Objectives, approach and methodology

Evaluation objectives and key questions

This evaluation is intended to examine IFAD's performance and impact in decentralizing environments with reference to Ethiopia, Tanzania and Uganda, and reflect on decentralization issues and the design and implementation of relevant rural development programmes in the three countries. These objectives are addressed by aiming to answer four sets of key questions:

1. What is the framework of decentralization policies and programmes within which IFAD works in each country? What opportunities and challenges did it create for rural development in general, and for IFAD in particular, since the introduction of decentralization?

2. What strategies and interventions did IFAD introduce in its Country Strategic Opportunities Papers (COSOPs) and projects to address the opportunities and challenges presented by decentralization? What was the purpose of these strategies and interventions, and what means did IFAD provide to implement them?

3. What has been the institutional impact of IFAD's strategies and interventions on the project institutions at which these were aimed? In what ways have these institutions enhanced their capacity for rural poverty alleviation and utilized it to this end?

4. Were IFAD's strategies and interventions relevant, and did they enhance the efficiency, effectiveness, impact and sustainability of IFAD's strategic and project objectives in each country? What outcomes did IFAD help achieve through advocacy and policy dialogue?

Evaluation approach and methodology

As would be apparent from the key questions, the evaluation methodology used in this report is an adaptation of the Office of Evaluation (OE) Methodological Framework for Project Evaluation (or MFE, for short). The MFE, however, is a methodology for project evaluation and is not applicable in a straightforward manner to this thematic evaluation. The subject of this evaluation—decentralization—has been defined (Rondinelli 1981) as "the transfer of responsibility for planning, management, and resource raising and allocation from the central government to: (a) fields units of central government ministries or agencies, (b) subordinate units or levels of government, (c) semi-autonomous public authorities or corporations, (d) area-wide regional or functional authorities, or (e) organizations of the private and voluntary sector"¹. Thus, decentralization, in the first instance, is about institutional change, which may then generate changes that influence the main evaluation criteria used in the MFE. The evaluation approach, therefore, focuses on the institutional impact of decentralization with key question number 3, before focusing on the MFE type of questions in key question number 4². While the evaluation missions fielded in the three countries focused on all four key questions, OE commissioned desk reviews prior to the missions that addressed the first two questions³ . OE also invited five selected projects to contribute self-assessment reports.

The desk review part of the evaluation included all those IFAD-assisted projects that have been approved since the mid-1990s and are not confined to a single commodity (e.g., a crop) or input (e.g., seed)⁴. This gave a total of 12 projects in the three countries that are listed below, out of which five projects (identified by an asterisk) were selected for the fieldwork:

Ethiopia	Tanzania	Uganda
* Special Country Programme, Phase II	* Kagera Ag. and Environmental Management Project	* District Development Support Programme
Agricultural Research and Training Project	* Participatory Irrigation Development Programme	Area-Based Agricultural Modernisation Programme
Rural Financial Intermediation Programme	Rural Financial Services Programme	* National Agricultural Advisory Services Programme
Pastoral Community Development Project	Agricultural Marketing Systems Development Programme	Rural Financial Services Programme

The criteria for project selection for fieldwork are explained in section I.B of the main report. Fieldwork in Tanzania and Uganda revolved mainly around a two-week mission to each country, which held meetings in the capital city with government officials and donor representatives, and visited parts of the project areas of selected projects to meet beneficiaries, project officials and elected representatives. In Ethiopia, OE conducted an interim evaluation of the selected project, and the preliminary results from that evaluation have been used in the preparation of this report in lieu of a separate decentralization evaluation mission⁵. In addition, OE fielded missions in all three countries for purposes of reconnaissance, discussion of country context and facilitating self-assessment, which focused on consultations with partners in the capital city.

Country contect of evaluation

The socio-economic Context

The three countries included in this evaluation have experienced faster economic growth than the average for low-income countries, and Tanzania and Uganda fare better than low-income countries as a whole in terms of their literacy. All three countries, however, confront challenges that few among the developing countries face with such severity. Notable among these are the following:

1. The people, societies and economies of Ethiopia and Uganda have been devastated by many years of dictatorship and civil war, while Tanzania has suffered from the economic, environmental and social consequences of conflict in neighbouring countries. Insurgency and conflict retain a potential in or around these countries for creating significant disruption;

2. Poverty is a serious and widespread problem that both stems from and contributes to other socio-economic problems. In Ethiopia, 44% of the people live below the basic needs poverty line. In Tanzania and Uganda, 39% of the rural people are considered poor;

3. The effects of the HIV/AIDS epidemic have been debilitating not only for human life and health but also in terms of economic growth and long-term social costs. For Ethiopia, the World Bank estimates that the epidemic is costing the country one percentage point of economic growth, each year, "and unless reversed, could erode the recent development gains." In Uganda, IFAD counts the cost in terms of "loss of time and resources among the many families that care for the ill and the children orphaned by the disease; reduced supply of labour and investment resources; and a rising number of orphaned children."; and

4. As in many other countries of Africa, women's lives represent a harsher reality than men's, whether viewed in economic or any other terms. Wide gender disparities persist in terms of almost all economic and human development indicators, as well as in relation to social and legal status and participation in politics and development.

The driving forces for decentralization: an overview

Unlike Tanzania and Uganda, Ethiopia was never a colony (with the exception of the Italian entr'acte). Ethiopia also stands out because of years of civil war, the end-result of which was the overthrow of a military regime in 1991, the creation of two countries—Ethiopia and Eritrea—out of one, and the adoption, in 1995, of a constitution that established a federation of nine ethnicity-based regions. Along the way, Ethiopia experienced, first, a quasi-feudal monarchy and then, from 1974 to 1991, a Marxist-socialist state dominated by central planning, state ownership and control under the Mengistu dictatorship. Together with Nigeria and Pakistan, Ethiopia is unique in that it adopted federalism so emphatically only after a civil war, which, in Ethiopia and Pakistan, led to the break up of the original state. The country subsequently initiated a second phase of decentralization, starting in 2002, which saw a massive devolution of finances and staff from the regional executives to woredas (districts). It is only since then that the woredas have become the focal points for much of the development work.

Political forces of the kind that have propelled Ethiopia towards federalism and devolution have been largely absent from Tanzania's long and varied experience with decentralization. Indeed, Tanzania appears unique because of its cohesion, the variety of its experiences with decentralization, and the long timeframe it has set for achieving its decentralization objectives. As in many other former British colonies, Tanzania's modern experience with decentralization includes a long period of what was called "indirect rule" through traditional leaders. But Tanzania embarked upon the path of devolution with renewed vigour during the mid-1990s. And in 1999 it launched an ambitious and wide-ranging Local Government Reform Programme (LGRP) to address the challenges of decentralization comprehensively over a period of about 12 years

Unlike Tanzania, and even more than Ethiopia, Uganda has moved rapidly from an extremely centralized state to a comprehensive decentralized system in recent years⁶. After 15 years of political strife, economic mismanagement, civil wars, mass emigration and mass murder (1971 - 1986) left the country devastated, the National Resistance Movement (NRM) captured Kampala and formed a government that remains in power to this day. The NRM, initially a military grouping, drew its strength from local Resistance Councils. In 1987 the Government enacted the Resistance Councils and Committees Statute, which transferred authority to plan, make decisions and provide services to the local levels, under the leadership of the Resistance Councils. The 1995 Constitution consolidated and extended the decentralization process and established five tiers of elected local councils.

Country goals and objectives for decentralization

In Ethiopia, "The explicit goals of decentralization are to bring government closer to the people, give political representation and voice to the diverse ethnic groups, and make governance and resource allocation sensitive to local needs and preferences" (World Bank 2003a). The government is committed to decentralization for economic as well as political reasons: according to the National Capacity Building Plan, one of the major objectives of decentralization "is to meet the needs and aspirations of the country's diverse populations in order to deepen the process of democratization". Decentralization is linked explicitly to the government's Sustainable Development and Poverty Reduction Programme (SDPRP), the overarching objective of which is to reduce poverty by enhancing rapid economic growth while at the same time maintaining macro-economic stability⁷. A national umbrella programme called the Public Service Delivery Capacity Building Programme aims to support institutional transformation at the federal, regional, and local levels.

The goal that leads the decentralization initiative in Tanzania is to "improve the quality of and access to public services provided through or facilitated by local government authorities". The LGRP elaborates "that the goal will be achieved through the reorganization programme and will have two components: (i) a shift of responsibility for managing and providing services from central to local authorities (districts); and, (ii) increase efficiency by reorganizing the district administration and allowing the councils greater freedom in organization their activities and managing their personnel." Policy makers have also mentioned the idea of inclusive governance as an expectation associated with decentralization. Moreover, decentralization is linked, directly or indirectly, to a number of broad poverty reduction initiatives as well as the Tanzania Development Vision 2025.

Uganda also offers a broad official rationale for decentralization, "defined as the transfer of planning, decision-making and administrative authority from the central government to regional branch offices, local governments, and/or non-governmental organizations" (Mulumba 2004). The intentions include: giving people a greater chance to participate in development planning, decision-making and implementation; improving the efficiency and co-ordination of services at the local level; enabling voters get to know and challenge policy makers as well as corrupt and inefficient officials; encouraging localities to come up with innovative means of resource mobilization and management; and encouraging a mutual and equitable approach to solving local problems. Related objectives are found in official plans for poverty eradication and agricultural modernization.

Empowerment of women is an explicit objective of decentralization in Tanzania and Uganda. Tanzania allocates 25% of the local council seats to women, and Uganda guarantees at least 33%. In Ethiopia, however, out of the 1,755 seats in the Regional Parliaments only 218, or 12%, are occupied by women. The ratio ranges from just over 1% in the Somali Region to 28% in Tigray. Women's representation at the woreda level appears to be at least as low.

The international donor community has been highly supportive of decentralization in all three countries, and the general assessment is that it will take many years for decentralization to attain its potential. Most of the donors have been extending project assistance as well as engaging in policy dialogue. While the World Bank is the major player in most respects, other multilateral agencies as well as the bilateral ones have also played key roles in capacity building and policy reform (and in the latter, most have been more active than IFAD). Increasingly, donors are pooling resources and coordinating their programmes through Sector Wide Approaches (SWAps), and IFAD is also part of this trend. Some donors, however, have taken the position that a high percentage of their support would be channelled through SWAps, but bilateral projects and NGOs would also be funded side by side. In one variant of this (Ireland, in Uganda), projects are important for working out and testing new approaches that can then be upscaled and mainstreamed through SWAps. Bilateral support is also emerging for associations of local governments, which represent a platform for lobbying, sharing experiences and strengthening decentralization.

Decentralization in IFAD strategy and project design

Overall IFAD and regional strategy in relation to decentralization

The country context of this evaluation represents one important point of departure for this evaluation, and the way IFAD approaches decentralization represents another. Unlike some other donors and governments, IFAD views decentralization not as an objective in itself but as a means to an end that is defined by its overall mandate. The Strategic Framework for IFAD 2002 – 2006 is an exceptionally perceptive example of international literature that seeks to identify the conditions under which decentralization could help the rural poor. Its assessment of rural poverty recognizes that decentralization could provide opportunities for the rural poor if— and only if—the rural poor can influence the institutions, policies and decisions that affect their lives (emphasis and "only if" added). This emerges from the very first page of The Strategic Framework.

The Strategic Framework views decentralization as a framework within which governments can respond more effectively to the needs of the rural poor, in particular, by increasing the accountability and transparency of rural service delivery. It also suggests, however, that this expectation can be met only if poor people are organized, if they participate in the project cycle, if they have the means to influence institutions and if governments assist them actively in these endeavours. Generating catalytic impact through field operations, policy dialogue and advocacy on behalf of the rural poor is an important part of the strategy that is expected to support decentralization.

16. The IFAD Regional Strategy Paper for Eastern and Southern Africa, finalized in March 2002, is built around three crosscutting principles, four means for addressing its catalytic role and four strategic thrusts for programming. These over-arching elements of the strategy are summarized in Box 1 of the main report. Decentralization is not a main objective in this strategy, but support for decentralization is invoked as part of the third principle, which states, " … the best guarantee that public policy and institutions will effectively facilitate the efforts of rural poor people to work themselves out of poverty is to ensure democratic accountability of governments." The strategy also suggests that identifying the poor in rural communities would be one of the steps leading to their empowerment, welfare and voice in the process of accountability.

Decentralization and related concerns in the IFAD country strategies

To varying degrees, the strategic directions outlined above are also reflected in the IFAD COSOPs for the three countries, in all of which decentralization is recognised fairly early in the process as an issue of importance for IFAD. More specifically, the 1999 Ethiopia COSOP draws the lesson that "early and full integration of project co-ordination, management and implementation arrangements into the decentralized federal and regional institutional framework is critical for success and sustainability." In support of decentralization, this COSOP calls for institutional capacity building particularly at the grassroots, district (woreda) and zonal levels. The 1998 Uganda COSOP signals the mainstreaming of IFAD assistance within decentralized structures. Support for strengthening decentralization and promoting beneficiary participation constitutes an important strategic thrust in this COSOP. The 1998 Tanzania COSOP addresses policy dialogue and also includes short and incisive analyses of centralization and decentralization

Given the different country contexts, it may not be surprising that the five COSOPs written during the timeframe selected for this evaluation differ in the importance they assign to decentralization and what IFAD might do for strengthening it. Differences in context, however, do not explain why only the Tanzania COSOPs (of 1998 and 2003) identify decentralization broadly as an area for policy dialogue, or why only one COSOP (Uganda 2004) highlights the need to strengthen the poverty orientation of local governments and initiate policy dialogue on mobilizing resources through service users and local governments⁸.

Gender concerns are also treated unevenly in the five COSOPs. The two Tanzania COSOPs are particularly sensitive to gender concerns and present specific analyses and directions for addressing these. The 1998 COSOP is unique in its attention to criteria for targeting women headed households. The other COSOPs also chart out approaches for the development of women.

In relation to targeting the poor, the COSOPs for Tanzania as well as the 1998 Uganda COSOP draw upon statistics to help identify geographical areas for a targeted approach to poverty. However, none of the five COSOPs elaborate upon the need or techniques for identifying the poor within rural communities.

Relevant aspects of the design of Sselectedprojects

The strongest feature of project design in the three countries is the inclusion of community organizations in implementation and the funding that IFAD provides for working with these organizations. Implementation through existing decentralized structures is also a particularly strong feature of these projects, most of which provide funds not only for service delivery but also for strengthening the capacity of new or emerging government structures. There is little evidence, however, that IFAD assistance was aimed at developing the capacity of the elected institutions of local government, as opposed to the civil service.

The empowerment of the poor is highlighted as a specific objective in only two projects⁹, neither of which is aimed at decentralization or local governance. Moreover, project design also lacks attention to targeting, as defined in the regional strategy: none of the 12 project designs appears to have mechanisms in place for "identifying the poor in rural communities." Only one project lists policy dialogue or policy reform as an objective, and only two provide funds for this purpose¹⁰. None of the IFAD-assisted projects provide funds for policy dialogue in Uganda, where decentralization is, by all accounts, proceeding more rapidly than in the other two countries.

Institutional change and impacts

Decentralized structures in the three countries

As indicated earlier, Ethiopia is structured as a federation. Its nine regions and two administrative areas are divided into 66 zones and 556 woredas; each woreda is divided into a number of kebele. At the local level, there are elected Woreda and Kebele Councils. Above the woreda there is the regional state, which has two councils, namely, the elected Regional Council (or Parliament) and a cabinet body called the Regional Administrative Council. Many of the regions have decentralized sector programmes to the woreda level, and there are sector-specific agencies within the executive branches of the regional and woreda administrations¹¹ .

The highest sub-national level of administration in Tanzania is also called a region. This is not, however, a separate tier of government but an extension of the central government. The sector ministries are represented at the regional level by technical officers who oversee the work of the sector ministries, in addition to providing policy guidelines and technical advice to the districts within the regions. There are 26 regions and 124 rural or urban districts in the country. The districts are sub-divided into divisions and wards. At the district level, local government authority is vested in the District Council elected from the wards in the district. Public servants working in the district are answerable for their work to the District Council through the District Executive Officer¹². At the sub-district level, there exists a Ward Development Committee.

Unlike Ethiopia and Tanzania, the sub-national institutional landscape in Uganda exists exclusively within the district: there is neither a political unit nor an administrative level between the centre and the district. There are currently 56 districts. In each district, there are five tiers of local government, and each of these carries a Local Council (or LC) appellation, together with a number from one to five. Thus, LC 1 is the village, LC 2 the parish, LC 3 and LC 4 are the sub-county and county, respectively, and LC 5 is the district. The LC 5 (district) and LC 3 (sub-county) are elected local councils and also the main implementing bodies, with significant staff and budgets. The district is the main planning authority, policy review and approval body. The district administration is headed by a Chief Administrative Officer (CAO) and includes the heads of line departments. The CAO is responsible to the District Council for the discipline and performance of administrative staff. At the sub-county level, the Sub-county Chief heads the local government and an Administrative Officer is incharge of the administration, which includes representatives of some of the line departments¹³.

Similarities and differences in the directions of change

Desk reviews and fieldwork show, unsurprisingly, that the three countries included in the evaluation have some striking similarities as well as perhaps equally striking differences. The most relevant of the broad similarities include the following:

1. (The recent and ongoing emphasis on decentralization entails a focus on the district level (called woreda in Ethiopia) for development planning and implementation. A tier below the district level is also important for decentralization, albeit, in varying degrees: in Ethiopia it is the kebele, in Tanzania the ward and in Uganda the sub-county level;

2. The capacity for planning, implementing and monitoring decentralized development programmes is generally weak and evolving. Central authorities and their international partners are supporting decentralization through large-scale initiatives for capacity building;

3. Decentralization is linked to countrywide initiatives for service delivery and poverty alleviation. Most of these as well as similar smaller initiatives are donor-supported;

4. There are well-defined processes in all three countries for local-level planning. Local governments, however, do not have anything resembling a poverty alleviation strategy; and

5. Local government finances depend almost entirely on transfers from senior governments, which, in turn, are financed mainly by international donors.

There are also, however, significant differences in governance arrangements among the three countries, including the following:

1. By virtue of their constitutions, Tanzania and Uganda have a unitary form of government, whereas Ethiopia is a federation. In Ethiopia, the regions are federating units, while Tanzania's regions are extensions of the central government and Uganda has no regions at all;

2. Power is concentrated at two levels in Uganda—the centre and the district—and is more diffused along the hierarchy in Ethiopia and Tanzania. Uganda, however, has a more extensive set of institutions for accountability within and outside the government, including institutions for personnel recruitment and anti-corruption, and private local radio stations;

3. Tanzania has adopted a gradual approach to decentralization, whereas change in Ethiopia and Uganda has been more decisive and abrupt. At this time, elected councils at the lower levels have more influence in Uganda than Tanzania, and they are weakest in Ethiopia;

4. In Uganda the local government contracts district administration employees. In Ethiopia and Tanzania, employees of senior governments staff the executive branch of district government; and

5. Ethiopia and Tanzania use formula-based grants for transferring funds from senior to local governments. Uganda employs three kinds of grants, each of which has a different purpose in relation to local administration and service delivery.

Staffing, planning and financing issues

In Ethiopia, in many of the decentralized sectors, specific components have actually been decentralized in terms of organization and implementation as well as operation and maintenance. There is still centralization in terms of resource allocation for equity purposes as well as in terms of policy making and partially in terms of planning. Day-to-day rules and regulations for planning and implementation rest with the Regional Administrative Councils and the respective bureaux. To a large extent, the rules and regulations seem to be supportive of capacity building and responsibilities being devolved to the local or woreda levels.

The planning process at the regional level is a relatively complicated one, involving the identification of priority development activities at the sub-kebele (or got) level and building up to final approval by the Regional Council. The initial identification at the got level is expected to broaden the direct participation of communities in the identification of their priority development needs. This results in a draft prioritized plan proposal by the gots that is submitted to the kebele for consideration. After review by the kebele development committee, the draft sectoral development proposals are submitted to the Kebele Council for review and subsequent submission to the Woreda Council for approval. The Woreda Council consolidates the kebele plans and budget proposals and then submits the approved budget and revenue plans to the region or zone, as appropriate.

In Tanzania, sector policies are initiated and developed by the respective sector ministries of the central government. Operational strategy and budgets for projects that are to be implemented within the framework of decentralization are normally prepared by the implementing bodies—in this case, the local government authorities—in consultation with the Ministry responsible for local government and the relevant coordinating and/or sector ministries. Consultations with prospective and other relevant institutions at the district and sub-district levels (including relevant non-governmental organizations, community-based organizations and village council leaders) do also take place. The law requires that the District Plan must be prepared through a bottom-up participatory process, starting from the village and coming up to the ward and then the district. The District Plan consists of various sector plans and is the guiding document for the allocation of development funds.

In Uganda, the District Technical Planning Committee (DTPC) in each of the districts of the Programme Area is responsible for the implementation of Programme components in that district. This is normally done through the line departments. The DTPC is chaired by the CAO, who is also the ultimate authority for the management of Programme affairs and is accountable to the government for Programme performance. The DTPC reports to the District Executive Committee of the District Council. The latter is expected to receive, review and submit planning proposals, and the annual work programmes and budgets of all departments involved in implementation. The Committee is also expected to receive and review all reports, supervise expenditures, receive and submit for audit all financial accounts and review all procurement and generally supervise implementation.

Ethiopia and Uganda have developed transfer programmes that account for around 30% of central revenues in Uganda and over 40% in Ethiopia. In Ethiopia, financial transfers to the regions are made on the basis of a formula that uses criteria and weights approved annually by the lower house of parliament. The formula is based on: (a) a region's population; (b) its level of development; and (c) its revenue generation capacity. In Uganda, the government supplements local revenues for service delivery with three types of grants, namely, unconditional, conditional and equalization grants, which are described as follows: (a) unconditional grant is the minimum grant that is paid to a local government to run its services; (b) conditional grants are given to finance programmes agreed upon between the central and local governments; (c) equalization grant is paid to local governments that are lagging behind the national average standard for a particular service. In Tanzania, there has been limited progress on the devolution of financial, fiscal and administrative authorities to the local governments. With regard to the development budget, fund allocation is based on: population (70%); regional poverty levels (20%); and the size of the area (10%).

All three countries have experienced large-scale changes in staffing patterns, and these are most visible in Ethiopia and Tanzania, and less so in Uganda, which went through this transition earlier. In Ethiopia, there is a serious problem of positions lying vacant at the woreda level: field investigations in three regions showed that only 20-44% of the positions had been filled. Large-scale retrenchment and repositioning of human resources has also impacted service delivery in Tanzania. It goes without saying that this has affected the implementation of IFAD-assisted projects.

IFAD contributions to local governance

A greater sense of local ownership of development programmes is evident in all three countries as a result of decentralization, and IFAD-assisted projects have both benefited from and contributed to this development¹⁴ . In Ethiopia and Uganda, in particular, local officials and communities report a pronounced positive effect in terms of service delivery in some sectors.

Although all three governments have initiated large-scale programmes for capacity building, it could take several years for decentralized structures to perform as expected. Lack of financial resources is affecting even basic aspects of service delivery, particularly in Ethiopia. And the paucity of resources for operation and maintenance in all three countries makes it highly unlikely that the infrastructure entrusted to the care of local governments will be sustained as designed.

Under the circumstances, IFAD assistance for staffing, logistics and capacity building generally has been a timely and valuable contribution to the strengthening of decentralized structures. This is particularly true for the two projects—the DDSP in Uganda (which is cofinanced by the BSF) and PIDP (cofinanced by Ireland) in Tanzania—in which lessons were available from earlier experiences, and the projects appeared at a time when there was an immediate need for strengthening emerging decentralized structures. Moreover, true to its signature approach to rural development, IFAD has also invested in a considerable amount of social capital formation in all three countries.

Omissions and challenges in capacity building

IFAD has done much to assist the three countries with decentralization, but five main omissions and challenges emerge when its approach is compared with the IFAD strategies and country needs. First, as mentioned earlier, IFAD assistance for capacity building is aimed largely at the civil service part of local governments and does not extend directly to elected officials.

Second, apart from organizing the beneficiaries and introducing the logical framework analysis, there is little capacity building in IFAD-assisted projects to enhance the pro-poor orientation of local government. At the strategic level, only one COSOP (namely, Uganda 2004) highlights the need to strengthen the poverty orientation of local governments. At the project level, there is no evidence of assisting local governments to: (a) develop local poverty alleviation strategies; (b) adopt techniques (such as wealth ranking) for identifying who the poor are in each community¹⁵; or (c) introduce flexibility in plans and budgets for responding with pro-poor interventions for identifiable groups of poor.

Third, the articulation of voice is also inhibited by the weak participatory orientation of projects in relation to the challenges they face, particularly for infrastructure development and promoting inclusiveness and accountability in service delivery. Examples of this are given from all three countries in the main report.

The above-mentioned examples also point to a fourth important problem in capacity building, and this may be described as the challenge of dealing with technicalism, that is, an approach that is driven by official technical specifications and procedural requirements rather than the human, social and financial capital of communities and relevant institutions. The consequences of this approach include weak community ownership of project-sponsored interventions and institutions, lack of community contribution to scheme implementation and maintenance, limited and unsustainable project impacts, and ambiguous poverty impact. Examples are given in the main report.

Finally, these and other observations point to policy issues that require discussion with the governments concerned, and this is another major challenge. As indicated earlier, however, only two of the 12 projects included in this evaluation have allocated funds specifically for policy dialogue.

Issues in accountability

The Strategic Framework and regional strategy emphasize accountability, but neither the COSOPs nor the project documents propose how IFAD would assist with this important aspect of local governance and service delivery. Perhaps the perception is that this is a sensitive subject that is best avoided in the context of IFAD-assisted projects. The sensitivities may be understandable if accountability is equated with certain instruments and attitudes, but many instruments are available (see Box 2 in the main report) from which IFAD-assisted projects could select appropriate ones.

Conclusions and recommendations

Relevance of IFAD strategies and project designs

Of the 12 projects reviewed during this evaluation, three are rural microfinance projects aimed more at setting up and supporting the architecture required for microfinance than at decentralization. Another project—the AMSDP in Tanzania—focuses on agricultural marketing, whereas a fifth—the ARTP in Ethiopia—aims to support central and regional agricultural research organizations. The remaining seven projects are implemented through local and regional governments and aim broadly at rural and agricultural development through community participation. The broad implementation approach of these seven projects is, therefore, relevant to the decentralizing environments in which they operate, as well as the IFAD strategy of supporting decentralized service delivery and grass roots institutions. These projects are also relevant to several of the capacity building needs of decentralized structures. The assistance they provided strengthened institutions that lacked human and financial resources, and sometimes even the basic requirements for running an organization, and were then shaken up by the large-scale changes brought about by decentralization.

Neither IFAD strategies nor the projects, however, have established direct relevance to the capacity building needs of the elected institutions of local government. This could be a serious omission to the extent that the voice of the poor is better articulated through elected rather than bureaucratic institutions: depending on the latter, with or without the introduction of participatory approaches, is considered insufficient for promoting the interests of the poor.

Moreover, the idea of empowering the poor is reflected as an objective in only two projects. And none of the five COSOPs and 12 projects designs propose to establish mechanisms for "identifying the poor in rural communities" as required by the regional strategy, or by effective initiatives for poverty alleviation. Only the 2004 Uganda COSOP highlights the need to strengthen the poverty orientation of local governments, and participate in a policy dialogue on mobilizing local resources through service users and local governments. Only one project lists policy dialogue or policy reform as an objective, and only two provide funds for this purpose. Only the Tanzania COSOPs identify decentralization broadly as an area for policy dialogue. Finally, neither the COSOPs nor the project documents propose how IFAD could assist with accountability and transparency in service delivery.

Implications for efficiency and effectiveness

As the opposite of centralization, decentralization is often supported on the grounds that it promotes efficient and effective service delivery. The evidence from the three countries, however, is mixed: it suggests that the disruption caused by the administrative and financial changes introduced for decentralization initially has negative effects for both efficiency and effectiveness. Matters improve, first, when the essential human resources required by local authorities are in place, and, subsequently, when new responsibilities for the flow of funds have been worked out. Without the kind of assistance that IFAD has brought to bear in support of decentralized structures, there is little doubt that project efficiency and effectiveness would have been even lower during critical stages of the reform process.

Implications for impact and sustainability

As elaborated in Appendix 3, there is no doubt that IFAD-assisted projects have generated impact in terms of a wide range of impact domains associated with rural and agricultural development. This is not, however, a major issue in this evaluation. The major impact issue is whether projects and other interventions generated the impacts that IFAD strategies and projects expected to achieve from and through support for decentralization. Thus, poverty alleviation is the main over-arching concern, accountability is considered important insofar as institutional impact is concerned, and sustainability is a key expectation associated with decentralization and community empowerment in the COSOPs and projects.

International evidence supports the view that decentralization rarely leads to successful poverty alleviation, and never by itself. Some of this evidence is cited in section V.C of the main report and includes assessments published in the IFAD Update in 1999 by IFAD's two Africa divisions. Viewed in the context of the IFAD and regional strategies, and the 1999 assessment, this points towards some of the interventions that IFAD-assisted projects logically should have made for assisting local governments through appropriate capacity building and pro-poor policy positions and advocacy. These aspects of IFAD assistance have been discussed in section IV.E.

As explained in the main report, the problems and challenges listed above also have implications for sustainability. This is important because the main expectation in the COSOPs and projects is that decentralized structures and grass roots organizations would lead to greater sustainability (quite apart from any other benefits in terms of efficiency and effectiveness). The evaluation highlights the fact that the combination of local government and community resources is not working satisfactorily as far as the sustainability of IFAD-assisted interventions is concerned.

While it is only natural to associate the lack of sustainability with lack of resources, this may simply be a roundabout way of avoiding a hard look at the root causes of the problem, and focusing on its symptoms. If lack of resources is a given—an obvious fact of life in the countries concerned—then the logical conclusion is that many if not most project interventions are unaffordable and unsustainable by the nature of their design or delivery. When there are widespread reports of lack of maintenance of infrastructure and the inability of communities to sustain IFAD-sponsored grass roots institutions, an invitation to seriously reconsider the way these important interventions are designed and delivered can only be considered overdue¹⁶.

It is impossible to state with any conviction that decentralization has enhanced the impact and sustainability of IFAD-assisted interventions, or made them more pro-poor or more accountable to the poor. Part of the problem lies in the wide range of capacity problems facing local governments and grass roots organizations at this time. This recalls an observation in IFAD Update (1999) that decentralization remains by and large work in progress. This is reflected, in the first instance, in low levels of efficiency and project effectiveness. Another part of the problem, however, is that projects are operationalized with little enthusiasm for some of the concerns that are stated to be priorities for IFAD, for example, empowerment, targeting, accountability, transparency and sustainability. While the projects have unquestionably had an impact in terms of rural and agricultural development, evidence of the impact on identifiable groups of rural poor has not been forthcoming, enhanced accountability is reported only occasionally, if at all, and the sustainability of infrastructure and institutional arrangements has remained an unresolved issue for a long time.

Recommendations

IFAD's approach to decentralization focuses not only on government structures but also on grass roots and non-governmental organizations (NGOs) that may be engaged for promoting grass roots development. It aims not only at improvement, accountability, transparency and sustainability in service delivery but also the more challenging goals of poverty alleviation and the empowerment of the poor. And it encompasses not only project assistance but also policy dialogue in some measure. International experience and the findings of this evaluation show that this is a combination of elements that would test the limits of most if not all implementers and policy makers.

But IFAD now has a wealth of relevant experience in decentralizing environments, including operational experience and reviews, and project and thematic evaluation reports, which it could utilize more systematically for meeting challenges. A distinctive feature of this experience is the degree to which IFAD emphasizes the symbiotic relationship between decentralization and grass roots organizations, in strategy as well as operations. The term "local governance" would be an appropriate way of describing the space that includes not only local government but also traditional rural institutions and the grass roots organizations sponsored by IFAD-assisted and other initiatives. At present, however, attention to local governance as a complete and useful concept is observed only rarely¹⁷ in IFAD documents. Policy dialogue that is adequately resourced and conducted in partnership with other stakeholders in a country is one way of sharing the depth and breadth of IFAD's experience in local governance. Systematic attention to policy dialogue would be ensured if it the idea is injected into country programmes through regional strategies and COSOPs, and realized through grant- and loan-funded projects.

This is not likely to happen, however, if the point of departure is provided by existing high-level strategies, COSOPs and project design documents. At present, the high-level strategies set up challenges that draw a wide range of responses from the COSOPs and the projects, ranging from ignoring the difficult challenges to addressing them selectively or incompletely¹⁸. Moreover, project design and implementation are driven more often than not by the imperatives of technicalism. An alternative point of departure, and one that would utilize IFAD's knowledge in the process, is to prepare a Local Governance and Poverty Alleviation (LGPA) strategy. The LGPA would be an operational strategy, and its main rationale would be to provide guidelines to help translate the overall IFAD and regional strategies into operational policies and procedures that are as free as possible from the limitations of technicalism. In other words, the LGPA would be a tool for circumscribing the permissiveness of strategic statements, helping the rural poor to find realistic ways out of technicalism, and encouraging project level innovations in these directions.

The evaluation report provides some food for thought regarding elements of the LGPA, as they have emerged during this evaluation. The evaluation suggests that IFAD-assisted interventions can be better attuned to the needs and circumstances of the poor if a number of steps are taken; these are outlined in Box 1.

The evaluation suggests that focusing on the organization and voice of the rural poor becomes particularly difficult in a context of rapid change and wide-ranging reform, when aspects of reform such as decentralization, privatization, downsizing and resource mobilization tend to pre-occupy implementers and policy makers. In these as well as more settled conditions, IFAD needs to be pro-active in "identifying the poor in rural communities" so as to ensure that they are included in project activities. This implies that every COSOP and project should include cost-effective mechanisms (such as wealth ranking) for identifying the poor within their communities and monitoring their participation in IFAD-assisted activities. This is a general recommendation, however, and is not limited by the context of this discussion on the proposed operational strategy for LGPA.

Even this operational strategy, however, would need to be reconciled with national and local priorities, operationalized as required and tested for its newness, and, if appropriate, upscaled and replicated in the given context. This suggests the need to approach local governance and poverty alleviation through a combination of instruments. Perhaps the first step would be to use grant funds to develop innovations and test local poverty alleviation strategies in selected areas, in partnership with local and senior governments and relevant donors.

Box 1: Suggestions for a Local Governance and Poverty Alleviation Strategy As food for thought, it is suggested that: COSOPs and projects actually implement what they are expected to do by the regional strategy in terms of targeting the poor within their communities; the poor should be actually consulted by the technical experts and their knowledge and priorities reflected in the design of activities; the design of interventions be driven not by technical blueprints but by the social, human and financial capital of the communities and local governments; the poor should be offered broad choices rather than menu-driven solutions; procedures for organizing communities and delivering services be simplified to the point of being accessible to the poor, instead of insisting that the poor must be educated and trained to comply with unrealistic requirements; projects be required to practice full disclosure of information to the beneficiaries and the mass media; accountability be strengthened by strengthening elected institutions to play their role more effectively; and IFAD should be willing to invest in the autonomy of the poor from state actors19.

The next steps would be aimed at upscaling and replicating appropriate lessons from grant-funded initiatives, and these could entail the use of loan funds for Sector Wide Approaches (SWAps) or more traditional projects. IFAD has begun to move in the direction of supporting SWAps in some of the countries of the region, including Tanzania and Uganda, most of which have emerged from national poverty reduction strategies. This offers the opportunity of taking crosscutting or sector-relevant innovations from grant-funded experimental projects into relevant sectors. The more traditional multi-sectoral projects offer the space for a broader set of lessons but within more limited geographical areas. Both types of projects offer the possibility of engaging partners whose support could be vital for IFAD to play a more effective role in policy dialogue.

---

1/ Within this broad definition three main forms of decentralization can be identified: (i) de-concentration: the transfer of some authority to lower bureaucratic levels within central government agencies; (ii) delegation: responsibility and resources for implementing specific tasks and delivering specific services are transferred to a public agency, a local government, a private enterprise, etc; (iii) devolution: local institutions have both high autonomy from the central government and high accountability to local service users.

2/ The methodology was developed with the participation of two external experts, namely, Mr Roger Slade, former Senior Manager at the World Bank Operations Evaluation Department, and Professor James Manor of the Institute of Development Studies, Sussex. Professor Manor also participated as an external reviewer at various stages of the evaluation, including the preparation of the final evaluation report.

3/ The evaluation was supervised by Mr Ashwani Muthoo, Senior Evaluator, and Mr Fabrizio Felloni, Evaluator. Dr Tariq Husain was the Evaluation Team Leader, and the Mission Leaders were Mr Abdul Rashid Khan for Tanzania and Dr Ramson Mbetu for Uganda. The desk reviews were undertaken by the Team Leader, three national experts (Mr Justin Maeda for Tanzania, Dr Ramson Mbetu for Ethiopia and Mr Daniel Mulumba for Uganda), and a junior consultant (Mr Bruno Cozzari).

4/ As elaborated later in this report, IFAD's strategy conceives decentralization as a framework for poverty alleviation. Single-commodity and single-input projects are not as closely aligned with IFAD's poverty reduction focus as the multi-sectoral or area-based projects that represent the norm at IFAD in recent years.

5/ The Ethiopia evaluation mission was led by Professor Richard Carter and included four other members.

6/ Before colonization, and for a short period after its independence, present-day Uganda consisted of a number of kingdoms including the well-known kingdom of Buganda.

7/ However, as pointed out by James Manor, external reviewer for this evaluation, decentralization seldom does much to promote rapid economic growth.

8/ The 1999 COSOP responded to regionalization and decentralization by emphasizing that implementation will be through the new structures, but it omitted any mention of policy dialogue. This was a critical time when the 1995 constitution was already in force and local governance was in a state of flux.

9/ Rural Financial Services Programme and Agricultural Marketing Systems Development Programme, both in Tanzania.

10/ Pastoral Community Development Project in Ethiopia and Agricultural Marketing Systems Development Programme in Tanzania; the latter includes policy reform as a project objective.
In the regions of Amhara, Tigray, Oromiya and the Southern Nations, Nationalities and Peoples Regional State (SNNPRS), the woreda administrations have offices for the following sectors: agriculture, health, education, rural primary roads, domestic water supplies and local administration.

11/ District government is responsible for service delivery in five sectors, namely, primary health care, basic/primary education, safe drinking water, feeder roads and agricultural extension.

12/ These officials include the Agricultural, Veterinary, Clinical and Community Development Officers.

13/ The main findings of the evaluation missions, focusing on five selected projects, are given in Appendix 3.

14/ The only reported exception is the Kibaale District in Uganda, which is one of the five districts in the DDSP project area.

15/ The Independent External Evaluation (IEE) of IFAD, the IFAD Annual Report on Results and Impact and previous evaluations by OE have also flagged sustainability as a major issue.

16/ The sustainability problems observed by the IEE (ITAD 2004) in various sectors are very similar to those identified here and elaborated in Appendix 3 of this report.

17/ For example, in the Tanzania CPE of 2003 and the project design of the Pastoral Community Development Programme in Ethiopia.

18/ As the IEE observes, "IFAD's strategic statements are highly permissive and sufficiently imprecise to make almost any work with the rural poor seem to be relevant."

19/ A recommendation for autonomous support organizations that perform as "honest brokers" between rural communities and service delivery organizations, including government agencies and NGOs, is explained in the 2002 Sri Lanka CPE. As explained in the Insight associated with this evaluation, such an organization would be "a self-governing, not-for-profit body to help organize the poor, promote participation, and ultimately strive for their empowerment and for rural poverty alleviation. Management would be strictly professional, its board to include some government officials (as a minority) and others known to be politically neutral and committed to development." An endowment fund contributed by the government and donors would cover recurrent costs. Where British law is in use, the organization may be established as a private company limited by guarantee.