Analysis Main Publications News The Yemen Review Publications Index

Commentary Flawed Data Continues to Plague Humanitarian Response in Yemen

اقرأ المحتوى باللغة العربية

In October 2021, I published a series of reports with the Sana’a Center revealing deep failures in the humanitarian aid system in Yemen. Based on interviews with 73 aid workers, UN staff, analysts and experts, the reports described an inefficient and inappropriate humanitarian response driven by flawed data and beholden to myths. Key issues raised included the control of Houthi authorities over the humanitarian response, and the reliance of aid actors on partial and biased data. Flawed data has been used to support exaggerated narratives describing Yemen as the “worst humanitarian crisis in the world,” and ever “on the brink of famine.” While such narratives have been highly effective for fundraising, they collapse under scrutiny.

The reports focused on aspects of the response that could be improved given sufficient will, with the goal of provoking honest and transparent discussions on how to reform the aid system in Yemen. While there are signs of nascent progress in some areas, such as more open debate on security management, fundamentally the response remains invested in false narratives. Worryingly, challenges to collecting independent, reliable, unbiased data appear to have worsened. This can be seen in the collection of data for the 2022 Integrated Food Phase Classification exercise (IPC), a system for classifying food insecurity, the results of which were published on March 14.

Questions of accuracy, timeliness and bias surrounded the IPC process in Yemen in 2021, and the situation appears to have deteriorated further. “If you think we had challenges collecting the data last year, this year the challenges have tripled,” an expert involved in the process told me. Fundamental issues with the data arose during analysis, forcing a quality review by an independent Famine Review Committee (FRC).

Conversations with several people involved with the food security and livelihoods assessment for the IPC process revealed serious concerns about interference, a lack of transparency and flawed analysis. In areas under the control of the armed Houthi movement (Ansar Allah), the World Food Programme (WFP) again outsourced data collection to a third-party contractor. Houthi authorities accompanied the contractor’s data enumerators on most field visits but did not grant permission for any INGOs or UN agencies to join them. According to one person involved in the broader data collection process, Houthi officials visited a training for data enumerators and instructed them: “It is your responsibility to collect the information. The information affects everyone so you have to show the truth. The truth is that the situation is really bad, and it is the fault of the blockade and the war.” Such interference raises concerns about bias and the neutrality of the results.

The IPC process for years has suffered from a lack of transparency. Houthi authorities routinely do not allow raw data collected in areas they control to be taken outside Yemen, therefore in previous years IPC staff outside the country could only view outcome data. In 2022, Houthi authorities also prevented staff inside Yemen from viewing the raw food security data, including the IPC technical working group in Yemen. The raw data is held on a password-protected server accessible only by the Houthis’ Supreme Council for the Management and Coordination of Humanitarian Aid (SCMCHA) and WFP, which refused to share the data.

Such restrictions on access to data inhibit quality analysis and prevent analysts from being able to judge the quality of the raw data. However, while raw data on food security was inaccessible, raw data on nutrition and livelihoods collected during the process was shared with UNICEF and the Food and Agriculture Organization (FAO). An analyst involved in the process said the IPC’s nutrition and livelihoods data sets were such poor quality that most had to be discarded; this does not bode well for the quality of the food security raw data that was not being shared.

Analysts involved in the IPC process say manipulation and coaching of survey respondents, which occurred in previous years, escalated during the 2022 data collection process. “[Houthi authorities] have learned, because we taught them on the indicators. We taught them what to collect and how it is asked, and they coached the people collecting on what the answers should be,” one source explained. Even with limited access to the data, there are indications of manipulation. For example, in Hudaydah governorate, the IPC survey found that around 300,000 people were newly facing emergency levels of food insecurity. This is a shocking deterioration given extensive support from WFP in Hudaydah, and it may point to manipulation of answers. A clear example of manipulation was raised by data collectors in Sana’a city, who said local authorities instructed them not to record any annual incomes above 50,000 Yemeni rials (about US$85). Following these reports, the income data was considered so skewed it was eliminated from the analysis.

This year, initial presentations of data during the analysis process indicated that almost all of Houthi-controlled Yemen should be classified as IPC 4, facing emergency levels of food security, with the number of people on the brink of famine tripling in one year. Some areas were projected to be in full-fledged famine, including Abs district in Hajjah governorate, Dhamar city in Dhamar governorate and parts of Amran governorate. These results did not line up with realities on the ground, prompting some members of the technical working group to propose adjustments that would increase food security levels in some districts and lower them in others, to reduce clear differences and obscure obvious quality problems. It was this attempt at artificial juggling as well as dubious projected outcomes that a source involved in the process said eventually led members of the IPC technical working group to request the FRC review the IPC results.

IPC data classifying Abs in fully-fledged famine presented a particular conundrum. Almost everyone in Abs receives food aid, while internally displaced people in Abs have received consistent humanitarian aid since 2018. There are a number of humanitarian actors present in Abs, including major INGOs. Absent serious failures in aid delivery and monitoring mechanisms, the finding of famine in Abs indicates the data collection is fundamentally flawed. Notably, organizations in Abs expressed surprise when presented with the potential findings of famine.

Ultimately, the FRC found no evidence of famine in Abs, or in Yemen as a whole, and the brief, published IPC key results ultimately reflected that. Instead, a source familiar with the FRC’s review and confirmed by the official findings of the FRC, said the committee found serious gaps and discrepancies in the data — and a need for further investigation to understand why areas that receive consistent humanitarian aid are assessed to be at emergency levels of food insecurity. This indicates problems with aid delivery or data collection; it is likely both.

Clearly, data collection in Yemen continues to be highly compromised and requires a fundamental change in approach. The current data coming out of Yemen is not credible, and as a result, the humanitarian response lacks credibility. Privately, two major donors have expressed frustration and a total lack of trust in the largest agencies working in Yemen.

This week, a pledging conference will take place in Geneva to raise funds for the humanitarian response in Yemen. Many questions remain about the credibility of the UN data on humanitarian needs in Yemen. Donors have a choice: They can continue to ignore fundamental flaws in the response and to accept narratives built on biased and manipulated data, or they can choose to listen to humanitarian actors asking for change. Donors must demand that the humanitarian response is able to operate independently without interference, and they must insist on independent, credible data collection to form the basis of an appropriate response. Funding should be contingent on it.