In brief

Humanitarian aid is expected to reach as many people as possible and to prioritise reaching those most in need. Making sure that the ‘right’ people receive aid was a top priority for the recipients we consulted. While the system is larger than ever, its ability to comprehensively and equitably get support to all those most in need remained compromised. In 2021, under the response plans and appeals where figures on reach were available, the system reached an estimated 46% of the people it identified to be in need and 69% of those it targeted for assistance.170

The COVID-19 pandemic altered the picture of reach, bringing a new scale of coverage while also creating new gaps. But many pre-pandemic challenges also persisted. Geographically remote populations were missed, government-imposed impediments and counter-terrorism measures prevented aid getting to certain areas, and insecurity – including a 54% rise in the number of aid workers attacked – hampered access.

Within the communities that humanitarians did reach, people were highly concerned about the fairness of decisions about who should receive aid. Only 36% of aid recipients surveyed thought that aid went to those who needed it most. Governments and local gatekeepers often sought to influence recipient lists, and a lack of trust in selection criteria was exacerbated by humanitarian agencies’ poor communication. Aid diversion is a high risk in the delivery of aid and, although strong anti-corruption measures were in place, it remained a top concern for communities.

Only 36% of aid recipients surveyed thought that aid went to those who needed it most.

Efforts to ensure equitable reach to the most marginalised community members had a limited focus on LGBTQI171 people, but did result in more attention to the needs of women, older people and people with disabilities. However, the system has little data on how well it is doing against these reinforced frameworks and there were many examples of good intentions not standing up under pressure.


At the heart of the humanitarian endeavour is the principle that aid should be directed according to need – that it should reach the ‘right’ people. Humanitarian action is expected to both prioritise those in the greatest need and respond to as many people in need as possible – in other words, provide both equitable and comprehensive coverage. However, determining and prioritising who is most in need in a given crisis is far from straightforward. It is bound to be an approximate and contested process; as well as being technically difficult, it is also often perceived as deeply political.  

When the system collectively assesses the reach of its aid, it often focuses on top-line numbers of people reached by programmes and proportions of target populations met – rather than whether these were the right people and populations. Not only are these statistics on ‘reach’ often questionable estimates, but they can also distract from important questions picked up in more detailed vulnerability assessments and mapping. Target-driven incentives may privilege the easier to reach majority over the hardest to reach minority. Headline figures also tell us little about whether the ‘right’ aid reached the ‘right’ people. As the last edition of the SOHS noted, ‘the issue of coverage is difficult to assess because, almost by definition, humanitarian actors know much less about the areas where they are not present than about the areas where they are. They know even less about areas that they have not noticed as requiring assistance’.172 As one humanitarian leader put it, the ‘metric of success’ for humanitarian coverage should not be the 60% of the target population who were accessed but how the system responded to the 15% to 20% in the hardest to access places.173

This matters deeply – not only for agencies, donors and their taxpayers, but also for affected people. In focus groups in Yemen, DRC and Venezuela,174 aid recipients said that needs-based aid allocation should be a top priority in how the SOHS assesses the performance of the system. Their concerns centred on how people were selected to receive aid, and the extent to which corruption and aid diversion stopped aid from reaching those for whom it was intended. 

Aid recipients said that needs-based aid allocation should be a top priority in how the SOHS assesses the performance of the system.

Although the system is larger than ever before, with more aid workers spending more money to assist more people in more crises, this is still falling short of the growing scale of identified needs. At the same time, many factors are converging to make it more difficult to ensure comprehensive and equitable coverage for many crisis-affected people, not least restrictions imposed by both national authorities and international governments, including counter-terrorism measures and sanctions regimes.  

This chapter examines the evidence on overall reach and explores the three most salient challenges identified in the research: access to affected populations, inclusion of marginalised groups and aid diversion or corruption.

What is the global picture of reach?

Estimating need and reach

Under the UN-coordinated Humanitarian Needs Overview (HNO), the system estimates how many people are in humanitarian need and how many of these it will target for assistance. Over the course of the year, OCHA also compiles data on the number of people that the response expects it was able to reach.238 This data is not available for all responses, but under those plans and appeals where it is available, the system reached an estimated 106 million people – 46% of those identified to be in need under these responses, and 69% of those they targeted for assistance.175 The OCHA-compiled global figures masked significant differences between countries. In the best case, Afghanistan, where the Taliban takeover brought an increase in need, access and funding levels, the system reported 95% reach of people in need. At the other end of the spectrum, in Burundi, low levels of funding meant that under 16% were reached. This mixed picture of coverage echoes the split in aid workers’ perceptions of their ability to reach people in need – 42% thought that the system did a good or excellent job of this, while 58% thought it poor or fair.176

The system reached an estimated 107 million people in 2021 – just over 70% of the people it targeted and 46% of the total number of people it estimated to be in humanitarian need.

Figure 27: Number of people in need targeted and reached in UN-coordinated appeals, 2021

According to 2021 Humanitarian Response Plans, the largest populations in need were in Ethiopia, Yemen, DRC and Afghanistan. However, there was considerable variation in the proportion of people in need that the system aimed to assist, as well as the number of people it estimated it could reach.

Source: Development Initiatives based on UNOCHA HPC API.

Notes: Expected reached data for El Salvador, Honduras, and Guatemala is not final, and is not shown. Comprehensive data only available for Humanitarian Response Plans. Figure excludes Flash Appeals, Regional Refugee Response Plans, and Other appeals. CAR is Central African Republic; DRC is Democratic Republic of Congo; oPt is occupied Palestinian territory.

These figures are of course only the system’s own estimates of its reach. As Chapter 3 has explored, the flaws behind appeals data are well known. The quality and clarity of the methods for arriving at numbers of people in need, targeted and reached varies between countries and sectors. Agencies each have their own methods and incentives for calculating how many people benefitted from their programmes, and measuring the number of people reached by, for example, cash payments is much easier than estimating the population reached by a water system. These estimates of reach are in any case a crude indicator of success – a study of aid in Syria reported that, in the COVID-19 response, the UN ‘deemed some facilities were reached with aid even when the aid delivered amounted to only a couple of disposable hospital gowns and shoe covers’, citing one donor’s complaint that the system has a very low bar for what counts as coverage and ‘tends to bury the quality of the response in number of tons’.177 

The quality and clarity of the methods for arriving at numbers of people in need, targeted and reached varies between countries and sectors.

Evaluations during the study period raised concerns around a lack of methodological consistency across the inputs which were aggregated into the HNOs,178 with some agencies and sectors making more sophisticated attempts than others to distinguish between the severity of needs.179 There was an ongoing lack of transparency in how clusters and agencies derived their target population from the total population in need, with one evaluation of the response in north-east Nigeria reporting frustration that a cluster lead agency was ‘not clearly voicing the actual needs, nor adequately advocating for appropriate levels of funding’.180 A multi-country evaluation queried how target populations were calculated, detailed and revised, noting that ‘estimates of people in need lack accuracy and they mostly fail to differentiate between different levels of vulnerability. The imbalance between humanitarian demand and supply, donor priorities and efficiency measures all reinforce an approach that prioritizes coverage over equity and quality’.181 

The system has been working to address these concerns and institute improvements. Pressure on the UN-coordinated appeals process to standardise, prioritise and clarify the data behind growing demand has resulted in refinements. Compared to the first generation of HNOs in 2014, and the consolidated appeals documents that preceded them, the latest iterations distinguish the wider population in need from the group that agencies aim to target, present their analysis in a common format and include a more transparent explanation of the methodology. Since 2020, this includes the use of the Joint Intersectoral Analysis Framework, as explored in Chapter 5, which has led to improvements in the HNOs of most of the countries where they have been piloted. 

Drawing up recipient lists  

When aid recipients were asked what they wanted this edition of the SOHS to cover, targeting was one of their top concerns. They had limited trust in the criteria and decision-making, and concerns about both the cultural premises that underpin selection and how the system thinks about both comprehensive and equitable coverage. Only 36% of aid recipients surveyed thought that aid went to those who needed it most.182 

The basic humanitarian practice of selecting individual households remained problematic in many contexts, stigmatising individuals and disrupting the social fabric. In Somalia, for example, programme objectives based on single household units ran counter to communities’ sharing culture, which ‘allows assistance to reach non-beneficiaries, both supporting the community, but also building relationships and acting as a kind of “safety net” for when they might be in need in future’.183 In Turkey, recipients expressed a preference for targeting a wider group of refugees with cash payments, even if it meant that they would receive smaller sums, on the basis that ‘at least this would be equal’.184.An evaluation of a resilience programme in Sudan that delegated beneficiary selection to Village Development Committees found that they had a perception of social equity based on balancing support to different tribal components in the community, which was at odds with the programme’s vulnerability-based criteria.185 

The basic humanitarian practice of selecting individual households remained problematic in many contexts, stigmatising individuals and disrupting the social fabric.

Pressures on the response also resulted in the arbitrary application of selection criteria, resulting in decisions that were even more socially divisive. In DRC, for example, where the sheer size and geographic spread of populations in need outstrip available resources,186 one aid worker explained how ‘when everyone is in the same situation, establishing selection criteria becomes impossible and this puts assessors in a bind – sometimes they’re forced to suspend identification when the number is reached’. One evaluation following Cyclone Idai in Mozambique found that only 36% of the people surveyed thought that aid went to those who needed it most. This was attributed to the extreme community pressure on the local chiefs tasked with targeting distributions in the immediate wake of the disaster, and the lack of awareness of selection criteria.16 

When everyone is in the same situation, establishing selection criteria becomes impossible and this puts assessors in a bind.

Poor engagement with affected communities was repeatedly found to undermine perceptions of fair targeting, fuelling mistrust of humanitarian agencies and the exclusion of people unaware of their entitlements. One system-wide global synthesis of cash programming lessons found widespread confusion about who had been targeted and why. In Syria, poor communication of complex eligibility criteria left space for social media rumours to spread, fuelling widespread suspicion of targeting choices.187 As Chapter 8 explores, there are documented instances of both good188 and poor practice in engaging communities in design and decision-making, but it is widely understood that the system as a whole has a long way to go.  

Transparency and engagement were all the more important in the light of evidence of external influences on selection lists by state authorities and other interest groups. In some cases – particularly around cash and social protection – disagreements between aid agencies and governments arise from genuine tensions between the desire to reach the most people, or the most vulnerable people. Elsewhere it is a more worrying form of control. A global WFP evaluation, for example, found multiple instances of government interference including reviewing humanitarian agencies’ lists and providing alternative lists.189 In Syria, there is evidence of intelligence officials vetting names,190 while aid workers in Yemen interviewed for this study reported a number of influences on who is selected to receive aid – including the terms of access negotiations with local sheiks and local council demands to approve recipient lists. At a local level – including in Somalia, Afghanistan and Mozambique – authorities and community leaders often act as gatekeepers to selection lists, and can introduce both intentional biases and ‘arbitrary prioritisation’,191 necessitating community validation and triangulation.192 One focus group participant in Yemen told our researchers that ‘dealings must take place directly between the organization and the beneficiary without referring to the community leaders who are the cause of our suffering and deprivation of aid’.

How well is aid reaching ‘hard-to reach’ populations?

Remote locations 

A lack of physical access affects the picture of needs as well as the reach of aid. Insecurity, logistics, assertive authorities and COVID-related restrictions on movement all prevented assessments from taking place in many contexts. While overviews of need could be extrapolated from remote methods and secondary or proxy information, there were concerns – for example in Borno State in Nigeria – that these can be misleading, failing to give a good enough picture of the situation in these areas or of the highly vulnerable groups within them.193 During the pandemic, aid workers in Bangladesh described the reliance on remote methods in lieu of participatory needs assessments and voiced concerns of bias towards the better educated and the exclusion of those without a phone. 

Populations in remote locations often find themselves under-served as the humanitarian effort centres on well-worn routes. Agencies still often tended to cluster together near ‘well-tarmacked roads’, creating geographic coverage gaps, as previous editions of the SOHS have noted.194 This can be due to a combination of logistical difficulties and preferential pathways – agencies identifying needs in the places where they are already active and donors funding programmes in places with established response capacity. In Somalia, there were concerns that poor coordination and a tendency to focus on locations close to existing interventions meant limited support to more remote areas.195 In the Cyclone Idai response in Mozambique and Malawi, activities were concentrated in more accessible and established response locations, while some severely affected areas that required the use of airlifts or boats were far less served.196 In DRC, aid workers told us how poor roads and infrastructure make reaching remote populations extremely difficult and expensive, leading agencies to concentrate their resources where they can reach the most people.

Insecure settings 

In highly insecure and politically constrained settings, threats to humanitarian space remained a major barrier to reaching populations. As Chapter 11 explores, there is a widespread sense that humanitarian space is shrinking,197 and that agencies’ risk tolerance or capacity to influence access is not adequately rising to this challenge.

Threats to humanitarian space remained a major barrier to reaching populations, with a 54% rise in aid worker attacks over the study period.

Attacks on aid workers

Attacks on humanitarian workers continued to rise. In the four years between 2017 and 2020, there was a 54% rise in the number of aid workers attacked – in total, over the period, 947 attacks were recorded, with 1,688 aid workers known to be victims.198 Targeted violence against humanitarians increased in several contexts.  These included Syria, where cases more than doubled, in Tigray, Ethiopia, where aid became a deliberate target, and in South Sudan, where despite the formal end to the war in 2018 rising tensions and the departure of peacekeeping forces fed into a spreading threat ‘in an atmosphere of increased lawlessness and opportunistic violent crime’.199 The only notable decline was in Afghanistan, where the establishment of Taliban control meant that large areas of the country were no longer violently contested; for the first time in the 20-year conflict, it was no longer among the five most violent contexts for aid workers. 

Figure 28: Victims of attacks on aid workers, 2015–2020

The number of attacks on aid workers has risen year on year between 2015 and 2020, as did the number of victims. While the number of people killed has declined in recent years, the number of humanitarian workers kidnapped has nearly doubled – and people wounded, more than doubled – since 2017.

Source: Aid Worker Security Database. 

The increase in the number of attacks is not just the result of growth in the humanitarian system. While there are more aid workers in crisis contexts, as Chapter 2 explains, the rate of incidents has clearly risen: in 2020 there were 73 attacks recorded per 100,000 aid workers – a 38% rise on the 2017 rate.200 Over the last decade there has also been a seven-fold increase in aid worker victims of shelling and airstrikes.201 This violence affects national and local staff disproportionately. While the number and rate of direct attacks on international staff fell over the period, they rose for their national and local counterparts as the system relied on them to deliver in highly constrained and violent contexts. This appears to be an ongoing trend predating the mass withdrawal of international staff' at the start of the COVID-19 pandemic. It is also part of a wider phenomenon of risk transfer to local and national NGOs (L/NNGOs), as we discuss in Chapter 9. This has an evident chilling effect on efforts to reach populations – as one staff member of a L/NNGO in Venezuela told us, ‘Our teams’ safety is first since we can’t send them as cannon fodder even when there are people in that community who need what organisations like ours do’. 

While the number and rate of direct attacks on international staff fell over the period, they rose for their national and local counterparts.

Figure 29: National and international victims of attacks on aid workers, 2015–2020 


Figure 30: Victims of attacks on aid workers by country, 2015–2022

The cases of targeted violence against humanitarians increased sharply in Syria, DRC, South Sudan, Ethiopia and Mali – and declined notably in Afghanistan – between the 2018 and the 2022 SOHS study periods.

Source: Aid Worker Security Database.

In many conflict-affected countries, there was a recurrent pattern of populations living under the control of non-state actors being less well-served by humanitarian aid. There are a number of reasons for this, including state-imposed restrictions, donors’ counter-terrorism conditions and the inability to negotiate safe access in active conflict. Before the 2022 Russian invasion, international humanitarian agencies in eastern Ukraine were able to meet some basic needs in areas not under government control, via local partners, but this was extremely limited.202 In Yemen, access to Houthi-controlled areas remained extremely compromised, with organisations required both to obtain an official permit from the government, and to negotiate access with local groups and influential leaders. Syria has seen a marked decline in access to areas outside government control since 2019, with Russia and China using their vetoes at the UN Security Council to close three out of the four original UN-mandated border crossings serving north-western Syria, and uncertainty over the last remaining one.203 

Bureaucratic impediments, sanctions and counter-terrorism measures 

Despite the increase in attacks, many aid workers (28% of SOHS survey respondents, the majority of them international staff) felt that bureaucratic obstacles or political interference were a far larger obstacle to accessing populations in need. This was twice the number of respondents who cited insecurity or attacks on aid workers. Although OCHA deems bureaucratic impediments to be a ‘lower order’ access constraint than logistical barriers, their impact on aid operations has been debilitating, including in Syria204 and Yemen. Bureaucratic tactics designed ‘to make it a headache for you to be there’ include denying or delaying entry visas, procrastinating on memorandums of understanding, holding back travel permits, adding new layers of approval and changing operating requirements. These daily preoccupations for frontline aid workers are starting to rise up the system’s collective agenda. 

Despite the increase in attacks, many aid workers (28% of SOHS survey respondents, the majority of them international staff) felt that bureaucratic obstacles or political interference were a far larger obstacle to accessing populations in need.

The effects of government delays and interference play out not only in the ability to access populations to deliver support, but also in who gets counted as being in need in the first place. Instances were reported of governments or de facto authorities objecting to the results of needs assessments, forcing amendments to the data, and delaying or preventing publication.205 In an IAHE of the 2019 drought response in Ethiopia, interviewees said that the number of people identified as being in need was influenced in both directions, ‘either inflated – to draw more resources to an area – or deflated – to paint a more favourable picture of an area and to uphold a narrative of economic success’.206 In extreme cases, such as government-controlled areas in Syria, analysts and practitioners concluded that independent needs assessments and monitoring are ‘virtually impossible’.207  

The effect of sanctions and counter-terrorism measures on humanitarians’ ability to reach people in need was a growing concern for aid workers. A 2021 survey by VOICE found that 42% of respondents said that these measures affected decisions relating to their programming in the field, by preventing them from carrying out certain humanitarian programmes and activities, or by impeding access to areas where needs are acute.208 In Somalia, counter-terrorism laws mean that people in al-Shabaab-controlled areas – constituting about 70% of the country – have long received only a fraction of what goes to those in government-controlled areas.209 The same is true of Islamic State-controlled areas in Syria and is even observable within specific IDP camps. According to Médecins Sans Frontières (MSF), IS families in al-Hol were excluded from the support afforded to the rest of the camp: ‘There was no health screening. Water provision is terrible. The 12,000 children there have no access to any kind of mental health services, toys, education. They can literally see – across the fence – that the other children have safe spaces and playgrounds … So, it’s active discrimination against that population who have been tagged as terrorists or ISIS people’.210 

The effect of sanctions and counter-terrorism measures on humanitarians’ ability to reach people in need was a growing concern for aid workers.

Counter-terrorism measures also have a wider ‘chilling effect’211 on humanitarian agencies’ appetite for risk: regulations are complicated and unclear and there are high risks associated with violating indistinctly defined ‘indirect support’ clauses, so humanitarians tend to err on the side of caution. In Syria, fears of financial or criminal liability have reportedly led aid workers to avoid some areas of acute need in the north-west of the country.212 In a 2019 report, the UN Secretary-General noted the ‘uncertainty and anxiety among humanitarian organisations and their staff regarding the threat of prosecution or other sanctions for carrying out their work’.213 Agencies also tended to adopt a ‘don’t ask, don’t tell’ attitude, which perpetuated the lack of clarity.214 Others have noted a ‘chain reaction’ of effects of suspending or reducing aid, including a loss of local acceptance, heightening security risks and further impeding access.215 An NGO report on counter-terrorism suggested that agencies had developed strategies to deal with regulations over time, but that the chilling effect remained strong.216  

Humanitarians did however note the significant time and effort invested in securing humanitarian exemptions to sanctions to enable operations in Afghanistan and Yemen, and some suggested that there was new understanding and engagement on the part of some donors, including the Biden-⁠Harris Administration in the US. The agreement of Security Council Resolution 2615 in December 2021 was an important moment for the Afghanistan response: after months of negotiation, it provided a humanitarian exemption to the international sanctions regime. As one advocacy leader put it, ‘It’s a tough nut to crack, but they’re trying to figure it out. So that’s one major area where we’ve seen some forward progress, albeit in small amounts, but I think they’re there. Their hearts and minds are in the right place’.

How well is the system reaching marginalised groups? 

There has been a notable increase in system-wide attention to the sections of society most commonly understood by the humanitarian system to be socially marginalised – women, people with disabilities and older people, as well as emerging awareness of LGBTQI people. In the four years since the last edition of the SOHS identified ongoing gaps in reaching these groups, there has been significant investment in commitments, frameworks and tools to ensure that these people are better ‘seen’ and included in programming, as Chapter 5 details. These efforts are filtering through into more concerted representation and differentiation in needs assessments. There have been multi-country pilots of gender analysis tools,217 Age and Disability Inclusion Needs Assessments among the Rohingya population218and the mobilisation of local civil society disability organisations in Vanuatu to join needs assessment teams in the wake of Tropical Cyclone Harold in April 2020.219 Multiple analytical studies examined the specific – and often disproportionate – impacts of COVID-19 on different social groups.220  

However, concerns remained about the system’s capacity to register the needs of all sections of society. The majority of respondents to our survey of aid practitioners remained less than positive about the system’s consideration of gender, age and disability.221 As Chapter 5 explains, there are fundamental concerns about preconceptions about ‘vulnerable groups’ which, taking women and older people alone, would constitute the significant majority of any affected population. Although there is growing awareness of the importance of a more nuanced and intersectional approach,222 default assumptions about needs profiles persist longer than justified by the initial response phase.223 

When it comes to translating needs assessments into reach, progress appeared to be more limited and inconsistent. Intentions and guidance did not routinely stand up against programming pressures. At the start of the war in Ukraine, wheelchair users reported futile searches for support and assistance with evacuation, only to be told by a prominent agency, ‘Oh, we don’t help people with disabilities’. These are not niche needs. According to one activist, ‘2.7 million Ukrainians have a documented disability; the real number is likely much higher. Fifteen per cent of the world’s population has a disability, and we have had these statistics for decades. These are not niche needs. So why hasn’t the humanitarian field learned to accommodate them? ’224 

Fifteen per cent of the world’s population has a disability, and we have had these statistics for decades. These are not niche needs. So why hasn’t the humanitarian field learned to accommodate them?

In Yemen and DRC, interviewees suggested that ensuring equitable reach to marginalised groups was a secondary consideration after the primary exigencies of the emergency had been met – that, in situations of extreme need or intense conflict, the focus reverted to what to deliver, rather than how to ensure reach and access. Performance was found to be variable even within multi-country initiatives: one programme demonstrated concerted efforts and success in reaching ‘the most vulnerable … and not the easiest to reach’ in the Philippines, but more mixed results in Nigeria and Pakistan, where time pressures were cited as a barrier to ensuring equity.30 Other people we spoke to pointed to prioritisation dilemmas between widespread acute needs and considering specific groups. As one aid worker in DRC put it, ‘There’s a lot of attention to disabilities, which is very important. But … you need to choose, is attention to disability a priority for me? On paper, yes of course. But the reality is that you can barely keep people alive, and so everything else becomes a luxury in certain contexts’. As Chapter 5 details, the lack of tailored design also makes it more difficult for marginalised groups to access support even where it is meant to be available to them.  

Much of the evidence around the system’s ability to reach marginalised groups is still anecdotal. There has been progress in producing gender-disaggregated data, but disability, age, LGBTQI identity, and other demographic data remains limited in programmes and clusters. Monitoring data is not sufficiently granular, making it is hard enough to know whose needs are being met,225 let alone whose aren’t. As one evaluation noted, ‘the vision for equity was also compromised by partial or limited disaggregation of data’, as without it agencies and clusters cannot know who is reached by interventions ‘or understand the extent to which equity targets are addressed and met’.226 Given the stigmatisation that makes identifying as LGBTQI risky in some contexts, it can be challenging to even assess needs under such categorisations. Indeed, over this study period, same-sex relationships have been illegal in several countries experiencing large-scale humanitarian crises.227 Identification challenges aside, there are critiques that humanitarian actors do not adequately consider the existence of LGBTQI needs in their planning.228 Another review noted an absence of clear evidence on the positive impacts or outcomes for people with disability and older people resulting from inclusive humanitarian response, and very little critical assessment of the use and effectiveness of existing inclusive approaches. There was also a notable lack of evidence on the costing of inclusion and of cost–benefit analyses to support agencies in making difficult decisions about who to prioritise and reach with overstretched resources.229  

Is aid being diverted from people in need? 

Aid diversion is a fact of life in the highly compromised settings in which humanitarians operate. Reaching affected populations involves pay-offs, trade-offs and leakages, which can have a real effect on whether the right people are reached with the support they require. Research on corruption indicates how it permeates the programme cycle – from needs assessments and targeting to design costing, contracting decisions to the delivery and monitoring of assistance. This is a reputationally, ethically and practically thorny issue that is rarely covered in evaluations and assessments of the system, but one that aid workers recognise as a live issue. When asked how significant a problem it was in the country where they worked, nearly three-quarters said it was a moderate or high concern.230 For affected communities, diversion was a major concern that clearly affected how they rated the aid endeavour – focus groups told us that it should be a priority issue for the SOHS and over a fifth (22%) of aid recipients responding to our survey said it was the biggest problem.231 As one participant in Yemen explained, ‘The aid provided in Yemen is estimated to be equal to billions of dollars, but it went all in vain and the situation is getting worse. There is a huge corruption here, but we do not know who is responsible for this corruption. We heard that there was a ferry full of food aid, the aid expired, and no one benefited. We do not know who is responsible and everyone is blaming each other’. 

When asked how significant a problem it was in the country where they worked, nearly three-quarters said it was a moderate or high concern.

There are no estimates of the global impacts or costs of aid diversion, but context-specific studies, including the refugee response in Lebanon232 and in the Puntland region of Somalia,233 show how public institutions, humanitarian agencies, local organisations and community representatives can all be implicated in different ways. In Syria, there was evidence of food items being systematically diverted to feed the military, while the government’s imposition of an unfavourable official exchange rate on aid agencies meant that, in 2020, an estimated 50% of funding was reportedly funnelled to the Syrian Central Bank.234 Interviewees in DRC, Venezuela and Yemen all suggested there was widespread and systemic diversion of aid. This involved both state and non-state armed groups directly appropriating aid for electioneering or profiteering and bribing or ‘taxing’ aid agencies in exchange for access or protection. In all three contexts, interviewees also noted instances of corruption during aid distribution, from falsification or favouritism on distribution lists to intercepting cash at the point of distribution and charging recipients for aid. In the Rohingya camps in Bangladesh, the authorities created the ‘majhi system,’ appointing individuals within the camps to act as an interface between the refugees and aid providers. Concerns among humanitarian workers about the lack of representativeness and abuse of power by the majhis were echoed in focus group discussions with refugees, who said that ‘NGO workers didn’t take anything from us in exchange for service they provide but majhis do; when the majhis collect the list of households for NGOs distribution, they collect 20 Bangladeshi taka per household, They do it because they don’t get any salary.’  

Donors and aid agencies have policies and mechanisms to address corruption. Interviews with aid agencies at headquarters and in country suggested that additional checks and balances and investigation mechanisms were in place to prevent or address the most egregious forms of corruption, and that this was contributing to an increase in reported incidents. For some agencies this was part of a wider re-evaluation of organisational risk and compliance approaches, with one NGO noting how this involved addressing issues around organisational culture alongside top-down, zero-tolerance policies, including co‑designing training with local staff and partners to reflect culturally specific perceptions and definitions of corruption. As one INGO leader explained: ‘We’ve tightened up on our own internal investigation measures, we’ve even now got an ex-policeman on our books just to make sure that we’re really tight on the way that we do business’. Corruption is part of the equation when it comes to providing urgent assistance in some of the most constrained and lawless parts of the world, and over the period, agencies became more rigorous and open about this fact (but remained wary of the reputational risks involved).

Box B: Corruption in rapid response programming in DRC 

In DRC, a multi-million dollar corruption scandal exposed the scale and endemic nature of the problem and the inadequacies of existing systems to address it. The corruption came to light in late 2018 after Congolese business owners were caught trying to bribe Mercy Corps staff with bags of cash. An internal investigation found that, while Mercy Corps had been affected for up to a year, similar schemes had likely been in operation for over a decade, affecting multiple NGOs in the UNICEF-administered Rapid Response to Population Movement (RRMP) programme. The organised corruption exploited reduced checks and tracing in the advance financing for rapid response, and involved community leaders reporting exaggerated numbers of affected people to NGOs. Local business owners would then buy ID cards from hundreds of people who had not been affected by the crisis and bribe corrupt aid workers to falsely register them for support. Cash payments for the false beneficiaries would then be collected and shared between the business owners and local leaders.

An investigation published in 2020 by The New Humanitarian estimated that around $6 million may have been lost to multiple aid organisations over two years, with approximately $639,000 lost in the space of a few months in 2018 by Mercy Crops and its partners.235 Analysts suggested that, while the rapid response mechanism was particularly vulnerable to fraud, it was symptomatic of a wider issue in contexts habituated to decades of humanitarian aid – that there were interests who knew how to ‘game the system’.

While Mercy Corps undertook a thorough internal investigation and UNICEF launched an audit of RRMP partner NGOs in DRC, The New Humanitarian suggested that several other potentially implicated organisations had not taken the same measures. At the time of its investigation, there had been no prosecutions in relation to the fraud, and although Mercy Corps terminated the employment of implicated staff, it had not shared their details with other NGOs for fear of violating Congolese labour law. 

In response to the allegations, the UN Humanitarian Coordinator in DRC outlined a set of anti-fraud measures including enhanced triangulation of information, analysis of targeting data by external teams and more rigorous audit, oversight and monitoring by senior managers.236 A joint Anti-Fraud Task Force was established, which commissioned an operational review of corruption exposure following the RRMP scandal. Among its findings were that, despite recent improvements, there was a persistent reluctance to share information on corruption incidents between aid organisations, let alone collectively address them. The review also suggested a link between the relevance of a programme and exposure to the risk of corruption and exploitation – in other words, if a programme is not seen as relevant by the host community, there will be greater motivation for corruption.237 


See footnote 175.


LGBTQI includes people who identify as lesbian, gay, bisexual, transgender, queer or intersex, as defined in Danielle Roth, Alexandra Blackwell, Mark Canavera and Kathryn Falb, Cycles of Displacement: Understanding Violence, Discrimination, and Exclusion of LGBTQI People in Humanitarian Contexts (International Rescue Committee: New York, 2021)


ALNAP, The State of the Humanitarian System 2018, ALNAP Study (London: ALNAP, 2018).



Charles Petrie speaking at OCHA’s 2021 Global Humanitarian Policy Forum, ‘Where to from here? Humanitarian action between rising needs and distant solutions’, 9 December 2021.


In the inception phase of the SOHS, local research teams held a series of focus group discussions across these three countries to understand aid recipients’ priority issues for the SOHS to focus on.


These figures cover HRPs, flash appeals and other UN coordinated appeals. These figures differ from the data used in chapters 3 and 10, which for reasons of data comparability and availability only use data from HRPs. Taking HRPs alone, available data showed that the total people in need in 2021 was 218.3 million. Of these, 143.1 million were targeted and at least 100.7 million were expected to be reached. For reference, in 2020 HRPs covered 207.8 million people in need, 141.2 million targeted and 98.4 million expected to be reached.


 In answer to the question ‘How well do you think your sector (or system as a whole) performed in your setting in reaching all people in need’, 33.4% rated it ‘Good’ and 8.2% ‘Excellent’.


A 2019 inter-agency evaluation of the Ethiopia drought response noted that quality had improved since the introduction of the HNO, but assessments were still falling short of minimum standards (Julia Steets et al., Inter-Agency Humanitarian Evaluation of the Drought Response in Ethiopia 2015-2018 (IASC, 2019), 35).


Featherstone et al. (2019) cites the 2019 Afghanistan appeal as a good example of a multi-layered approach to differentiating levels of severity and distinguishing from chronic needs (Andy Featherstone, Evaluation of the Coverage and Quality of the UNICEF Humanitarian Response in Complex Humanitarian Emergencies (New York: UNICEF, 2019), 51).


Nick Maunder et al., Evaluation of WFP’s Corporate Emergency Response in Northeast Nigeria (2016-2018) (Rome: WFP, 2019), 38.


 In the SOHS survey of affected populations, when asked if they thought aid went to those who needed it most, 36% of respondent answered ‘Yes’, and 45% answered ‘Partially’.


C. Mike Daniels, Georgina Anderson and Badra Yusuf Ali, Evaluation of the 2017 Somalia Humanitarian Cash-Based Response (Oxford/Rome: CALP and WFP, 2018), 33.


WFP, WFP Evidence Summary Cash-Based Transfers Lessons from Evaluations (Rome: WFP, 2021), 5.


FAO, Final Evaluation of the Joint Resilience Project in Kassala (Rome: FAO, 2018), 16.


Global Affairs Canada, Evaluation of International Assistance Programming in the Democratic Republic of Congo 2012–13 to 2018–19 (Global Affairs Canada, 2020), 9.

WFP, WFP Evidence Summary Cash-Based Transfers Lessons from Evaluations (Rome: WFP), 5.


See, e.g., Key Aid Consulting, British Red Cross Final Evaluation: Nepal Earthquake Recovery Programme (British Red Cross, 2018), 2.


Julia Steets, et al., Evaluation of WFP Policies on Humanitarian Principles and Access in Humanitarian Contexts (Rome: WFP, 2018), 63.


Hall– (2022) reports that, ‘In early 2019, SARC employees admitted to denying World Food Programme (WFP) food baskets to beneficiaries. If the intelligence branch wrote the word “security” next to a beneficiary’s name, the family had to visit a security branch before being eligible for aid’ (Rescuing Aid in Syria, 37).


Thirty-six per cent thought that aid reached those who needed it most. Evidence from FGDs and secondary data indicate that this was due to the fact that distributions were mainly managed by local chiefs.


Jock Baker et al., Inter-Agency Humanitarian Evaluation of the Response to Cyclone Idai in Mozambique (New York: OCHA, 2020), 39.; Glyn Taylor, Corinna Kreidler and Paul Harvey, Global Cash Evaluation (Oslo: Norwegian Refugee Council, 2019), 58.;  Daniels, Anderson and Yusuf Ali, 2017 Somalia Response, 16.


ALNAP, The State of the Humanitarian System 2015 (London: ALNAP, 2015), 8.


Daniels, Anderson and Yusuf Ali, 2017 Somalia Response, 14–15.


Of the aid workers who responded to our survey, 45% felt that humanitarian space had declined and 24% that it had stayed the same. Only 30% felt it had improved.


 Aid Worker Security Database (see methodology in annex 3).


Abby Stoddard et al., Aid Worker Security Report 2021: Crime Risks and Responses in Humanitarian Operations (Humanitarian Outcomes, 2021).


Analysis by Humanitarian Outcomes, see methodology in annex 3.


Analysis by Humanitarian Outcomes based on AWSD (Aid Worker Security Database)


Ecorys Polska, Evaluation of Canadian International Assistance Programming in Ukraine 2009-2018 (Global Affairs Canada, 2019), 21.


Steets et al. (2018) refer to OCHA’s access severity constraints methodology (Evaluation of WFP Policies).


Steets et al., Evaluation of WFP Policies, 62–63; SOHS field interviews in Yemen.


Gillian McCarthy, Adding to The Evidence the Impacts of Sanctions and Restrictive Measures on Humanitarian Action (VOICE, 2021).


Michiel Hofman, ‘I’m a humanitarian. Don’t prosecute me for doing my job’, The New Humanitarian, 12 November 2019; O’Leary, Principles Under Pressure.


MSF informant cited in Vittoria Elliott and Ben Parker, ‘Balancing act: Anti-terror efforts and humanitarian principles’, The New Humanitarian, 26 November 2019.


Alice Debarre, Safeguarding Humanitarian Action in Sanctions Regimes (Policy Brief) (New York: International Peace Institute, 2019).


David McKeever, ‘International humanitarian law and counterterrorism: Fundamental values, conflicting obligations’, International and Comparative Law Quarterly 69, no. 1 (January 2020): 43–78.


Lindsay Hamsik and Lissette Almanza, Detrimental Impacts: How Counter-Terror Measures Impede Humanitarian Action (Washington DC: InterAction, 2021).


See Women’s Refugee Commission, Piloting the IASC Guidelines on Inclusion of Persons with Disabilities in Humanitarian Action: Disability Inclusion in Gender-Based Violence Programming in Jordan, Sri Lanka, and Uganda (Women’s Refugee Commission, 2019).


REACH, Age and Disability Inclusion Needs Assessment: Rohingya Refugee Response (REACH, 2021).


 S. Bula, E. Morgan, and T. Thomson, ‘Pacific People with Disability Shaping the Agenda for Inclusive Humanitarian Action’ (London: HPN/ODI, 2020).


See, e.g., Tal Rafaeli and Geraldine Hutchinson, The Secondary Impacts of COVID-19 on Women and Girls in Sub-Saharan Africa (K4D, 2020).;
Kasey Ochiltree and Iulia Andreea Toma, Gender Analysis of The Impact of Recent Humanitarian Crises on Women, Men, Girls, And Boys In Puntland State In Somalia (Oxford: Oxfam, 2021).;
Samuel Hall, The Impact of COVID-19 on Older Persons in Afghanistan (Help Age International, 2020).;
Innovation to Inclusion, Impact of COVID-19 on the Lives of People with Disabilities: Insights and Stories from Bangladesh and Kenya (Leonard Cheshire Disability, 2020).


According to our survey results, 43% thought it ‘Fair’ and 17% ‘Poor’.


See, e.g., Catherine Jones et al., Applying an Inclusive and Equitable Approach to Anticipatory Action (Rome: FAO, 2020).;
Valentina Shafina and Pauline Thivillier, Inclusive Client Responsiveness: Focus on People with Disabilities and Older People (New York: IRC, 2021).


Sophie Van Eetvelt, ‘What does the evidence say? A literature review of the evidence on including people with disabilities and older people in humanitarian response’, Humanitarian Exchange 78, no. 7 (October 2020).


Anna Landre, ‘The disabled Ukrainians doing what the UN can’t (or won’t)’ (Blog), From Poverty to Power, 9 March 2022.


Robina Shaheen, Hannah Wichterich and Deepak Sardiwal, Final Evaluation Report of Norwegian Church Aid’s Emergency Preparedness and Response Programme in South Sudan (Action Against Hunger and Norwegian Church Aid, 2018);
Bizzarri et al., Evaluation on Gender Equality; Nyasha Tirivayi et al., Evaluation of the School Meals Programme in Malawi with Financial Support from United States Department of Agriculture (USDA) 2016 to 2018: Volume I (Rome: WFP, 2019).; Phuong Pham et al., DEPP Evaluation Summative Phase Report (Action Against Hunger, CDAC Network, Start Network, DFID, 2018).;
Florence Tercier Holst-Roness et al., Evaluation of UNHCR Prevention of, and Response to, SGBV in Brazil Focusing on the Population of Concern from Venezuela (2017-2018) (Geneva: UNHCR, 2019).;
Taylor, Kreidler and Harvey, Global Cash Evaluation.;
DFAT, Performance of Australian Aid 2017–2018 (Canberra: Australian Department of Foreign Affairs and Trade, 2019).


UNICEF, Global Evaluation of UNICEF Water, Sanitation and Hygiene Programming in Protracted Crises, 2014–2019 (New York: UNICEF, 2021), 7.


Nicolas Salazar Godoy, Sexual Orientation and Gender Identity in Contexts Affected by Fragility, Conflict, and Violence: Discussion Paper (Washington DC: World Bank Group, 2020).


 Roth et al., Cycles of Displacement.;
and M. Daigle, ‘How Humanitarians should consider LGBT+ issues in their work’ (London: ODI, 2021).


44% said it was ‘Moderate’ and 29% ‘High’.


 Overall, this made it the second largest concern for aid recipients, after ‘Not enough aid’ (34%).


Soha BouChabke and Gloria Haddad, ‘Ineffectiveness, poor coordination, and corruption in humanitarian aid: The Syrian refugee crisis in Lebanon’, International Journal of Voluntary and Nonprofit Organizations 32, no. 4 (2021): 894–909.


Abdinur Abdirisak Sofe, ‘Assessment of corruption in the humanitarian assistance in Puntland State of Somalia’, Journal of Financial Crime 27, no. 1 (2020): 104–18.


Natasha Hall, Karam Shaar and Munqeth Othman Agha, ‘How the Assad Regime systematically diverts tens of millions in aid’, Center for Strategic and International Studies, 20 October 2021.


Philip Kleinfeld, ‘EXCLUSIVE: Congo aid scam triggers sector-wide alarm’, The New Humanitarian, 11 June 2020.


OCHA, ‘Statement by the Humanitarian Coordinator in the Democratic Republic of the Congo, David McLachlan-Karr’, The New Humanitarian , 12 June 2020.


Nicole Henze, François Grünewald, Sharanjeet Parmar, Operational Review of Exposure to Corrupt Practices in Humanitarian Aid Implementation Mechanisms in the DRC (London: DIFID, 2020).


These figures cover those HRPs, flash appeals and other UN-coordinated appeals for which data is available on the number of people targeted and the number of people expected to be reached, as reported on OCHA’s Humanitarian Insights database ( global-humanitarian-overview-2021 – data downloaded August 2022). The data therefore differs from that used in chapters 3 and 10 which refer either to global estimates of need and people targeted for assistance, or – for reasons of data comparability – only those included under the HRPs.