Participants agreed that there were still many areas that needed further clarification. These included:
SL approaches may help us to understand the poor, but do they help us reach them?
What are the most effective entry points for SL approaches?
What is the best way to achieve policy changes?
If the SL framework helps us widen the range of possible project interventions, does it give adequate guidance on how to prioritize them?
What indicators can we use to assess the impact of SL approaches?
What are the perceived omissions or inadequacies of the DFID SL framework?
There was substantial agreement on what the unresolved issues were, but no consensus on how to address them. Participants thought that more discussion, informed by feedback from applying SL approaches to project or programme implementation, would shed light on these areas.
Do SL approaches always imply a focus on the poorest? There was little consensus on this issue. Agencies committed to poverty alleviation are interested in SL approaches because they are seen as a way of contributing directly to meeting internationally agreed upon development targets.6 Other participants felt that the sectoral mandates of certain agencies would restrict their ability to reach the poorest.7
SL APPROACHES AND THE POOR
Although the Mongolia project was intended as a poverty alleviation project, its main activity - credit for reconstituting herds - failed to reach the worst-off, 50 percent of the rural poor. This was due to its concern that credit be repaid. The group felt that an SL approach would have improved the project's poverty orientation.
Although the Mali project assisted dryland farmers in one of the world's poorest countries, it largely bypassed the poorest households within the project area. The case study group concluded that adoption of SL approaches would have highlighted socio-economic differences within the project area, thereby offering the project the opportunity to target the poorest.
SL approaches may help us to understand the poor, but do they help us reach them? Five of the case studies seem to indicate that there is a fundamental issue regarding targeting any project interventions to the poorest. They point out that the projects have been successful with the "middle poor" and upwards.
The consensus of Forum participants was that it was still too soon to answer this question. The only SL-type projects analysed were Malawi, DELIVERI and Honduras. The remainder shared only some SL elements. However, the groups felt that none of them succeeded in reaching the poorest sectors of the communities with which they worked.
Although the projects reached people who were poor by international standards, in the local context these people could have been characterized as the "middle poor" or "borderline non-poor" (Bangladesh, Bolivia, Mali and Mongolia). The Honduras project promoted technologies suitable to different socio-economic strata, including those that were of interest to and were adopted by poor tenant farmers, yet it is not clear if they reached the poorest.
As illustrated by the Pakistan mini-case study, the existing pattern of incentives for line agency staff may not reward them for focusing on the poorest.
INSTITUTIONAL DISINCENTIVES FOR WORKING WITH THE POOR - PAKISTAN
Working with the poorest is not glamorous. The poorest are the least willing to change because they can least afford to take risks, and because they live in a state of basic survival and have the least time available for discussion or planning. Although women staff of this project made a concerted effort to work with the poorest villages, they found that they had to spend twice as much time in those villages and made less "progress" than they did in wealthier villages. Since staff performance was evaluated on positive results, and such results were easier to achieve among the non-poor, incentives were biased against working with the poor.
There can often be trade-offs between sustainable natural resource management and reaching the poor.
"It is important to bear in mind that natural resource management interventions that have public benefits do not always have direct benefits for the poor. If the poor are not involved in project activities, then consideration must be given to add components that address their livelihood needs. These needs may be addressed by other partner organizations and not directly by the project."
Tim Frankenberger, CARE
When projects take land or natural resource-based activities as their starting point, the implication is that they will work mainly with households owning or managing that resource. In spite of diagnostic studies and PRAs, the two participatory upland watershed management projects in Bolivia and Pakistan bypassed the landless poor because their strategies were land-based. Because the Bangladesh INTERFISH project was based on integrating fish and rice farming, it could not reach the functionally landless. According to its project document, the primary beneficiaries of the DFID/FAO 25-country West Africa artisanal fisheries project are to be: "the resource users in artisanal fishing communities, particularly the poorer groups, including small-scale traders and processors (mostly women) and consumers. Through a 'process approach' these beneficiaries will be helped ... to identify, implement and evaluate development activities that correspond to their own needs and aspirations". It will be important to monitor whether the project manages to reach the poorest households, whose members do not own boats but eke out a living by shore fishing, fish processing, crop farming, working as casual labourers or working on other people's fishing boats.
There was consensus that SL projects or programmes should be either single-sector or multisectoral. A range of options that can be applied, depending on where the project is in the programme cycle, is more important. Ongoing projects can incorporate SL perspectives during critical moments of the project cycle, such as during mid-term reviews or evaluations. This would help to clarify whether other factors, beyond the sector-specific constraints already being addressed by the project, could help or hinder it in achieving its objectives. The main grey areas for entry points are:
Who selects the entry point (government, donors or poor people)? Ideally, the intended beneficiaries - rather than the government or donors - should select the entry point. This will require a change from the current practice, whereby governments and donors select the entry points before beneficiaries can be consulted. Currently, agencies are not entirely free to empower communities to generate their own project ideas and are expected to respond to government requests. Before any beneficiary consultation can begin, the project must feature in the government's public expenditure programme. If a project enters the pipeline as a livestock project - as occurred in Mongolia - it may be difficult to change the institutional entry point from the livestock department, even if the project title is changed.
What is the best level of entry point for SL approaches? Some of the DFID participants at the Forum stated that SLapproaches could start at any level, but it was more cost-effective to work at the highest level possible. The higher the level, the higher the leverage to reorient policies, institutions and processes in order to make them more responsive, client-oriented and people-centred. However, that view is not supported by the experience of the Malawi project. This project focused its efforts at the top, attempting to mainstream SL approach concepts in all relevant national development plans and programmes (household food security, employment and sustainable NRM), but found it difficult to bring about concrete livelihood improvements at the village level.
Do SL approaches always need a policy dimension, and if so, what is the best way of bringing about policy change? The consensus among participants of both the Forum and the Web/E-Conference was, yes, SL approaches always need to consider the interaction between livelihood systems at the micro level and the policy environment in which they operate. Understanding these relationships is critical both at initial diagnosis and during project implementation. However, SL-type projects need not have an explicit policy objective.
There is also emerging consensus that using SL approaches upstream may add value, if they are appropriately linked to micro-level ground-truthing.
The DELIVERI project in Indonesia is an example of an institutional reform project whose strategy was to use information gained from piloting new approaches at the village level to press for policy changes at the provincial level and above. Conversely, DFID's approach to the Uganda Agricultural Advisory Services Project may illustrate the danger of donors embarking on a high-level policy dialogue aimed at institutional change without adequate ground-truthing at the level of livelihood systems diagnosis.
Participants acknowledged that although all agencies needed to be aware of linkages between livelihood issues and macro policies, not all agencies had a comparative advantage for policy dialogue at the national level. Agencies with a comparative advantage at the community or district level could also make a valuable contribution in their own right. They could also increase their influence on policy by establishing partnerships with other agencies that had a tradition of policy dialogue.
An important debate centred on the ethics of attempting to effect changes in policy. Although some DFID participants emphasized that it was more cost-effective to go directly to the top (to the Minister or even to the President) to influence policy, other participants at the Forum and Web/E-Conference emphasized the need to change the policy-making process to make it more bottom-up and demand-driven. They cautioned that policy changes imposed from the top down were unlikely to be lasting because they depended too much for their implementation on continuity at the top and were likely to be reversed when the Minister or head of government changed.
Policy is not exclusively owned by governments but is the outcome of a pluralistic consultation and negotiation process. Policy changes that are driven from below - that have a strong grassroots power base and are supported by civil society - are likely to be more lasting.
As the Forum discussions and Web/E-Conference contributions highlighted, there is clearly a need for further discussion on which institutions to target, and at which level - local, meso or macro - in order to establish SL-enhancing policies.
There was consensus that new types of outcome indicators are needed to assess the impact of SL approaches. Standard output-type indicators fail to capture the most important changes, such as project-assisted increases in the problem-solving capacity of beneficiaries. In particular, more work needs to be done on capacity-building indicators. Currently, we have few examples of indicators for measuring institutional improvements.8
BANGLADESH - FROM OUTPUT INDICATORS TO FARMER-DEFINED OUTCOME INDICATORS
Because of its initial concern with output rather than outcome indicators, the Bangladesh INTERFISH project's M&E system tended to underestimate project impact. Important impacts such as the empowerment of community members to identify their own needs, access outside resources and make informed decisions, were not captured by the original M&E system. For example, farmer field schools not only trained farmers in fish/rice integrated pest management (IPM) but also offered them the opportunity to gain experience in problem-solving, critical thinking and general field ecology.
The introduction of a participatory M&E system fundamentally transformed the project management's relationship with participants. For the first time, participants were asked to define how the project should measure success, and as a result the project began to think beyond the completion of its activities towards the quality of the activities and the longer-term outcomes. At the end of each rice season, farmers assess whether the learning process has helped them achieve the goal that they themselves have set.
There was less agreement among Forum participants on how to address the problem of relevant indicators. What is the best way to capture a project's impact on human and social capital development? How holistic should impact evaluations be? How can one avoid collecting too much unnecessary data? By involving the intended beneficiaries in selecting impact indicators relevant to their own priorities, the Bangladesh project was able to limit the range of outcome indicators monitored. But if donors are to compare project impact across villages, they cannot rely exclusively on the location-specific criteria suggested by beneficiaries. They will also need a minimum set of standardized criteria.
"The indicators that are used for monitoring and evaluation must be clearly linked to the problem analysis and the objectives. The project should not collect unnecessary data that are not clearly linked to the objective or the problem analysis. Programme information systems should be set up to capture both the intended and unintended consequences of programme activities. These lessons learned can be derived from participatory monitoring systems and other aspects of the M&E system. Capturing the lessons learned will be critical to programme improvements. One of the key problems that implementing agencies have is allocating time and resources to document the lessons learned.
"To measure the impact of a livelihood programme, it is important to measure both criteria relevant to communities as well as normative criteria. Criteria derived from participatory approaches are the changes that are meaningful to communities. If these changes do not occur then the project has not brought about the kinds of improvements that are significant to the community. These measures may be very location-specific. Normative measures are important for targeting and allowing for cross-regional comparisons. Such measures are critical for donors and governments that need to make resource-allocation decisions across regions or across countries. Thus, both types of information need to be included in SLA M&E systems."
- Tim Frankenberger, CARE
Several different versions of the SL framework have been developed. All are intended as tools to help people investigate livelihood strategies and understand them better. The frameworks were never intended as blueprints for analysis. Developers of the frameworks emphasize that no framework can contain every factor that affects livelihoods, and that the frameworks should be adapted and developed as practical experience is gained in working with SL approaches.
Nonetheless, Forum participants made several important observations about the framework and perceived inadequacies of its present form, which they felt could benefit from further development.
Making people visible. While the first principle of SL approaches is that they are "people-centred", people are not "visible" in the current framework. The fact that they are central to SL approaches could be made more explicit.
Levels of analysis of assets. The level at which livelihood assets are to be analysed is not immediately clear. It is important to distinguish between individually owned assets, such as land and livestock, and community assets, such as forests, grazing land, wells, roads and schools. Some means of distinguishing among these different levels needs to be introduced into the framework.
Socio-economic differentiation. It was also felt that the framework did not highlight differences between socio-economic strata within communities in their resource base, livelihood strategies and living standards. Members of poor households tend to be more vulnerable to external shocks than those of non-poor households because they own fewer assets. Policies, institutions and processes at the national level can result in highly skewed asset ownership. The poor usually have fewer livelihood options than the non-poor.
Incorporating historical and dynamic elements. Participants also felt that the SL framework was rather static. It was not immediately clear how dynamic elements, such as historical changes that did not necessarily contribute to vulnerability, might be incorporated into the analysis. It would be useful to emphasize the livelihood trajectories (ascending or declining) of different socio-economic categories of the population.
The framework needs to be sufficiently flexible to identify social networks that bind different livelihood systems together (Forum and Web/E-Conference). In many rural societies, socio-economic strata are bound together in interdependent livelihood systems, primarily where richer households with capital turn to poorer households for labour. Thus, livelihood analysis requires an understanding of how people depend on cross-strata social networks for asset-sharing, renting and co-ownership (e.g. of land, livestock, and fodder). Patron-client relations and reciprocal agreements for sharing labour or capital are important coping strategies for poor people in times of illness or other stresses and emergencies. Moreover, they lend legitimacy to the "rich" in positions of local authority. Horizontal social networks are also pivotal for addressing critical capital shortages among the poor, particularly through traditional labour-sharing and other reciprocal arrangements.
SL APPROACHES AND EXISTING SOCIAL NETWORKS
When DFID proposes a partnership analysis, it focuses on other international organizations, whereas it is equally important to analyse the partnerships that exist between stakeholders at the local level. Although SL approaches incorporate "everything", they tend to overlook the existing social networks that link households at the village level and connect them with the outside world. Any proposal on natural resources and poverty corresponds not only to the household but also to a network of interrelated actors, such as local authorities, small enterprises, community organizations, NGOs and the private sector. These networks are social structures that exist independently of projects. Projects that identify these existing social networks and build on them are more likely to be sustainable than those that attempt to create their own dynamics and organizations. However, since poor people may not necessarily have an equal voice in these local networks, there is a need for local capacity-building to empower the poor to build up their influence in the local context. - Web/E-Conference
Cultural dimensions. The SL framework is based largely on experience from English-speaking countries. French- and Spanish-speaking practitioners experience some difficulty in working with terms such as livelihoods, which cannot be translated easily into one word in their languages. At the same time, there is a significant body
of experience in both Spanish- and French-speaking countries that deals with the same issues but, perhaps, they have been conceptualized in different ways. A way of incorporating this experience into the framework and generating more culturally appropriate forms of the framework needs to be developed. Although sustainable livelihoods can be translated into Spanish as sistemas de vida y desarrollo sostenible, there seems to be no adequate French equivalent. In fact, none of the possible French translations below captures the concept fully.
Incorporating the political dimension. Most participants felt that, while the framework helped practitioners focus on policies, institutions and processes, the political dimension of these structures and processes needed to be made more explicit. Experience from the case studies suggests that projects or programmes attempting to address livelihoods in a holistic way will need to engage in the political sphere in order to have positive and sustainable impacts. Political elements that need to be reflected include:
transparency and governance;
the impacts of structural adjustment;
the impacts of globalization.
These all have clear political dimensions and potential impacts on livelihood strategies. Modifications to the framework should make it easier for people to incorporate these dimensions into their overall analysis. The impact of macro-level processes on livelihood trajectories needs more emphasis. In its SL framework, UNDP uses a hexagon rather than a pentagon, with the additional segment representing political capital.
Means of analysing policy, institutions and processes. The focus on policy, institutions and processes was seen as a valuable element of the SL framework, but practical methods that might be used to analyse this area were felt to be lacking and needing development.
SL FRAMEWORK AND PRIORITIZATION OF INTERVENTIONS
"The SL framework shows the interlinkages and helps us to understand the complex context, i.e. by broadening the way we can look at a situation ... but it doesn't help to prioritize or to come up with valid interventions. The biggest grey area is on the question of methods to use in the post-design phase. More work needs to be done to provide guidance on the process of getting from A to B. The SL framework increases the list of things to do but doesn't provide guidance on which one or how to tackle them."
- Zambia case study group
"The SL framework is an aid to good decision-making, and clearly does not negate the need for sound judgement. The holistic nature of SL approaches draws us to examine and prioritize options, but leaves us with no better tool for balancing the two imperatives of priority and feasibility than good judgement."
- Bangladesh case study group
The SL framework helps to suggest possible project actions but does not provide adequate guidance on how to prioritize among them.
The measurement of assets within the framework does not allow for comparison between livelihood systems or for quantification against targets.
ASSET PENTAGON ISSUES - MALI CASE STUDY GROUP
The SL asset pentagon has five dimensions: natural, physical, financial, human and social capital. In attempting to apply the pentagon, the Mali case study group discovered that the perception of project planners differed from that of the case study group and that three pentagons were needed - one for each socio-economic stratum. Moreover, it was difficult to combine several dimensions of one type of capital on the same axis. Regional and community physical infrastructure did not fit easily on the same axis with household assets. How do we rate natural capital on a single axis when land is abundant but of poor quality? How do we rate human capital when health and literacy are low but labour supply is abundant? When we rate social capital as high, medium or low, what is to be our yardstick of comparison (regions within the country, the world)?
The SL framework is not easy to explain. DFID and CARE report that although governments, technicians and project staff readily grasp the SL guiding principles, it is much more difficult to communicate the SL framework to such an audience. CARE has addressed the problem by using participatory workshops for staff training. Each workshop starts from the SL guiding principles, and participants are encouraged to derive their own framework on the basis of the principles. DFID has found that beginning with the detailed framework, it is not easy to get messages across, but building up to the framework by drawing on examples allows it to be well understood.
6 DFID and other agencies hope and expect that adopting SL approaches will contribute directly to the international development target of reducing by half the number of people living in absolute poverty by 2015, thereby providing a means of focusing on the root causes of poverty.
7 Depending on the local context, the poorest include the bottom two to three quintiles of the population, but not permanently disabled and very old people (for whom safety nets are more appropriate).
8 Tim Frankenberger, CARE