Analysis Main Publications News The Yemen Review Publications Index

The Myth of Data in Yemen


In recent years, Yemen has come to the forefront as one of the world’s best-known crises. It is portrayed as the world’s biggest humanitarian crisis and the biggest humanitarian response world-wide. It is well known for being on the brink of the biggest famine in years, decades or perhaps a century,[1] and for having the largest cholera outbreak ever recorded,[2] among other headlines. This visibility has attracted an immense amount of interest and subsequent funding, making Yemen the second-highest-funded response in the world. This is particularly the case since 2017; response funding for Yemen more than doubled in 2018, with a high of US$5.2 billion being allocated to the response in 2019.[3] If one success can be accorded to the Yemen response, it is that it is among the best fundraisers worldwide, with more than US$17 billion raised for the response between 2015 and 2020.[4]

The financial success of the narrative has created a deep institutional investment in it. Big money is an attractive prospect for numerous reasons. Firstly, it is easier to scale up than to scale down, which requires divesting of offices and staff and rationalizing operations, and the administrative headache of reducing programming due to constrained funding is greater than actual fundraising. Secondly, money bestows importance. An organization’s leadership may be reluctant to lose the image and status that accompany appearing able to cover many needs, even if those needs are only being superficially addressed. For example, following a reduction in funding, a cluster coordinator’s request in 2020 to INGOs to streamline operations for better quality coverage by reducing geographical scope went ignored.[5] A staff member of one of the INGOs confirmed their organization had decided not to heed the request, saying that it looked better to their donors to have wider geographical coverage.[6] A UN agency staff member also pointed out, “there is a difference in status in being a program manager with a budget of tens of thousands of dollars and millions of dollars.”[7] Lastly, a certain part of each UN agency’s budget is allocated for institutional support at regional and head offices. This is part of how overhead costs in New York, Geneva and Rome are managed. The larger the response, the more money allocated to headquarters. As a result, from the bottom to the highest levels, there is a vested interest in keeping funding levels up.

But this narrative of “the biggest humanitarian crisis in the world” detracts from the real issues at hand. Countless anecdotes and reports of access challenges, diversion of aid and struggles to deliver assistance have and continue to be given. Failures in the Yemen response are often excused as stemming from a difficult operating environment, restrictions imposed by authorities, insecurity or insufficient funding. But rather than streamlining the response and addressing inefficiencies in the system, the Yemen response has focused overwhelmingly on fundraising alone, which has led to the need to perpetuate the narrative. Yet, as a majority of key informants pointed out, the amount of money is not the issue, but rather how to more effectively spend it.[8]

This contradiction can be seen in the tension that currently exists between two competing access narratives international aid organizations have pushed in recent years. The narrative in Yemen has been that access is difficult and that people are increasingly less accessible whether due to conflict or for logistical reasons, with up to 19.1 million people (64 percent of the population) living in hard-to-reach areas by the end of 2020.[9] This narrative has been the bedrock of the explanation as to why there is such limited humanitarian presence outside of Sana’a, Aden and the hubs, and why so much of the implementation is outsourced. USAID cited this outsourcing and the accompanying lack of accountability that ensued as being a large part of why it cut humanitarian funding in 2020.[10] Desperate to have this funding reinstated, INGOs and UN agencies lobbied the US, claiming improved conditions and access,[11] even though staff on the ground were finding the opposite — less access and more bureaucratic challenges.[12] The contradiction between the two narratives led a UN agency to block publication of access data in 2020 because it undercut the narrative to reinstate funding.[13]

This vested interest has led to a persistent inability to correct one of the most fundamental problems within the Yemen response, that of data, its use and interpretation. To continue to fundraise, the response must continue to captivate the interest of the world. To do so, the narrative must be sustained even if it is based on an exaggeration of needs, which are not evidenced by solid data and are doubted by humanitarian practitioners on the ground. Data must continue to support the narrative, even if the data used to justify doing so is limited, contradictory or flawed.[14] This has resulted in several problems. It has led to a warped understanding of needs and is diverting attention from needs and root causes to money for the system.[15] It has also led the response to become captive to its own narrative, thereby limiting its ability to innovate, affecting how data is discussed and preventing development of a more effective response. Lastly, it has led to the conflict being hyped as a deeply insecure environment, which is not in line with reality on the ground.[16] This report delves into the problems and challenges of data collection and its use in Yemen, and how its limitations are exploited to continue the prevailing narrative sustaining the response. A full discussion of security can be found in the next report, ‘To Stay and Deliver: Security.

The Qahfa al-Hamra IDP camp
The Qahfa al-Hamra IDP camp in Al-Shamayateen district of Taiz governorate, March 18, 2018. // Sana’a Center photo by Ahmed Al-Basha
Establishing Need: The Fundamental Baseline of Humanitarian Response

Any humanitarian response is based on the understanding of needs — the gaps that exist within populations affected by crises that must be filled to ensure survival and to maintain life at a minimal standard.[17] Needs are grouped into sectors for classification purposes: food and nutrition; shelter and non-food items; water, sanitation and hygiene (WASH); health and protection, among others. Without establishing needs, there is no clear rationale for a humanitarian response.

“We require needs assessments that are impartial, unbiased, comprehensive, context-sensitive, timely and up-to-date. … The needs assessment process must be coordinated, impartial, collaborative and fully transparent with a clear distinction between the analysis of data and the subsequent prioritization and decision-making.” —“The Grand Bargain,” struck at the 2016 Humanitarian World Summit in Istanbul

Accurately establishing needs in a response is crucial as it forms the baseline of any further action. Needs assessments, therefore, must be conducted independently, undertaken by those who have no vested interest in the outcomes. The fundamental necessity of impartiality and transparency has consistently been acknowledged and was specifically included as part of “The Grand Bargain” agreement, adopted at the 2016 Humanitarian World Summit in Istanbul by key aid providers and donors.[18] Assessments should be guided by the four humanitarian principles, a key aspect of which is that needs are established by independent, technically competent humanitarian actors without interference by parties to the conflict or others with a vested interest in the outcomes.[19]

The only exception to this relates to input by affected populations and (future) beneficiaries. As decisions based on outcomes from assessments directly affect the people any aid is intended to support, it is recommended to involve affected communities in the process. Doing so in the needs assessment process not only allows people a degree of control over their own situation, which promotes a key element of dignity, it also allows for greater acceptance of the results and subsequent interventions. In addition, such inclusion can help shed light on particularities of the environment that will need to be taken into account when planning a response and improve the appropriateness of the response in a given area. This contributes to an increased general acceptance of aid and aid workers, which has corresponding longer-term benefits for both access and security.[20] Finally, inclusion of beneficiaries within the process also improves the response’s accountability to the populations it serves. Accountability to affected populations has become a key concept within humanitarian aid and is an active commitment to using power responsibly.[21] Out of 31 Yemenis involved in the humanitarian response who were interviewed during this research, 10 directly indicated that a lack of involvement by local communities in needs assessments has led to a wrong or inappropriate response.[22]

It is also important to note that needs cannot be identified properly unless contextualized. This requires additional analysis of prevailing socio-economic conditions as well as cultural elements, and gender, power and conflict dynamics. Only then will vulnerabilities and relationships among needs as well as their root causes be well enough understood to focus the response so it effectively addresses the needs identified.[23] The importance of properly analyzing and understanding the wider environment within which humanitarian response takes place is seen in the next reports, ‘To Stay and Deliver: Security,’ and ‘To Stay and Deliver: Sustainable Access and Redlines’.

Understanding How Needs are Established

To establish needs, assessments are conducted. This crucial first part of the humanitarian program cycle provides a body of evidence that allows response objectives to be identified; it also brings into focus a picture of the crisis and the required response, helping to quantify the amount of aid needed. Assessments guide planning within relevant sectors. Such evidence-based strategic plans ultimately also support follow-up monitoring, making it easier to track whether commitments are being fulfilled and aid is reaching recipients as intended.[24] On the basis of needs assessments, human resources and financial requirements are calculated, setting the scene for the operational response.

Needs assessments are technical and are typically carried out for each sector by individuals who have expertise in or understanding of the data being collected. Good practice calls for a coordinated approach to avoid duplication and gaps, ensuring a coherent and efficient response. As humanitarian operations have evolved, standardized tools[25] and templates have been developed to guide system-wide responses for establishing need in accordance with minimum standards, to ensure a level of comparison across crises and to save time in responding to emergencies. The output of (coordinated) needs assessments form the humanitarian needs overview (HNO),[26] which contains the consolidated evidence base needed for the joint strategic response. Along with a shared understanding of the specified crisis, the HNO includes the most pressing needs and an estimated number of people who require assistance.[27] The HNO leads to development of a humanitarian response plan (HRP), which articulates this shared vision.[28]

While initial needs assessments are usually quick to ensure a basic understanding for a rapid response, as time goes on, needs assessments should become continuous processes throughout the humanitarian response cycle. This ensures better understanding of needs as they evolve over time and allows for adapting the response to changing circumstances. In a protracted crisis, for example, the depth and volume of information needed for an effective response increases as it evolves.[29] For longer-term emergencies, needs overviews are typically conducted twice a year.

Data and Needs in Yemen

The 2021 HNO for Yemen determined that 24.3 million people were in need out of 29.8 million (81.5 percent). Out of this figure, 19 million people were to be targeted for response.[30]

Figure 2.1

Source: UNOCHA, Global Humanitarian Overview 2021

While the data seems straightforward and fits in with the general narrative of Yemen being the worst crisis in the world and one of the biggest responses, the figures presented hide the fine print: that the data used to inform and present them is flawed. At times, the figures even appear arbitrary. For example, between the December 2020 release of the 2021 HNO and the 2021 Yemen HRP published in March 2021, need figures were revised downward from 24.3 million to 20.7 million allowing for fewer people to be targeted, 16 million instead of 19 million.[31] Yet both reports indicated the severity of the crisis was worsening.[32] All key informants interviewed in the course of this research, when asked about the veracity of the figures presented in HNOs and HRPs, indicated that they harbored significant doubts regarding their accuracy or what the actual level of need is across Yemen. All believed that the needs represented were overblown and, in reality, would be lower.[33] This presents a major problem at the root of the response. As noted above, an accurate understanding of needs is necessary to plan a response and to effectively allocate resources. And this does not exist in Yemen.

Beginning with a Poor Baseline

The baseline for the response was shaky to begin with. Six weeks after the 2015 evacuation of international humanitarian agencies from Yemen, limited staff returned to Sana’a. No movement was permitted outside the city, and the aid network and staff outside Sana’a was minimal.[34] Considering the prevailing security situation — heavy fighting, airstrikes, risky roads, civilians and warring parties on the move — conducting a quality needs assessment was impossible. According to a UN staff member who was present during this time, “we had no idea what was going on outside Sana’a. We had absolutely no idea whether [the] response was ongoing or done.”[35] As a result, and with only limited national staff present in Hudaydah, Aden and Sana’a, the needs assessment in 2015 for the 2016 HRP was based on a phone survey with no sound systematic or methodological assessment.[36] This was made clear in the 2016 HRP with many of the baseline indicators reflecting “no data.”[37] The inaccuracy and the scarcity of data was acknowledged in the next year’s HRP, for 2017. The humanitarian coordinator at the time directly referred to the scarcity of data,[38] and the response plan reduced the estimate of people in need to 18.8 million from the 21.2 million cited in 2016, when data collection and analysis were considered more accurate.[39]

Over time, however, data collection has remained problematic. Figures for the 2021 HNO were rolled over from the 2020 HRP because of a lack of available data, with some updating as available for the 2021 report.[40] What is important to note is that the 2020 HRP itself was based on a collation of data from 2016–2018 that had been used to extrapolate and estimate data for 2020.[41] This is in part because 2019 saw a 60 percent reduction in completed assessments, even though a concerted effort to improve the quality of the data was made in 2018 with the rollout of a multi-cluster location assessment (MCLA) in most of Yemen’s districts. Three crucial countrywide assessments — the MCLA, the Food Security and Livelihoods Assessments (FSLA) and the Standardized Monitoring and Assessment of Relief and Transitions (SMART) nutrition survey — were unable to be completed in 2019 because of obstacles imposed by Houthi authorities.[42] Digging deeper into the data of 2020 calls into question its usefulness and relevance (see Table 2.1). According to the references provided within the 2020 HRP, the majority of the data on which the HRP was based was at least two years old, while some data dated back as far as 2016.

Table 2.1

Source: 2020 Humanitarian Response Plan, UNOCHA[43]

Another example involves 2017 IPC data, which relied on an Emergency Food Security and Nutrition Assessment (EFSNA) for food security, nutrition and mortality data. The EFSNA did not include Al-Mahra, Socotra, Taiz and Sa’ada governorates, so the 2017 IPC data, which was used in planning the 2018 response, instead included 2014 data for Al-Mahra and Socotra.[44] The source of data used for Taiz and Sa’ada in the IPC analysis remains unclear, but it is most likely based on spot reports from various sources.[45] The problem is not new and has plagued response plans even before the current crisis. For example, the 2011 HRP references nutrition data from 2005 as its source for stunting.[46] As shown above, data in Yemen is consistently out of date and incomplete, compromising the reliability of the information used to make decisions on response and resource allocations and undermining trust in the foundation of the entire response.

The 2021 HRP was able to benefit from several key assessments finally carried out at the end of 2020, most importantly the FSLA.[47] This has gone some way to improving the data quality, at least for the food security sector. But, as the HRP explicitly points out, “comprehensive countrywide data are not available in Yemen,”[48] a recognition that the response in Yemen is still not based on fully evidenced data.

The response’s understanding of needs has been questioned on several occasions. The number of reported people in need was more than 60 percent of the population early in the response and has topped 80 percent in recent years; the number of people targeted has grown even when there has been a lack of success in meeting the targets. The number of people in need is calculated based on highest estimates rather than being an accurate reflection of the data, and the methodology for how the figure is calculated varies per sector.[49]

As far back as January 2016, an Inter-Agency Standing Committee (IASC) peer review noted the Yemen response’s emphasis was “on some form of humanitarian assistance. There needs to be a more focused approach on prioritization in the response.” It said the response must understand the humanitarian needs in Yemen through assessment, information-gathering and listening to Yemenis affected by the crisis, then prioritize what it could do based on people’s needs and the capacities of agencies and non-governmental organizations. “The Strategic Objectives of the HRP … need to be adjusted to ensure they clearly reflect what is needed and achievable in the response, rather than bland statements which are highly principled, but excessively aspirational.”[50] Despite this advice, HRPs have continued to reflect higher numbers and, to date, continue to refer to Yemenis requiring “some form” of need[51] without clarity or prioritization. Out of 31 Yemenis working on the humanitarian response who were interviewed, 12 directly indicated that the humanitarian response does not understand needs in Yemen appropriately.[52] Despite these shortcomings, this has not prevented the response, and particularly senior leadership in Sana’a and New York, from using this unnuanced data and projection of needs when making its case to donors.

Data Flaws Outside the Strategic Planning Reports

It is not only the data for the HNOs and the HRPs that is questionable. Other aspects of the response also lack adequate data, undermining reliability. One major UN agency, for example, does not know where a significant proportion of its distribution points are, and at least two others provide support to schools and health centers they have never visited and are not sure are functional.[53] The explanations often given for this relate to distribution points either being managed by authorities acting as implementing partners who refuse to share locations or to distribution being in hard-to-reach areas. There is not always evidence to support this. For example, distribution points for the above-mentioned UN agency in Al-Dhalea, which were managed by a local NGO, remained unknown even though the area was fully accessible for the majority of time until clashes broke out in the area in 2019.[54]

In addition, output data for the response in Yemen is often misleading. The 2019 HRP end-of-year report indicated 13.7 million people were reached during the year.[55] This figure, according to some key informants, is generated by using the highest monthly total in the calendar year of people who had received food distribution, which is the most widely dispersed service.[56] The end-of-year report stated 13.3 million people had received emergency food aid in 2019, and it noted that reach across clusters also had been considered in determining a total reach of 13.7 million people.[57]

It was not clear, however, how it was determined that there were 400,000 people who were not receiving food aid, but were receiving other assistance. In addition, WFP data from throughout the year contradicted this figure, indicating the highest number of people reached with food aid in a single month was 12.73 million in November 2019.[58] And even the veracity of that figure is questionable. Since the end of 2018, widespread diversion of food aid has been reported both internally and externally, culminating in the suspension of food distribution in Sana’a in June 2019.[59] It is implausible to believe that all, or even most of the aid reaches its intended targets considering the well-documented instances of diversion, the inability to monitor and the fact that the Houthi-led Ministry of Education — which has a vested interest in the allocation of those resources — implements a significant share of the food distribution. Figures for those reached merely indicate the amount of food dispatched for a certain number of beneficiaries; there is no credible data to verify correct delivery of that aid.[60] All of these discrepancies, combined with the fact that a one-month high — whether 13.3 million or 12.73 million — does not reflect consistent or inter-sectoral reach, only casts further doubt on both the quality of aid provision (whether, for example, people are receiving both food aid and health care) and on the overall figure used to make planning decisions and inform donors about the response’s reach.

The following section examines why data in Yemen is so problematic. It is important to note that access is often used to excuse a lack of good quality data in Yemen. Access, however, will be considered at length later in this series of reports, in terms of how a flawed approach to security risk management is limiting aid worker access in To Stay and Deliver: Security, and in terms of access limitations imposed by conflict actors and how those have been handled in To Stay and Deliver: Sustainable Access and Red Lines.

The Problems with Data in Yemen

Problem 1: A Lack of Field Presence

One of the main reasons for the lack of quality data is an issue that undercuts many areas of concern raised in this series of reports: the lack of presence of humanitarians in the field. Yemen is often touted as one of the largest humanitarian responses in the world. Yet, it has among the fewest international aid workers present (See Figure 2.2).[61]

Figure 2.2

Source: 2018 Yemen HRP End of Year Report, UNOCHA, August 31, 2019

In addition, the majority of international aid workers have been and remain centralized in Sana’a and, to a lesser extent, in Aden with some presence in field hubs.[62] All of the 23 aid workers interviewed in the course of this research cited the lack of presence in the field as a key issue, with all aid workers and 14 other informants from various sectors indicating it directly impacts their understanding of needs. This was identified as far back as the January 2016 peer review of Yemen operations, which found that “limited field presence and proximity to people affected by the crisis is impeding a clear understanding of humanitarian needs.” Expanding field presence in priority areas “will result in better information analysis [and] stronger evidence-based assessments … to accurately understand needs.”[63] Failed efforts to decentralize and the impacts this has had beyond assessing need are explored later in this series, in ‘A Centralized Response is a Slow, Ineffective Response’.

The only way to truly understand needs and the operational environment is to be present where the needs exist. Indicators can be useful, but to fully understand root causes of the needs established and to ensure a meaningful way to alleviate suffering, any data collected not only has to be correct, it also has to be understood in context, taking into account environmental, socio-economic and cultural factors. For example, one interviewee explained that in 2019, food production in areas under the control of the armed Houthi movement declined due to a ban on fertilizer imports that was enforced through coalition forces’ ports blockade.[64] Data captured the decline in productivity, and the reduction of availability in markets,[65] but the root causes were never properly identified and, therefore, no attempt was made to address them in a sustainable manner.

Problem 2: The Lack of Independently Sourced Data

Another key issue is the outsourcing of data collection. As discussed above, a crucial aspect of establishing needs is to do so in an independent manner, with the necessary information gathered by those who do not have a vested interest in the outcome or resource allocation. This is especially necessary in humanitarian responses such as Yemen, which operate in a conflict and where parties to the conflict, for myriad reasons, are vested in how aid resources are allocated.[66] In Yemen, the majority of data is collected by authorities, both in areas controlled by the armed Houthi movement and in those under the control of the internationally backed Yemeni government. These authorities hold control over all needs assessments and data collection.[67]

The lack of independence has been a problem from the very beginning of the process. Beneficiary lists for emergency response are compiled by the Supreme Council for the Management and Coordination of Humanitarian Affairs (SCMCHA) in the north, and the Executive Unit in the south.[68] For regular programming, the implementing partner is responsible for registration. Often, implementing partners are the authorities themselves. Other times, implementing partners are national organizations; in this case, their loyalties and interests must be taken into account. For example, at least one national non-governmental organization (NNGO) is directly linked to the Houthi-controlled Ministry of Education: The School Feeding and Humanitarian Relief Project (SFHRP), which in 2021 was reportedly distributing 60 percent of food aid.[69] Though lists are supposed to be verified independently, the verification process is often marred by interference and replete with mistakes and duplications.[70] This process has led to the exclusion of legitimate beneficiaries, including some of the most vulnerable, who are sidelined in favor of interests to the authorities.[71] Even when program staff from humanitarian agencies conduct needs assessments, they are frequently accompanied by authorities who act as spokespersons for the community, making it extremely difficult to gather independent data.[72]

Authorities’ influence in earlier stages of the process was also reflected in the implementation data of the humanitarian response. For example, in 2017, 54 percent of the total general food distribution caseload was handled by the Houthi-run Ministry of Education and SFHRP.[73] This is not just confined to the food security sector. The 2019 HRP end-of-year report, for example, indicated that in the WASH sector, response by NGOs decreased by 20 percent compared to 2018, with the activities being taken over by authorities who then came to implement 78 percent of the humanitarian response within the WASH sector.[74] This practice of relying on authorities to identify needs as well as who to help is apparent across all other sectors. Combined with a lack of staff presence in the field to enable proper monitoring, the door opens for interference and data manipulation.

Accompanying the lack of independence in data collection in Yemen is a lack of transparency about the potential bias within this data. Data may be collected, but it is usually not clear who collected the data, where the source is and whether there has been any accounting for potential bias. If authorities collect data on their own, without oversight and independent data verification, the likelihood of accurate data is low. This is supported by other research on data collection in Yemen. For example, research on food security data noted that “humanitarian agencies using data from politically motivated parties to inform their reports are not upfront about potential political biases and frequently are not even aware of the potential biases in the information on which they base programs and activities.”[75] This study noted that most data was collected by or with authorities,[76] that data was not allowed to be taken out of the country and that there are extreme limits to even how much data is shared within Yemen. Donors complained they were required to make allocation decisions without being able to see the data needed, and often the only data available was outcomes, without underlying data to evidence the results.[77] In addition, those providing the data were often not transparent about where the data came from and its potential bias.[78]

Problem 3: Methodology, Transparency and Concept Definition

Understanding how data is gathered is key to its correct interpretation. A 2019 study of data in Yemen found that 64 percent of data-based products available to the public that it had reviewed did not discuss the methodology used to collect the data,[79] indicating a fundamental flaw of data collection in the response. Recognizing these challenges and aiming for a more streamlined response in line with global policy, efforts have been made to collect better data and to systematize data collection, such as the new Joint Inter-Sector Analysis framework that aims to improve the measurement and analysis of needs.[80] Yet, even within these efforts as well as more broadly, contradictions exist due to methodology — in how data is collected, the limited amount of information available, what is included and how concepts are defined.

A clear example of contradictory data relates to the collection of data on fatalities in Yemen. Both the Civilian Impact Monitoring Project (CIMP), which falls under the UN Protection Cluster for Yemen, and the US-based non-profit Armed Conflict Location & Event Data Project (ACLED) collect data on civilian fatalities, or civilian deaths due to the conflict. Both organizations are respected, use sound methodologies and record data systematically. Still, their totals from December 1, 2017, through December 31, 2020, differed by more than 1,000 people: CIMP recorded 4,415 civilian fatalities in that time,[81] while ACLED counted 5,466 over the same time period.[82]

One difference between the two organizations is how they define “civilian” fatalities. CIMP records all civilian fatalities, no matter whether the death was due to collateral damage from an event related to the conflict (for example, a civilian killed in crossfire between two warring parties).[83] ACLED, on the other hand, only counts civilian fatalities when the main target of the attack is civilian.[84] However, data between the two organizations also differs because of the information sources available. Information in Yemen is heavily reliant (much more so than in other contexts) on indirect data collection as few partners are present on the ground. All data from CIMP and ACLED on casualties, fatalities and violent events is collected through open source means, which can be difficult to verify. Information is filtered and cross-referenced as much as possible, but there are times when it is difficult to establish correct numbers and disaggregate data.[85] Different sources can quote different fatality or casualty numbers for the same incident, and factors such as age or sex are not necessarily mentioned. If only conflicting numbers or unclear information is available, it is up to the methodology established by the researchers to make a call on how to record the number. For example, if mention is made of the word “victims” or “fatalities” but no number is given, and there is no other evidence to form an estimate, CIMP then takes the most conservative estimate and puts the fatality count at two.[86] ACLED, on the other hand, usually uses 10 as the common denominator in these cases. This means that both organizations count consistently but have diverging numbers. The sources also hold inherent bias. Verifying casualty and fatality data is complicated when, especially in remote or inaccessible areas, there are no independent partners on the ground. Media, and especially social media, also suffer from reporting biases, and accurate data collection is challenging without an ability to verify such reports.[87]

Methodology is also key to understanding to what extent data is representative of a situation. In normal circumstances, most data is collected at the household level. Due to an inability to properly conduct household-level surveys, key informant interviews have become a main avenue for data collection in Yemen. For example, the 2018 MCLA in Yemen, which by design is a household-level survey, started off as focus group discussions then developed into a large-scale key informant interview exercise because household surveys have been blocked by Houthi authorities.[88] While key informant interviews are an important method of data collection, they cannot replace randomized widespread surveys that sample a larger population. The selection of key informants also has the potential to be biased,[89] which needs to be taken into account when analyzing interview data.

This is particularly problematic as the MCLA is intended to be the foundation of how the number of people in need is determined. In theory, this number is based on a representative survey of the population. But because the MCLA was not rolled out properly in Yemen, and data was collected through key informant interviews, it could not accurately reflect the number of people in need. An additional problem with MCLA data was that it required piecing together bits of assessments collected from different areas of the country, and questions asked were not always the same. Neither was the data collection methodology consistent, which has made the comparison and compilation of data problematic.[90] Combined with the fact that there had not been an MCLA since 2018, the basis was thin to assume accuracy of its projections for 2020 and 2021. This alone greatly undercuts the validity of the entire humanitarian response plan. Yet these flaws have not been made clear in official documentation setting out the population in need in Yemen.

Transparency is the key difference among the examples discussed above. Both CIMP and ACLED are transparent about their methodology and the challenges they face around collecting data. This provides users the ability to better interpret and understand the data provided, and to use it in a more appropriate manner to inform advocacy and response decisions. However, this cannot happen when data limitations are not made clear, as in the case of the MCLA data collection.

Ill-defined Concepts Lose Meaning, Skew the Data and the Response

Narrative frameworks and labels based on shared definitions are ways to implicitly trigger a broader understanding within the humanitarian sector, whether referring to violence in South Sudan[91] or cholera cases, food security and famine in Yemen. This shared understanding, then, informs explanations given and responses proposed.

An Ill-Defined Epidemic: The Impact of Flawed Cholera Data

Among the biggest Yemen headlines has been “the largest cholera outbreak on record,” with more than 2 million cases reported since 2016. This claim continues to be made in the media and by UN agencies in Yemen,[92] even though the data behind it is flawed.

In 2019, a staff member of an international medical organization highlighted that the cholera definition as applied in the Yemen cholera response was problematic,[93] a sentiment affirmed by an aid worker within the cholera response.[94] In places where an outbreak has been declared, such as in Yemen, the World Health Organization (WHO) defines a suspected cholera case as any person presenting with or dying from acute watery diarrhoea (which it defines as three or more loose or watery, non-bloody stools within a 24-hour period).[95] The case definition laid out by the response in 2017 for suspected cholera was generally similar: “any patient presenting three or more liquid stools with or without vomiting for the last 24 hours.”[96] Both definitions are broad, meaning other diarrhoea-related illnesses endemic to an area can get noted as suspected cholera, so observation and confirmation of such cases help provide a more complete understanding of the severity of the outbreak. WHO considers any suspected case to be confirmed if a culture or polymerase chain reaction (PCR) test confirms Vibrio cholerae O1 or O139.[97]

A December 2018 review of cholera in Yemen noted issues with the lack of systematic clinical confirmation of cases and low-quality data, but it also found that the number of suspect cases reported had likely been much higher than the actual number that met the suspect case definition.[98] Researchers from Johns Hopkins Center for Humanitarian Health found that there was a tendency to report on the cholera line lists any consultations in which a patient reported diarrhoea, even if the suspect case definition — which requires meeting the specific indicators of acute watery diarrhoea — was not met.[99] They also noted that logs of all consultations, regardless of whether diarrhoea was among the patient’s complaints, were “frequently” used to fill in cholera line lists, meaning a patient could have been seeking treatment for an unrelated ailment.[100] This situation, which the report said worsened over time, could have been due in part to the incentive structure put in place by UN agencies for the cholera response. It noted that WHO and UNICEF paid the wages of health workers staffing cholera treatment centers because they had not been receiving their government salaries since 2016, which the authors said inadvertently incentivized ensuring the centers appeared busy so they would be kept open and salaries would continue. [101] In addition, there was limited observation time of those suspected to have cholera.[102] As a result, the already liberal definitions of cholera were rendered meaningless, and the system was flooded with false cholera cases, leading to massive inflation of numbers as reflected in WHO data. In 2019, for example, 825,000 suspected cholera cases were reported. Out of these, 9,694 were officially tested – with 5,298 being confirmed as cholera cases (54.7 percent).[103]

The startling headlines were based solely on dubious figures for suspected cases. While they helped attract funding, they also negatively impacted response. As aid workers involved in the response attested, they were left without accurate data needed to target and respond to true cholera.[104] Improvements were only seen in 2019, when proper criteria began to be applied and observation time respected. Testing has remained an issue.

In Yemen, the term “food security” has become one of the leading terms of the response following years of predictions that Yemen is “on the brink” of famine.[105] But “food security” and “famine” are terms often bandied about without any clear understanding of what they mean, how they are measured and who means what by them.

Food security is generally defined as the situation that exists “when all people, at all times, have physical and economic access to sufficient, safe and nutritious food that meets their dietary needs and food preferences for an active and healthy life.”[106] This definition, adopted at the World Food Summit in 1996, is the most commonly accepted for food security, even though it is not universally considered sufficient.

Defining food security within the humanitarian sphere is important because it prescribes measurement, which determines response. The global definition noted above, while providing general guidance, does not account for differences and levels of food insecurity. By this definition, which is often used by agencies including FAO, the whole population of Yemen in the current context could be considered food insecure.[107]

A woman prepares food at an IDP camp
A woman prepares food at an IDP camp in the Abs district of Hajjah governorate on August 18, 2018. //Sana’a Center photo

Beyond having little practical value, such a vague, sweeping definition opens the door for varying interpretations. As a result, different organizations use different techniques in their attempts to measure food insecurity or its various forms or degrees of severity. This leads to different outcomes. It is, therefore, not a given that all those involved in discussions referring to the term “food insecurity” are actually considering the same measurements or sharing similar interpretations of what those measures indicate.[108] If there is no common definition of food security that includes common standards and benchmarks, conversations about the concept become increasingly difficult. Several key informants noted this poor shared understanding of “food insecurity”, and considered it detrimental to the response.[109] Most importantly, it has hindered efforts to analyze and understand underlying factors of food insecurity.

A more precise and useful way of measuring food insecurity is through the use of the Integrated Food Security Phase Classification (IPC), also employed in Yemen, which follows international standards to scientifically determine the severity and extent of acute food insecurity and acute malnutrition. It differentiates among “phases” of food insecurity with the aim of improving analysis used to inform emergency response planning, future policies and programming.[110] As such, its definition is far more specific than that used by the FAO, which was adopted in 1996 for the very different purpose of informing the UN’s overarching sustainable development goals.

Questions of definition become all the more important when the term to be defined has far-reaching consequences and triggers a certain type of response, such as the word famine. While many countries face food security crises, with large numbers of people who go hungry and are unable to obtain enough food, only rarely do the conditions meet the humanitarian community’s formal criteria for a famine. Famine, which is not in itself an easy term to quantify,[111] is often described as defined by the UN’s refugee agency: “A situation in which a substantial proportion of the population of a country or region is unable to access adequate food, resulting in widespread acute malnutrition and loss of life by starvation and disease.”[112] The definition used in the IPC classification differs. It defines famine as, “the absolute inaccessibility of food to an entire population or sub-group of a population, potentially causing death in the short term.” And it lays out specific requirements, all three of which must be met, to measure this:

  • at least 20 percent of households in an area face extreme food shortages with a limited ability to cope;
  • acute malnutrition rates exceed 30 percent; and
  • a crude death rate exceeding two persons a day per 10,000 people.[113]

Yemen has not reached this threshold at any point in the current conflict, and best efforts to analyze the available data have indicated that, although significant food insecurity exists, famine remains an unlikely scenario (see:A Data Case Study: Famine in Yemen). This, however, has not altered fundraising appeals warning Yemen is on the verge of famine, raising the question of what definition is being used. It falls short based on the IPC definition intended for making such calls and requires a generous interpretation of even the more generic UNHCR definition above.

Problem 4: A Lack of Innovation

Despite longstanding challenges to data collection in Yemen — the lack of field presence, the inability to undertake surveys, the widespread lack of independence that is partly due to complicated access — the humanitarian community has until recently largely accepted the situation. For example, a common issue long discussed among humanitarians has been the lack of mortality data in Yemen. This has deeply impacted important analyses such as the actual effect of the conflict on civilian mortality, and it has hindered a proper assessment of the consequences of food insecurity.[114] Yet, Yemen is not the only country where mortality data is problematic, and mortality data has been able to be collected in a representative manner in other locations. In South Sudan, the London School of Hygiene and Tropical Medicine conducted a study that established a representative estimation of excess deaths since the start of the conflict in 2013.[115] In addition, the Geneva-based REACH Resource Centre developed a protocol for measuring mortality in complex conflicts that was trialed in South Sudan. Despite its success there and an initial agreement in 2020 to trial the protocol in Yemen,[116] this hasn’t happened due to what an aid worker described as an unwillingness to try it despite any good alternative options.[117]

Innovation is something that the wider community has been looking at over time. The use of geospatial data, for example, can help locate places of displacement, and even the number of people displaced. It can map watersheds, destruction of infrastructure, mortality through counting graves, and help identify patterns of migration. It also allows for the overlaying of data to see how different types of data correlate and influence each other.[118] In Yemen, into the seventh year of the response, attempts to approach information-gathering in different ways are only slowly developing. For example, there has been some use of global imaging software in the past year to map watersheds, and organizations such as ACAPS, a Geneva-based independent provider of humanitarian data and analysis, as well as REACH and the Sana’a-based security provider Safer Yemen have initiated a conversation around the gaps in data collection with some of them undertaking operational response on the ground. In general, efforts to improve data for the Yemen response are coming from outside of the response itself, which is cause for concern. The lack of internal impetus for change has led to continued reliance on poor, incomplete and unreliable data to inform the response.

A family carries shelter supplies distributed
A family carries shelter supplies distributed June 24, 2021, at an IDP camp in Al-Sha’ab, Aden governorate. //Sana’a Center photo by Ahmed Waqqas

The Consequence of Using Flawed Data: The Wrong Response

Despite its flaws — whether the amount of data collected is too limited to be representative, is of poor quality and lacks contextual analysis, or is biased and lacking transparency — this problematic data is what is used to inform response decisions on funding, modalities and allocation of resources. It and the warped narrative it supports have real consequences on the appropriateness and effectiveness of the response.

The trouble begins with how need is defined so broadly that targeting becomes almost impossible. When, as noted above, more than 80 percent of the population is in “some form” of need, the possibility of proper prioritization within the response dissipates.[119] As a result, resources have been largely spread in a superficial manner without regard to response quality and with little incentive to focus on hard-to-reach and complex areas. As one key informant stated:

“If 80 percent of the population meets your target, you can design any program you want that would meet the criteria within 10 kilometers of your office and secure funding. There is no incentive for anyone to push out much farther than that.”[120]
Humanitarian Issues Analyst

Data is one input into decision-making with regard to response. Others are analysis of the wider context, resources available, response capacity and potential future scenarios. This has been best practice for years; it most recently was confirmed by the development of a global-level joint intersectoral analysis framework. This framework, spelled out in August 2020, advocates for needs assessments to consider the context of a crisis, the main drivers and their impacts, and then, most importantly, how and why these factors are affecting humanitarian conditions overall. It recognizes that people don’t experience the effects of shocks and crises in isolation, but that their context and situation are relevant, and that needs are linked and can exacerbate each other.[121] For example, someone living in a rural area will experience shocks differently than someone living in an urban area and will require a different kind of response.

Yet in Yemen, the simple collection of data, regardless of its quality, is used as the basis for continued fundraising and for decisions made without taking into account the surrounding context. The presentation of such isolated data has in itself become a proxy for analysis and decision-making, with a belief that this somehow provides sufficient answers — ones that are then used to uphold the preferred narrative. For example, the determination of food insecurity and threat of famine continues to lead to attempts to fundraise to increase food distributions. The root causes of the food insecurity, though, are not addressed, resulting in a less impactful response.

It is no secret that the underlying causes of the Yemen crisis lie in pre-existing development issues. Pre-conflict Yemen was already the poorest country in the Middle East with worrying development indicators.[122] The conflict has exacerbated those problems, but the issues are systemic, long term and rooted in the collapse of the economy. As a result, all 73 persons surveyed in the course of the research for this paper posited that humanitarian aid may not be the most appropriate form of assistance because it does not address root causes or offer anything for the long term.

At best, humanitarian aid provides some immediate lifesaving assistance; it has few longer term benefits. Food distributions, for example, will not fix a lack of income that led to food insecurity. Water trucking can be done for decades to come but will not result in the sustainable supply of clean water. Long-term displaced people do not need a steady supply of emergency shelter kits and are better off with the provision of transitional shelter kits. Yet these will not be issued because moving from emergency to transitional kits does not fit into the programming and narrative set for the crisis.[123] There are humanitarian needs in Yemen, and a true emergency response providing immediate lifesaving support is required in situations such as displacements, floods or drought. But a very different, more structured and sustained response is required for longer term impactful change. Even though humanitarian aid is not necessarily the appropriate response for the Yemen crisis, the system is deeply invested in promoting it as the solution because change will impact the bottom line and gains for organizations and individuals.

The investment in this narrative in which humanitarian aid is promoted as the solution for needs in Yemen has deeply constricted the debate surrounding data and closed off the space for innovation. Any broader analysis or solutions that lean toward acknowledging the limitations of humanitarian assistance within the response are strongly discouraged or shut down, according to several humanitarians, because they do not fit into the narrative that has proved so successful for fundraising.[124] As one analyst pointed out, “there is no innovation without failure, and there is no safety to fail within the Yemen response.”[125] A culture of improvement does not exist because it requires acknowledging the failure in recent years of the response to appropriately analyze and respond to needs.

Addressing data transparency and a flawed narrative will come at a personal cost to those who have the courage to open that box. It undermines investments made in past years and decisions taken by some of the most senior humanitarian leadership globally. It would mean admitting that the world and the response have misguidedly focused on a narrow aspect of the problem of Yemen and have thrown an incredible amount of donors’ money at it. Strong leadership will be required of those who choose to change course as immense pushback can be expected from those unwilling to admit mistakes and past failures. Yet without this room to fail and learn from mistakes made, innovation and improvement in the response cannot happen. In the end, avoiding responsibility only curtails the international community’s ability to improve conditions for the people of Yemen.


There is a great need to improve and innovate data collection in Yemen, which necessarily requires a willingness to abandon dubious narratives that surround the Yemen conflict. At a minimum, it is paramount that there is full transparency surrounding the flawed and severely limited data collected and being used, ensuring its value is not overstated. Transparency not only could lead to improving data quality, it also would ensure more accurate interpretation as well as inform and promote mitigation measures to help prevent the misuse of humanitarian aid. Even inadequate data has its uses, but it can only be used properly when there is transparency about its limitations. A shift toward a broader analysis of data also is important, taking into account context, social and environmental factors and main drivers of conflict. In this way, true needs can begin to be understood in Yemen and an appropriate response be developed. Conflict has exacerbated Yemen’s deeply rooted economic problems, but the simple mechanisms of humanitarian aid are not capable of addressing these problems, and at minimum a discussion about potentially better solutions is needed. Toward this end, specific recommendations are as follows.

To Senior Leaders of the Yemen Response:

  • Acknowledge that the current humanitarian response is suboptimal for Yemen, and begin an extensive analysis with the aim of instituting a more appropriate and targeted response. A reformulation should take into consideration root causes of the crisis and help foster a more effective humanitarian response when and where lifesaving assistance is required.
  • Separate data collection and needs analysis from fundraising and agency interests to allow a more accurate representation of needs to inform the response, aid in strategic planning and programming and to restore donor faith in the response. This necessitates a willingness to abandon exaggerated or false narratives and to relay to donors and the public more honest, evidence-based assessments of the situation. Clearly indicate when such assessments have been hampered or impossible.
  • Ensure humanitarian interventions when and where necessary are effective by providing the best quality data possible. To that end:
    • Openly and transparently share the potential bias, limited representivity, and methodology by which data is collected to ensure it can be interpreted correctly. If the only available data is flawed and dated, ensure its limitations are made clear and its usefulness is not overstated;
    • increase the use of alternative and innovative ways of data collection to increase understanding and help provide a clearer picture. Specifically,
      • employ alternative protocols, such as those developed by REACH for measuring mortality in complex conflicts, and
      • ensure innovative efforts being seen outside the response to improve data collection and analysis in Yemen are encouraged within it so quality and timely data specific to operational needs is generated; and
    • undertake as soon as is feasible a representative countrywide assessment to establish an accurate picture of the needs of the Yemeni population.

To All Humanitarian Sector Actors within the Yemen Response:

  • Ensure data and information gathered is as independently sourced as possible by avoiding reliance on information provided by actors with vested interests in geographical allocation, type of response and inclusion of certain segments of society. To this end,
    • do not allow an inability to roll out improved data collection in all parts of the country to hamper doing so where possible. For example, more quality and independent data collection can be done in the south, where security and access are less problematic than in northern areas under Houthi control.
  • Be more open and transparent about indicators and figures used to validate the response, so donors and others are informed of actual and realistic achievements. The amount of aid moved across the country, for example, is not equivalent to the beneficiaries reached. One approach would be to introduce a rating system that reflects data quality, with scales indicating the reliability and independence of data gathered.
  • Take a more nuanced approach when considering and establishing the needs of the Yemeni population. In particular:
    • analyze data collected within context and environment to correctly understand root causes and indicate an appropriate response. Using more economic and anthropological analysis to frame root causes would better inform data interpretation and subsequent response;
    • ensure any reference to the term “in need” accurately reflects the form of need to ensure appropriate response; and
    • involve local communities in needs assessments and program design.
  • Step away from quantity and focus on quality. To do this:
    • ensure aid provided is appropriate for the needs (for example, food aid is not always the correct form of aid to address food insecurity), and is not chosen purely based on ease of distribution or donor preferences;
    • ensure data being used for inclusion criteria is collected independently, not by authorities or actors with vested interests; and
    • ensure resources are spent wisely by focusing on activities that create a substantive, positive and lasting impact on communities in need, rather than continuing those offering wide but superficial coverage.
  • Increase aid workers’ field presence outside of Sana’a to inform a better understanding of needs and the operational environment as well as of the drivers and contributing factors to the needs of affected populations.

To Donors:

  • Invest in funding independent field reviews of aid efficiency. Without a physical presence of donors in Yemen, there is no independent mechanism to verify information from the humanitarian sector. This is necessary to retain the integrity of the response.
  • Ensure that any funding is based on transparent and unbiased data, representing the most accurate reflection of how to appropriately address needs in Yemen.
  • Be exacting in questioning the humanitarian leadership regarding how it frames the humanitarian narrative, and demand specific examples and high quality data substantiation before accepting the veracity of statements regarding the response.
  • Ensure funding modalities are appropriate to the response needed in Yemen. To do this:
    • establish multi-year funding cycles for longer-term responses;
    • move away from humanitarian funding in situations where needs are better addressed by longer-term and development interventions, and reserve humanitarian funding for true emergency response needs.
  • Support higher quality programming that may have a narrower geographical scope but can ensure positive lasting change over temporary, superficial coverage that does not address needs in a sustainable manner.

Up Next in this series of reports examining fundamental issues of concern in the Yemen humanitarian response is To Stay and Deliver: Security, which examines how the security apparatus impedes rather than enables aid delivery.

The Sana’a Center for Strategic Studies is an independent think tank that seeks to foster change through knowledge production with a focus on Yemen and the surrounding region. The Center’s publications and programs, offered in both Arabic and English, cover diplomatic, political, social, economic and security-related developments, aiming to impact policy locally, regionally, and internationally.

This report is part of the Sana’a Center project Monitoring Humanitarian Aid and its Micro and Macroeconomic Effects in Yemen, funded by the Swiss Agency for Development and Cooperation. The project explores the processes and modalities used to deliver aid in Yemen, identifies mechanisms to improve their efficiency and impact, and advocates for increased transparency and efficiency in aid delivery.

The views and information contained in this report are not representative of the Government of Switzerland, which holds no responsibility for the information included in this report. The views participants expressed in this report are their own and are not intended to represent the views of the Sana’a Center.


  1. See, for example, “Yemen facing largest famine the world has seen for decades, warns UN aid chief,” UN News, New York, November 9, 2017,; Hannah Summers, “Yemen on brink of ‘world’s worst famine in 100 years’ if war continues,” The Guardian, October 15, 2018,; Hillary Clarke, “12 million face famine in Yemen if the bombs don’t stop warns WFP,” CNN, October 16, 2018,; “10 million Yemenis ‘one step away from famine’, UN food relief agency calls for ‘unhindered access’ to frontline regions,” UN News, New York, March 26, 2019,; Peter Beaumont, “’Open your eyes’: Yemen on brink of famine again, UN agencies warn,” The Guardian, July 22, 2020,
  2. Christina Potter, “Largest Cholera Outbreak on Record Continues,” Outbreak Observatory, Johns Hopkins Bloomberg School of Public Health, Center for Health Security, New York, January 16, 2020,
  3. “Yemen: Funding trends inside and outside the response plan/appeal,” UNOCHA Financial Tracking Service, accessed January 17, 2021,
  4. Ibid.
  5. Interview with UN agency staff member #5, December 8, 2020.
  6. Interview with INGO staff member #5, November 16, 2020.
  7. Interview with UN agency staff member #5, December 8, 2020.
  8. Twenty-eight out of the 43 key informants expressed this view.
  9. “Global Humanitarian Overview 2021. Part 2: Inter Agency Coordinated Appeals. Yemen HRP,” UNOCHA, New York, accessed January 19, 2021,
  10. Sarah El Deeb and Maggie Michael, “Curfews extended as USAID declares aid suspension in Yemen,” The Associated Press, March 27, 2020,
  11. David Milliband, Jan Egeland, Abby Maxman, Michelle Nunn, Janti Soeripto and Beth deHamel, “NGOs call on USAID to end suspension of humanitarian aid to north Yemen,” open letter via International Rescue Committee, New York, August 20, 2020,
  12. Interviews with UN staff member #1, November 12, 2020; UN agency staff member #1, November 13, 2020, and December 2, 2020; INGO staff member #2, November 13, 2020; INGO staff member #5, November 16, 2020; and INGO staff member #10, December 3, 2020.
  13. Interviews with UN staff member #1, November 12, 2020, and humanitarian analyst #2, December 15, 2020.
  14. Interviews with UN senior staff member #3, November 30, 2020, and UN agency staff member #4, December 7, 2020; on data limitations, see also: “Yemen Analysis Ecosystem,” ACAPS, Amman, April 2019,
  15. “Yemen Advocacy Working Group Strategy,” Internal document, Sana’a, September 2019 (document shared with author during the research period in December 2020); Interviews with UN agency staff member, December 7, 2020; INGO staff members #3 on November 14, 2020, #4 on November 16, 2020, and #7 on November, 2020; INGO humanitarian adviser, November 20, 2020; and donor #3, December 14, 2020.
  16. Interviews with UN staff member #1, November 12, 2020; UN agency staff member #1, November 13, 2020; senior UN staff member #3, November 30, 2020; UN Agency staff member #4, December 7, 2020; UN agency staff member #5, December 8, 2020; INGO staff members #4, #5 and #6 on November 16, 2020; humanitarian analyst #2, December 15, 2020; donor #3, December 14, 2020; and journalist #1, January 14, 2020.
  17. For more information on minimum standards in humanitarian response, see the Sphere Standards:
  18. “The Grand Bargain: A Shared Commitment to Better Serve People in Need,” World Humanitarian Summit, Istanbul, Turkey, May 23, 2016, p. 8,
  19. For background on the four principles – humanity, neutrality, impartiality and independence – see: “OCHA on Message: Humanitarian Principles,” UNOCHA, Geneva, June 2012,
  20. “Accountability to Affected Populations. Using AAP indicators in Multi Sector Needs Assessments (MSNAs): A training module for assessment teams and enumerators,” REACH, Geneva, May 2020, p. 9,
  21. Ibid, p. 4.
  22. Interviews with UN agency staff member #2, November 26, 2020; UN staff member #4, December 9, 2020; and with Yemeni aid worker #s 2, 3, 7, 8, 9, 22 and 23, and community member #1.
  23. “Needs Assessment and Analysis,” UNOCHA, accessed on January 13, 2021,
  24. “Humanitarian Programme Cycle. Needs Assessment: Overview,” Humanitarian Response, accessed January 13, 2021,
  25. For example, multi-cluster assessments have become standardized, and they continue to be improved by lessons learned over time. For more information on the multi-cluster assessment, see:
  26. The HNO is a globally standardized tool. For more detail on its structure and layout, see: “2021 Humanitarian Needs Overview – Annotated Template,” UNOCHA, August 15, 2020,
  27. Ibid.
  28. The HRP is a globally standardised tool. For further explanation of the HRP, see:
  29. Ibid.
  30. “Global Humanitarian Overview 2021,” UNOCHA, December 1, 2020, p. 111,
  31. “Humanitarian Response Plan Yemen,” UNOCHA, Sana’a, March 2021, p. 7,
  32. Ibid., p. 5; “Global Humanitarian Overview 2021,” p. 113.
  33. All informants interviewed during the course of this research indicated their overall lack of trust in figures presented.
  34. Interviews with senior UN expert, November 5, 2020, and UN staff member #5, December 15, 2020.
  35. Interview with UN staff member #5, December 15, 2020.
  36. Ibid.
  37. “Humanitarian Response Plan. January – December 2016,” UNOCHA, Yemen, January 2016, pp. 48-49,
  38. “Humanitarian Response Plan. January – December 2017,” UNOCHA, Yemen, January 2017, p.2,
  39. Ibid, p. 6.
  40. “Yemen Situation Report. Analysis: Global Humanitarian Overview indicates increasing humanitarian needs,” UNOCHA, Yemen, December 11, 2020,
  41. “Extension: Humanitarian Response Plan. June – December 2020,” UNOCHA, Yemen, May 28, 2020, p.3,
  42. “2019 Humanitarian Response Plan End of Year Report,” UNOCHA, Yemen, June 2020, p. 18,
  43. Additional sources: “Extension: Humanitarian Response Plan. June – December 2020,” pp. 6-7 and “2019 Humanitarian Response Plan End of Year Report,” pp. 27, 36.
  44. Daniel Maxwell, Peter Hailey, Lindsay Spainhour Baker and Jeeyon Janet Kim, “Constraints and Complexities of Information and Analysis in Humanitarian Emergencies. Evidence from Yemen,” Feinstein International Center, Tufts University, Boston, May 2019, p. 16,; “Emergency Food Security and Nutrition Assessment (EFSNA), Yemen,” World Food Programme (WFP), UN Food and Agriculture Organization (FAO), United Nations International Children’s Emergency Fund (UNICEF) and the Yemen Food Security and Agriculture Cluster (FSAC), June 2017, pp. 8-9.
  45. The 2017 IPC analysis references data sources from 2016 and 2017 from the EFSNA, WFP’s mVAM surveys (conducted via mobile phones) and market monitoring system, the UN Food and Agriculture Organization and Yemeni government Ministry of Planning and International Cooperation’s FSIS/FSTS market data, UNOCHA, UNICEF/Yemeni Ministry of Public Health SMART survey data, World Health Organization, FSAC, government institutions, recent Task Force on Population Movements reports, rapid assessments from Save the Children, Acted, Oxfam, the Norwegian Refugee Council and others: “IPC Analysis. Republic of Yemen. Summary of Findings Part 1: Acute Food Insecurity Projected Situation Overview,” IPC, March 1, 2017, p. 8,
  46. “2011 Yemen Humanitarian Response Plan,” UNOCHA, Yemen, November 15, 2010, p. 16,
  47. ¨Humanitarian Response Plan, Yemen,” UNOCHA, Yemen, March 16, 2021, p. 8, 58,
  48. Ibid. p. 10.
  49. “2019 Humanitarian Needs Overview: Yemen,” UNOCHA, Sana’a, December 2019, pp. 55-58,
  50. Panos Moumtzis, Kate Halff, Zlatan Milišić and Roberto Mignone, “Operational Peer Review, Response to the Yemen Crisis,” Inter-Agency Standing Committee, January 26, 2016, p. 8.
  51. See, for example, “Humanitarian Response Plan, Yemen (2021),” UNOCHA, March 2021, p. 5,
  52. Interviews with UN agency staff member #2, November 26, 2020; UN staff member #4, December 9, 2020; and with Yemeni aid workers #2 on September 21, 2020, #9 on January 14, 2021, #13 on December 8, 2020, #15 and #16 on January 29, 2021, #23 on January 18, 2021, and #26 on February 11, 2021, and with community organization representatives #1 on January 16, 2021, and #2 on January 27, 2021.
  53. Interviews with senior UN staff member #3, November 30, 2020, and UN agency staff member #4, December 7, 2020, and confirmed by internal UN agency maps shared with the author during the research period that indicated known distribution points and areas where specific points are unknown. The author became aware of similar situations while working in Yemen in 2019.
  54. Ibid.
  55. “2019 Humanitarian Response Plan End of Year Report,” p. 6.
  56. Interviews with UN senior staff member #1, November 13, 2020, and UN agency staff member #5, December 8, 2020.
  57. “2019 Humanitarian Response Plan End of Year Report,” p. 9.
  58. Monthly figures can be accessed through the WFP Emergency Dashboard for Yemen:; the November 2019 figure is reflected in the December 2019 dashboard,
  59. Diversion picked up through the WFP complaints hotline is compiled internally. Public accounts of diversion include: “WFP demands action after uncovering misuse of food relief intended for hungry people in Yemen,” WFP, Rome, December 31, 2019,; Karen McVeigh, “Yemen’s Houthi rebels accused of diverting food aid from hungry,” The Guardian, June 17, 2019,; Ben Parker, “Criminal obstruction of aid in Yemen, says watchdog,” The New Humanitarian, London, September 14, 2020,; and “Deadly Consequences. Obstruction of Aid in Yemen During Covid-19,” Human Rights Watch, New York, September 14, 2020, pp. 31-33,
  60. Interviews with UN staff member #1, November 12, 2020; UN agency staff member #1 and UN senior staff member #1 on November 13, 2020; UN staff member #2, November 27, 2020; UN senior staff member #3, December 2, 2020; UN agency staff #4, December 7, 2020; UN agency staff #5, December 8, 2020; journalist #1, January 14, 2021; INGO staff members #4 and #5, November 16, 2020; INGO humanitarian adviser, November 18, 2020; INGO staff member #7, November 20, 2020; donors #2 on December 8, 2020, and #3 on December 14, 2020; and humanitarian analyst #1, December 4, 2020.
  61. This has been reduced even further due to the COVID-19 pandemic, which prompted the evacuation of the majority of UN international staff, many of whom were still to return in 2021.
  62. By 2019, field hubs were present in Ibb, Sa’ada and Hudaydah. The IOM had a presence in Marib, and field offices were in the process of being established in Mukalla, Al-Mokha and Hajjah. By 2021, most agencies were present in Marib; only OCHA had a national staff presence in Al-Mokha. Presence generally is restricted to one or two international staff per agency, and mobile staff of smaller agencies still are often not permanently tied to hubs.
  63. Moumtzis et al., “Operational Peer Review,” p. 3.
  64. Interview with UN agency staff member #2, November 26, 2020.
  65. Market Watch Reports are produced on a monthly basis by WFP. See, for example,; the issue also was reflected by FEWSNET in “Yemen Food Security Outlook Update,” FEWSNET, December 2019,, and in “Report: Yemen’s Agriculture in Distress,” Conflict and Environment Observatory, West Yorkshire, October 2020,
  66. Examples of such motivations include to buy loyalty from communities or to divert aid resources toward the war effort.
  67. “Deadly Consequences;” author’s experience in Yemen within the humanitarian response, confirmed by other aid workers, including: UN agency staff member #1, November 13, 2020; UN senior staff member #3, November 30, 2020; and UN agency staff member #5, December 8, 2020.
  68. Author’s experience in Yemen, consistently affirmed by humanitarian aid workers interviewed during the course of this research.
  69. The SFHRP falls under the authority of the Houthi education minister, Yahya al-Houthi, brother of top Houthi leader Abdelmalek al-Houthi. See: Farouk Al-Kamali and Moath Rajeh, “Special investigation: How Houthis weaponize humanitarian aid,” Al-Masdar, March 15, 2021,
  70. Author’s experience in Yemen, and raised by multiple key informants during interviews, including UN senior staff member #3, November 30, 2020; UN agency staff member #4, December 7, 2020; INGO staff member #5, November 16, 2020; and donor #2, December 8, 2020.
  71. “For Us but Not Ours. Exclusion from Humanitarian Aid in Yemen,” Danish Refugee Council and the Protection Cluster, November 2020, p. 19; and Farouk Al-Kamali and Moath Rajeh, “Special investigation: How Houthis weaponise humanitarian aid,” Al-Masdar, March 15, 2021,
  72. Author’s experience in Yemen; Lindsay Spainhour Baker, Peter Hailey, Jeeyon Kim and Daniel Maxwell, “The challenges of humanitarian information and analysis: evidence from Yemen,” Humanitarian Practice Network, January 2020,
  73. Internal UN report, February 2019, provided to the author during the research phase in 2020.
  74. “2019 Humanitarian Response Plan, End of Year Report,” p. 28.
  75. Maxwell et al., “Constraints and Complexities,” p. 20, based on Sameer Bhutani, Annie Bliss, Lian Najjar, Andrea Silkoset, and Nicola Toffelmire, “The Data Debacle. Public Health Data in Yemen: Poor Numbers or Poor Politics? Yemen,” 2018, London, London School of Economics and Chatham House.
  76. This refers both to Houthi authorities and those of the internationally recognized Yemeni government.
  77. Maxwell et al., “Constraints and Complexities,” pp. 23-24; Interview with food security analyst #1, November 25, 2020; confirmed in an interview with donor #3, December 14, 2020.
  78. Maxwell et al., “Constraints and Complexities,” p. 20.
  79. “Yemen Analysis Ecosystem,” p. 9.
  80. “Humanitarian Response Plan Yemen. Humanitarian Programme Cycle 2021,” UNOCHA, pp. 8, 139.
  81. “Civilian Impact Monitoring Project Yemen. Key Statistics,” CIMP, accessed January 14, 2021,
  82. “Dashboard, Yemen,” ACLED, custom search, January 14, 2021,
  83. Interview with a CIMP staff member, November 26, 2020.
  84. Interview with an ACLED staff member, November 26, 2020; “ACLED Methodology and Coding Decisions around the Yemen Civil War,” ACLED, 2020, p. 7,
  85. Interview with CIMP staff member, November 26, 2020.
  86. Ibid.
  87. “ACLED Methodology and Coding Decisions around the Yemen Civil War,” p. 1.
  88. Interview with INGO staff member #11, December 14, 2020; “2018 Humanitarian Response Plan End of Year Report,” UNOCHA, Sana’a, August 2019, p. 12,
  89. For example, choices can be influenced by those selecting the key informants; there is a tendency to select more educated key informants who may be more articulate, but are not necessarily representative of the wider community.
  90. Interview with INGO staff member #11, December 14, 2020.
  91. “Adjusting Terminology for Organised Violence in South Sudan,” Conflict Sensitivity Resource Facility South Sudan and WFP, South Sudan, November 2020,
  92. See, for example, WHO monthly cholera updates: and “WHO: Speed of Yemen cholera outbreak ‘unprecedented’,” Al Jazeera, May 19, 2017,; Potter, “Largest Cholera Outbreak.”
  93. Author’s discussion with a humanitarian aid worker, March 2019.
  94. Interview with UN Agency staff member #5, December 8, 2020.
  95. “Vaccine-Preventable Diseases Surveillance Standards: Cholera,” WHO, September 5, 2018, p. 5,
  96. “Joint Cholera Response Plan, Yemen, July 2017,” Health luster, WASH cluster, WHO, June 29, 2017, p. 3,
  97. “Vaccine-Preventable Diseases,” p. 5.
  98. Paul B. Spiegel, Ruwan Ratnayake, Nora Hellman, Daniele S. Lantagne, Mija Ververs, Moise Ngwa and Paul Wise, “Cholera In Yemen: A Case Study of Epidemic Preparedness and Response,” Johns Hopkins Center for Humanitarian Health, Maryland, December 4, 2018, pp. 5, 22-27,
  99. Ibid., p. 23; confirmed by interview with UN agency staff member #5, December 8, 2020.
  100. Spiegel et al., “Cholera In Yemen,” p. 23.
  101. Ibid., pp. 23-24, 26.
  102. Interview with UN Agency staff member #5, December 8, 2020
  103. “Cholera situation in Yemen, January 2019,” WHO, February 2019, and “Cholera situation in Yemen, December 2019,” WHO, January 2020,
  104. Interviews with UN agency staff member #5, December 8, 2020, and INGO humanitarian adviser, November 18, 2020.
  105. Examples of numerous predictions since 2018: Summers, “Yemen on the brink;” “10 million Yemenis;” and “Yemen teeters on the brink as conflict and economic crisis grinds on,” WFP, September 23, 2020,
  106. “Policy Brief: Food Security,” FAO, Issue 2, June 2006,
  107. Ibid.; interviews with food security analysts #1 on November 25, 2020, and #2 on January, 15, 2021.
  108. Interview with food security analyst #2, January 15, 2021.
  109. Interviews with food security analyst #1, November 25, 2020; food security analyst #2, January 15, 2021; UN agency staff member #4, December 7, 2020; INGO staff member #4, November 16, 2020; INGO humanitarian adviser, November 18, 2020.
  110. “IPC Overview and Classification System,” IPC,
  111. “What is a famine?” IRIN News (The New Humanitarian), Johannesburg, May 13, 2010,
  112. “Famine explained: Definition, causes and facts,” UNHCR website, accessed January 19, 2021,
  113. “IPC and Famine: Using the appropriate terminology and scientific thresholds,” IPC Global Support Unit, March 2012,
  114. Maxwell et al., “Constraints and Complexities,” pp. 18-19, 26; Interviews with UN agency staff member #4, December 7, 2020, and INGO staff member #11, December 14, 2020.
  115. Francesco Checchi, Adrienne Testa, Abdihamid Warsame, Le Quach and Rachel Burns, “Estimates of Crisis Attributable Mortality in South Sudan December 2013 – April 2018, London School of Hygiene and Tropical Medicine, September 2018,
  116. “Yemen – Terms of Reference. Secondary Desk Review and Methodology Development for Estimating Mortality in Yemen,” REACH, Amman, October 2020,
  117. Interview with INGO staff member #13, October 30, 2020.
  118. Interviews with INGO staff member #13, October 30, 2020, and INGO staff member #11, December 14, 2020.
  119. “Humanitarian Response Plan. January – December 2019,” UNOCHA, February 2019, p. 5,; Recommendations to drop the use of this terminology were made in the January 2016 peer review of the response, but went unheeded. See, Moumtzis et al., “Operational Peer Review,” p. 8.
  120. Interview with humanitarian analyst #2, December 15, 2020.
  121. “Global Humanitarian Overview 2021. Part 3: Delivering Better. Joint Intersectoral Analysis Framework,” UNOCHA, New York, accessed January 19, 2021,
  122. Yemen ranked 154 out of 187 in the United Nations Development Programme (UNDP) Development Index: “Summary Human Development Report 2011; Sustainability and Equity: A Better Future for All,” UNDP, 2012, p.16,
  123. Interviews with INGO staff member #1, November 5, 2020; INGO staff member #5, November 16, 2020; and INGO staff member #10, December 3, 2020.
  124. Interviews with UN senior staff member #3, November 30, 2020; UN agency staff member #1, November 13, 2020; UN agency staff member #4, December 7, 2020; UN program staff member #6, December 18, 2020; INGO staff members #4 and #5 on November 16 2020; INGO humanitarian adviser, November 18, 2020; donor #3, December 14, 2020; humanitarian data analyst, November 26, 2020, humanitarian analyst #1, December 4, 2020; and humanitarian analyst #2, December 15, 2020.
  125. Interview with humanitarian analyst #2, December 15, 2020.