Every month, nearly one million people flee their homes because of conflicts or natural disasters. With few wars ending, and new wars starting, the number of people displaced by conflict now exceeds 50 million. Not since World War II have people sought refuge—in their own countries or in neighboring states—on such a scale. The disorder driving mass displacement is unlikely to be transitory. In a growing number of countries, the glue of national identity and state authority is unable to patch ethnic, sectarian, or tribal divisions, all of which are exacerbated by regional rivalries. Faced with such threats, multilateral institutions may be strong enough to prevent interstate war but too weak or divided to stop the fighting, as is the case in Iraq, South Sudan, Syria, and Ukraine. Further adding to the tide of humanitarian misery are climate change and demographic pressures.

The humanitarian sector sustains and improves life for people caught in these crises. Its staff are heroic, skillful, and inspiring. Yet the sector is struggling to cope with new realities, and there is a growing gulf between the needs of people affected by crises and the help they are receiving. One measure is humanitarian aid flows: the amounts pledged to address the consequences of crises now regularly fail to reach 40 percent of the UN’s targets. It is logical, then, to argue for increases in humanitarian aid funding. The total global budget for humanitarian aid stands at just $22 billion. And at 0.3 percent of GDP, the $135 billion spent on all aid last year—on poverty reduction as well as humanitarian crises—by members of the Organization for Economic Cooperation and Development (OECD) fell well short of the UN’s target of 0.7 percent. But with Europe and the United States still emerging from economic crisis, overall aid is unlikely to grow much in the short term.

More resources would help. But resources also need to be used for greater impact. Environmentalists like to speak of “factor 4” improvements, which cut resource use in half while doubling productivity so as to quadruple overall efficiency. The concept provides a useful benchmark for the humanitarian sector, too. Over the next decade, donors need to not just double the amount of aid directed to the places of greatest need but also undertake reforms that seek to double the productivity of aid spending. Doing that would require significant shifts in practices and assumptions. Given the growing scale of the problem, however, anything less risks being marginal.


Aid is given a variety of labels. Money used for saving lives and alleviating suffering in war zones and after natural disasters is categorized as humanitarian relief, whereas funds that support economic growth and long-term improvements in quality of life in poor countries are classified as development assistance. Donors also distinguish between aid for low-income countries and that for middle-income countries.

Those categories are increasingly unhelpful. For example, as António Guterres, the UN high commissioner for refugees, pointed out in February, it is “absurd” that Jordan and Lebanon, despite dealing with millions of refugees from Syria’s civil war, are ineligible for World Bank support because they are considered middle-income countries. Similarly, when wars last for decades, as is the case in Afghanistan, the Democratic Republic of the Congo, and Somalia, it is strange to separate supposedly short-term humanitarian relief from long-term development aid and senseless to have so little alignment among the objectives, processes, conceptual frames, and institutions for the two types of aid. The short-term timeline that often characterizes humanitarian projects, for example, creates disincentives for the type of planning, implementation, and measurement necessary to offer lasting help to people. Yet donors continue to funnel money through different institutions to many of the same places: the $5.5 billion in humanitarian relief they spent on the top 20 crises in 2013 sits alongside, and is overshadowed by, the $28.6 billion they spent in the same countries on development assistance. Add in the false separation of services—between health, education, and women’s protection, for example—and the result is frustration at the failure to meet people’s needs.

A better prism for considering how to focus spending is fragility—defined not just by how much a state lacks the capacity, will, or legitimacy to provide basic services or enforce the rule of law but also by the extent to which that country is exposed to violence, poverty, economic instability, and environmental shocks. The concept of fragility covers both low- and middle-income states, and it cuts across national borders, since relatively stable countries can contain pockets of instability. Statistics show that fragile places are falling behind their stable counterparts. When it comes to the UN’s Millennium Development Goals (MDGs), four-fifths of the states classified as fragile by the OECD are not on track to achieve universal primary schooling by the end of this year, and two-thirds will fail to cut poverty in half. The proportion of poor people living in fragile states is on the rise; the OECD estimates that even in a best-case scenario, more than 62 percent of the world’s extreme poor (defined as those living on less than $1.25 a day) will reside in fragile states by 2030, up from 43 percent today. With just 38 percent of aid spent in fragile states today, the growing concentration of poverty, along with the current mismatch between need and provision, makes the case for doubling aid spending in these places.

A medical officer prepares to vaccinate an infant at a health clinic run by the medical charity Medecins Sans Frontieres (MSF) Holland at Gassire, a camp for displaced Chadians, June 2008.
Finbarr O’Reilly / REUTERS

The changing geography of poverty in recent decades makes it possible to rebalance funding from stable states to fragile places. The majority of the global poor reside in stable middle-income countries, with a third in India alone. But in India, China, the rest of East Asia, and the more stable parts of Africa, aid is not the primary tool for tackling poverty; economic growth, domestic revenues, foreign investment, and remittances are. Rather than larger aid budgets, eradicating poverty there requires sharing the proceeds of growth more equally. In stable low-income countries, aid will continue to play an important role, but the focus should go beyond aid to policy changes. Expanding foreign direct investment, removing international agricultural trade subsidies and tariffs, and facilitating migration and remittances would do far more to eradicate poverty than development assistance.

A new, competing pressure on aid budgets is the challenge of financing action to tackle climate change. The 2009 Copenhagen accord set out the expectation that by 2020, developed countries would give developing countries $100 billion a year for climate change mitigation and adaptation. The aid budget is an obvious short-term resource. But it is not a sustainable long-term solution. That would require, for example, a global emissions-trading system that built on the growing number of regional carbon markets. By allowing polluters to pay for emissions reductions wherever they were cheapest, such a system would reduce the costs of mitigation. What’s more, a dedicated source of finance would be created that did not rely on the discretion of policymakers in aid ministries.


To improve the impact of aid by anything like a factor of four, a doubling of the funding in fragile places needs to be matched by a doubling of productivity. Every dollar of aid needs to deliver greater impact, reaching more people and achieving more profound and longer-lasting change in their lives. Donors can drive this shift by creating incentives for generating and applying evidence of what works.

In the past decade or so, aid agencies and donors have partnered with economists and political scientists to analyze various development programs with a new level of rigor. They have embraced randomized impact evaluations (akin to the randomized controlled trials used in health care), which randomly assign communities, schools, and so on to either a control group or a treatment group so as to estimate the net effect of a particular intervention. In the cold light of scientific evidence, some popular development ideas have been shown to have limited impact. Take microfinance, once seen as the game-changing solution to extreme poverty. A body of evidence now shows that although providing small loans increases people’s consumption of basic goods and services in the short term, it does not achieve its transformative purpose of significantly lifting their incomes.

A database compiled by the International Initiative for Impact Evaluation counts more than 2,500 studies of various approaches to tackling poverty in relatively stable low- or middle-income countries. Thanks to such evidence, donors have been able to focus their investments on services that they are confident can make a difference—for example, vaccinations, HIV/AIDS treatment, and efforts to lower the financial and geographic barriers to school attendance.

Every dollar of aid needs to deliver greater impact.

A particularly promising area of emerging research relates to behavioral change. The decision to send one’s children to school, save money to start a business, or desist from violence is rarely made by rationally assessing the costs and benefits. People may weigh losses more heavily than gains, value rewards today far more than rewards tomorrow, and base their decisions on their perceptions of what others do. One insight to come from behavioral economics is that one need not change attitudes to change behavior. A study conducted by the psychologist Elizabeth Levy Paluck in Rwanda found that getting people to listen to a reconciliation-themed radio soap opera made them more likely to cooperate and dissent openly in a group conversation and less likely to forbid their children from intermarrying—even as their personal beliefs remained unaltered. The implication is that rather than try to upend deeply held attitudes, programs should focus specifically on behavior.

Where this research falls short, however, is in scope. Put bluntly, it has missed out on fragile places, in many cases due to the perception of donors and academics that conducting research there is less predictable and more expensive than it is in more stable contexts. Fewer than 100 impact evaluations have been conducted in crisis-affected places. In these zones, the humanitarian imperative to act has often overshadowed the need to measure, learn, and improve.

Many of the lessons from more stable contexts may seem transferable. Vaccinations, clean water, and good hygiene will help tackle the spread of infectious diseases whatever the context. Children will always learn better when they enjoy nutritious food and a caregiver who can provide warmth and stimulation. But in fragile contexts, problems impede the delivery of interventions. And even though the humanitarian sector may know that a particular intervention works, it knows much less about how to nurture the institutions, systems, and staff that can actually implement it in challenging environments. In Pakistan, for example, annual floods devastate whole communities, which need cash quickly to replenish basic goods. Despite the predictability of the crisis, it still takes six to eight weeks for money to get from a donor, via UN agencies, nongovernmental organizations (NGOs), and local partners, to an actual recipient. And in the first year after a flood, each dollar can cost up to 46 cents to transfer.

A similar problem affects vaccine delivery. Over the last two decades, improvements in vaccine technology, refrigerated supply chains, and global financing have dramatically raised immunization rates. Yet in developing countries, rates have plateaued at 80 percent, causing more than 1.5 million children to die each year from preventable diseases. Part of the problem is behavioral: health workers not showing up at clinics or families failing to make the long journey there, because of either the cost or the distance. In fragile places, however, there are also security threats. Given the burden of disease in such areas, there needs to be greater investment in finding ways to improve coverage and the use of essential medical services.

Evidence-based solutions create a major opportunity for donors and agencies to make aid more productive. Yet in many instances, their practices remain out of step with the data. Take cash transfers. As a substantial body of evidence now shows, money not only helps people buy essential items; it also helps families raise their income, so they can send their children to school rather than work. Yet between 2009 and 2013, only an estimated 1.5 to 3.5 percent of humanitarian aid went to cash.

Donors should adopt a simple principle: fund only those programs that are based on the best available evidence or, in cases where impact evaluations (or their equivalent) have yet to be conducted, that are supporting the generation of evidence. In the same way that donors expect programs to comply with financial and legal rules, they should ensure that their own programs and those they fund are justified by high-quality, relevant data. The prize is immense: closing the evidence gap in the humanitarian sector’s highest-priority areas. This will require investment. Randomized impact evaluations and similar rigorous studies can cost several hundred thousand dollars each, and it would take up to $1 billion over ten years to conduct the same number of impact evaluations in crisis places as have been conducted in stable low-income places. But without that investment, donors will not be able to deliver the innovations needed to match rising demand or properly direct their spending toward the highest impact.

  A displaced Iraqi Sunni man fleeing from Islamic State militants in Anbar Province sits near aid provided by the United Nations Refugee Agency (UNHCR) in Baghdad, February 2015. 
Thaier Al-Sudani / REUTERS

The focus on evidence has its critics. The most powerful argument is not that emergencies preclude evidence gathering; it is that evaluations are of limited use because the findings from one setting do not necessarily apply to another. Yet this is an argument not for ignoring evidence but for generating more of it across a range of contexts and then ensuring that each program is carefully tailored to the local situation. The Ebola outbreak in West Africa and the violence perpetrated by the self-declared Islamic State in the Middle East have highlighted the increasingly complex and unpredictable nature of humanitarian crises. In order to be effective in such environments, programs must be customized to the local context. But they also need to draw on good evidence.


Better evidence is not only the foundation for making programs more effective; it can also help make them more efficient. The starting point for that is transparency. The websites of many NGOs and UN institutions, including our own, the International Rescue Committee, tell the public what they can “get” for a donation of $20, $50, or $100. Yet it is not always easy to make sense of those statistics, since there are remarkably few published benchmarks for the actual cost of delivering water, sanitation, food, and so on. In the International Rescue Committee’s own recent analysis, we found large variations in costs, not just between programs—it stands to reason that vaccinating a child may cost $10, whereas providing time-intensive counseling to a rape survivor could cost several hundred dollars—but also within them. Unless donors begin requiring aid organizations to report the cost of delivering services, those groups will lack an important incentive to improve efficiency and share best practices.

The good news is that costs can be reduced without sacrificing quality. Aid travels on a convoluted journey to recipients. Taxpayers fund aid ministries, which give grants or contracts, sometimes via intermediaries called “fund managers,” to NGOs, which often subcontract the work out to local groups. Or aid ministries hand the money to UN agencies, which spend it through a combination of implementing projects on their own and giving grants to NGOs. The attendant accountability systems are multiple, overlapping, and divergent, all with their own costs. Cutting out some of the layers of bureaucracy could radically reduce transaction costs. And greater transparency about costs would force each link in the aid-delivery chain to justify its role.

The humanitarian sector could direct funding toward outcomes rather than outputs.

NGOs face pressure from public and private donors to keep their overhead costs down and deliver the maximum amount of resources possible to programs. These organizations have made important commitments to employ local, rather than expatriate, staff, which lowers costs. But the fragmentation of the sector overall—among both implementing agencies and donors—raises costs. A greater degree of consolidation could reap significant savings, as would joining forces for procurement. The same goes for setting up shared service centers in such areas as finance, security, human resources, and transportation, as well as for such tasks as developing curricula and treatment protocols.

Cheaper is not always better, of course, which is why donors should combine data on costs with measures of effectiveness. That way, they won’t just do the same things more efficiently; they will be able to choose whatever intervention extracts the greatest impact from each dollar. The potential dividends are huge. In Kenya, for example, there is good evidence that a number of programs can improve children’s literacy, but they vary widely in their cost-effectiveness. According to a review by the economist Patrick McEwan, $8,900 could upgrade the reading skills of 100 students by 20 percent if the money were spent on computer-assisted literacy instruction, but that money could get the same result for 423 students if it were spent on performance incentives for teachers and for 695 students if it were spent on remedial tutoring. Despite the value of such information, however, studies on cost-effectiveness—even among those researchers conducting randomized impact evaluations—remain exceptionally rare.

Such studies can help determine the most productive ways of improving education or health, but they won’t help donors decide how to allocate funding between competing priorities. It is not easy to compare vastly different outcomes. What is the value of therapy to a victim of gender-based violence, and how does it compare with the value of teaching a child to read, raising a family’s income, or helping someone survive famine? How much should preventive measures, such as climate change adaptation, be discounted, given that the benefits will accrue only many years from now? Yet these are exactly the questions faced by policymakers and donors when allocating billions of dollars and by NGOs when they decide whether to bid on various projects. So even though it may seem jarring to place a monetary value on life, conducting cost-benefit analyses leads to a more rational debate about where to allocate aid.


Over the past three decades, policymakers in Europe and the United States have recognized the power of incentives when it comes to improving government programs. In the United Kingdom, public-service reform over the last 25 years has come in three parts: top-down incentives, including national minimum standards to ensure that those in disadvantaged neighborhoods or schools were not left behind; bottom-up pressure, through mechanisms to allow recipients greater choice and voice; and horizontal pressure, as monopolies were broken up and barriers for private companies and charities removed.

Humanitarian relief would benefit from a similar focus on incentives. First, the goals that shape donors’ behavior need to be clearer, which would mean specific global aspirations for citizens in fragile states. When the MDGs were established in 2000, the individual targets for each goal were set at a global level. Those targets then became the de facto standard for every country, whatever its starting point. As a result, the incentive was to focus on those countries that were closer to the threshold and those with the largest concentrations of poverty. So even as overall living standards have increased since the goals were introduced, the MDGs reinforced the gap between those living in stable states and those living in fragile places.

The 17 Sustainable Development Goals, which will replace the MDGs at the end of this year, risk perpetuating this dynamic. Despite the excessive number of individual targets, 169 in all, the specific needs of civilians living in conflict zones are all but ignored. That is true even for the proposed goal dedicated to peace and even for the proposed goal devoted to equality for women and girls. It would be best if the key Sustainable Development Goals were translated into national targets tailored to each country’s starting point and potential. To ensure that fragile states get the attention they deserve, the goals—for example, on education, health, and the empowerment of women and girls—should be accompanied by floor targets for the most fragile states, creating a commitment to raise levels of a given metric to a minimum standard.

Second, the humanitarian sector could direct funding toward outcomes rather than outputs. Ideally, government donors would take an independent, overarching, and long-term view of countries’ needs and award taxpayers’ money to the agencies best able to meet those needs. But the reality is more confused, with changing priorities, budgetary pressures, and a lack of coordination among and within governments. Meanwhile, NGOs are incentivized to spend less on management, including on measuring programs’ quality.

Donors should make NGOs and private contractors demonstrate measurable improvements in clients’ lives.

Donors should shift away from holding NGOs and private contractors accountable for spending money on specified outputs and instead make them demonstrate measurable improvements in clients’ lives. Rather than paying organizations to build a certain number of schools and train a certain number of teachers, for example, donors should award grants to the programs that can deliver the biggest improvements in functional literacy and numeracy for the most people. Linking funding to results would encourage or­ganizations to focus as much on high-quality implementation and continuous improvement as on well-drafted proposals. Even in emergency settings, focusing on outcomes can help. In fact, doing so is all the more important in acute, life-threatening situations, where the imperative to save lives comes first.

At the same time, donors should be less prescriptive about how outcomes are achieved. All proposals ought to draw on relevant evidence or generate rigorous learning. But organizations should be encouraged to assess which intervention would work best in a given context and adapt accordingly. Instead of aid destined for the same country being split into humanitarian and development channels, there needs to be far greater coordination among aid providers, building on experiments with pooled funding. And of course, if different donors aligned their efforts, the potential gains would be all the greater.

A third reform involves bottom-up pressure: programs need to respond less to what donors want and more to what recipients need. The most powerful way to do this is, again, through funding. Giving cash directly to the poor allows them to choose what to spend it on. In that scenario, organizations providing goods and services would be driven to attract and serve clients. And in places where populations have fewer organizations to choose from, the organizations that are present should collect real-time data on the satisfaction levels of those they are attempting to help and make the data public. Doing so would introduce pressure to adapt to the perspectives of clients.

There is a delicate balance between cooperation and competition in the humanitarian sector. Organizations work closely together on the ground, often in coordination with the UN. Yet for incentives to work, aid providers need to know that they can be replaced. Indeed, the fact that the aid sector is fragmented means that it is also competitive. But as the main donor agencies face pressure to reduce their administrative costs, they are making fewer, larger grants and giving them to consortiums that cover wider geographic areas. This funnels money out at a lower cost, but it limits competition, because the main players in a given area are encouraged to join forces.

Donors need to encourage organizations to collaborate on areas of efficiency (such as procurement and other back-office functions) yet compete on quality. This will involve giving individual organizations, particularly local NGOs, enough resources to compete. It will involve leveling the playing field—not assuming that the private sector is more efficient or that local groups are more sustainable but making choices based on cost-effectiveness. And it will involve subjecting organizations to competition not just from other organizations providing the same products and services but also from ones that offer substitutes, such as cash transfers. Some of the greatest economic leaps have been spurred not by competition between companies but by competition between technologies. The same logic should apply to providing aid in fragile places.


For over a century, the humanitarian sector has been a symbol of compassion, ingenuity, and heroism. It remains so in many ways. The convoys of UN officials shepherding civilians out of Homs, Syria, under sniper fire, the aid workers suiting up to fight Ebola, the women’s groups organizing against violence in the Democratic Republic of the Congo—they remain sources of inspiration in the midst of indifference. The principles the humanitarian sector lives by—not least, independence and impartiality—are extraordinarily powerful.

But the sector is also increasingly a place of missed targets: appeals that are not met, pledges that are not delivered, ideals that are not translated into action. That need not be the case. The sector requires more funding, but it also has to embrace new ways of doing business: more joined together, more evidence-based, more outcome-focused, more hardheaded.

This agenda is not a substitute for political action to prevent and stop wars. After all, the humanitarian sector stanches the dying, but it takes states and politics to stop the killing. In the midst of multiple global crises and pressure across the world for governments to focus on the home front, however, the prospects for renewed political will are slim. That forces the humanitarian sector onto the frontlines. The best armor will be best practices.

You are reading a free article.

Subscribe to Foreign Affairs to get unlimited access.

  • Paywall-free reading of new articles and a century of archives
  • Unlock access to iOS/Android apps to save editions for offline reading
  • Six issues a year in print, online, and audio editions
Subscribe Now