(introduction...)

One of the reasons why survival strategies seem so important is that predicted famines, with millions said to be facing starvation, have often not materialized, at least in the form predicted and in the form that donors and the general public recognise as 'famine'. It is important, then, to ask on what basis such predictions are made. Equally, it is important to ask whether the absence of mass mortality and an instantly recognizable 'famine' really means that there has been no significant crisis and no significant need for relief.

2.1.1 Estimating production, food availability and population

Appeals for emergency relief in Africa tend to be greatly shaped by annual assessments of food needs that are made by the UN's Food and Agriculture Organisation (FAO). FAO estimates of food needs are obtained by calculating the total quantity of food available for consumption in a country in a given year (allowing for imports, exports, use of food for Animals, seeding, industrial purposes, and losses during storage and transportation, as well as changes in stocks), and then subtracting this figure from the total food consumption needs of the country's population. In performing these calculations, the FAO generally relies heavily on government production estimates, whose methodologies are often not explained.

Difficulties surrounding the assessment of food production in Africa are formidable indeed. African countries are often sparsely populated, with many different climatic zones and cropping systems. Non-cereal crops may be particularly neglected in of ficial estimates. Animals and undomesticated plants are likely to provide a substantial portion of the diet. Non-cereal foods, including tubers, pulses, fruit and fish, may be ignored. Indeed, the neglect of relatively 'invisible' crops such as root crops has a long history. Colonial of ficials in Malawi in the 1940s came to the conclusion that households in one area had a chronic maize deficit, but took no account of sorghum, cassava and root crops (Vaughan).

Aerial photography may underestimate acreage, for example, if two or more crops are planted on the same land. On the other hand, satellite photographs create the danger that weeds like striga will be mistaken for crops, and that areas seen to be green may yet fail to flower and may yield virtually no crop (as happened recently in parts of Eritrea). Significantly, US Department of Agriculture estimates of crop production tend to make the weaknesses and uncertainty in the data more explicit than do FAO estimates. There is an element of false certainty about much of the FAO data.

The origins of the data on losses and uses of food other than for human consumption are not clear. Figures on stocks are likely to be unreliable. Official trade figures may fail to take account of widespread smuggling, and may in any case be inaccurate: one recent study of African trade flows found that a particular country's records of imports and exports to another country rarely bore much relation to the records of these same trade flows that were kept in this second country. Population data are also likely to be inaccurate, not least because regions can hope to gain more government resources by overestimating their populations. Inaccurate production estimates may not simply be the result of chance errors, but also of repeated, and perhaps rational, under-reporting of production by a number of parties. Rural producers may underestimate their own production and resources as a result of a desire to avoid taxation or compulsory government purchase of crops. Many people (at local, regional and national levels) may be influenced by a desire to solicit aid.

It is arguable that the FAO itself has an interest in accepting exaggerated shortfalls in production, since the notion of large-scale food problems in Africa provides much of the justification for the organisation's existence. Aid agencies may have a number of pragmatic reasons of their own for supporting major appeals: at ground level, endorsing UN appeals may give improved access to scarce UN resources like transport and radios; back at headquarters, large-scale emergencies are acknowledged as an important trigger for raising funds. Meanwhile, the consequences—in terms of human suffering—of 'getting it wrong' are likely to be far less grave when needs are overestimated than when they are underestimated.

2.1.2 The 'food balance-sheet' approach

Current international assessments of African food needs rely heavily on the 'food balancesheet' approach, which implicitly portrays regions of African countries as being selfsufficient in grain in 'normal' years. Grain production and grain requirements are typically calculated on a regional basis, and conclusions are drawn about the 'food deficit' in individual regions. When this 'food deficit' is not filled by adequate relief, and yet people still survive, this appears something of a mystery, commonly bringing forth the 'explanation' that people have a range of special survival strategies which they employ in time of famine. However, even in 'normal' years, many regions of Africa would not be expected to produce enough grain for all their needs; such strategies as selling livestock and selling labour, while they may be pursued on a greater scale during time of famine, may well take place year after year. As Sen has argued convincingly, a proper understanding of famine should involve not only examining production but also looking at the means by which people attempt to secure access to whatever food exists. It is important to look at these means both in normal years and in famine years. This kind of perspective is still not fully taken into account in UN emergency appeals.

The idea of the subsistence farmer—producing all his or her needs—loomed large in the British colonial mind, and the idea continues to influence much official thinking on relief and development. Yet the true subsistence farmer appears to be very rare, and he or she is probably also very poor. Vaughan has written in relation to Malawi:

When colonial agricultural of ricers painstakingly calculated the carrying capacity of the land..., they failed to bear in mind the fact that never in the accessible past had villages, let alone households, provided all their needs from their own labour on their own plots of land.

From a detailed study of Wollo, Tigray and Eritrea, de Waal concluded that most of the farming population do not survive by agriculture alone, even in normal years. Livestockfarming, petty trading, and casual labouring were all significant. Reliance on off-farm incomes increased from south to north within Ethiopia and from highlands to lowlands. Pastoralists also depended on trade and casual labour to a considerable extent. A Leeds University team that visited rebel- and government-held areas of Eritrea in 1987 calculated that, even in a 'normal, non-war' year, production of staple foods would be enough to feed the population for only seven to seven-and-a-half months. In a 'normal, war' year, the figure fell to 4.6-4.8 months.

One study of Burkina Faso in 1984 has shown how much greater quantities of food aid were targeted to the Sahelian zone compared with the Sudanian zone, on the basis of the former's lower rainfall and lower yields estimates. Yet people in the Sahelian zone had access to a much wider range of economic opportunities (including trade and large livestock herds) than those in the Sudanian zone, and their purchasing power was significantly greater (Reardon, Matlon and Delgado).

All this is not to say that food balance-sheets are uninformative or unhelpful; it is the excessive reliance on them that is misleading. Some Oxfam field staff have stressed that the food balance-sheet approach, although a blunt instrument, is far from useless, since it provides an early warning of potential problems, it can reflect trends in production, and it may offer some kind of consensus view (at least among officials) about what is happening. Nevertheless, most feel this approach is unhelpful in indicating the actual level of food production, or the numbers of people that are likely to require assistance.

2.1.3 Donor fatigue/scepticism

It is important to distinguish between the processes of needs-assessment that find their way into written UN appeals, and those that are made informally by donors. If donors are sceptical about levels of need in written appeals, if they feel that relief is unlikely to be distributed to the intended beneficiaries, or if they are simply unwilling to allocate major resources to a particular emergency, they may not allocate resources in line with those requested in official appeals. All these factors are likely to influence donors' private assessments of need. In the section of needs-assessment in specific countries, evidence of donor 'fatigue' is presented in relation to Mozambique.

2.1.4 Gender biases

A significant source of bias in needs-assessments—perhaps tending to encourage underestimation of needs in some contexts—is the fact that such assessments are typically made by men. Whether information comes from key informants in central or local government or at village level, it is very likely that the information comes from men. Oxfam's Gender Adviser for Africa and the Middle East, Bridget Walker, offers an example of the distortions this can create. As part of Oxfam's ongoing nutritionalassessment programme in Darfur, Sudan, separate teams of interviewers were used (men interviewing men, mostly village notables, and women interviewing women). It was found that village notables appeared to be overestimating the harvest substantially, whilst women were making much lower estimates. The reasons for this were not clear. It may be that the men interviewed were unrepresentatively wealthy, and relatively unconcerned about harvest shortfalls. Or it may be that they were out of touch with what was going on in the fields. The fact remains that differences in assessments of production were dramatic.

WaLker also notes that there has been some tendency for women to give greater estimates of household needs, including in their definition of the household many people who were on the fringes of the household. By contrast, men have tended to give a more limited definition of the household, including only those people for whom they felt directly responsible. Women's direct involvement with processes of food production and consumption can be contrasted with their habitual exclusion from processes of needsassessment, and constitutes a major weakness in current systems of assessment.

2.2 Needs-assessment in some specific countries

In relation to Sudan, the inter-agency appeals of the Special Emergency Programme for the Horn of Africa (SEPHA) give breakdowns of 'drought-affected' and 'displaced' people by region and province. In the documents I have seen, it is not made clear where these estimates come from, but FAO production estimates seem to provide the main basis for estimates of drought-affected people.

Relief needs in Darfur for 1991 were assessed by FAG at 144,000mt. This was based on a pre-harvest assessment in 1990 by the Darfur regional government's Agricultural Planning Unit (APU), which employed a 'food balance-sheet' approach. The APU reported that cereal production as a percentage of requirements varied from 0-40 per cent in different rural councils within Darfur. Oxfam field staff in Sudan reported that:

In general this assessment is considered as reliable... Although the results can only be considered as rough estimates (based on rapid assessments, key informant interviews, observations), ... (they) discriminate between rural councils and give an indication of the seriousness of the situation.

Significantly, in September 1990, NGOs, the UN, and the Sudanese Relief and Rehabilitation Commission undertook a rapid assessment in North Darfur and came up with similar conclusions to those of the APU.

It is true that the 'food balance-sheet' approach is likely to be misleading to some extent, since some areas of Darfur, particularly in the north, would not expect to produce all their own grain even in normal years. Nevertheless, given the generally limited movements of commercial grain into Darfur from the rest of Sudan, low grain production within Darfur is likely to mean high grain prices in the region and corresponding hardship (Keen, 1986). In relation to Ethiopia, SEPHA gives a more disaggregated picture of needs. It gives more regional subdivisions, and it provides estimates for each region of populations affected by failure of the last meher rains, by failure of the belg rains, numbers displaced, families of ax-soldiers, and returnees from Somalia. There are few clues about where these figures come from in the reports I have seen. However, the format and figures match those in reports of the Ethiopian government's Relief and Rehabilitation Commission, which made the assessments along with the Relief Society of Tigray (REST) and the Ethiopian Relief Organisation (ERO) in their respective areas of operation.

In Angola, devastation accompanying the civil war was exacerbated by government agricultural policies and by drought in 1989-90, which affected the southern and central provinces of the country. The drought caused sharp rises of grain prices in relation to livestock prices in some areas, and appears to have encouraged theft of crops. The UN undertook programmes of needs-assessment in co-ordination with the Angolan Government, a leading role being played by the UN Development Programme, the World Food Programm e and UNICEF. These assessments formed the basis of emergency appeals and were part of the Special Relief Programme for Angola (SRPA). However, field investigation seems to have been minimal, and much of the data appears to be extracted from rather dubious official sources.

Oxfam's country representative in Angola commented: 'Relief assessments in Angola have been minimal at best and non-existent at worst during the past few years.' Aid agencies have been discouraged from working in Angola by a political and bureaucratic climate which makes it very difficult for them to operate effectively. Government officials involved in relief assessment have often lacked the necessary experience. The war made many areas inaccessible. Angolan government assessments of needs have apparently been based on needs-assessments by provincial authorities. The assessments have often been treated with some cynicism by international donors. Government statistics on the displaced are rarely updated and are thought to be rather unreliable. There is some feeling among Oxfam staff in Angola that donors respond not only to formal assessments but also to informal contacts and impressions. The province where assessments of need were most clearly articulated was Huila province, where a number of NGOs carried out a joint assessment during the drought. The relatively clear articulation of needs in Huila appears to have been reflected in relatively large amounts of international assistance to this province, although the province has not been as severely affected by the war as some others.

In Mozambique, aid agencies and the government have evolved a way of arriving at requirements, but again few serious attempts have been made to relate needs-assessments to detailed surveys on needs at local level. Performing this kind of survey is difficult or impossible in much of Mozambique. Aid-agency staff are sceptical about the accuracy of needs-assessments. They point out that different criteria have been used in different provinces. UN and Government of Mozambique discussions with provincial officials are unlikely to give an accurate picture of needs: these officials often lack appropriate training; they often have no access to areas they are asked to assess; and again they may have an interest in inflating assessments of needs in order to attract relief to their areas. One agency worker said that the UN/government missions would visit only two or three provinces during the assessment period, clearly an inadequate basis for nationwide assessments. NGOs have also played an important role in relief assessments (for example, Oxfam in Niassa province), benefiting from access to transport and to rural areas where they are working.

Although government officials may have an interest in overestimating needs, other factors may have encouraged underestimation. For example, the need for free emergency relief is currently considered to disappear after a person has been receiving assistance for 18 months. Yet stopping free relief will not necessarily be justified at this point: the attainment of 'serf-sufficiency' will depend on favourable climatic conditions, on access to adequate labour and fertile land, and on receipt of adequate relief (including seeds and tools) before the planting season. One Oxfam report noted that:

The government had only requested the very base minimum for a displaced population of 1.5 million in 1990/91 as opposed to the several millions of displaced people.

Furthermore, while the 'displaced' (that is the recently displaced) are supposed to get 100 per cent of their food needs, the agreed ration (including only 350 grammes of maize, 40 grammes of beans and 10 grammes of oil) is actually insufficient to live on. As a December 1991 UN/Government of Mozambique report noted:

This... represents a survival ration only on the assumption that most displaced populations have access to some food resources beyond Emergency supplies, at least for part of the year.

Meanwhile, the 'affected' are supposed to get only 60 per cent of their needs. Another possible source of underestimation of needs is the fact that even the theoretical ration of 10.5kg per month is intended for households of five people. However, reports from the field say most households consist of more than five people.

Assessments of needs —in practice, and possibly also at the level of official UN/government appeals—also appear to have been affected by donor 'fatigue' and scepticism. Analysis of donor response to appeals 1987-89 shows a marked downward trend. Pledging of food aid, relief items, and logistical support fell progressively, apparently hit by stagnant emergency and development budgets in donor countries in the face of increasing numbers of emergencies in the world, and by donor scepticism about the prospects for a proper distribution in Mozambique. The security situation in the country also appears to have discouraged donors from funding rehabilitation projects. In 1987/88, 480,000mt of cereals for the market were reported to be required, and 475,800mt pledged. The following year, the total required had risen to 500,000mt, but the total pledged fell to only 335,600mt. And in 1989/90, the total required rose again to 560,400mt, and the total pledged fell to 303,600mt (only 54.1 per cent of agreed requirements). Support for logistics fell dramatically in 1989/90.

The trend for emergency food aid was different, but the shortfalls were still significant. For emergencies, the total required in 1987/88 was 410,000mt, and the total pledged 148,600mt. The next year, 230,000mt were required and 190,400mt pledged. In 1989/90, 195,000mt was required and 144,000mt pledged (74 per cent of agreed requirements). Such shortfalls have continued. In 1991/92, total requirements were put at some 263 million US dollars, of which some 109 million remained unpledged by November 1991. The logistics requirement was 26 million US dollars, but pledges totalled only 11 million. Pledges for therapeutic feeding and nutritional rehabilitation of displaced people were particularly low, in relation to agreed requirements, covering only 49 per cent of requirements for dried skimmed milk, 16 per cent for sugar and 18 per cent for oil.

2.3 Limited relief

Whatever the distortions involved in assessing needs, there is no doubt that actual delivery of relief has tended to fall far short of these assessments. Consider the case of Sudan in 1991. The actual receipt of relief in Darfur by no means matched assessments of needs. In Kebkabiya area council, North Darfur, households had receded an average of 29kg per person by the end of 1991. This quantity, although significant, would not have lasted longer than six weeks. Moreover, the major part of this food arrived late, in the period October-December, whilst it was most need in the period before the harvest. In North Darfur as a whole, significant quantities of relief were not received by households until October. Between October and December the flow of relief from El Fasher to districts of North Darfur was substantial, but still only around one-third of estimated requirements (1,000 mt per week for the North Darfur population of 1.2 million, compared with an estimated need of 400 gm per person per day). In Red Sea Hills, individuals were reported to have received some 12-14 kg per month since June 1991, when distributions began.

In Angola, relief is thought to have played a much smaller role in ensuring people's survival than people's own survival strategies, which have been assisted by the fertility of much of Angola and the general absence of severe population pressures (except in certain urban areas where people have been artificially concentrated as a result of the war). Oxfam field staff in Angola report that people displaced by the civil war have rarely received relief assistance during the period immediately after their displacement. Some food and other relief items have eventually reached these people, with the government, international donors and NGOs tending to channel assistance through the government organization SEAS (Secretaria de Estado dos Assuntos Socials). Oxfam's country representative comments:

Until a stock of humanitarian emergency goods was established in Luanda, such assistance was generally extremely slow in arriving, given the procurement, maritime shipping delays, port bureaucracy and internal logistics of delivering assistance to would-be beneficiaries.

Relief for those in southern and central provinces who were affected by drought in 1989-90 was, for the most part, 'too little, too late'. In general, a large portion of international assistance has been directed at urban rather than rural populations. Most international relief has been directed at government-held areas of Angola, although a number of US NGOs have worked in areas controlled by the insurgent UNITA forces.

In Mozambique, the shortfalls in delivery of relief are well illustrated by the fact that 1991 saw the delivery of quantities of relief that had been pledged for 1989. One agency worker said distributions had generally been at less than half rations, and this appears to be confirmed by detailed information on distributions in the worst-hit province of Zambezia. The 1990-91 appeal envisaged a ration of 12.17kg per month offood aid (mostly maize, with some beans and oil) for 429,000 beneficiaries in Zambezia province. Given that only 73 per cent had reached district level by October 1990, and given that the real beneficiary population was then being estimated at some 854,000 people, the average quantities received must have been very much lower than the intended levels, the report stated. It calculated the average ration at 4.49kg per person, adding that it was observed in some of the districts visited that the average distributed was 1.2kg per person. Of nonfood relief items, only clothing was provided in important quantities, much of it from Oxfam/UK, the report said. Nor was Zambezia by any means the worst case, in terms of food deliveries. Between May and August 1990, 88 per cent of estimated need was distributed in Zambezia province. Yet distributions of emergency food in other provinces over the same period varied from 124 per cent of estimated needs in Maputo province to 45 per cent in Sofala and Niassa. Logistical and security constraints impeded distribution in most provinces.

2.4 A real crisis in Africa

While outright starvation has been fairly rare, it has not been altogether absent. Further, there is abundant evidence of many kinds of suffering and long-term damage to livelihoods and the environment. While aid agencies and the UN might be accused of 'crying wolf', the crisis has been real enough. What needs emphasising is that the wolf has not always appeared in the form which Western stereotypes about 'famine' lead people to expect. Where are the mass deaths? And where are the mass migrations?' These questions have quite rightly been asked in relation to recent crises in Africa, when Western publics were told that 'millions faced starvation'.

The evidence tends to suggest that mortality has not occurred on the scale that many were predicting. On the other hand, there are dangers in down-playing mortality too severely. There are also incentives for doing so. One senior agency worker suggested that while 'donors want blood' at the time when appeals are being made (that is, they want to hear endorsements that lives are on the line), by contrast, after a crisis has past its peak (and relief, typically, has been minimal), donors seek to down-play the incidence of mortality: the relief was inadequate, but this did not really matter because there was no real crisis in the first place. In fact, it can be very dangerous to make sweeping statements about how many people did or did not die in a particular crisis, as Ken Wilson, of the Oxford-based Refugee Studies Programme, emphasizes. Mortality in a famine may often be very patchy, and it may be largely invisible if people remain in their home villages.

Largely anecdotal evidence suggests that mortality has been significant in some areas, though not on anything like the scale of Ethiopia in 1984-85. Moreover, there has been significant mass migration in some areas. And there is evidence of alarming levels of malnutrition. The picture is a vary variable one. In Niassa province, in the far north of Mozambique, Oxfam representatives report that little evidence of increased malnutrition or mortality has been reported there, despite the fact that relief del*eries had by no means matched the level of needs that was assessed, and that diversion of relief supplies was significant. (Representatives acknowledge that mortality records are not kept.)

On the other hand, there is some evidence of very patchy elevated mortality, notably in Mozambique's Zambezia province, in recent years, Wilson reports. Thousands died in government centres in Zambezia province in 1990, in particular. Administrat*e and political obstacles to relief were significant, and there was a certain feeling among agencies that the government had compelled people to go to government-held towns, and they were reluctant to provide assistance in these circumstances. Many were fed up with emergency interventions, preferring development interventions. A major exception was World Vision, which saved a lot of lives. Oxfam field staff in Mozambique reported increases in mortality, malnutrition and disease rates over the period 1989-91, notably in areas where the government had regained control from Renamo forces. Forced removals of people led to overcrowding in camps where relief assistance was inadequate. One Oxfam worker reported that 'In some camps the nutritional deterioration was drastic', particularly among young children.

A 1990 UN/Government of Mozambique report noted:

Food availability and distribution during the 1989-90 appeal year was considerably below the requirements, and as a consequence there has been a notable increase in malnutrition nation-wide.

Low levels of pledges of market food aid, and for logistics, had led to increased malnutrition among large sections of the population. By mid-November 1989, acute food shortages were reported in Inhambane, Zambezia and Nampula:

The intended reactivation of the market network in rural areas has been severely restricted, in part owing to shortfalls in stock levels.

Pledges were only 56 per cent of cereal needs and 26 per cent of non-cereal needs. A December 1990 Government of Mozambique/lJN report noted:

A comprehensive study within 24 districts of Inhambane, Tete, Zambezia, Nampula and Niassa, finalized in July, showed an average low birth-weight of 26 per cent and chronic malnutrition (low height-for-age) of 53 per cent among children under three years. Cities and urban centres are excluded from free food distribution, but as urban poverty is increasing, urban nutritional indicators are worsening. A survey made in July in Tete city shows the level of acute malnutrition to be 12 per cent.... A recent nutritional survey, carried out in Zambezia (Murrua) in mid-October, showed acute malnutrition in 43 per cent of children under five. Another such situation in Mugulama (also in Zambezia province) in April found acute malnutrition rates of 20 per cent.

In Angola, anecdotal evidence suggests significant numbers of deaths occurred during the emergency there. There have also been substantial migrations of people to urban areas in search of food. Many of these people were in very poor condition. Increases in the level of malnutrition among children arriving in Huila, Cubal and Kaluquembe have been observed.

In Darfur, Sudan, in 1991, most people remained in their villages, where there are no proper records kept of mortality and disease. However, village leaders in Kebkabiya area council reported mortality at significantly higher levels, rising from an average of one death per village council in July 1991 to seven deaths per village council in September. Despite the limitations of such data, if the figures are extrapolated for the whole of North Darfur, they imply tens of thousands of deaths in the three months of July to September alone. Save the Children Fund officials say they are aware of clear excess mortality in certain parts of Darfur in 1991. Levels of malnutrition (below 80 per cent weight-forheight) of between 19 and 25 per cent were found in parts of North Darfur. Mass outmigration was not observed. Yet this may indicate a learning of lessons from 1984-85, rather than the absence of a crisis. In the earlier famine, migration did not necessarily bring access to food, whilst exposing people to a variety of major health risks. John Seaman of Save the Children Fund said: 'Memories of Nyala and Geneina in 1985 are fresh in people's minds. They don't want to do it again.'

The position appears to have been rather different in the Red Sea Hills area of northeastern Sudan. Many people are reported to have stayed in their home villages in 1984 in the hope that good rains would fall. It is reported that when the rains failed, this strategy increased the loss of human life during the famine. This experience appears to have encouraged many Beja to migrate at a relatively early stage in 1990. Another very significant factor was that by this time many households had lost almost all their animals. Large-scale migration, first by men and then by women and children, to major urban areas of Red Sea Hills has been observed. People also moved en masse to the side of the road between Port Sudan and Khartoum, where relief was being distributed. An Oxfam survey in the Red Sea Hills area in August 1991 found 6.6 per cent of rural people were severely malnourished, while 20.9 per cent were moderately malnourished. In the major urban centres of Port Sudan and Halayib, malnutrition rates had risen since 1990. This was attributed to the large numbers of people who had migrated there from Arbaat area. Malnutrition-related diseases have been widespread.