This analysis focuses on the institutional production of data concerning food security. It is a continuation of an experience with MSF in Ethiopia, where I worked as Head of Mission between January 2000 and October 2002. Confronted during this period with several nutritional emergencies, our intervention criteria often left me perplexed. To begin with, the fact that our knowledge of nutrition was still riddled with uncertainties forced us to make many adjustments in order to ensure that our programmes were coherent. Given the pressure of responding to emergencies, we sometimes had to choose intervention sites by balancing a ‘scientific’ approach based on the results of incomplete nutritional surveys and the ‘empirical’ application of in-house emergency response skills, such as therapeutic feeding centres (TFC). Furthermore, the Ethiopian officials with whom we dealt tended to question the validity of survey results and the relevance of TFC type programmes. The medico-nutritional approach that we had taken such pains to develop was at odds with approaches that seemed to stem from different forms of logic. Did officials fear losing control of the people we were treating in the feeding centres? Did they fear that our action would highlight the failure of Ethiopia’s food security policy, or that the media would focus exclusively on images of emaciated children congregating in a TFC? Whatever the case, our nutritional criteria were often of little use when trying to convince the authorities of the sound basis of our operations.

On our side, the implementation of nutritional programmes was accompanied by similar suspicions – justified or unjustified – that information regarding food security was being manipulated to serve political aims. Preconceived and sometimes obscure ideas about the dangerous liaison between famine and politics were always likely to nurture suspicions of this kind, but they were certainly reinforced by the incoherence of available data on food insecurity. We were faced with the difficulty of identifying pockets of food insecurity and of defining the priorities for action, problems which stemmed as much from the proliferation of sources – NGOs, government, international agencies – as from the multitude of often contradictory, if not incomplete, evaluation reports that were passed around at meetings.

Finally, after having spent vast sums of money on responses to ‘emergencies’, we were still left with a nagging doubt: had we actually helped to exaggerate the drama and the scale of the action? How could we admit without flinching the widespread excesses of emergency programmes? There were over 400 NGOs present in the country; the United Nations maintained a massive presence there; aid had been pouring in for many years, and innumerable officials and experts were working to assess food situations. Therefore, it was not unreasonable to expect a crisis to be identified before it escalated into a full-blown catastrophe. And there was also the question of whether the crises precipitated by poor harvests had been exacerbated by ‘dysfunctional’ elements in the aid system, flaws which facilitated the reproduction of institutional emergencies. The NGO, the last resort, joins the end of the chain and attempts to compensate for the inadequacies of the system and address the shortcomings of the so-called ‘early warning’ evaluation systems. Non-governmental organisations relieve decision-makers of the burden of responsibility. In this respect, they perpetuate the aid system; they fully adhere to it and consolidate it by maintaining its exponential dynamic.

The crisis that occurred in 2003 once again emphasised the black holes in the procedures designed to identify food requirements in Ethiopia. Following the national evaluations conducted by ‘multi-agency’ teams in November 2002, the situation in some parts of the country, mainly the highlands (Tigray, Wollo) and the lowlands (Ogaden, Hararge), was described as catastrophic. Wolayta zone was not considered to be at risk until MSF revealed the existence of high rates of malnutrition. At that time, national aid and beneficiary totals had already reached record levels and food was being distributed in the rest of the country. The omission of Wolayta was hard to justify: the zone was secure, easily accessible, and noted for the recurrence of ‘green famines’; several NGOs were operating in its districts. Within weeks, therapeutic feeding centres sprang up like mushrooms. The region’s capital became the centre of the 2003 emergency season, attracting NGOs, agencies and the media.

What are the reasons for such disparities in the distribution of aid? An initial temptation would be to go along with the ‘conspiracy’ theories which claim that hunger is a weapon and that food aid can be instrumentalised (Brunel, 2002). But the clear absence of political stakes in Wolayta in 2003 does not support such assumptions. Conspiracy theories not only oversimplify the complexity of the Ethiopian political landscape and the issues involved, they also ignore the operation of early warning systems (EWS) which, when subjected to analysis, suggests that such theories are too radical.

I therefore intend to look more closely at the role that early warning systems play in the mechanisms for allocating food aid. My approach may be summed up as follows: EWS are tools for the regulation of the political logic underlying the allocation of food aid. Before going on to support this argument with material acquired during a field study of multi-agency evaluation practices in Ethiopia, I will briefly examine the context and processes which led to the introduction of early warning systems.