See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/237468428 Methods for understanding urban poverty and livelihoods Article CITATIONS READS 8 34 4 authors, including: Arjan de Haan Michael Drinkwater 90 PUBLICATIONS 1,223 CITATIONS 12 PUBLICATIONS 143 CITATIONS International Development Research Centre SEE PROFILE WayFair Associates SEE PROFILE Some of the authors of this publication are also working on these related projects: Employment and Growth View project All content following this page was uploaded by Michael Drinkwater on 18 February 2015. The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document and are linked to publications on ResearchGate, letting you access and read them immediately. Methods for understanding urban poverty and livelihoods Arjan de Haan, Michael Drinkwater, Carole Rakodi and Karen Westley Introduction A livelihoods approach places households and their members at the centre of analysis and decision making, with the implication that household-centred methods of analysis must play a central role in developing an understanding of livelihood strategies and in programme and project planning and evaluation1. Knowledge is needed about the situation of and strategies adopted by poor households, in relation to both their characteristics and external opportunities and constraints. The methodological approach in such data collection and analysis is first, contextual and, second participatory. Methods that are contextual are those that attempt to capture a social phenomenon within its social, economic and cultural context (Booth et al, 1998). They are likely to generate qualitative and in-depth data. Rather than being purely extractive, such methods aim to link into better programme and project design by ensuring that respondents are (at least joint) owners of the knowledge and data generated, enabling them to participate in policy debate and project planning. However, such methods need to be complemented by larger scale data collection and quantitative analysis, in order to reveal the characteristics of the context, the overall dimensions of and trends in poverty, and the extent to which household characteristics revealed in in-depth relatively small scale studies are ‘typical’. Such data also provide a basis for assessing the impact of broader macro-, meso- and city wide policies and in helping prioritisation for resource allocation. Increasingly, both at national and local urban level, the limits of single strand analysis and the importance of triangulation are acknowledged (Booth et al, 1998; Moser et al, 1996). In this chapter, therefore, both approaches are discussed. First, Michael Drinkwater and Karen Westley describe a possible approach to developing an understanding of household livelihoods at the local level. Second, Arjan de Haan discusses the strengths and weaknesses of a variety of sources of quantitative data. That the discussion is organised in this way reflects the experience of the contributing authors rather than a desirable sequence: in practice data collection for analysis and monitoring of household livelihoods, deprivation and wellbeing, and for project and programme planning and evaluation should be an iterative process. In such a process different types and sources of data are used to generate concepts, questions and explanations which inform and challenge analysis based on alternative methodological approaches. 1 The material in this paper was prepared as part of a larger project commission by the UK Department for International Development’s Infrastructure and Urban Development Department to review current understanding of urban livelihoods, its policy implications and some recent experience of attempts to reduce urban poverty and deprivation. The project was coordinated by Carole Rakodi and Tony Lloyd-Jones and the results published in Rakodi with Lloyd-Jones (eds) (2002). The views expressed in this paper are those of the authors and not necessarily of DFID. 1 Household perspectives and qualitative methods This introductory section on methods for understanding urban livelihoods outlines the types of analytical processes and tools that have been used by the NGO CARE, for which the authors work, in different programmes in a range of countries. While both quantitative and qualitative tools are used, this section will emphasise the qualitative tools, locating them in an appropriate sequence with other methods. In order to place these methods in context, the first part of the section shows how the Household Livelihood Security (HLS) model that CARE uses provides an analytical framework that guides assessments in an holistic but flexible manner. The second part outlines the two main types of analytical processes that have been used in practice – a short participatory assessment during design and a longer process, with a wider range of purposes, during programme start up, with an example of each. Since urban livelihoods are complex, this account emphasises that, if analytical work is to be used to inform urban programmes, which people and organisations are involved and how are critical aspect of the overall process. Use of a livelihood model in urban programme design and implementation From the mid-1990s CARE International has been developing an expanding urban programme portfolio based on a household livelihood security (HLS) framework. This has been a particular emphasis in the southern and west Africa regions, although the urban portfolio now encompasses projects from all regions of the world where CARE operates. Many of the early urban projects focused on infrastructure provision, often in association with food for work. For example, CARE Ethiopia had a large urban roads improvement project using food-for-work, and in Lusaka, CARE Zambia’s PUSH (Peri-Urban Self Help) project began as a drought relief food-for-work activity, designed to ameliorate the environmental sanitation conditions that had led to a cholera outbreak (Sanderson and Hedley, 2002). Adopting a livelihoods approach, in both urban and rural contexts, led to two main outcomes. First, it introduced a holistic analytical process of programme design. And second, it resulted in the evolution of programmes -- whether multi or single sector -- that were oriented towards improving livelihoods. For example, in Lomé, Togo, a participatory livelihood analysis led to the design of a girls’ empowerment project that focused on girls’ informal education and the development of broader life skills which would help create more opportunities in their lives. These qualities are amongst those captured in the livelihoods programme design framework which has gradually evolved, largely through reflective negotiation, over the last few years. The representation of this framework in Figure 1 was used in a programme design workshop for Southern and West Africa, held in Malawi in January 2000, but is now also used more broadly in the organisation. 2 CARE’s Design Framework for Livelihood Projects Holistic Analysis O per at ional Envir onment Livelihoods A ssessment Dif f er ent iat ion/ disaggr egat ion S t akeholder Lens I nst it ut ional A ssessment H uman Right s Synthesis H ier ar chical Analysis Cause- Ef f ect Logic Visioning Pr ospect ive par t ner s Pot ent ial component ar eas Improved Livelihoods Reflective Practice Focused Strategy I nt ended/ Unint ended Changes Review impact on dif f er ent gr oups eg gender , yout h, poor I nst it ut ional Lear ning Change M anagement S har ing Coherent Information Systems I nt er vent ion Design I nst it ut ional S t r engt hening S yner gism Goal Def init ion Benef it / H ar ms Analysis I ndicat or select ion Logic M odeling S equencing of act ivit ies M &E Planning Benchmar king Figure 1 CARE’s livelihoods programme design framework What Figure 1 shows is how CARE has operationalised the HLS approach throughout the project cycle, with an emphasis on the assessment and design stages. A livelihoods approach is not intrinsically necessary to the process, and an alternative conceptual framework could certainly be used. However, in its application of the HLS framework in various contexts, CARE’s experience has demonstrated that it does allow a holistic perspective to be brought to bear on analytical and programme design events, through a diverse and adaptable range of tools. The nature of this framework is discussed below. Analytical features of the livelihood model in the urban context2 Living in an urban environment is clearly a distinct experience from life in a rural setting. Yet despite the contrasts in terms of context, there is one factor that remains unchanged: people themselves. Wherever people live, they retain essentially the same human needs, and the desire for the same entitlements or rights. They require access to productive resources such as land, knowledge and capital, and from these an income to support consumption needs. They require food, shelter, clothing, access to medical facilities, the ability to educate children, and the ability to participate, in all senses (socially, politically, intellectually and spiritually), in the society of which they are part. Thus these requirements amount to the entitlement each person has to lead a life that is fundamentally secure in respect both of the basic needs and broader social and psychological senses of a livelihood. 2 The text for this section is largely drawn from a powerpoint presentation produced by Chris Dunston, Urban Programme Director, for CARE’s Southern and West Africa Regional Management Unit Design Workshop, Lilongwe, Malawi, 24-28 January 2000. 3 The focal unit of a HLS analysis is quite clearly the household. For instance, in the analysis of assets, those held at other levels are not neglected, such as the natural resources or infrastructural services to which individual households have some form of entitlement. But what the analysis highlights is the exact nature of this entitlement to the household: what form of tenure, and de jure and de facto access rights do households and individuals actually have? CARE’s HLS framework has evolved through the organic accretion of lessons learned by different but collaborating practitioners across the organisation. In this process, three key analytical principles have emerged. They should be understood as cross-cutting, with some overlaps between them. The first key principle attempts to capture the key characteristics of an holistic analysis’. In CARE’s lexicon, this has three levels: it is contextualised, differentiated and disaggregated. · To be contextualized, the analysis has to explore the major political economic, social cultural and resource-based issues affecting households in a particular context; · To show differentiation, the analysis must provide a good understanding of social and economic differentiation between households; · And to grasp the disaggregated situations of diverse individuals, gender and generational roles and issues within the household or basic social unit requires critical but sensitive investigation. The second analytical principle refers to the ‘vulnerability context’. An emphasis is placed on understanding the wider shocks and stresses to which livelihoods are subject, since these factors provide some of the major challenges to the ongoing maintenance of basic consumption and asset levels, and indeed survival itself. Since much of our lives is often spent seeking to mitigate or cope with present or likely future stresses and shocks, understanding these is key to grasping what trends people are responding to, and thus what motivates them to engage in and adjust their particular livelihood strategies, both as individuals and as members of social collectivities. As shown in Figure 2, there are three major elements in CARE’s household livelihood model: context, livelihood strategy and livelihood outcomes. To understand these distinctly is the final analytical principle. Contextual factors place the household and community into a situated perspective. Chiefly it is factors around governance, government and policies; markets and macro-economic linkages; and civil society and broader support networks that are analysed – the ‘policies, institutions and processes’ which mediate access to the resources and assets required to sustain a livelihood. A contextual analysis aims to produce, on the one hand, an understanding of the key contextual factors affecting livelihoods, and on the other, an identification of the major shock and stress factors affecting livelihoods (see also Rakodi, 2002; Meikle, 2002). At the level of livelihood strategy, the aim of the analysis is to understand the typical levels of human, social, economic and natural capital that are possessed by different types of households, and the nature of production, income and exchange activities to which these give rise. Consumption activities for each household member can then be summarised in terms of the livelihood outcomes status for different areas of livelihood security. 4 CARE ’S LIVE LIHOOD SE CURITY MODE L ASSETS Natural Resources Institutions Natural Capital (resources) Human Capital (Livelihood Capabilities) Security of: Social Capital E conomic Capital (Claims & (Stores & Resources) Access) F ood Nutrition Infrastructure Health History E conomic, Cultural and Political E nvironment Demography SHOCKS & STRESSES CONTEXT Production & Income Activities Consumption Activities HOUSEHOLD Water Shelter E ducation Processing, E xchange, Marketing Activities LIVELIHOOD STRATEGY Community Participation Personal Safety LIVELIHOOD OUTCOMES A fter Swift, 1989; Drink water, 1994; Carney, 1998; F rank enberger and Drink water, 1999 Figure 2 CARE’s household livelihood security model Types of participatory livelihood methodologies and tools used in urban areas The types of analytical methodologies that CARE has been using over the past five years fall into two distinct types. Short duration participatory livelihood assessments are usually employed as part of the early assessment phase of a programme. Most often the intention is to generate a basic though holistic analysis of livelihoods, which will validate and add insight to already available secondary information, as well as produce strategic constraints, opportunities and recommendations from the participating communities. The second type of methodology is less intensive in nature and is generally implemented over a longer period of time, as part of a programme’s start-up phase. There may be initial intent to generate detailed diagnostic and project design information, but since these activities need to involve key stakeholders, ideally they should give rise to an on-going, interactive approach to project implementation, monitoring and evaluation. Often there is a desire to build partnerships and to develop ownership, not only by community based institutions, but also by broader municipal agencies. Thus, who is involved and how they are involved is as critical to these types of longer duration analytical processes as is the nature of the information produced. 5 In this section, a generic outline of participatory livelihood assessment methodologies and tools which can be used in both initial and ongoing investigations is provided and some of the practical issues which arise when using them in urban situations identified. Second, a specific illustration is provided of an overall design process in Maputo. Finally, an illustration of a sophisticated, longer duration start-up analytical process is provided from the Mahavita project in Antananarivo, Madagascar. Participatory livelihood assessment An account of participatory livelihood assessment methods will be provided in this section, with examples of how initial participatory diagnostic work can lead to on-going participation in all project activities. Assessment exercises run the risk of becoming exhausting, extractive, depressing and acontextual exercises with little application to programming, if careful attention is not given to their planning and the means of conducting them. From experience, we have found the following basic principles to be important. · Qualitative neighbourhood-based assessments need to be linked to the ‘big picture’ through good secondary data analysis. If secondary data is not available then it may be appropriate to collect quantitative primary data beyond the community level. · Tools do not stand alone; combined, they contribute to an iterative analytical process, as well as allowing the validation of data through triangulation and cross-checking. · There is no ‘checklist;’ Tools are flexible and provide a framework for dialogue and discussion. Therefore, additional information will arise and some may not come up at all, depending on participants’ priorities and interests. It is not possible to prescribe or predict all the possible information that will emerge. · The validity of data is proportional to the quality of the interactions between facilitators and participants. If participants do not understand the objective of an assessment and do not trust the facilitators, then the information that emerges may not present the real picture. An assessment may be the first step in developing a long-term relationship with a group. The success of that relationship depends on the first impressions. Table 1 provides a framework and description of research tools or methods that can be used to develop an understanding of household livelihoods in a specific area. Both area-level and household level issues are explored, and then compared to generate an overall picture of the context in which households operate, the linkages between the livelihood context, the household assets and livelihood strategies and livelihood outcomes. In general, community-level analytical methods such as mapping, venn diagrams, and historical profiles are used to initiate interaction. These are followed by more specific social analysis, usually through developing livelihood (or wealth) profiles for different socio- economic groups, on the basis of a set of key livelihood indicators identified by the community. More detailed household interviews add insight to the dynamic nature of the livelihoods of each wealth category. Key informant interviews may take place before, during and after. Focus group discussions around issues and themes may go on 6 throughout, but are often used following the livelihood profiling to discuss issues affecting different social economic groups. Table 1: Tools used for participatory livelihoods assessment Livelihood Component Themes for discussion and analysis Principal tool Livelihood context Presence and importance of community level institutions; interaction of population with external organizations; control of resources by organizations; formal versus informal institutions and organizations, e.g. Crime rings, gangs, slum lords Location of community with respect to topography –e.g. flood prone areas; slopes and hillsides; environmental issues: contaminated areas, dump sites; access to green space; traffic and safety. Availability of education, health, social services; water and sanitation infrastructure, roads and transport, markets; electricity, access of population and households to infrastructure Ethnicity; religion and gender; urbanisation patterns – did villages move ‘en masse’ to a specific neighbourhood, are there ethnic ‘ghettos;’ are there ‘indigenous’ people (villages swallowed by the city) Political parties; access to voting; feelings of insecurity/uncertainty at household and community level; informal controls through gangs/’mafias’ etc; police harassment; other harassment by state or informal structures. Impact of rules, regulations and policies on households and communities; access to identification documents; taxation (formal and informal); tenancy laws; regulations on ‘hawking;’ influence of zoning Macro-economic trends; urban economic base and activity mix; employment and cost of living (inflation) trends; policies and attitudes towards informal sector activity; micro finance regulations, frameworks and practices Skills; entrepreneurial ability; education level; ability to work; security of employment; income earnerdependency ratio Exchanges of goods and services; assistance to or from extended family networks in rural areas, other urban areas or overseas; membership in community groups; nature of interactions with other households; level of social isolation Land; home ownership; transport; equipment; shops; market stalls; household water and sanitation facilities; savings; salary; money from income generating activities; remittances; access to credit Venn diagram Type of activities undertaken by each household member, level of contribution to household economy; access to employment; income generating activities; access to credit; diversification vs. dependence on single earner; flows of money, people and goods from rural to urban areas. Occurrence, intensity and duration of flooding, such as earthquakes, war, riots, strikes, gangs, police harassment; increased levels of crime, power cuts. Nature and origin of neighbourhood associations; activities; external assistance and relief activities Nature of impact of external shocks on household; loss of assets due to shock; unemployment; illness; imprisonment; personal security Coping mechanisms, such as diversification of livelihood strategies; sale of assets; migration, etc. Shelter, food, nutrition, health, water, education, community participation, personal safety Organizations Natural environment Infrastructure Cultural environment Political environment Economic environment Household Human Assets (their nature and Social how they are used affects households’ ability to Economic recover (includes from physical and stresses and financial) shocks) Livelihood strategies (production, processing, exchange and income generating activities) Nature of shocks and stresses and responses Area level Household level Livelihood Outcomes Source: Sanderson and Westley, 2000 7 Neighbourhood mapping Tools for triangulation Key informant interviews; household interviews; secondary data Secondary data; key informant interviews Neighbourhood mapping Secondary data; key informant interviews Historical profile Secondary data; key informant interviews Key informant interviews Household interviews; Venn diagram; historical profile Secondary data; key informant interviews; group discussions Household interview Household interviews; economic activities matrix; Livelihood profile Household interview Secondary data; key informant interviews; livelihood profile Household interview Livelihood profile; key informant interviews Household interviews Livelihood profile; secondary data Household interviews; Historical profile; secondary data, key informant interviews Household interviews; Household interviews Secondary data; key informant interviews These core tools are often used more productively, at least initially, with men and women in separate groups. Groups may also be divided by age, marital status, occupation, or livelihood category if necessary and appropriate. Groups of participants should preferably come together at the end of each day to discuss findings and key differences. The type of process discussed here usually requires four to five days in each selected area to be completed well. Once the livelihood status of the population and the individual households is understood, a second, synthesis phase of problem identification and prioritisation, cause and effect analysis, opportunity analysis and visioning can take place. Experience has shown that ending an assessment at the problem identification stage leads to feelings of despondency and hopelessness amongst facilitators and participants. It is therefore valuable in terms of generating a more constructive outcome, in which participants are more likely to feel that they can directly use the analysis and synthesis themselves, to take assessments to a final level of identifying opportunities around which local action can be initiated, even if external institutions and resources are required as part of the process. A number of practical issues arise in planning participatory urban livelihood assessments. The most important are listed below, with guidelines on how might be addressed from CARE’s experience. Selecting areas: Before commencing an urban assessment, selection of the primary geographical units of analysis needs to be undertaken. In CARE’s urban work, a combination of two methods has usually been used: either to use existing information to select the areas that seem most appropriate for the purposes at hand, or to consult with key stakeholders. In both cases the criteria involved will affect the decision. For CARE, these criteria are often poverty related, but may include other factors such as access, the nature of the local and municipal institutions involved, and the presence or absence of other development agencies. A related issue is defining the limits of the area to be assessed. The boundaries of a village are often easier to delineate than those of an urban neighborhood. What scale should the assessment cover? Is it more appropriate to undertake a broad scan of an area, or enter into more depth in a selected neighbourhood or group of people? Generally, CARE conducts assessments within neighbourhoods, but often with extrapolation to the lowest level of formal local government. Even at this level, the population may reach tens of thousands: significantly larger that that of many rural villages. How committed is the organisation to working with the population? Before involving people in an intensive assessment process, the purpose of the assessment must be very clear to all stakeholders. If there is little or no prospect for on-going work in the area, then conducting this type of assessment is not appropriate unless the community has requested it, or clearly acknowledged the specific objectives to be of value to them. What is the best way to bring people together? In urban areas, there may be limited social interaction between family units in a given geographical area, and limited involvement in 8 neighbourhood activities. There is rarely a single obvious leader or ‘mobiliser’ to bring people together for discussion. Municipal authorities may have little contact with the population. Yet it remains important to find an acceptable entry point before starting activities. This may be a residents’ association, a women’s collective, or an NGO with a long-standing relationship with people in the area. Some urban areas may have a traditional leadership structure. For example, In Lomé, Togo, several villages were ‘swallowed’ by the city. In these areas, indigenous populations have retained their traditional chiefs and advisors. In Bangladesh, an assessment team used occupational groups as the entry point. Whatever the entry point, it is necessary that the survey team is able to use it to go beyond the initial group and explore other perceptions and experiences. Traditional or government authorities may be able to call a group together, but sometimes their presence can inhibit a participatory process. Another strategy would be to start with household interviews and key informant interviews, or to use school children, as a way to let people know about the assessment and arrange times for facilitators to meet with a larger group. This would help to reach people who may not already be involved in neighbourhood activities. Creating a neutral space: Political organisations should preferably not be used to arrange meetings, and meetings should not be held at sites associated with a political party or other interest group that might prevent participation of a broader group of people. Timing of activities: Activities may need to be held in the afternoons, evenings or early mornings, to accommodate the different work schedules of people in urban areas. For example, it may be easier to work with women street food vendors during the day and young men doing casual labour in the informal sector in the evenings. Afternoons are generally better than mornings. An assessment team will need to plan carefully if it wishes to meet with people of different ages and occupations. Extrapolating from group discussions to the wider urban context: In urban areas, due to the lower levels of social cohesion and organisation, it may be difficult for people in group interviews to talk generally about others in their area. Thus in filling out matrices on economic activities and livelihood strategies, or for wealth ranking, it may be necessary to restrict the information to those present at the meeting rather than using the group discussion to develop a generalised understanding of issues at the area level. This means, for instance, asking those present to list their own economic activities rather than listing general economic activities for the area. Linking qualitative and quantitative methods in participatory project design: the Kuyukana Project, Maputo As suggested above, participatory methods for generating qualitative and quantitative information are often combined with the analysis of secondary data, or with the strategic collection of more formally collected, primary quantitative data. An example from Maputo shows the relationship between the quantitative and qualitative work, as well as the steps that lead to an initial participatory assessment (Figure 3). 9 KUYUKANA PROJECT, MAPUTO, DESIGN PROCESS Figure 3 Phase One: Selection of Target Urban Area Secondary data review including statistics and other studies Document summary of secondary data and selection of urban districts with greatest vulnerability characteristics Individual meetings with stakeholders Stakeholders workshop to share findings and identify four neighbourhoods for detailed assessment work Phase Two: Community Level Environmental Analysis Carry out institutional mapping exercise Identify vulnerability gaps with communities Cross validate assessment findings with quantitative survey conducted by government Phase Three: Social analysis – livelihood strategies of vulnerable households Quantitative dietary/health assessment in two neighborhoods Undertake participatory social analysis in multiple stakeholder survey team (including municipal government) Present results of environmental assessment and social assessment and identify priorities for interventions with community members Prepare final report and disseminate findings 10 Participatory analysis for programme start-up: Community Approach Pilot for Programme Mahavita, Antananarivo The Mahavita urban programme in Madagascar provides a sophisticated example of an analytical process used for a broad range of purposes during a programme start-up phase. In Mahavita, this process was used to: · Orient and train staff and in particular to build teamwork as a way of working for the programme as a whole. · Pilot a community-based approach which facilitates discussions with communities, empowers them to analyse their problems and assists them in designing projects to improve their livelihoods. · Develop an approach for identifying and forming partnerships, as well as systems for strengthening institutions and providing technical assistance to social/public service providers and NGOs that affect the lives of the target population. · Link the above two to design systems of public/social service delivery for meeting priority consumption needs. The intention of the community approach which was piloted over an 8 month period in 1999 in three of the more than 30 areas in which the programme will eventually operate, was to develop a start-up process which could then be replicated more efficiently in the other communities. This process was termed FAMOA, meaning in Malagache, ‘identifying the problems and solutions together’. More specifically, the goal of the FAMOA process was seen as being, ‘to mobilise the community to take ownership of their situation and problems and to design strategies for improving them’. During this process, a representative community structure was developed with the capability to design, implement and coordinate local development activities, and, led by this structure, a common vision and plan for the design of community projects was then developed. The model for the FAMOA process, as refined from the pilot phase, has five stages: Stage 1: An opportunity to participate in the urban livelihood security programme is offered to previously selected communities. The requirements to begin participating are a letter of request and a calendar for proceeding into Stage 2. The result is an agreement between the community and Mahavita programme to continue. Stage 2: Initial establishment of a representative structure to facilitate the process and be the design leaders, on behalf of the community, takes place. This structure is formed to represent, in an inclusive way, all organisations, geographic areas and socio-economic groups within the area. An analytical exercise designed to gain a holistic view of livelihoods is begun in the community. Stage 3: This stage entails a more thorough ‘handing over the stick’ to the community structure, as it becomes both more representative of and more accepted by the community as a whole. A more exhaustive participatory analysis, problem diagnosis and needs assessment is conducted, 11 with CARE field agents acting as trainers and mentors to the process, but with the members of the representative structure facilitating all the activities. Stage 4: A common vision for development is created at this point, entailing the identification of points of leverage, exploration of potential interventions through opportunity analysis, and the elaboration of strategies. Goals, objectives, interventions and indicators are then agreed for the community development plan. Stage 5: Finally, a marketing plan is designed, and community members are mobilised to support and contribute to planned community development activities. Emphasis is placed on assisting the representative structure to effect linkages to service providers and other agencies that can contribute funding, technical assistance or services. A detailed implementation plan is finalised. Quantitative methods for understanding poverty and livelihoods This section discusses common methods that describe in a quantitative way issues related to livelihoods, including the assets available to households, their employment position , livelihood outcomes (or indicators of well or ill-being) and the context in which households live. The main aim of quantitative methods is to ensure representativeness: one of the main reasons why household sample surveys are needed, for example, is to obtain a representative overview of the welfare of the population. However, it follows from the complexity of livelihoods context and strategies that there are trade-offs: measurement of the many indicators that may be relevant for understanding of people’s livelihoods has costs, and in many cases there is a shortage of monitoring capacity. This implies that the approach here is rather different from the one in the preceding section, which focused on tools and methods in CARE programmes. Whereas an integrated and multidimensional approach may be feasible and desirable in planning and implementing specific interventions, policy making at a macro level - such as city planning or allocation decisions at national level - requires data at that level. This is not to say that information at the macro-level cannot be multi-dimensional, but practical planning considerations do require an approach that focuses on key indicators, as well as an understanding of trends and disparities. The different approaches are not alternatives: as noted above, household-centred qualitative and participatory approaches need to be linked to the bigger picture through secondary data analysis, and also qualitative methods are usually required to inform the design of quantitative surveys, or to help interpret their results (Carvalho and White, 1997). Therefore, the approach in this section is to discuss some of the more traditional methods of analysing poverty and labour markets and to indicate how these can be used for understanding livelihoods3. This is structured as follows: censuses and household surveys are discussed first. Methods of measuring relating to health and nutrition are then referred to, as these provide important measures of livelihood outcomes. As the livelihoods of people in urban areas are to a large extent dependent on labour markets, the final section focuses on this. 12 The basis for planning: censuses and household surveys Most developing countries have a fairly regular census: a quick count showed that at least 34 of the 50 countries in Sub-Saharan Africa had one during the 1990s.4 Censuses normally collect basic demographic, employment and shelter-related data. Their main strengths are their universal coverage and the potential for detailed disaggregation. However, they are implemented infrequently, take a long time to process and do not include income or expenditure data. For basic demographic data on people and their households, some key social indicators such as literacy and school attendance, information on economic activities (see below), description of housing conditions and access to basic infrastructure and long term trends, census data are invaluable. They monitor population trends including urbanisation and provide a crucial source of data for various planning purposes. They may also provide the sample frame for household surveys. If other analysis demonstrates that poverty is related to demographic characteristics, census data can be useful to analyse its location. Moreover, triangulation - comparing the results of different methods of data collection - may provide useful additional information. Hentschel (1999) explores the possibility of ‘inputting’ household consumption into census data, to form the basis of a ‘poverty map’. He concludes that the errors in poverty estimates warn against targeting using censuses as a basis, but that they may be useful to create a poverty map at a disaggregated level. He also suggests that they may be most useful for comparison with spatial patterns of other welfare indicators, such as healthcare centres. The problems with censuses are well-known (see, for example, Rakodi, 2002b). First, they are too infrequent to reveal short-term changes. Also it is extremely difficult to adequately register the whole population. In particular, mobile populations, like temporary or circular migrants in urban areas and homeless people, are easily missed out - though some national surveys do include information that allows comparison of the well-being of migrants and non-migrants (e.g. India, de Haan, 1997). Estimates for urban populations are very sensitive to definitions of boundaries. Moreover, what constitutes ‘urban’ differs greatly across countries. Redefinitions of rural as urban areas - sometimes for political purposes - causes problems for measuring trends adequately. Projections of population trends have been notoriously unreliable: in 1980 Mexico City’s population was expected to be 30 million in 2000 - the present 18 million or so is widely off the mark (Brockerhoff, 1999). For governments, poverty data need to be based on nationally representative household surveys. Without this, conclusions about poverty at the national level and about trends become meaningless. Such data also are essential to provide disaggregated information about the poor and are essential for targeting. International organisations have renewed their interest in such surveys, because of a move towards using outcome-based measures of progress and increased emphasis on quantifiable targets such as the International Development Targets. To obtain reliable information on poverty, expenditure surveys are usually preferred over income surveys, because of under-reporting of income and its fluctuations. Such surveys measure household per capita consumption – they do not measure inequalities within households. The measure of poverty is the number of people who fall below a poverty line, set in absolute terms (e.g. $1/day, minimum calories or food requirements) or in relative terms (as a percentage of mean income for example). Measures also exist to indicate inequality among the poor, and the 13 ‘depth’ of poverty, e.g. the distance below the poverty line of the average poor household. Household surveys focus on levels of consumption, but can include information about, for example, levels of education and economic activities. Household data usually show that rural poverty incidence and the depth of poverty tend to be much higher than in urban areas, particularly larger cities. Lipton and Ravallion (1995) show that rural poverty incidence tends to be 1.3 to 5 times as high as the urban incidence,5 although Satterthwaite (1997) argues that urban poverty in Africa, Asia and Latin America tends to be underestimated and its character misunderstood.6 Household surveys in India have provided extremely useful information about urban poverty, showing how the incidence of poverty has been higher in rural than in urban areas since the 1950s, how the gap slowly closed, but opened again during the 1990s. Research using these data has shown how urbanisation and industrial growth did less than rural growth to reduce overall poverty. Household survey data in India is exceptional in the South in terms of coverage and trends. In Africa, despite improvements in household survey techniques and capacity, lack of data remains a major hindrance to our understanding of indicators of livelihoods and well-being. As with all forms of research and data collection, such surveys have disadvantages and limitations as well as strengths: It takes time for the results to become available, because of the time required to process the data, but also because of the need to collect data throughout the year, to capture the effects of seasonality. World Bank Living Standard Measurement Surveys, the main instrument for poverty measurement in Africa, take at least two years to be completed, although the more restricted ‘Priority Surveys’ take less than a year. Indian household surveys similarly take at least two years to become available. Disaggregation is problematic. Though generally national household surveys allow for describing differences between rural and urban areas, and sometimes between major categories of urban areas (eg metropolitan and other cities), limited sample size means that further disaggregation is usually not possible. They therefore do not help in specifying the most vulnerable groups or areas within a city - specific surveys are required for such purposes. Definition of the Poverty Line is dependent on a range of assumptions about what constitutes the minimum expenditure needed to sustain a household. These assumptions may reflect technocrats’ rather than poor people’s views. They may not adequately encompass differences in living requirements between regions and between urban and rural areas and may not be adjusted to reflect changes in consumption patterns over time. Periodic re-definitions of minimum consumption requirements, however necessary, make analysis of trends problematic. Bar-On (2001), for example, shows how the technocratic redefinition of PLs in Botswana has exaggerated trend figures showing decreasing poverty and as a result reduced the proportion of people eligible, in principle, for social assistance. In addition to these technical problems, the definition of PLs is vulnerable to political manipulation. The data themselves do not ‘explain’ poverty, they merely record it. They do not describe livelihoods, but for the most part provide indicators of the outcomes of livelihood strategies. The 14 emphasis on consumption, by definition, limits the information on livelihood sources. It is not impossible in principle to extend questionnaires and include these issues, but practical considerations of questionnaire length make this difficult in practice. Although analysis of the data generated can provide some understanding of the causes of poverty, for a full understanding and to assess the effects of policies, qualitative information is essential, to shed light where survey data has not (regarding, for example, vulnerability, reasons for asset depletion, or survival strategies adopted). Income or consumption poverty is only one of the aspects of well- and ill-being. The data needs to be complemented with other indicators, relating to health, education (as described below) and less tangible issues like rights and empowerment. It cannot be assumed that poverty and other data overlap. A recent paper showed for six countries that income poverty and human development indicators correlate in most cases, but income explains very little of the variation in ‘non-money metric welfare indicators’ - many other factors determine knowledge and health (Appleton and Song, 1999). Methods can only imperfectly substitute for each other (Ravallion, 1996). · A focus on households implies that inequalities within households are not accounted for, although it is in principle possible to carry out surveys that focus on individuals, or have smaller sub-samples with more extended questionnaires. · Finally, as with census information, there are questions about recording, whether for example homeless people are included in such surveys. In China, where official urban poverty rates have been very low, migrants have been excluded from surveys and the household registration system has contributed to under-recording of urban poverty. Relatively new areas of work in the area of household surveys relate to vulnerability and a dynamic understanding of poverty, an issue covered quite well in more qualitative approaches. A limitation of household surveys is that they provide a static picture of well-being, especially if successive surveys are not carried out. Panel surveys, which have been developed recently in a small number of countries in the South, can indicate whether households move in and out of poverty, and how well-being changes over life cycles and generations.7 However, finally, it needs emphasis that quantitative measures do not exclude holistic approaches. Research on social exclusion in London (London Research Centre, 1996), or the recent Human Development Report for the UK for example (UNED-UK), show how residents in particular areas suffer from multiple and overlapping forms of deprivation with respect to unemployment, health and education. Measures of health and nutrition In addition to health indicators, which show the human capital available to households (Harpham and Grant, 2002), another important indicator of well- or ill-being is nutritional status, particularly among the poorest. Recent research by the International Food Policy Research Institute shows that food security and nutrition indicators in urban areas do not necessarily show that urban populations are better off than rural (Ruel et al (ed), 1999). Nutrition monitoring 15 generally measures height-for-age (indicating chronic problems for children) or weight-forheight (pointing at acute problems, also among adults). It is a common way of measuring wellbeing and tracking changes over time, in both rural and urban areas. It enables the impacts of factors such as market changes to be traded. Such characteristics are relatively easy to measure, objective and reliable, and cheaper to collect than expenditure or income data. Nutrition monitoring can build on existing institutional mechanisms, such as schools and data from health centres or clinics (though vulnerable groups may not use these facilities). Nevertheless, at a national level only half of the African countries seem to have data available. Like household surveys, nutrition monitoring does not explain the causes behind deprivation. Also, there are concerns about the quality of existing data in many countries, and in many cases figures are imputed. Often, information is available only on the distribution between households, though it is known that in some areas intra-household bias is a problem. At an international level, data availability has improved recently through Demographic and Health Surveys (DHS). These have been implemented in most African countries – though in relatively few more than once – since the mid-1980s, funded primarily by USAID. These surveys include information on population and population planning, under 5 and infant mortality, and child malnourishment. Analysis has allowed, for example, conclusions about differences in child mortality rates among migrants in cities (Brockerhoff, 1995). Data on child anthropometry are also becoming increasingly available through World Bank sponsored surveys. But these data suffer from common concerns about comparability, as different or successive surveys, even within the same countries, use different methodologies, sample frames, and reference agegroups, which are of crucial relevance for their outcomes. Finally, though data on child or infant mortality, life expectancy, literacy or enrolment rates, according to some sources, are available for almost all countries in the South, they can seldom be disaggregated below the city level and are reasonably reliable only for the survey year. Labour markets Within urban areas, labour and labour markets are crucial for people’s livelihoods (see also Amis, 2002). Whereas poverty, nutrition and health indicators provide information about the outcomes of people’s livelihood strategies, labour market information provides information about their livelihood activities: about levels and quality of employment and about the characteristics of the economic context in which they seek to earn a living.8 Labour market information may be obtained from censuses, labour market surveys, employment and earnings surveys or enterprise studies (e.g. surveys of small and micro-enterprises). Particularly in SubSaharan Africa, urban (as well as rural) labour market analysis has been relatively neglected.9 At the most general level, employment data - as reported by the ILO for example10 - provides information about levels of employment, usually disaggregated by divisions of economic activities. Censuses include virtually the whole population and can, if well designed, capture work in both the formal and informal sectors, paid and unpaid. Labour force participation rates provide an indication about what percentage of the population (of working age) is engaged in paid work. However, in Africa, according to the ILO, such data are available for only half of the countries, some are more than a decade old, and some are not disaggregated by sex (de Haan and Koch Laier, 1997). 16 A general concern with employment data is that the recording of characteristics of enterprises through surveys has had limited coverage, tending to focus on large-scale enterprises and the ‘formal’ sector. There is now agreement that employment definitions ought to be wide, including work in the urban ‘informal sector’11 - but problems of recording remain. Data on the informal sector are available for many countries, but often they are too old to be used for policy purposes, although some recent surveys of microenterprises have been sponsored by USAID’s GEMINI programme. Many urban employment activities are invisible or not perceived as work and never get recorded or adequately estimated. Nevertheless, data indicate that, at a global level, during the last two decades, there has been shift from formal towards informal employment (van Ginneken, ed, 1999). Though some organisations have compiled data on unpaid family workers (including unpaid workers in the subsistence and informal sectors), employment data focus on paid employment, and definitions often do not include a variety of activities pursued by household members, most notably reproductive and community work, which is largely performed by women. The 1998 SPA Status Report, which emphasised the need for national budgets as well as measures and monitoring of well-being to be gender-sensitive, also indicated the importance of time budgets which show that women tend to be more ‘time-poor’ (Blackden and Bhanu, 1999). Studies of crises and economic adjustment have shown that the burden falls unequally upon women: women have intensified both the number of activities in which they are engaged and the time spent on them, including both formal employment as well as activities which do not fall neatly into existing employment categories, but include domestic and community work. As a result of these biases, two-thirds of women's and one-third of men's total work time (rural and urban) may go unrecorded (UNDP, 1996). Data on wage levels and trends are available for many countries, but coverage is limited. Again the focus is on formal sector employment. Data on wages in the formal sector indicate that during recent decades wages increased as often as they decreased. For example, they declined in 26 out 33 non-Asian developing countries.12 Informal sector studies have provided information about income there as well - data collection for the informal sector is in principle not different, but it is of course much more difficult to obtain wide coverage. Most analyses have shown labour market differences between the formal and informal sectors, for example in terms of income, job security and working conditions, particularly for women, although economic adjustment has in many places been accompanied by casualisation of labour even in the formal sector. Unemployment data exist, but unemployment is generally thought to be an inappropriate or unreliable concept. In several countries the data does not distinguish appropriately between unemployment and informal-sector activity. Levels of recorded unemployment are often low, particularly in the poorest areas, indicating that poor people cannot afford to be unemployed. As with household surveys, India provides the best data on unemployment (both rural and urban), focusing on the proportion of days or weeks spent work-less and looking for work - this ‘time rate of unemployment’ is higher among the poor, and sharply so among the poorest (Lipton, 1995). 17 Though sector of employment is usually not a good predictor of poverty, research in Coimbatore indicated that a refined concept of ‘labour status’ may provide a guide to understanding livelihoods and their outcomes (Harriss et al, 1990). The study identified, with a simple questionnaire, seven forms of labour status, clearly distinguishing different jobs. Statistical analysis showed a strong correlation between labour status and various poverty indicators, providing insight into the relationship between different types of measurement, particularly the measure of ‘outcomes’ (income, consumption), assets and capabilities. This study was designed as a methodological pilot, but has not been followed up elsewhere, and its main findings can therefore not be generalised. Given the clear but complex inter-relationships between employment, education and poverty, employment data are vital to understanding urban people’s situation, but there are both practical and principal limitations to existing methods. Coverage of many urban economic activities, particularly those of the poor and of women and children, is limited. Unemployment data only provides a good indicator of hardship in a few cases. Moreover, employment data tend to be sectoral. It may provide information about, say, wage rates, in a particular sector. This does not present information about people’s or households’ income from various sectors – either within urban areas or between rural and urban areas - a crucial element of the understanding of livelihoods. Conclusion To understand urban livelihoods is a complex analytical process, for which, in CARE’s experience, a livelihoods framework provides a vital key. Methodologically both qualitative and quantitative methods are required, preferably with the quantitative methods informed by a qualitatively generated understanding, although often participatory methodologies build on a synthesis of formal surveys. Whichever approach to data collection is being used, urban analytical processes should involve all major stakeholders as much as possible. They are opportunities to develop the commitment of local communities and urban government, as well as a range of private, public and civil society stakeholders to any subsequent interventions that the analytical work is used to inform. Where analytical work on urban livelihoods is designed to provide inputs to programme and project design, who understands is as if not more important than what is understood. The description of various quantitative methods indicated above is also based on an assumption that there is no one best method for identifying livelihood components and monitoring the outcomes of household strategies and policy interventions.13 Censuses provide information on a basic set of characteristics, some of which may be directly relevant to well-being and livelihoods. They allow detailed disaggregation and the study of long term demographic trends. Household surveys may be used for a variety of purposes, particularly to provide a representative indication of well-being but can rarely be adequately disaggregated for use in urban policy and programming. Nutrition monitoring provides data on an alternative outcome indicator, and is a useful input for a variety of policies. Labour market data provide essential information about changes in the economic context in which urban residents seek to earn a living and about levels of economic activity. Each of these has specific uses and limitations, and in most countries improvements to their design, quality and frequency are needed. 18 Livelihoods analyses call for more nuanced and gender-sensitive definitions of poverty in household surveys and better recognition of the livelihood activities most crucial to the poor. However, because of lack of resources and capacity in poor countries, monitoring should concentrate on a relatively small number of indicators that are not too difficult to measure. A complex livelihood analysis, through qualitative methods, can help to identify which the most relevant indicators are, and these can then be taken up in quantitative surveys. The bottom line is that, whatever the shortcomings of quantitative approaches, policy makers cannot easily do without representative information about well-being and livelihoods in different areas and among different groups. Suggestions or recommendations regarding the optimal methods for monitoring livelihoods and well-being need to be practical and context specific. They need to start from an understanding of the most crucial gaps in data availability. National and sectoral policy making in many developing countries continues to be seriously hampered by the absence of regular monitoring and lack of capacity to carry this out. Holistic methods of livelihood assessment, as described in the first section of this paper fill one such gap, crucial for both increasing in-depth understanding and for particular interventions. Traditional quantitative methods are not holistic. Both qualitative and quantitative approaches are essential for various planning purposes, and much remains to be done to improve their quality and develop local capacity to carry them out. References Amis, P. (2002) ‘Municipal government, urban economic growth and poverty reduction – identifying the transmission mechanisms between growth and poverty’ in Rakodi, C. with LloydJones, T. (eds) Urban Livelihoods: A People-Centred Approach to Reducing Poverty, Earthscan, London, pp. 97-111. Amsden, A. and van der Hoeven, R. (1996) ‘Manufacturing output, employment and real wages in the 1980s: Labour’s loss until century’s end’, Journal of Development Studies, Vol.32, No.4, pp.506-30 Appleton, S. and Song, L. (1999) ‘Income and Human Development at the Household Level: Evidence from Six Countries’, Mimeo, University of Bath, Bath. Bar-On, A. (2001) ‘Providing so little for so few: Botswana’s social assistance scheme’, in Wilson, F., Kanji, N. and Braathen, E. (eds) Poverty Reduction: What Role for the State in Today’s Globalized Economy?, Zed Books, London and New York, pp.246-68. Blackden, C.M. and Bhanu, C. (1999) Gender, Growth and Poverty Reduction. Special Program of Assistance for Africa, 1998 Status Report on Poverty in Sub-Saharan Africa, World Bank Technical Paper No.428, Washington DC. Booth, D., Holland, J., Hentschel, J., Lanjouw, P. and Herbert, A. (1998) Participation and Combined Methods in African Poverty Assessment: Renewing the Agenda, Department for International Development, Social Development Division, Africa Division, London. 19 Brockerhoff, M. (1995) ‘Child survival in big cities: The disadvantages of migrants’, Social Science and Medicine, Vol.40, No.10, pp.757-778. Brockerhoff, M. (1999) ‘Urban growth in developing countries: A review of projects and predictions’, Population and Development Review, Vol. 25, No. 4, pp. 757-78. Carvalho, S. and White, H. (1997) Combining the Quantitative and Qualitative Approaches to Poverty Measurement and Analysis: The Practice and the Potential. World Bank Technical Paper No. 366, Washington DC. Datt, G. and Ravallion, M. (1996) Why Have Some Indian States Done Better Than Others at Reducing Rural Poverty? World Bank, Policy Research Working Paper 1594, Washington DC. de Haan, A. (1997) ‘Rural-urban migration and poverty. The case of India’, IDS Bulletin, Vol.28, No.2, pp. 47-54 de Haan, A. and Koch Laier, J. (1997) Employment and Poverty Monitoring, Issues in Development, Discussion Paper 19, International Labour Office, Geneva. De Haan, A., Lipton, M., Maxwell, S. and Martin, F. (2000) Poverty Monitoring in Sub-Sahara Africa, Background paper for the SPA Poverty Status Report 2000. de Haan, A. and Lipton, M. (1998) ‘Poverty in emerging Asia: Progress, setbacks and log-jams’, Asian Development Review, Vol.16, No.2, pp 1-42. Harpham, T. and Grant, E. (2002) ‘Health, health services and environmental health’ in Rakodi, C. with Lloyd-Jones, T. (eds) Urban Livelihoods: A People-Centred Approach to Reducing Poverty, Earthscan, London, pp. 165-79. Harriss, J., Kannan, K.P. and Rodgers, G. (1990) Urban Labour Market Structure and Job Access in India: A Study of Coimbatore, International Institute for Labour Studies, Research Series No. 92, Geneva. Hentschel, J. (1999) Combining Census and Survey Data to Study Spatial Dimensions of Poverty; a Case Study of Ecuador, World Bank Policy Research Working Papers No.1928, Washington DC. Hentschel, J. and Seshagiri, R. (2000) The City Poverty Assessment: A Primer, World Bank Technical Paper No. 490, Washington DC. Horton, S., Kanbur, R. and Mazumdar, D. (eds) (1994) Labor Markets in an Era of Adjustment. Washington: World Bank. 20 Lipton, M. (1995) ‘Growing points in poverty research: Labour issues’, in Rodgers, G. (ed) The Poverty Agenda and the ILO. Issues for Research and Action, International Labour Office, Geneva, pp.? Lipton, M. and Ravallion, M. (1995) ‘Poverty and policy’ in Behrman, J. and Srinabasan, T.N. (eds) Handbook of Development Economics, Vol.3B, Elsevier Science, Amsterdam, pp.22512637. London Research Centre (1996) The Capital Divided. Mapping Poverty and Social Exclusion in London. London Research Centre, London. Meikle, S. (2002) ‘The urban context and poor people’ in Rakodi, C. with Lloyd-Jones, T. (eds) Urban Livelihoods: A People-Centred Approach to Reducing Poverty, Earthscan, London, pp. 37-51. Moser, C., Gatehouse, M. and Garcia, M. (1996) Urban Poverty Research Sourcebook, Module I: sub-City Level Household Survey, Urban Management Program WP 5, World Bank, UNDP and UNCHS, Washington DC, New York and Nairobi. Rakodi, C. (2002a) ‘A livelihoods approach – conceptual issues and definitions’ in Rakodi, C. with Lloyd-Jones, T. (eds) Urban Livelihoods: A People-Centred Approach to Reducing Poverty, Earthscan, London, pp. 3-22. Rakodi, C. (2002b) ‘Economic development, urbanization and poverty’ in Rakodi, C. with Lloyd-Jones, T. (eds) Urban Livelihoods: A People-Centred Approach to Reducing Poverty, Earthscan, London, pp. 23-35. Rakodi, C. with Lloyd-Jones, T. (eds) (2002) Urban Livelihoods: A People-Centred Approach to Reducing Poverty, Earthscan, London. Ravallion, M (1996) ‘How well can method substitute for data? Five experiments in poverty analysis’, The World Bank Research Observer, Vol. 11, No. 2, pp? Ruel, M.T., Garrett, J. and Haddad, L. (eds) (1999) ‘Urban challenges to food and nutrition security in the developing world’ World Development, Vol. 27, No. 11 (special issue). Sanderson, D. and Hedley, D. (2002) ‘Strengthening urban livelihoods in Zambia: PUSH II and PROSPECT’ in Rakodi, C. with Lloyd-Jones, T. (eds) Urban Livelihoods: A People-Centre Approach to Reducing Poverty, Earthscan, London, pp.247-56. Sanderson, D. and Westley, K. (2000) Participatory Urban Livelihoods Assessment, CARE International UK, Urban Briefing Note, London. Satterthwaite, D. (1997) ‘Urban poverty: Reconsidering its scale and nature’, IDS Bulletin, Vol.28, No.2, pp.9-23. 21 UNDP (1996) Human Development Report 1996, Oxford University Press, New York. UNED-UK (United Nations Environment and Development UK Committee) (2000) Poverty in Plenty: A Human Development Report for the UK, Earthscan, London. van Ginneken, W. (ed) (1999) Social Security for the Excluded Majority. Case Studies of Developing Countries, ILO, Geneva. Yaqub, S. (2000) ‘Poverty dynamics in developing countries’, IDS Development Bibliography, Institute for Development Studies, Brighton. The authors Arjan de Haan is a Social Development Adviser for the UK Department for International Development. He previously worked in the Poverty Research Unit at Sussex University. He has carried out research related to poverty, urbanisation and migration. Michael Drinkwater is presently a regional programme coordinator with CARE International’s regional office for Southern and West Africa. He has an undergraduate background in urban and regional planning, a PhD in development sociology and 20 years experience of practical work and research in Africa. Carole Rakodi was for many years in the Department of City and Regional Planning, Cardiff University, she is now Professor of International Urban Development in the International Development Department, School of Public Policy, University of Birmingham, UK. She has worked as an urban planner and researcher mainly in Zambia, Zimbabwe and Kenya and also in Ghana and India. Karen Westley is Programme Manager for the Shell Foundation in London. Previously she was the programme advisor for design, monitoring and evaluation at CARE International UK. She holds an MSc from Yale University. Her expertise encompasses training and implementation of participatory livelihoods assessment, and design of monitoring and evaluation systems. She has worked primarily in West Africa, and also in Asia, Southern and Eastern Africa and the Balkans. 22