Right‑Sizing Food Assistance: New Approaches to Estimating Food Gaps in Acute Crises

The number of people facing crisis-level or worse acute food insecurity has nearly doubled over the past 5 years, to encompass 300 million people. During the same period, however, funding for humanitarian assistance has almost halved. While these opposing trends have put humanitarian agencies in deficit and struggling to respond to urgent global needs, there is also the question of whether we know how much food assistance is actually needed to prevent starvation. Existing early warning systems for food crises merely identify the number of people in need of food assistance. They do not identify the actual shortfall in food intake of those in need.

In a new IFPRI Discussion Paper, Exploring food gaps for “right‑sizing” food assistance, we explore ways to fill this void and assess various approaches to estimate shortfalls in food consumption. This analysis lays the groundwork for more precise and context‑appropriate humanitarian food assistance.

Why Measuring “Food Gaps” Matters

The Integrated Phase Classification (IPC)[RV1]  is an inter-agency mechanism for monitoring acute food insecurity and is considered by many the “gold standard” in the business. IPC categorizes populations into five phases of severity ranging from “minimal” to “catastrophe/famine.” However, these five phases represent broad categories. They classify entire areas within countries as in Phase 3 (“Crisis”), 4 (“Emergency”), or 5 (“Catastrophe/Famine”) if more than 20 percent of the population has been assessed to face the corresponding severity in acute food insecurity. The phase classification does not identify the degree of food deficiency of the affected populations. Not knowing this deficiency makes it difficult for policymakers and humanitarian agencies to target the right quantities of food assistance to meet actual needs.

This lack of adequate data is the key constraint. IPC analyses can rarely rely on household surveys that measure actual food consumption to identify caloric and other food-intake deficiencies. Conducting such surveys is not only costly and time-consuming, but it is often also hampered by conflict and mass displacement in the fragile settings where most food crises emerge. Instead, IPC assessments need to rely in good measure on proxy indicators for which data are easier to collect through simple and short questionnaires, such as dietary diversity scores or experience‑based measures of hunger. These indicators have proven useful in broadly classifying populations in need but inadequate for quantifying actual shortfalls in food intake and in means to access food.

The new IFPRI study explores three possible ways to estimate food gaps despite these data limitations: (1) using a “naïve” approach to estimate caloric gaps; (2) using correlates among proxy food insecurity indicators; and (3) identifying dietary deficiencies across multiple dimensions.

A “naïve” calorie‑gap approximation

The first method assumes an average caloric deficit for each IPC Phase, drawing on the Household Economy Approach (HEA), one of the tools used within IPC assessments. The HEA provides presumed thresholds for calorie deficits in each IPC Phase (see Table 1). We then apply the upper and lower bound of this deficit uniformly across the population in each phase (so each individual in an area classified as IPC Phase 3 is assumed to have the same calorie deficit).

Table 1 “Naïve approach” to calculate food gaps by IPC Phase 

Source: Rice, Brendan; and Vos, Rob. 2026. Exploring food gaps for right-sizing food assistance: Methods, data challenges, and lessons learned. IFPRI Discussion Paper 2414. Washington, DC: International Food Policy Research Institute.

Although simplified, this approach provides a first estimate of how much food assistance, expressed in calories or cereal equivalent, would be required to eliminate these deficits. The range of caloric food gap and food assistance requirements across selected food crisis countries is presented in Table 2.

Assuming a uniform caloric deficit across all households within each IPC phase and using the mid-points of the ranges in Table 1, the required daily food assistance in 2025 for Afghanistan, for instance, would be 1,029 MT of cereals. This figure is illustrative of how food assistance needs can be derived from this framework, rather than from operational estimates, as the uniform per-phase deficit assumption is a rather strong simplification.

Table 2 Estimated food assistance requirements (in MT of cereals) for selected countries, 2024

Source: Rice, Brendan; and Vos, Rob. 2026. Exploring food gaps for right-sizing food assistance: Methods, data challenges, and lessons learned. IFPRI Discussion Paper 2414. Washington, DC: International Food Policy Research Institute.

Using correlates between alternative food security indicators to measure caloric deficits

The second approach taken in the paper investigates whether proxy indicators used in IPC assessments – including the Food Consumption Score (FCS), Household Dietary Diversity Score (HDDS), Coping Strategy Index (CSI/rCSI), Household Hunger Score (HHS), and Food Insecurity Experience Scale (FIES) – can reliably approximate calorie deficits.

The analysis draws on FAO’s Data in Emergencies (DIEM) household‑level data matched with IPC area classifications. Across thousands of geographic units and several survey waves, we test how well these indicators correlate with each other and with the calorie‑gap estimates of the “naïve approach.”

We find correlations to be generally very low, even between indicators that are conceptually similar. For example, FCS and HDDS – both measures of dietary diversity – show only moderate correlation. FIES and rCSI show stronger alignment, but both relate to coping and hunger experiences rather than intake levels. None of the indicators correlate strongly with the assumed caloric gaps.

These findings imply that dietary diversity or hunger experience indicators cannot reliably be used as proxies for calorie deficits. This is an important insight for practitioners who often rely on these indicators under the assumption that they convey similar information.

Identifying “food gaps” across multiple dimensions

Since proxies cannot be used to derive caloric deficits with the available data, we propose a third approach, which is to estimate deficits against the conceptual scale of each of the food insecurity indicators referred to earlier. Again using FAO’s DIEM database, we estimate alternative food gaps as the difference between each household’s indicator score and the IPC‑referenced thresholds that define “minimum adequacy” for each proxy indicator. This yields, for example, a dietary diversity gap (using FCS or HDDS), a hunger experience gap (HHS), and a reduced coping capacity gap (rCSI).

Crucially, the analysis reveals that significant shortfalls exist across all IPC phases, even in IPC Phase 1 or 2 areas: categories with presumably no acute food insecurity or no serious food deficiencies (see Table 3). Conversely, many households in IPC 3+ areas show no dietary or hunger shortfalls, illustrating how area‑level IPC phases mask substantial within‑area variation. It also suggests that geographic targeting of food assistance (i.e., by area) likely will imply large targeting errors of both inclusion (providing assistance to people not really in need) and exclusion (people in need who do not receive assistance). 

Table 3 Dietary and coping capacity FGT indices by IPC Phase

Source: Rice, Brendan; and Vos, Rob. 2026. Exploring food gaps for right-sizing food assistance: Methods, data challenges, and lessons learned. IFPRI Discussion Paper 2414. Washington, DC: International Food Policy Research Institute.

We then use these “food gap” or “dietary deficiency” estimates to calculate the possible cost of certain interventions based on IPC phase classifications and the population in those phases in each food crisis country with available data. The average gaps by phase can then be multiplied by whatever unit is considered relevant in a particular response context. For example, one stopgap “solution” would be to cover the gap through food distribution, cash transfers, or asset protection. To get a rough approximation of the cost to fill the gaps, we use the average cost of a basic WFP balanced food aid package, which WFP values at US$0.50 per person per day. We multiply this cost by 365 days and multiply that value by the population with a shortfall and the average estimated food gap in each IPC phase for each country. Early warning systems generally consider populations in IPC Phase 3 or worse as being in need of food assistance. Table 4 shows the resulting food assistance needs for people classified in IPC Phase 3 (“Crisis”) and IPC Phase 4 (“Emergency”) for the food gaps measured through each of the proxy acute food insecurity indicators. 

As the estimated shortfalls vary depending on which acute food insecurity indicator is used, so does the estimated cost of required humanitarian assistance. For instance, the required food assistance for people in need would be US$ 510 million in Afghanistan in 2025 when using the food consumption score and US$ 653 million when using the household dietary deficiency score. However, the estimated assistance would be US$360 million when using the reduced coping score index and US$ 311 million when using the household hunger score. Again, as with the caloric deficit approach, the estimates presented in Table 4 should thus be taken as merely illustrative of how the approach could be deployed. Because of the data limitations encountered, they should not be considered adequate for operational use.

Table 4 Annual costs estimates of food assistance needs per country and indicator, 2024-2025

Source: Rice, Brendan; and Vos, Rob. 2026. Exploring food gaps for right-sizing food assistance: Methods, data challenges, and lessons learned. IFPRI Discussion Paper 2414. Washington, DC: International Food Policy Research Institute.

The way forward

The study concludes that accurate, actionable food‑gap estimation requires better data – particularly directly measured calorie deficits in areas facing acute food insecurity, as well as representative household consumption surveys linked to IPC assessments. Still, in the absence of that data, the indicator‑based approach in the third method offers a practical starting point and a transparent way to communicate the magnitude of need across contexts.

We highlight four priorities for further work:

1. Improve survey representativeness and sampling documentation
2. Validate threshold values for all IPC indicators
3. Conduct research linking caloric intake with acute and chronic food insecurity dynamics
4. Use machine‑learning and integrated datasets to better understand drivers of both the severity and depth of acute food insecurity

With humanitarian budgets tightening, knowing how much assistance populations need, not only how many people are affected, is urgently needed. Our study provides first proxies of food gaps and required food assistance. However, given the important data limitations that remain, further research and better data collection is needed before we can make definitive recommendations for how to “right size” food assistance.

Brendan Rice is a Research Specialist in IFPRI's Markets, Trade, and Institutions Unit. Rob Vos is a Senior Research Fellow in IFPRI's Markets, Trade, and Institutions Unit.

 [RV1]Suggest we add a hyperlink here