Assessing hunger: what do the FAO’s revised undernourishment figures tell us?

18 March 2013

The UN Food and Agriculture Organization's picture of how the world "hunger problem" has evolved over time has been turned on its head; from a secular deterioration to steady improvement.

For the first time since 1996, the UN agency responsible for food and agriculture has significantly revised how it arrives at an estimate of the number of hungry people in the world. The FAO has been estimating this figure, the Prevalence of Undernourishment (PoU), for decades now. Notably, when compared with the recent past, the picture of hunger offered by the organisation now shows steady improvement since the 1990s, instead of secular deterioration.

The global "hunger problem" has evolved over time, as has our understanding of its complexities. The revised estimate of undernourishment for the most recent year, 2009, is 1.5 percent lower than it would have been if the FAO's old methodology and data had been used. Significantly, however, when the new methodology is used to generate estimates for the 1990s, the figures are about 18 percent higher than those presented earlier. This implies that the global long-term trend in PoU has declined steadily since the early 1990s up to the 2008-10 period. This runs counter to previous estimates, which showed a continual increase in the number of undernourished people from the mid-1990s up to the late 2000s. In other words, the entire situation has been turned on its head (Figure 1). This is quite a drastic change with important implications.

The PoU estimates are intended to help governments and other national and international agencies reduce the "hunger problem" and measure progress towards reaching the first millennium development goal (MDG), which calls for halving the proportion of people in the world without enough food between 1990 and 2015. With the new trend estimate, the prospects for reaching this goal seem to have improved, although a large number of countries in Sub-Saharan Africa, as well as populous India, are unlikely to attain it.

The revised estimates show that, as a proportion of the growing population, the number of undernourished people declined from 23.2 percent in 1990-92 to 14.9 percent in 2010-12 (FAO 2010b, Annex 1). Reaching the first MDG target of 11.6 percent by 2015 no longer seems impossible. The new FAO global trend estimates also square better with the trends indicated by other nutritional measures, most notably anthropometric indicators, such as the prevalence of childhood stunting, and the incidence of underweight children and women aged 15-49 years.

The FAO's PoU estimates are based on the average amount of food available over three years, the calorie content of that food, and the distribution of the calories across households. The FAO further sets up norms for the minimum calorie intake for different gender and age groups. Those estimated to have a habitual calorie intake below the minimum norms are classified as undernourished. All these entities have been estimated on data that are incomplete and subject to errors (Svedberg 1999).

What is new?

What, then, lies behind the FAO's revised PoU estimates? Two things: changes in the methodology used to arrive at the estimate and newer, more complete data used for the building blocks of the FAO model.

On the methodological front, the innovations are in (a) the assumed distribution of dietary energy consumption and (b) the way in which variations in habitual food consumption are estimated (FAO 2012a: 10-11). On data, the FAO has revised figures from the UN population division, updated the heights of adults from recent anthropometric surveys, and refined assessments of caloric availability for the new PoU numbers. The methodological changes account for a rather minor share of the overall changes. The data on population size and adult height also have marginal impacts, as do the "improved" gross calorie availability estimates, except for the latest years (the 2008-10 average), when upward-revised food supply data induced a reduction of PoU by some 60 million people, or by 8 percent from the previous estimate for the same years.

The most important data change is that estimates of food losses at the retail distribution level, not only at the production and storage levels, are taken into account. According to the FAO's calculations, incorporating such losses in the PoU estimates raises the number of undernourished people in the world by 125 million in the 1990s, if everything else is held constant. Proportionally, this means an increase of about 15 percent for that period (FAO 2012a, table A2). These losses reduce the quantity of food actually consumed and, hence, increase the number of undernourished people.

The inclusion of food losses at the retail level is undoubtedly called for. The problem is that the size of these losses is extremely difficult to put numbers on. This is the very reason why they were not incorporated in earlier PoU estimates. The reliability of the FAO's new data on losses will surely be closely scrutinised in due time. Indeed, the FAO itself acknowledges that the new loss estimates are "still tentative" (FAO 2012a: 10).

For all the inherent uncertainties and limitations of the FAO PoU estimates, there are two welcome improvements in the 2012 Technical Note.  One "novelty" is that the problems with the basic estimation method and the data used, are more openly admitted and discussed than in most earlier FAO publications. In previous reports and documents most critiques of the organisation's PoU estimation method were either ignored or dismissed off hand.

A second novelty is that the FAO is in the process of developing a range of additional food security indicators, intended to reflect changes in "determinants of (or inputs to) food security", as opposed to outcome indicators (such as PoU or anthropometric measurements). Especially welcome is the proposed food-price-level index. This index aims to capture how food prices evolve in relation to consumer prices in general in developing countries. Such an index would be a more relevant, reliable and more quickly up-dated indicator of food scarcity than the food availability data now underlying the PoU estimates, data that are utterly unreliable and often become available only after three to four years. Some of the other proposed additional food security indicators are of doubtful value (e.g. rail-line and road density) and others are already available from other international organisations (e.g. household access to improved water and sanitation).


All in all, the 2012 revisions of the methodology and data used by the FAO for estimating PoU are to be welcomed. This should not distract from the fact that the basic approach used by the FAO to assess the nutrition situation in the world and in individual countries is fraught with limitations and ambiguities. First, the notion that nutritional status is chiefly determined by calorie intake is, to put it mildly, a gross simplification. Different calorie expenditure requirements in individuals are modelled very crudely, micro-nutrient deficiencies are omitted and so are curable illnesses that often go hand in hand with both the absence of food and poor nutrition. Moreover, the number of people overweight or obese - facets of malnutrition that are on the rise - are ignored.

Second, as the 2012 revisions testify, the required data remain of doubtful quality and small alterations in parameter values induce large differences in the ensuing PoU estimates (Svedberg 1999; 2011). It is also telling that in its 2009 Food Insecurity report, written after the world-market food prices had started to raise notably in late 2007, the FAO projected a drastic increase of undernourished people in the world, to above 1 billion.  As Figure 1 reveals, this number was revised downwards in 2012, to about 850 million.

Third, but not least important, even if the basic data could be vastly improved, the FAO estimates of PoU cannot provide answers to questions that are imperative from a policy perspective. In order to directly inform policy interventions it must be known who the malnourished are (infants, young children, adults or the elderly), where they are located (urban/rural, by region, etc.) and when they are malnourished (seasonally?). The FAO method is, and will remain, silent on these policy-relevant questions. There is therefore little doubt that anthropometric measurements, which can help answer these kinds of questions, are and will continue to be the main indicators for assessing the nutritional status of populations, globally as well as in individual countries and sub-regions.


FAO. 2012a. The State of Food Insecurity in the World 2012. Technical Note.  FAO. October 7, 2012

FAO. 2012b. The State of Food Insecurity in the World 2012 (

Svedberg, P. 1999. "841 million undernourished?". World Development 27, 2081-98

Svedberg, P. 2011. "How many people are malnourished?".  Annual Review of Nutrition 31, 263-84

This article is published under
18 March 2013
Cross-border trade below the radar of government oversight is both a boon and a curse for many African economies. COMESA, the Eastern and Southern African trade bloc, has recently established...
18 March 2013
The three major West African importers of fish products - Nigeria, Ghana and the Ivory Coast - can consume the total exports of other countries in the region without satisfying their own needs...