House of Commons - International Development

Select Committee on International Development Third Report

IV. EARLY WARNINGS, INFORMATION, AND DONOR RESPONSE

53. Governments across Southern Africa declared emergencies and called for international assistance at different times between February and May 2002.[95] However, signals of an impending food crisis were evident from many months before, when harvests failed in mid2001. NGOs were reporting on localised household food insecurity in Malawi and Zimbabwe by August 2001. In her oral evidence, Clare Short emphasised that the response in Malawi was not "extraordinarily slow", but acknowledged that: "There was a general lack of urgent response to the crisis in Southern Africa".[96] NGOs[97] and other commentators[98] suggested to us a number of reasons why the signals were not acted on promptly, and why the donor response was late. In general, the obstacles to an effective humanitarian response are either informational, political-institutional, or logistical. The humanitarian response can be derailed at various stages, from the initial collection of information for early warning systems, to the assessments made on the basis of that information, to the warnings issued or not on the basis of those assessments, to the response of donors to those warnings, to the delivery of assistance (see figure 7).

54. Problems at each stage resulted in the failure to prevent a tragedy in southern Africa, especially in Malawi, in early 2002. Official information about food availability and access was confused or inadequate, governments and donors were slow to react to signals of impending food crisis, and the humanitarian response was delayed by logistical bottlenecks. We were encouraged to hear that information flows are stronger now than in late 2001, and the political and logistical failures of 2001/02 have been addressed by stronger commitment of key actors and a more timely response to the evolving 2002/03 emergency. There remain some concerns about whether the international community has sufficiently recognised the underlying problems of poverty and vulnerability of which the immediate crisis is a symptom. This is the subject of later chapters.

Famine Early Warning Systems

55. Famine early warning systems were established across most of sub-Saharan Africa following the drought-triggered famines of the 1980s. In southern Africa, national early warning systems are coordinated by the Southern African Development Community (SADC) Regional Early Warning Unit (REWU). This early warning unit draws on data collected or analysed by: SADC's Regional Remote Sensing Unit and Regional Food Security Database Project; the USAID-funded Famine Early Warning System (FEWSNET); WFP and other donors; and NGOs, as well as Ministries of Agriculture. However, the southern African systems are less comprehensive and are institutionally

Figure 7: Obstacles to the effective operation of early warning systems

Source: Committee's own

weaker than those in the Horn of Africa, where recurrent droughts and conflict-triggered food emergencies necessitate constant vigilance. Consequently, although national and regional food security information systems are in place, REWU has not been adequately resourced, either financially or technically.[99] The USAID-funded FEWSNET project synthesises available information on food security in most countries of the region, but draws mainly on other sources, such as Ministry of Agriculture crop estimates, which are not always reliable. Because of these limitations, conflicting signals about food shortages in the region during 2001 were difficult for policymakers to reconcile and interpret.

56. A second problem faced by early warning systems is that estimating crop production is only the first step in the famine prevention process. Understanding how crop failure will affect rural livelihoods and survival prospects is far more complex, but is critical for mobilising an adequate, appropriate, and timely response. Most of the investment in early warning systems in Africa has gone into predicting harvest failures. Recognising the limitations of early warning systems based on rainfall and crop forecasts, DFID and other donors are beginning to invest more of their resources into "vulnerability assessment" approaches such as the "household economy" approach pioneered by SCF-UK. This will be supported throughout the SADC region, and will emphasise bottom-up reporting from local "vulnerability committees".[100] Work needs to be done to establish why the needs assessment in Zambia lacked geographical nuances and therefore over-estimated problems in some areas. DFID is also drawing up a programme of support to SADC in food security which will, among other things, improve the ability of SADC member states to acquire and use information on vulnerability.[101]

57. We strongly endorse these encouraging developments. Food crises are likely to recur in southern Africa in the coming years. It is vital therefore that famine early warning systems are strengthened at both regional and national levels:

Firstly, by improving crop production forecasts, especially of nonmaize food-crops including cassava, and assessments of their contribution to food availability and consumption;

Secondly, by strengthening vulnerability assessment at national and sub-national levels, including systematic monitoring of market prices and better understanding of the analytical linkages between poverty, seasonality, and access to food;

Thirdly, by incorporating non-official data sources, such as qualitative NGO reports and monitoring by vulnerability committees, into official early warning systems.

58. Implementing these improvements to early warning systems in Southern Africa will require a commitment of financial resources and technical expertise from the donor community, national governments and regional organisations, notably SADC. We urge DFID to support reasonable requests for financial and technical assistance.

59. Early warning systems should provide information to policymakers about food supply (availability) as well as the demand for food. In late 2001, three sources of information about the evolving food crisis in southern Africa were available to policymakers: crop production estimates, reports from rural areas, and maize prices. These information sources are discussed below.

Supply-side (Food availability)

60. A primary function of conventional early warning systems is to forecast national levels of food availability, especially the annual harvest. As Save the Children Fund (UK) told us: "The estimation of supply is a proper, necessary and fundamental step."[102] In 2001, harvests in southern Africa were reduced by erratic weather, compounded by policy failures such as the "fast track resettlement" programme in Zimbabwe, which reduced the area planted to crops by an estimated 40%. In January 2001, WFP announced that 500,000 tonnes of food aid would be needed for Southern Africa because of flooding of crops, and Swaziland appealed for international assistance for flood victims. In Malawi, the early warning system proved to be reasonably accurate in terms of predicting a 32% fall in the maize harvest (though only after the mid-season and interim forecasts were each revised downwards). But it was highly unreliable with respect to other food crops, resulting in inaccurate projections of total national food availability.

61. The problem of over-optimistic crop forecasts leading to complacency and sluggish response was most acute in Malawi. Although it was known that maize harvests were lower than the previous (bumper) year, the scale of the food deficit was not accurately predicted. Specifically, the Famine Early Warning System used Ministry of Agriculture final crop production figures to conclude, in its Monthly Food Security Report of July 2001, that the projected maize deficit of 273,000 MT would be more than adequately covered by other cereal crops and tubers, especially cassava, leaving "a national food surplus of 438,000 MT."[103] This forecast turned out to be grossly exaggerated.[104] One reason why forecasts proved to be wrong may well be that as cassava and other tubers grow underground, it is inherently difficult to predict harvests. One reason why the forecasts were so exaggerated might be the fact that the agricultural extension workers who were involved in making the assessments have an incentive to over-estimate, to demonstrate their success in encouraging cassava production.

62. An unanswered question is why these estimates were accepted as credible, and used by donors as a justification for late and minimal intervention, when it was well known that their methodological foundations were seriously flawed. DFID's Food Security Adviser acknowledged the low credibility of cassava production estimates in Malawi, but explained to us that: "What we did not expect was that the figures we were given were going to be as exaggerated as they turned out to be."[105] DFID also wondered in late 2001 whether the "fixation on maize" in the countries of southern Africa might have led to cassava being dismissed as a snack food, with the result that actual availability of cassava would be greater than that reported,[106] and the food shortage less severe than predicted.

63. Conversely, information about food shortages from NGOs active at community level was ignored, presumably because the estimation of crop production is seen as a technical task that can only be undertaken by experts and agricultural specialists. In December 2001, for example, SCF-UK commissioned a national study on the availability of roots and tubers in Malawi. The report found that surpluses were produced in those areas where roots and tubers are predominantly grown (along the lakeshore and in northern region), but that poor transport networks and weak markets made it prohibitive to shift this food to maize deficit areas.[107]

Demand-side (Food accessibility and affordability)

64. A second function of famine early warning systems is to predict the likely consequences of food shortages on affected population groups, to answer the key policy question: "What proportion of households of what type will be unable to acquire sufficient food to meet their requirements?"[108] In the second half of 2001, community-level reports from NGOs and community-based organisations highlighted emerging problems including poverty and demand failure, problems which were publicised in the international media by October. Rapidly rising maize prices in the same period—maize prices in rural Malawi quadrupled between July and October 2001—provided a further indication of emerging problems. Nutrition data and, later, reports of hunger-related mortality in some communities, provided late indicators of food stress.

65. NGO reports failed to mobilise an official response, perhaps because their information was perceived as localised and/or not credible. SCF-UK conducted household economy assessments in two Malawi districts during October and November 2001 that reached "alarming" conclusions about the inability of large proportions of the rural population to meet their annual food energy requirements. In mid-November, SCF-UK convened a meeting of government, donor and NGO representatives in Malawi to convince them of the severity of the food crisis. Although their predictions were subsequently confirmed by later tragic events, SCF-UK was unable to mobilise a concerted humanitarian response. Instead, FEWSNET, WFP and the European Commission Food Security Programme conducted a rapid food availability assessment, after which WFP approved an Emergency Operation (EMOP) that targeted only 255,000 beneficiaries over a period of three months.

66. Maize prices were monitored and reported in Malawi by FEWSNET as well as by NGOs, but received surprisingly little attention, despite a widespread recognition that rapid food price rises are both a symptom and a cause of household food insecurity. As Professor Kydd, Dr. Dorward and Professor Vaughan explained in their memorandum:

rapid rises in staple food prices relative to purchasing ability are the key problem and symptom of famine situations, and therefore need to be taken very seriously. They are also relatively easy to monitor. […] It therefore seems extraordinary that the dramatic rise in maize prices was not recognised as a key indicator of widespread and rapid deterioration in food security over the period June to October 2001.[109]

67. Following the abolition of price controls and food subsidies in most southern African countries in the 1980s and 1990s,[110] price seasonality has returned as a major cause of household food insecurity. Food price movements in rural areas follow a predictable seasonal pattern, with prices being lowest around the harvest period when supplies are abundant and market demand is low, then rising gradually through the dry season and into the next farming season, as granary stocks and market supplies dwindle. Food prices typically peak in the pre-harvest "hungry season", which is associated with peak under-nutrition rates. As such it is extremely important to monitor food price movements closely, as a "demand-side" component of famine early warning systems. As the Nobel Prize-winning economist Professor Amartya Sen has demonstrated, it is possible for people to die of starvation simply because food prices rise to levels that are unaffordable for the poor, irrespective of the food availability situation at local or national level.[111] The Indian Famine Codes of the late 19^th century—a famine early warning system established by the British—included food price monitoring as an early warning indicator, and contemporary early warning systems in East Africa use price rises above seasonal norms to trigger relief interventions.

68. Finally, nutrition surveys conducted in Malawi found clear evidence of deteriorating nutrition status—the outcome of acute food insecurity. In Salima District, global malnutrition rates rose from 9.3% in December to 19% in February 2002. These statistics provided quantitative support for the anecdotal evidence from NGOs and community-based organisations of severe coping strategies being adopted by crisis-affected households, including distress sales of livestock and other assets, withdrawal of children from school, and dietary adjustments including food rationing.

69. We agree with Stephen Carr[112] that: "donors have to be prepared to seek advice from members of the local community with a broader knowledge of the situation than that available to their own staff, many of whom have only brief local experience."[113] We endorse too the recommendation made by Professor Kydd, Dr. Dorward and Professor Vaughan, in their written submission: "Early warning systems should place more emphasis on maize price monitoring and on reports from rural areas."[114] But we would go further still: Agricultural market information systems should be introduced or strengthened as a matter of urgency in all southern African countries. Lessons should be learned from the Indian Famine Codes and Kenya's Turkana District drought monitoring system, about how to incorporate price information into national and regional early warning systems. In addition to monitoring food prices and supplies in local markets, these systems should also collect data on agricultural input supplies and prices (especially fertiliser), and possibly also livestock prices and volumes, as "distress" sales of livestock at low prices are widely acknowledged as a robust indicator of livelihood stress.

70. More generally, however, we conclude that lack of information was not a critical constraint in triggering the food crisis of 2001/02. If anything, the problem was too much contradictory and uncoordinated information, to which the appropriate actors failed to respond. Rob Holden of DFID's Conflict and Humanitarian Affairs Department, explained to us the importance of institution building at national and regional levels. He stated that: "More work needs to be done in terms of building institutions, particularly in Africa, so we need good baseline data and regular data coming in so that when we get blips in the system we can respond rapidly to check that and to obtain some more detailed information coming through. More important is to have an institution and mechanism that will give us the level of analysis and give us credible data on which we can base a response in a more timely manner."[115] We share Mr. Holden's views; improvements to the capacity of early warning mechanisms, and Ministries of Agriculture, are very important. However, the Secretary of State cautioned against the naïve view that these complex problems can be easily resolved through technocratic interventions. In a context of collapsing state capacity—as in Zimbabwe—building or rebuilding effective early warning systems may be politically unfeasible for the foreseeable future. As she put it: "In many very poor countries we are experiencing a collapse in state capacity so you cannot put in a good early warning system on hunger if everything else is weakening. We have to put it in a context that works."[116]

The humanitarian response

71. The donor response to the humanitarian crisis in southern Africa has gone through three phases. At first (late 2001) the donors denied there was a crisis at all; then they responded but slowly and inadequately (early 2002); and finally they launched an enormous international appeal (mid-2002) and moved rapidly to prevent the recurrence of a tragedy. As SCF-UK wrote in their memorandum: "Until mid-2002, the international humanitarian response to this crisis was inadequate, lacking due attention and in-depth analysis. […] the international humanitarian community is now orchestrating what may become the largest humanitarian effort in the southern Africa region."[117]

72. In 1995, a book published about famine early warning systems had the subtitle: "The Missing Link".[118] This book focused on policymakers' failure to act on available information as a key factor contributing to famines. Several witnesses made this point in the context of the current southern African emergency:

early warning systems are only as good as the information that goes into them; but they are also as good as the response that they generate. What we have seen increasingly is not a failure of information but a failure of response. […] for whatever reason governments and donors do not act on the signals.[119]
inadequate food security and livelihood monitoring systems existed, and there was poor integration of these systems with effective national, regional and international policy response mechanisms.[120]
some of the donors—in particular the EU Food Security Office—did not take the early warning signals about food crisis seriously in the 2001-02 season. As a result, much valuable time and many lives were lost. […] There is a need to create a system for translating early food crisis warnings into action. The system should agree on indicators which would ring alarm bells—such as a low maize harvest and sharp increases in key food prices.[121]

Fund-raising: Food and non-food aid

73. In the last quarter of 2001, the World Food Programme (WFP) changed their senior management in Malawi and Zambia and opened a new office in Zimbabwe to improve and increase their capacity in the region. In March 2002 the United Nations Inter-Agency Standing Committee Working Group convened a special meeting in Rome on the emerging Southern Africa Food Crisis. Following this, in April/May 2002, WFP co-ordinated a series of inter-agency vulnerability assessments in the six most affected countries in the region, together with OCHA, UNICEF and FAO. In May, WFP set up a regional coordination and logistics unit in Johannesburg, to provide an overall humanitarian coordination role. On 67 June, a Humanitarian Needs Meeting was held in Johannesburg, and in July 2002 WFP launched an international appeal (EMOP 10200) for almost one million tonnes of food worth $507 million, with the aim of assisting 10,255,880 people throughout the region, including a General Food Distribution to 9,958,000 people. The initial focus of the appeal was on food, with a slightly delayed recognition of the centrality of the HIV/AIDS pandemic. In its November 2002 update, the UN remarks that although the international community was acutely aware of the HIV/AIDS pandemic, its depth and breadth "was not fully factored into the response."[122] Nevertheless, as part of the July 2002 appeal, $104 million was requested for non-food items, including support to health, water supplies and other vital social services, as well as the provision of agricultural inputs.

Figure 8: Funding for the southern African humanitarian response

Sector	Requirements (US$)	Contributions
(% of requirements)
Agriculture	29,783,796	35%
Coordination and support	9,814,183	42%
Economic recovery and infrastructure	1,949,000	Not available
Education	11,016,731	Not available
Family shelter and non-food items	900,000	Not available
Food	539,378,619	70%
Health	64,339,161	15%
Multi-sector	557,000	378%
Protection/human rights	1,425,000	Not available
Water and sanitation	8,511,385	8%
Total	671,864,225	60%

Source: UN-OCHA, reported in UN Regional Inter-Agency Coordination Support Office (RIACSO) update, 24 February 2003 - Available at www.reliefweb.int

74. By 21 February 2003, contributions to the southern Africa appeal (EMOP 10200) amounted to 60% of requirements[123] This is a significant improvement from the situation in October 2002, when only $200 million of the total appeal for $611 million had been secured.[124] The food aid component of the appeal is now—March 2003—70% funded, with the largest contributions coming from the United States and the United Kingdom. As is typical in most famine relief programmes, the response to the non-food appeals has been much lower than for food. For instance, the health component and the water and sanitation components of the appeal are 15% and 8% funded respectively. In November 2002, UN-OCHA complained that non-food needs, although of equal importance to food, "cannot be described as 'adequately addressed.'"[125] Most worryingly, UN-OCHA noted too that: "the impact of HIV/AIDS both as a cause and an effect of the current food crisis has been underestimated in terms of response required"[126] (see paragraphs 141-153). Also in November 2002, the UN summed up the response to the food appeal to date as "positive", but the response to the non-food requests as "limited" and the timing of contributions as "disappointing".[127] These remain fair assessments, and were re-iterated in February 2003 by the UN's Mid-Term Review, which emphasised the need to fund non-food items including water and sanitation, educational supplies, agricultural inputs and medicines.[128]

Logistics and delivery

75. Transport and infrastructure bottlenecks throughout southern Africa imposed constraints on the mobilisation and delivery of both commercial food imports and food aid in 2001/02. As we were told: "Simultaneous shortages elsewhere in the region, together with difficulties in managing the demands placed on an already difficult transport system meant that maize imports came into [Malawi] very slowly."[129] Limited port handling capacity, deteriorating rail and road infrastructure and lengthy customs clearance procedures were identified as serious impediments.[130] Nacala port in Mozambique proved unable to cope with the volume of food imports, rail routes from South Africa were affected by a derailment on the Zimbabwe border, and trucks were already engaged in transporting other commodities (including fertilisers, later Christmas goods), leaving little spare capacity. Within Mozambique, "the almost complete lack of adequate road and railway infrastructure linking the maize deficit south from the maize surplus north" inhibited the movement of food within the country.[131]

76. Following the problems with food imports and food aid distribution in 2001/02, substantial investments have been made in upgrading ports and rail lines to expedite the response to the crisis of 2002/03. In particular, DFID deserves praise for its support for the major improvements made to the Nacala rail link from Mozambique to Malawi. Nonetheless, it remains the case, that: "what is happening in Southern Africa is as much a logistical crisis right now as anything else, because they do not have enough transport; they do not have the institutional support; they do not have the presence of aid agencies like WFP on the same scale; they do not have the history or institutional memory for dealing with these kinds of crises."[132] The Government of Mozambique has set up a National Institute of Disaster Management (INGC) to coordinate emergency preparedness, with financial support from UN Development Programme (UNDP). However, information flows to and from the provinces and districts are extremely weak, because the INCG has been under-resourced and lacks physical infrastructure. Local NGOs have been working with the INCG to strengthen its role, with support from Christian Aid, which concludes: "Despite a current lack of capacity, such national bodies could play a key role in assisting communities to cope with situations of stress before they occur. However, they need to be resourced, well planned and supported by local structures to be effective."[133]

77. Despite the financial and logistical constraints which it has faced, the WFP has achieved impressive—if somewhat patchy—results, particularly as regards the proportion of its intended beneficiaries that it has managed to deliver food aid to (see figure 9). At one extreme, in Zambia, where the GM issue has added another layer of constraints, WFP had by December 2002 reached 40% of its intended beneficiaries. In Swaziland, WFP had met 93% of its target number of beneficiaries. For the region as a whole, WFP had delivered food aid to more than 6 million people, 62% of the target number.

Figure 9: WFP Delivery Coverage, July-December 2002
	Lesotho	Malawi	Mozam-bique	Swaz- iland	Zambia	Zimba-bwe	Regional Total
Beneficiary target	445,000	3,188,000	440,000	231,000	1,706,000	3,903,000	9,913,000
Beneficiaries actual	330,440	2,339,527	247,130	214,589	680,876	2,354,210	6,166,772
Coverage of beneficiaries	74%	73%	56%	93%	40%	60%	62%
Metric Tonnes target	40,175	140,943	42,435	16,548	90,024	290,852	620,976
Metric Tonnes actual	13,307	100,821	14,301	9,775	31,506	96,263	265,973
Metric Tonnes coverage achieved	33%	72%	34%	59%	35%	33%	43%