In ‘tackling hunger in a world full of food’ (WFP, 1996), the traditional food aid strategy has essentially been one of transferring food from surplus to deficit countries and then targeting this resource at the most vulnerable groups in these countries. Yet the dramatic global decline in food aid as a share of total ODA in recent years, and the declarations of the World Food Summit and parallel statements made by governments and NGOs (FAO, 1996a), suggest that all is not well in the ‘surplus transfer’ paradigm. Meanwhile, the supply-side analysis of food security finds new converts among those who fear globalisation and inequity in international trade. Here, so the argument goes, the only way to ensure food security and autonomous development is through a renewed promotion of local, national and regional self-sufficiency in food (ODI, 2000).
There is an alternative, counter-intuitive view of food insecurity which rests on the assumption that those short of food will only be able to obtain it in the long run if they are able to pay for it. Global statistics bear this out: the poorest 1.5 billion people in the world live on less than $US 1 per day and typically spend 80 percent of their income on food (ODI, 1997). Food self-reliance is thus a macro and micro-economic issue, highlighting the importance of markets, employment and distribution at household level, as well as food production. In developmental terms, food security requires the creation of assets, the generation of income, and the possibility for people to step beyond the cycle of poverty and indebtedness towards greater economic self-reliance.
The World Food Programme (WFP) has carved itself a specific niche within the UN system as the agency examining food security from the viewpoint of access and entitlements, as opposed production (making food available) or utilisation (using it in the best way). With limited resources, however, WFP requires realistic targets, clearly defined indicators and a thorough analysis of the ‘additionality’ that relatively small amounts of food can bring. The agency’s mandate, if not compromised, is certainly ‘stretched’ by realities on the ground. Once beyond the immediate ‘saving lives’ phase of food assistance, the use of food aid as a recovery or development tool can only be fully effective if placed within a cohesive, better-funded, and more comprehensive development strategy pursued by a broad range of actors. Critics of food aid too often neglect to mention that WFP’s developmental claims might be more qualified if the agency was not so frequently alone in the field.
Food and human security
International development cooperation since the early 1990s has seen a gradual shift in emphasis whereby traditional objectives – economic and social development, with its stress on poverty alleviation – have been supplemented with a set of new objectives – the promotion of human rights, democracy and good governance. The way in which food aid objectives and outcomes are now defined reflects the more holistic notion of human security. Food security - the adequate availability of and access to food by vulnerable people – is increasingly seen as one of the necessary preconditions to full participation in the process of development. UNDP’s Human Development Report 1994 first introduced the concept of human security as having two aspects: safety from chronic threats such as hunger, disease, physical and psychological abuse, and protection from harmful disruptions to daily life (UNDP, 1994). Famine, disease, ethnic disputes and social breakdowns were no longer distinct separable problems with their own individual set of solutions. Moreover, prevention and protection were easier to undertake and more successful than reactive interventions. From the mid-1990s onwards the components of human security were seen as interdependent, people-centred and fundamentally about sustainable livelihoods.
Sustainable livelihoods is a systemic and adaptive approach that links issues of poverty reduction, sustainability and empowerment (e.g., participation, gender empowerment, and good governance). Those living in extreme poverty and outside the formal labour market, for example, constantly improvise their livelihood strategies due to high uncertainty and limited options. Subsistence farmers in Cambodia will, in the off-season, become wage labourers and could later in the season revert back to farming. Similarly, in urban areas job security in the traditional sense is rare for the poorest people as they change jobs several times in their life time. Underlying these complex issues of human sustenance and livelihoods, is peoples’ relationship with and dependence on the natural environment.
In defining livelihoods, we refer to the dynamic relationship between opportunities and assets on the one hand and a range of beneficial or harmful variables such as ecological, social, economic and political variables on the other hand. In a recovery context we should also distinguish between coping and adaptive strategies. The former refers usually to short-term responses to specific shocks – floods, drought, displacement – whereby temporary changes in lifestyle require innovative adoption of new ways of surviving under stress. The latter refers to longer-term behavioural changes, such as adopting new income-generating skills, patterns of cultivation or gender relations as a result of new earning capacities in the family.
Can such adaptive responses be assessed and utilized as entry points for policy making? When it comes to evaluating the impact of project interventions within a sustainable livelihoods framework, the complexities are endless (Ezemenari, 1999). With its sophisticated Cambodia database, WFP has amassed much knowledge about beneficiary numbers, gender breakdown, food tonnages shipped, roads and canals built with food-for-work, and so on. Yet we still know little about the accumulative effect of all this. Although it is anticipated in project documents, measuring impact is a qualitatively different science. Placed within our broader definition of human security, there is a danger that food aid interventions will have a multiplicity of goals with potentially inconsistent and competing claims on limited resources. It is not unusual to find the following food aid goals explicitly stated even within one country programme:
- food aid to counteract the adverse human and economic effects of natural disasters, structural adjustment or economic shocks;
- food aid as a contribution towards human security in situations of conflict, political and/or social disintegration; or a contribution towards ‘social cohesion’, peace building or the like in the aftermath of conflict;
- food aid for more limited food security and poverty reduction projects for targeted vulnerable groups
It is inevitable that a multi-faceted programme, especially in a post-conflict situation, will deal with different levels of poverty and different stages of rehabilitation. What is remarkable is the similarity in food aid projects applied to relief, rehabilitation and development. Among newly settled populations in Northeast Cambodia, for instance, food-for-work as a form of ‘general distribution’ (with the caveat that the most needy will be “self-selected”) has been the only acceptable alternative to ill-advised free distribution  . Elsewhere in the country, the same tool apparently lends itself towards building local capacity, aiding reconciliation and developing economic assets. While not devaluing the potential role of food aid in each of these circumstances, we should be wary of over-ambitious goals. We have scant data for Cambodia and impact analysis is still in its infancy. As one commentator has suggested, there is a danger that in realising that we cannot actually do much, the development process becomes an unceasing series of agency simulation exercises coupled with periodic reinvention of identity when failure threatens to reveal itself (Duffield, 1999).
This said, in spite of well-rehearsed objections to food aid in principle (see, for example, Reutlinger, 1999) and persuasive evidence that it has little significant impact on sustainable development objectives (Pillai, 2000) we should at least continue to recognise its ‘second-best advantage’: that donors will continue to use it as a quick and fairly large-scale answer to problems when more permanent measures against the roots of poverty are not available or fundable (Schulthes, 2000). The terms and focus of the debate on food aid is now much narrower than before: it may be an effective safety net for livelihoods and food security in circumstances of short-term food scarcity or market breakdown; it could be a useful income transfer to the most needy families suffering chronic or transitory food insecurity; and it might under certain circumstances ‘enable’ development (WFP, 1999). Since Cambodia represents a ‘graduation’ from food assistance to more developmental approaches, we now look at what kind of additionality food aid brings to this process.
We begin by examining some of the underlying assumptions of WFP’s rationale for food aid in Cambodia in the light of recent evaluative findings. The argument runs thus: in the absence of substantial private and public investment in the country, WFP - the only food agency with anything approaching a countrywide coverage – provides essential medium-term assistance prior to, and in anticipation of, more regular development assistance. It also – through its geographical presence and unique data on poverty at commune and household level – acts as a catalyst and advocate for future development interventions. More specifically:
- recovery is an incremental and localised process in which a food-based safety net offsets the worst effects of chronic and transitory food insecurity at household and village levels.
- food aid interventions mitigate the worst effects of chronic poverty, economic shocks and/or disasters;
- to maximise their impact they should stimulate or complement other development measures which together lead towards sustainable food security;
- as a tangible resource, food aid can stimulate recovery by helping build and reinforce capacity at national and local levels.
There are three broad categories of food insecure people in Cambodia. The largest group are some 1.8 million people only recently recovering from displacement and conflict, and whom we generally refer to as ‘chronically food insecure’. Of these, 1.43 million currently receive WFP food assistance through an ongoing recovery programme.  The second group is the ‘vulnerable population’ numbering perhaps 500,000, including orphans, physically handicapped, elderly, HIV/AIDs and TB patients, of whom 150,000 are currently being assisted through WFP/Government/NGO institutional sector mechanisms (WFP, 2000a). A third and new group are the ‘transient food insecure’ such as those affected by the 2000 floods, the worst in 40 years. Of the 3 million affected, about 500,000 were displaced and received short and medium-term assistance from WFP under an additional emergency operation (WFP, 2000b).
WFP’s role in emergencies is uncontested and will not be covered here. Rather, our focus is on the role of food aid in a transitional context, where recovery makes way for development and food aid becomes institutionalised. Much of WFP’s experience worldwide points to the conclusion that food aid has a much more natural place in social development than in economic development. A basic cost-benefit analysis of, for example, a food-for-work scheme, would invariably reveal disappointing results if measured purely in terms of economic return. We must therefore look at the manner in which food aid can help create, or at least maintain and rehabilitate, assets – the building blocks of human security.
There is growing evidence that livelihood security in Cambodia depends more on retaining or buying rice than on producing more. A UNICEF-WFP report in 1998 showed that less than 25 percent of rice growing communes, representing about 15 percent of the population, produce 75 percent of the country’s surplus (UNICEF-WFP, 1998). Moving food around the country and encouraging markets is the first issue; the second is securing what people already grow (through investments in draught animals, credit schemes, improved health care, etc); and the third is focussing on how people earn money and how important this is vis a vis subsistence agriculture. Murshid (1998) surveyed three provinces and noted, for instance, that non-agricultural household income sources in Kompong Speu and Kandal were as high as 50 percent of the total. By contrast, they were only 15 percent in Prey Veng. The conclusion is hardly surprising: food security depends not solely on production but also on the relationship between how poor people make a living and what assets they have to secure their ability to obtain food.
A sustainable livelihoods approach would define assets as more than simply the physical results of a food-supported project. Rural people can access, build and draw upon five types of capital assets: human, natural, financial, social and physical (Turton, 2000). The adoption and investment in these are driven not only by personal preferences and priorities but also by external structures and interventions (organisations, institutions and policies). The following is a broad outline of the current status and challenges facing each of the five in Cambodia.
- the legacy of the past and its destruction of a generation of leaders; disruption to knowledge networks caused by displacement;
- the highest level of under-5 malnutrition in SE Asia; generally poor levels of health and education;
- large number of single-parent households where women shoulder a large burden of agricultural work.;
- a significant ‘brain/labour drain’ to urban centres.
- the wide variety of agro-ecological zones and productive natural resources – rich forest lands, coastal and inland fisheries;
- little consensus on the area of Cambodia under forests or on the current rate of deforestation and poor analysis of the impact and inter-relationships between deforestation, flooding, drought, fisheries depletion, etc.  ;
- poor analysis of how these wider trends affect household livelihood strategies;
- property rights weakened by privatisation and over-exploitation of forestry and fisheries;
- increasing levels of landlessness, irregular tenure agreements and hence uncertainty leading to low investment in land, and sometimes forced take-overs;
- constrained access to water for irrigation.
- loans mostly taken from moneylenders at high interest rates – few affordable credit alternatives;
- the necessity for poor farmers to often sell a major proportion of their produce immediately after harvest to repay debts;
- an increase in land sales  ;
- an increase of young men seeking alternative income in the cities – casual labour, cyclos, etc.
- strong patron-client relationships;
- variation in the forms of social capital depending on (a) whether communities are homogenous (original residents, even if recently re-settled), heterogeneous (new settlers, a mix of returning refugees and original residents) and/or previously under Khmer Rouge administration;
- the extent to which collective action is grounded in a genuine participatory planning process (e.g. the successes and failures of the Seila process). But participation is still very new, depending very much on external inputs and relationships with outsiders or with new political structures.
- Basic infrastructure – roads, water supplies, health, schools – is still very weak. However, in some areas projects to improve inter and intra-commune infrastructure have proven successful, particularly where commune ownership is demonstrated.
To this list might be added a sixth asset: the legal or regulatory framework which protects people from exploitation and resolves issues of property and land ownership, access to public services and security. This leads several donors to invest in the legislative, policy and institutional framework of national and local government which, although essential for the longer-term development of Cambodia, has so far had little impact on the majority of the rural poor.
Food-for-work: a strategy for generating and maintaining assets?
The majority of WFP Cambodia’s project portfolio is in village-based food-for-work assistance in targeted areas. With an average of 38,000 tons disbursed annually since 1999, food-for-work has absorbed about 80 percent of this. Rural roads, schools and health centres have been constructed or rehabilitated and land cleared for resettlement. In order to enhance agricultural yields, irrigation facilities, dikes, dams, ponds, water gates and canals have also been rehabilitated or constructed. In an interesting – and expanding – pilot programme in the five Northwest provinces, a certain tonnage of food has been allocated to designated communes through the ‘Seila’ process as a development resource to be utilised in conjunction with funds provided by other donors, including government (Bennett & Wallgren, 2000). Emergency feeding and social support for vulnerable groups has absorbed the remaining 20 percent of WFP’s assistance (relief distributions temporarily increased during the 2000 flooding).
Evaluations of WFP’s food-for-work have identified several constraints  :
- Inclusion/exclusion errors (i.e. how to concentrate limited quantities of food aid on the neediest people). Given the general scarcity of employment opportunities, there is a tendency for households with enough labour to diversify their income base; this makes food-for-work a welcome addition, not an alternative, to other employment. Also food-for-work participants work in shifts, allowing one family to sometimes have several members working on one scheme without disrupting regular farm work. In the absence of household-specific targeting criteria  , the participation of all income groups cannot be avoided and self-targeting will not always ensure participation of the poorest.
- WFP recognises that the very poor rarely participate in FFW schemes. There may be several reasons for this. Single-headed households (especially women with young children) cannot find childcare. Another reason is the late arrival of food. Although a scheme might be approved early in the year and the work undertaken mid-year, first payments are rarely made until September/October (the justification being that this is the “hungry” season). The very poor will meanwhile borrow food with interest, making the WFP capped rate of no more than 150kg family/year unattractive.
- Where there is an established CDC (Commune Development Committee, a central component of the ‘Seila’ (local participatory planning) structure in five northwest provinces), this should select villages for food-for-work based on their being the poorest within the commune. In spite of wealth ranking methods to identify the poorest villages, the CDC will sometimes choose the ‘richer’ villages, not least because these will be more likely to attain the 3 percent cash local contribution required (a condition of the Seila process) before the project is approved (for further details, see Bennett & Wallgren, 2000).
- A significant number of projects are, for various reasons, not completed, despite contractual undertaking by community leaders at the outset. Sometimes the project duration interferes with planting/harvest obligations and the labour force dissipates. It does, however, raise the question of whether the infrastructure itself is less important than the food, whether community ‘ownership’ is not truly established, or whether project decisions lie in the hands of an unrepresentative few at commune level. Certainly, there are cases where ‘imported’ labour from neighbouring communes underlines the food-as-wage ethos – and universally poor voluntary maintenance of infrastructure confirms this. It could, however, also reflect deeply imbedded short-term thinking among people for whom ownership and participatory planning are very new concepts.
- There is little information on the long-term impact of assets generated through food-for-work. Regular technical reviews comment on the quality, longevity, work norms and completion schedules of physical infrastructure, but not on their wider impact.
How targeting is done
Because food resources are limited, and because it is undesirable to have a blanket distribution which may flood markets and create dependency, targeting is central to WFP’s strategy. Designing any targeting system requires choices in three areas:
- the mechanism – who does it ( will beneficiaries be selected by administrative procedures, by their own choice, or by community decisions?);
- the targeting level (should the unit of targeting be a geographical area or population group, and/or should we discriminate between recipient and non-recipient households/individuals);
- the criteria for defining the target group ( demographic, nutritional, socio-economic, gender, etc) 
In Cambodia, WFP targeting is primarily based on a set of indicators of poverty and food insecurity at commune level. Until recently, WFP has been less concerned about particular households or individuals, not least because participation in food-for-work activities is high, reaching between 60 and 80 percent of the male and female population in the vicinity of the work.
In order to identify relevant beneficiary-groups, WFP in 1996 established the Cambodian VAM (Vulnerability Assessment Mapping) Unit which continuously assesses risks of food insecurity and the population’s coping ability. These assessments are based on series of indicators which include survey results for crop production, landlessness, income diversity, individual and village assets, and coping strategies. Relevant measuring factors include indebtedness, migration, access to basic services and markets as well as security. To more precisely identify the impact of factors like droughts, deforestation and poorly developed water resources on food security, food economy zones have been identified in Cambodia, based on land cover types, agricultural statistics and socio-economic survey data. Two of these zones - lowland rain-fed and scrub/degraded forest – are primarily targeted by WFP. There is, however, some concern over the adequacy, validity, and reliability of these indicators, particularly as they provide little information on income from non-rice and non-forest sources.
Lessons learned through the targeting process include the following:
- Targeting has often been a complex process, demanding a significant amount of staff time and attention. The process has helped to sensitise staff and partners to the targeting process and may have shielded WFP and its staff from undue political influence on the selection of target areas.
- Changing targeting criteria and procedures almost annually has hitherto produced a degree of discontinuity among target communes and their respective villages. From 2000 onwards, therefore, WFP moved towards the concept of clustering communes and integrated regional planning.
- Food aid benefits are maximised through complementary and parallel interventions by other development partners. In cooperation with the RGC, WFP has expanded links with partners such as the UN system (especially UNDP, UNICEF, IFAD, FAO, WHO, UNFPA, and ILO), World Bank, Asian Development Bank, EU, KfW and GTZ as well as NGOs in areas where integrated projects are planned.
- Some partner development organisations will work in the same district or regional, but not necessarily in WFP’s target communes. Such partnerships will sometimes create contradictions between competing criteria and methodologies of the various agencies. For example, the site for a World Bank or ADB project may be selected according to minimum economic viability, not necessarily where the poorest live.
- The same might be true of future planning priorities of the government. Building an integrated road network for the medium-term goal of economic growth may assume greater importance than assuring food entitlements (short-term goal) for those working on the road. This is a classic dilemma in a developing country where WFP is often (wrongly) perceived as a ‘road building’ agency.
- Too much emphasis may have been placed on geographic targeting at the expense of targeting more distinct groups of food-insecure beneficiaries with specific vulnerabilities, based more on demographic criteria or livelihood strategies. In the new recovery phase (2001-2003), WFP is increasing its Social Support projects alongside FFW schemes to ensure that particularly vulnerable people (e.g. young single mothers, handicapped persons or the elderly) are not marginalized;
- A comprehensive social welfare strategy is, however, beyond the capacity of WFP and its partners. It depends ultimately on mainstream policy support, good public administration and civic responsibility. WFP can merely act as a catalyst and advocate through a series of pilot schemes.
This last point raises the issue of absorptive capacity. One of the biggest problems in development aid is that, irrespective of need, disbursements are usually slow. In general, this has not been the case with food assistance, where effective and visible quick disbursements have encouraged donors. However, there is a point at which administrative overheads and recurrent costs, in addition to already high transaction costs, will make a general food safety net programme beyond the scope of most developing countries, not least Cambodia  . If, therefore, selectivity is required, are we still able to retain the notion of a safety net?
What do we mean by safety nets?
The ‘safety net’ that food aid provides for the most vulnerable cannot be treated as an anachronism. The necessity for social welfare may be the Achilles’ heel of the relief-development continuum, but it remains a central element of food assistance. In general, a food-based safety net is one that provides direct transfers to the poor in kind (employment-based food distribution or subsidies). Such transfers support those who are likely, within the time-line of the programme, to remain absolutely poor – either because their incomes will not rise rapidly enough as a result of the wider growth interventions, or because they will not be reached by them. In a recovery context, it is recognized that some programmes – public works and credit, for example – may simultaneously serve both transfer and income-growth objectives. In terms of measurable outcomes, however, the safety net concept does not include programmes designed to raise incomes more permanently.
To some extent this is problematic for WFP, for among its recovery objectives is the ‘building and rehabilitation of essential assets’ (WFP, 1999). Nevertheless, such an objective is qualified by the notion that this is a temporary stop gap measure, complementary to, and in anticipation of, more substantial and durable development investments which cannot be provided through food aid. Moreover, WFP’s increasingly refined targeting is aimed at the ‘poorest of the poor’, implying that food is not a general resource for development but rather a selectively applied intervention.
It is precisely this selectivity which has come under most scrutiny, as we have seen above. Selectivity has two aspects: the first is that interventions will need to be selective of certain sub-groups from among the poor; the second is that ways need to be found to ‘leverage’ the limited expenditures that will be possible on safety nets so that they have the greatest welfare impact.
The most likely way of doing this is by using the transfers they provide to reduce risk. Judiciously supplied inputs can help cushion these risks, improving the welfare of the poor without necessarily transferring large amounts of resources to them. For Cambodia, the analysis of risk might include some of the following categories:
- Seasonal risks associated with the cyclical food shortages and price rises that occur between September and February each year;
- Catastrophic risks, associated with drought, flooding or other external conditions;
- Price shocks - not yet widespread, but may result from depleted crops following flooding; and
- Idiosyncratic risks – associated with the characteristics of a particular household – such as illness, loosing a breadwinner to AIDS, etc.
It might be argued that changing our perspective to risk analysis is simply rearranging furniture: it makes little difference to the outcome. However, it does have a bearing on policy in two respects.
First, it reminds donors and the government that the choice of food aid instruments should reflect a balance between technical cost-efficiency and assessment of risk. Is there a role, for example, in selective food subsidies in urban areas where the rural poor become economic migrants? Are there aspects of food insecurity, that are distinct from poverty more generally, and which would warrant a range of different food-based interventions?
Second, it lends itself to preparedness as well as response. Some of the above risk categories are predictable, both qualitatively and quantitatively. WFP’s advocacy strategy – an explicit component of the current three-year recovery programme – highlights issues not only of emergency preparedness, but also the awareness of, and responsibility for, food security from village communities to district and provincial government authorities. The effectiveness of advocacy will depend on strengthening coordination and information exchange between all stakeholders in Cambodia so that a ‘common platform’ on food security evolves. Are non-food donors, for example, willing to invest in information, monitoring/evaluation and administrative networks necessary for this to become a reality?
Conclusion: measuring asset creation
The current shift in thinking and action towards a more people-centred, human development paradigm has necessitated a concurrent re-orientation of the policies and programmes pursued by development agencies, NGOs and governments. Evaluating these efforts has meant looking beyond conventional quantitative indicators to more qualitative ones. This has been, by no means, an easy task. Qualitative processes such as empowerment, for example, do not easily lend themselves to being objectively measured.
For WFP there has always been a difficult balance between process and product, and between project quality and the quest to distribute a pledged food tonnage within a set period of time. WFP is a victim of its own success in having pared down staff and administrative overheads over the years: it simply cannot ‘follow the food’ with the dedication intimated in copious policy documents, monitoring guidelines and the like. It will always be as strong as its weakest link – the number and capacity of monitors, partners and community implementers. If food-for-work creates some medium-term assets, WFP needs to be much clearer over what these assets are, how sustainable they are, and how we can trace a direct correlation between inputs and outcomes of such projects.
Labour-intensive rural works have income-generation by means of employment creation as a primary objective. Output indicators are thus assured: food wages impose few transaction costs on the beneficiaries and are proof against variations in prices, particularly important during the ‘lean’ months leading up to the harvest. Secondary objectives are less easily accounted for. If, as we have suggested above, assets are more than simply physical infrastructure, qualitative indicators must be developed as part of the ongoing monitoring role.
We have already questioned whether sustainability in the traditional development sense, should or could be within the remit of food assistance programmes such as those in Cambodia. But even our safety-net concept involves cross-cutting objectives pertaining to livelihoods. Livelihood systems and groups (i.e., individuals, households, communities) are dynamic in nature. Sustainability and vulnerability are processes, not events. They are not necessarily either/or terms. Notwithstanding its institutional constraints, WFP should begin developing indicators for projects undertaken in rural areas which combine elements particular to poverty, environmental sustainability and empowerment, thus linking more constructively both social and economic aspects of food aid interventions.
Bennett, J (1999) ‘North Korea: the Politics of Food Aid’, RRN Network Paper, No.28, Overseas Development Institute: London.
Bennett, J & Wallgren, P
(2000) ‘Evaluation of the WFP-CARERE-Seila collaboration, Cambodia’, WFP Cambodia, Phnom Penh: October 2000.
Clay, E & Stokke, O (2000) ‘Introduction: Food Aid and Human Security’ in Clay, E & Stokke, O (eds) Food Aid and Human Security, Frank Cass: London, 2000.
Duffield, M (1999) Unpublished paper on Sudan (draft) made available to the author.
Ezemenari, K et al (1999) ‘Impact Evaluation: A Note on Concepts and Methods’, Poverty Reduction and Economic Management Network, The World Bank, May 1999.
FAO (1996a) Rome Declaration on World Food Security and World Food Summit Plan of Action, Rome: FAO.
FAO (1996b) Background paper for the 1996 World Food Summit, FAO: Rome.
McAndrew, J (1998) Interdependence in household livelihood strategies in two Cambodian villages’, Cambodia Development Research Institute (CDRI), Working Paper 7.
Murshid, K (1998) ‘Food security in an Asian transitional economy: the Cambodian experience’, Cambodia Development Research Institute, CDRI, Working Paper 6.
ODI (1997) ‘Global Hunger and Food Security after the World Food Summit’ ODI Briefing Paper (1), 1997. Overseas Development Institute: London.
ODI (2000) ‘Reforming Food Aid: Time to Grasp the Nettle’ ODI Briefing Paper 1, January 2000. Overseas Development Institute: London
Pillai, N (2000) ‘Food Aid for Development? A Review of the Evidence’ in Clay, E & Stokke, O (eds) Food Aid and Human Security, Frank Cass: London, 2000.
Poh, L.K. (1999) ‘Manufacturing a Crisis: The Politics of Food Aid to Indonesia’, Development Report, No.13, Oakland, CA: Institute of Food and Development Policy.
Reutlinger, S (1999) ‘From “food aid” to “aid for food”: into the 21st century’, Food Policy, Vol.24, No1
RGC (2000) Interim Poverty Reduction Strategy Paper, Royal Government of Cambodia, Phnom Penh, October 2000.
Schulthes, J (2000) ‘Is there a future for the WFP as a development agency? Or does food aid still have a comparative advantage?’, in Clay, E & Stokke, O (eds) Food Aid and Human Security, Frank Cass: London, 2000.
Sharp, K (1998) ‘Between Relief and Development: targeting food aid in disaster prevention in Ethiopia’, RRN Network paper 27, Overseas Development Institute: London, September 1998.
Turton, C (2000) ‘The Sustainable Livelihoods Approach and Programme Development in Cambodia’, Working Paper 130, Overseas Development Institute, London.
UNICEF-WFP (1998) ‘Report on the joint UNICEF-WFP Baseline Survey of CASD project and WFP target areas’, WFP, Phnom Penh.
WFP (1996) Tackling Hunger in a World Full of Food: Tasks Ahead for Food Aid, Rome: World Food Programme, Public Affairs Division.
WFP (1999) ‘Enabling Development’
WFP (2000a) ‘Food Aid for Recovery and Rehabilitation: PRRO Cambodia 6038.01’, WFP Phnom Penh, October 2000.
WFP (2000b) ‘Emergency Food Assistance to Flood Victims in Cambodia: EMOP 6297.00’, WFP Phnom Penh, October 2000.
 On a much larger scale, the same rationale was applied to the WFP “food-for-work” programmes in Indonesia in 1998-99 where the politics of food aid has come under much critical scrutiny – see, for example Poh (1999) and the evaluation of WFP’s emergency operation, lead by the author (available on WFP website).
 From January 2001 – December 2003, WFP’s PRRO (Protracted Relief and Recovery Operation) plans to disburse 113,000 tons of food (41,300 tons in the first year, reducing to 32,500 tons in the third year), two-thirds of which is for income generation (FFW) schemes, the remaining third for social support to vulnerable groups and emergencies (WFP, 2000a).
 Though not quantified, few dispute that excessive logging and deforestation contributed to recent heavy flooding, increasing salination and eutrophication of the Tonle Sap Lake.
 In a limited survey undertaken in 1998, it was found that more than half of the households had sold land since the dissolution of the krom samaki, the system of agricultural collectives established during the PRK (post-1979) period (McAndrew, 1998).
 The following points are drawn from two evaluations lead by the author: a 1999 mid-term evaluation of PRRO 6038 (the first phase of the recovery programme) and a 2000 evaluation of the WFP/Seila/CARERE (local planning process) collaboration. A summary of the first is available on the WFP website; the second is published (Bennett & Wallgren, 2000).
 WFP elsewhere (e.g. Africa) has increasingly used Household Food Economy analyses as a basis for targeting in relief programmes; the extent to which this is feasible within FFW projects is doubtful.
 These interdependent choices may be partly pre-determined by the context and project objectives. A useful overview of targeting issues is found in Sharp (1998).
 According to the Ministry of Planning’s Socio-Economic Survey, 1999, it would cost a minimum of $126 million per year to eliminate poverty with perfectly targeted transfers. The social intervention budget of the same year amounted to $27, approximately 0.9 percent of GDP, almost all of which was absorbed by civil servants (RGC, 2000).