular medical services and hygienic water supplies was intended to benefit the local population generally (estimated at 24,000 people). It was assumed in the project design that (i) coordination of the financing arrangements would be organised by the donors and would not hinder implementation, and (ii) output prices would be set at a level that was sufficiently attractive to the settlers to encourage production of the government's priority crops.

Evaluation

The mid-term evaluation mission consisted of a rural engineer/irrigation specialist and an agricultural economist; they spent approximately two months in Burundi, from November 1986 to January 1987, and overlapped with a AfDB/IFAD supervision mission. The mission held extensive meetings with government agencies, project staff and beneficiaries, visited the project sites, and reviewed the reports on the project.

Implementation context

The implementation of project activities was directly affected by poor coordination between the funding agencies, with resulting delays in disbursements. At the same time a number of studies, which were included in the design, pointed to the need to revise some of the specifications of the individual components, and also change the balance between the components. These factors eventually led, in 1985, to a review of the project's financial situation which highlighted the strong imbalance of disbursements between 3 of the cofinanciers (ADF, IFAD and the OPEC fund). As a result, a revised list of goods and services and a new financing plan (in terms of foreign exchange, local currency costs distribution and percentage of financing by each cofinancier) was approved in October 1985.

In addition, implementation of the irrigation and rice production plans were directly affected by: (i) the government's decision to liberalise paddy and rice marketing, (which removed SRDI's monopoly, and had implications both for loan recovery and potential volumes processed by the rice mill); (ii) illegal imports of rice from Tanzania which, under parallel currency exchange circumstances, benefited from an unfair and significant market advantage; and (iii) the sales of Japanese food aid, in the form of rice, by SRDI on behalf of the Government of Burundi.

Project achievements

SRDI employs some 200 permanent national staff and 140,000 man-days/year of temporary labourers, 40-60% of which are assigned to East Mpanda. Various personnel changes at management level have taken place, and although their effect on project progress is difficult to assess, it noticeably contributed to the slow start-up of the project. Technical Assistance staff consisted of 3 experts (Agronomist, Financial Director, and Land Leveling Specialist) for a total of 11 man-years, a significant reduction compared to the originally planned 59 man-years for 10 experts. The counterpart situation was found to be satisfactory except in the case of the Financial Director, who did not have a counterpart.

The analytical accounting system adopted by the project was in line with procedures and systems introduced by other projects in Burundi and proved to be an effective control and verification tool. However, while it met all requirements of SDRI as the Development Company of the Ruzizi/Mpanda area, it did not allow for sufficient separation of the actual costs of the East Mpanda project alone, and did not give sufficient insight into unit costs and cost accounting for individual services/activities of SRDI.

Technical studies, entrusted to the engineering consultant DHV, were carried out in 1981-83, and resulted in the detailed design and tender documents for: (a) construction of buildings; (b) training of rivers and drains: (c) the primary irrigation network; (d) drinking water supply; (e) social infrastructure; and (f) feeder roads. Studies for secondary canals and on-farm irrigation have been carried out by SRDI. Progress of the studies was slow, which together with administrative delays, contributed to the project falling significantly behind schedule. Implementation of civil works started in 1984, 2 years later than foreseen at appraisal.

River alignment measures and drainage collectors, which included dredging and dyking of certain sections of the main rivers in the project area, were started in July 1984, and completed ahead of schedule in September 1985. The evaluation mission reviewed a number of engineering issues which required further attention, including preventive/protective measure against erosion, additional dredging etc. Construction of the Primary Irrigation Network started in November 1984, and was also completed ahead of schedule in late 1985. Primary networks have been completed for 571 ha of new schemes, and are substantially completed for the rehabilitation of the Bujumbura rice fields. However, in 1985 a (20 years high) flood destroyed two of the existing dams on which some primary canals depended, requiring some modifications of the network.

On-farm irrigation including construction of secondary and tertiary canals, and land leveling, have been undertaken directly by SRDI. Considerable delays were experienced with respect to on-farm irrigation, mainly due to the late arrival of machinery (end 1984), the need for revisions of studies, and unfavourable weather conditions in 1986. Construction of project buildings was completed in 1985, and construction of 5 village depots (under EDF financing) has recently started. The power line with transformers was completed in 1984, the line links the centres of Mugerero and Maramvya to the Bujumbura power supply.

Agricultural extension. At the time of the MTE, extension activities had been heavily biased towards rice farmers operating on irrigated schemes, with few activities for rainfed agriculture. Suitable, well documented technical packages were available for single cropped rice and for cotton, and were being disseminated among farmers, but more work was required for the development of technical packages for double cropped rice and other food crops. Demonstrations have been carried-out on irrigated plots assigned to the extension workers, but more systematic demonstration/trials on farmers' fields, were planned for 1987. The extension workers' tasks include (i) managing the input supply system to ensures timely availability of paddy seeds, fertilizers, agro-chemicals, tools and implements, at the village level, and (ii) organising the very effective collection and marketing system for paddy. The project started to organize farmers groups in late 1986; these groups were expected to improve, at field level, coordination of canal maintenance, cultivation operations and water management.

The supply of farm inputs is directly linked to the Credit Programme operated by SRDI, under which all inputs for paddy cultivation on irrigation plots are made available through individual crop loans. Virtually all rice farmers have made use of this facility, and repayments have dramatically improved compared to earlier years, reaching close to 90% in 1985-86. Introduced in February 1986, the social programme focuses on nutrition and sanitation education. The programme was found to be dynamic, but has suffered from a lack of resources and teaching materials (which were due to be provided under the 1987-89 budget). So far, the programme has reached well over 1,000 beneficiaries, including infants suffering from malnutrition. SRDI purchases paddy in the villages through its extension workers, and pays farmers on a cash basis, after deduction of crop loans and farmer's fees. Purchases in Mugerero and East Mpanda have increased from 2,700 MT in 1979 to 7,800 MT in 1986. The volumes of paddy received at the SRDI mill (including purchases from other areas) had also increased, from 4,300 MT in 1979 to 10,600 MT in 1986.

Monitoring and evaluation system The M&E unit produces monthly and quarterly M&E reports, and contributes to the preparation of general project progress reports. There is only one officer responsible for this unit, and the unit is understaffed and short of equipment. It does benefit from the occasional inputs from the M & E service of the Ministry of Agriculture (DGPA).

Effects assessment and sustainability

The first implementation activities with regard to the social infrastructure programme were started in November 1986, and civil works were launched in 1984. Other activities, such as agricultural extension, social animation, and monitoring and evaluation, were initiated in 1985 and 1986. Visible effects of the project were therefore very limited by MTE, and an attempt to assess effects would have been premature because of the implementation delays. Hence the comments below are circumspect.

The project has contributed very positively to the strengthening of SRDI, and has helped this institution put into place a sound administrative organisation supported by adequate resources in terms of personnel, physical facilities and equipment, thus creating significant employment opportunities in the area. In spite of various shortcomings and problems, the project has to be commended for the dedication of its staff and credited with having in place most elements (such as farmers' groups, credit, savings programme) which should now facilitate a speedy and successful implementation of the various activities under the project. This has been demonstrated on the 352 ha of irrigated lands developed by the project in 1985-86, where the systems put in place for input supply, paddy collection, and marketing have proved effective.

Yields achieved on SRDI schemes in 1985-86 for single cropped rice averaged 3.56 tons paddy/ha (actual sales by farmers to SRDI). Yields recorded by the monitoring and evaluation unit, based on 193 sample crop cuttings, averaged 5.7 tons paddy/ha. These results compare favourably with yields of 1.2 tons/ha outside the project area, and appraisal projections of 3 tons/ha at full development. The income of farmers from their 0.5 ha rice plot were, on average, USD 325, compared to appraisal estimates of 260 USD. Cotton yields averaged slightly below 1 ton/ha (c.f. appraisal projection of 1.35 ton/ha) and will only be reached with improved technical packages.

Recurrent Costs and Sustainability of the Project The total recurrent costs of SRDI for 1985 were approximately Fbu 175 million; for East Mpanda, after completion of physical infrastructure works, recurrent costs are likely to be approximately Fbu 70 million per year. SRDI's foreseeable annual revenues from the project area are of the order of Fbu 110 million (Fbu 44 million from farmers' fees and some Fbu 75 million from profits of the rice mill), and would therefore be sufficient to ensure sustainability in the long run. However, such revenues would not allow for the continuation of direct and indirect subsidies paid by SRDI (totaling in 1985 Fbu 62.5 million towards seeds, fertilizers, overheads, and former farmers' bad debts).

Main issues and recommendations

The major recommendations from the mission were as follows:

Project Management and Administration. With regard to general management the mission recommended: (i) a more participative form of management with representation, consultation, and involvement of farmers in SRDI's decision making; (ii) improved inter-departmental coordination so as to avoid imputation errors in accounting; (iii) streamlining of the accounting system, so as to separate the East Mpanda project from other SRDI activities and to generate essential cost accounting information; iv) immediate recruitment of a counterpart to the Financial and Administrative Director of SRDI; (v) computerisation of SRDI's data management and accounting system. For the Engineering Department, there is a need to: (i) create a separate service responsible for the coordination and management of all hydraulic aspects of the irrigation schemes; (ii) immediately recruit a rural engineer to head the service. With regard to the Extension Department, the need is to: (i) put in place, as planned, a service for agricultural extension, distinct from the service for agricultural production; (ii) introduce some degree of specialisation among field workers to avoid over-emphasis of administrative tasks; (iii) reallocate field personnel to provide a more balanced distribution of support services to rice and other farmer crops.

Concerning technical assistance, it was recommended to extend the contracts of the extension advisor and of the topography expert by, respectively, 2 years and 1 year and to provide up to 23 person-months short-terms technical assistance in disciplines such as monitoring and evaluation, audit, irrigation and hydraulic calculations, integration of women, and in-service training.

The mission recommended the extension of the project until 1989 to ensure completion of the irrigation works.

Women's participation in the project is limited and needs to be increased; in the first instance a study should be undertaken to prepare detailed profiles of women's activities and expectations, and then by means of a seminar and a sensitisation programme within the project, the role of women should be given greater prominence..

The M & E Unit should be strengthened and given more resources; additional information and indicators should be collected to help with more precise analysis of the effects of project activities.

Group formation has so far only been applied to rice cultivators. This should be given higher priority in the development process and further extended to include dryland farmers. Also, as women constitute 53% of the agricultural work force, group structures, concerns and activities should respond to women's needs.

Lessons learned

The financing plan of a project should be as straightforward as possible, so as to avoid administrative complexities and adverse effects on achievement of overall project implementation and objectives due to dependence on parallel financed projects. Most of the social components of the project aiming at improving the socio-economic conditions and the standard of living in the area under a parallel financing from EDF have yet to be implemented.

In this design of project, where detailed specifications are to be prepared as part of the early stages of implementation, or where in-country experience is limited, it is useful to have an outside technical back-stopping agency which can be called upon to provide support when necessary. This can also arbitrate between agencies, especially in the early stages of a project (for instance, during the preparation of tenders and selection of consultants). Such an arrangement could have reduced the considerable delays encountered in the project, which have led to the civil works being more than two years behind schedule.

A clear and detailed plan of operation is vital to implementation. The executing agency should have a clear mandate, there should be a well-defined project area and target group, and the project management should have the benefit of previous pertinent experience. Without these necessary precursors the initial stages of project implementation are likely to be marked by delay and confusion.

It is valuable to the project if the monitoring system is capable of taking on a quality control function for project implementation; this requires adequate resources and qualified staff with practical as well as theoretical training in monitoring methodology and knowledge an experience of the various aspects of development.

Paddy sales were adversely affected by changes in government policies and other factors over which the government exercised some control. These changes at the macro level were a threat to the viability of the small rice producers on the resettlement schemes. The CI should have made the government aware of these problems, and the coordination mechanism for the project should have been able to influence these decisions. In general, where the incomes of poor farmers are dependent on the sales of items where the price (or quantities) are controlled, then some assurance should be required from government that the situation will be maintained.