With rigorous economic research and practical policy solutions, we focus on the issues and institutions that are critical to global development. Explore our core themes and topics to learn more about our work.
In timely and incisive analysis, our experts parse the latest development news and devise practical solutions to new and emerging challenges. Our events convene the top thinkers and doers in global development.
Cash on Delivery is an approach to foreign aid that focuses on results, encourages innovation, and strengthens government accountability to citizens rather than donors. Under COD Aid, donors would pay for measurable and verifiable progress on specific outcomes, such as $100 dollars for every child above baseline expectations who completes primary school and takes a test. CGD is working with technical experts and potential donors and partner countries to design COD Aid pilots and research programs.
Cash on Delivery Aid is designed to overcome the problems of traditional aid, which often focuses more on disbursements and verifying expenditures than on results, undermines a government’s accountability to its citizens, and undervalues local experimentation and learning. COD Aid’s advantage is in linking payments directly to a single specific outcome, allowing the recipient to reach the outcome however it sees fit, and assuring that progress is transparent and visible to the recipient’s own citizens. These features rebalance accountability, reduce transaction costs, and encourage innovation.
COD Aid can be applied to any sector in which donors and recipients can agree upon measurable, verifiable outcomes and commit to making progress toward those shared goals. The approach is fully explained in Cash on Delivery: A New Approach to Foreign Aid (CGD, 2010). Listen to more about COD Aid in these Wonkcasts. Explore the links to the right for more information on specific sectors and countries.
Get Cash on Delivery Updates
CGD experts offer timely research, analysis, and policy ideas for the world’s emerging development challenges. Sign up to get the latest updates from CGD!
It’s quite the buzz phrase: results-based development. But what is actually meant by "results"? Here at CGD, we think of achieved, verified development outcomes, not outputs, i.e. children actually learning more in school as opposed to just more schools built, or fewer deaths from malaria as opposed to increased roll-out of bed nets, or real-time satellite data showing more areas of tropical forest still standing.
In a new CGD Podcast I asked the same question of my guests, Dr. Raj Shah, former Administrator of USAID under President Obama, and Michael Gerson, former presidential speechwriter and Assistant for Policy and Strategic Planning under George W. Bush, and now a well-known columnist for the Washington Post. The two have reached across a generational and political divide to collaborate on "Foreign Assistance and the Revolution of Rigor," a chapter in the book Moneyball for Government, which underlines the importance of data-driven decision-making in development.
“Results are about, in my view, using American assistance to create human opportunity and to end extreme poverty around the world,” says Shah. “You just have to know what you’re trying to achieve, measure it with rigor and sophistication and make adaptations to ensure you’re delivering those results.”
Listen to the podcast to hear the response from Michael Gerson, who, during his years in the White House, helped create PEPFAR, President Bush’s celebrated scaling-up of efforts to tackle HIV/AIDS. How does Gerson know that initiative has achieved results?
“One of the most moving things I’ve ever seen in Africa were empty hospital beds,” he says of a visit to a rural clinic in Rwanda in recent years. Gerson recalls asking the head of the hospital how many patients were being treated for opportunistic infections related to AIDS. “We don’t have any,” Gerson recalls the man telling him.
Climate change will have profound effects on development, poverty, health, and well-being in coming years. Rejuvenated by the recent Paris agreements, efforts to channel the international funding commitments need channels for cost-effective mitigation. The Green Climate Fund (GCF) represents the best current opportunity to address climate change effectively with international funding. Unlike other institutions, the GCF is relatively new and is still developing its policies and procedures.
Those who follow the Center for Global Development will be familiar with our branded meme: “Cash on Delivery” aid, or COD. As early as 2006, Owen Barder and Nancy Birdsall authored a working paper on the COD approach to foreign assistance, which was distinguished from “input-financing” and proposed as a mode of payment that would enhance efficiency and, by revealing government performance to local populations, improve the recipient government’s accountability to its own citizens. Subsequent work from CGD includes a book by Savedoff and Birdsall and most recently a paper by Savedoff and Perakis. Many are enthusiastic about COD’s potential to revolutionize aid effectiveness. Yet within some global development organizations, leadership and staff alike express common concerns: is COD practical in the real world? Have you thought about this problem, or that constraint? How would this work in the context of our organization? And if we decided to move forward, how would we design a COD grant?
To help answer these questions, our new report Aligning Incentives, Accelerating Impact differs from CGD’s past efforts in two important ways. First, due to the Global Fund’s engagement with our working group over many months, this report reflects and responds to the concerns raised by the staff of a single donor institution about the ways that cash on delivery could fail to improve results or could even create perverse incentives within their specific institutional context. As a result, the report avoids promoting COD as a panacea applicable to all of the Global Fund’s financing. Instead, it stresses the need for contractible, externally verifiable indicators and for a willing grant recipient that is able to pre-finance some aspects of service delivery in order to successfully implement a COD strategy. We therefore hope the report will prevent those with a more distal view of Global Fund activities, such as outside observers, board members, and even higher management within the fund, from pushing Global Fund staff to implement COD in situations where it is doomed to fail.
Second, more than any of its predecessors, this report explicitly adopts the “principal-agent” framework for understanding the relationship between the Global Fund (“principal”) and its counterpart in the recipient country called the Principal Recipient (“agent”). This framework comes from the fields of contract theory and mechanism design—which were recognized by the Swedish Nobel Prize committee in 2014 when they awarded the prize in economics to Jean Tirole, one of their most important contributors. (For a more technical presentation see here.) Importantly, the framework allows us to acknowledge that the Global Fund will always have less information than the Principal Recipient about the cost of service delivery—and especially about alternative ways personnel could be managed to reduce costs. Our report, therefore, proposes contract designs that will better align the Principal Recipient’s financial interests with its efficient expansion of health service delivery. In addition, forthcoming background papers by Liam Wren-Lewis and Han Ye suggest how specific examples of efficient contracts, drawn from the extensive literature and practice on industrial regulation and contracting, can be adapted to improve the efficiency of Global Fund expenditure. For instance, a particularly promising contract, the “Fixed Price/Cost Reimbursement” design, allows the Principal Recipient to choose the most advantageous option from a menu of contracts. This clever design could improve health service output per dollar of Global Fund financing.
We hope our new report will be useful not only to the Global Fund, but also to other donors in the global health arena. (PEPFAR, are you listening?) However, we caution that any other donor’s application of COD deserves the same detailed attention to their own institutional capabilities and constraints as the Global Fund received for this report. The last sentence of the Nobel Committee’s summary of Jean Tirole’s contributions warns that "desirable [mechanism designs] are different from market to market.” Similarly, we warn that desirable results-based contract designs differ from donor to donor—and, for any given donor, from recipient to recipient. Worst case scenario: failure from an inappropriate application of a COD approach could lead critics to reject incentives altogether, simply because a poorly designed project proved unworkable.
So please: say no to a cookie-cutter approach! And say yes to the careful, thoughtful, and tailored application of incentives as a strategy to improve the efficiency and effectiveness of global health investments.
Can short–term unconditional cash transfers (UCT) create longer-term impacts? In a new paper, Berk Özler and co-authors study a group of young women in Malawi, who participated in a two-year cash transfer experiment as adolescents, in order to understand the long-term impacts of these short-term cash transfers. More than two years after the end of transfers, they find that the substantial short-term benefits of the program have largely evaporated. Unconditional cash transfers (UCT) caused short-term reductions in marriage, fertility, and HIV infection, but the cessation of cash transfers is immediately followed by a wave of marriages and pregnancies, accompanied with a catch-up to the control group in HIV prevalence. For those who had already dropped out of school at the outset of the experiment, two years of conditional cash transfers produced a meaningful long-term increase in educational attainment, delays in marriage, declines in fertility, and a more educated pool of husbands; however they see no increase in employment rates, earnings, real-life capabilities, or empowerment, suggesting that schooling itself has not improved the medium-term welfare of young women in this context.
I’m always a little anxious introducing a topic at a workshop without knowing if the presentations that follow will support or contradict my points. So it was with some trepidation that I spoke earlier this month at a SIDA workshop in Stockholm, associated with the Swedish International Development Cooperation Agency’s launch of “Results Based Financing Approaches (RBFA) — What Are They?”.
I emphasized two aspects of pay-for-results programs which I have come to see as essential to effective design and implementation:
the need to independently verify results, and
the importance of recipient autonomy.
The workshop was intended to introduce the idea of paying for results — a concept which is still unfamiliar to a lot of people working on development issues. Conventional aid programs tend to plan a series of activities and then disburse money on completion of those activities, whether or not they achieve intended outcomes. Paying for results turns this equation around and disburses funds when results are achieved. The details of the program design will vary depending on the goals, sectors, availability of information, and recipients, but the basic logic is very different from conventional approaches. In my research, I’ve come to see the independent verification of results and recipient’s autonomy to be key features of a well-designed and implemented program because these features are most consistent with the way development, innovation and progress typically occurs.
All three presentations explained why independent verification of results is essential. For one thing, the credibility of the program depends on whether or not the results are real. But more than that, the benefits of a pay-for-results program depend on recipients having feedback on how well their efforts are succeeding. If information about results is inaccurate, the feedback loop is severed and progress is less likely.
Multiple sources of independent information to verify performance of Solar Home Systems in Bangladesh, from presentation by S.M. Monirul Islam, IDCOL, on Oct. 8, 2015, in Stockholm.
The three presentations also explained how recipient autonomy is critical to success. Centralized programs can certainly encourage and support the expansion of a market for solar energy systems or the delivery of health-care services, but they cannot fully realize these aims without relying on local agency and innovation. The presenters showed how well-designed programs that pay for results opened space for exactly this kind of decentralized action and local response to feedback.
The importance of recipient autonomy highlighted by E. Schoffelen, Cordaid, on Oct. 8, 2015, in Stockholm with regard to results-based financing for healthcare in Africa.
In the discussion that followed, I noticed that the most common concerns related to the possibility of manipulating the reported results and constraining local autonomy. These are valid concerns. However, if the solar energy and health-care programs presented at the SIDA workshop are any indication of the future, new programs will increasingly incorporate independent verification and strengthen local autonomy — which bodes well for the future.
The Green Climate Fund (GCF) is the newest funding source to address climate change in developing countries. With $10 billion in pledges – and $5 billion committed to mitigation – the GCF is at a critical juncture because its Board is considering the rules and protocols it will follow when it pays for results. The risk for a multilateral fund like the GCF is that its Board will be more concerned about its reputation than about climate change; that it will demand assurances which compromise effectiveness; and that it will treat payments for slowing climate change as if they were charitable contributions from a patronizing philanthropist.
We believe the GCF can learn a lot from existing results-based aid agreements and the state of REDD+ finance (summarized in the forthcoming report of a CGD Working Group) which demonstrate the strengths and weaknesses of pay for results approaches. In particular, we think the GCF has a unique opportunity to make its grants and loans highly effective by leaving the financing of “readiness” investments to others and following three basic principles:
focus its pay for results agreements in relatively few countries;
protect the autonomy of its recipients; and
link payments to independently verified emission reductions.
Other initiatives have fallen short by hesitating to enter into pay for results agreements that follow these three simple rules.
First, programs that pay for results need commitments that are large enough to draw the attention of policymakers to the task. When funds are limited, this means allocating money to the countries or sectors which would have the biggest impact – where the biggest tropical forests are. This contrasts with multilateral funds, like the Forest Carbon Partnership Facility (FCPF) and BioCarbon Initiative for Sustainable Forest Landscapes, that have faced pressures from their funders and beneficiaries to spread money across a wide array of participants in ways that dilute impact.
Second, programs that pay for results need to preserve the recipient country’s autonomy and agency over its own strategies. The aid system has recognized the importance of country ownership to program success. Yet, aid organizations don’t acknowledge that their conventional approach – involving detailed technical negotiation of country plans and strategies and active involvement in program designs that require external approval – undermines country ownership. Subsequently, when aid organizations monitor adherence to these plans, they introduce rigidities that interfere with the process of experimentation and adaptation that is necessary for real progress and real development.
Third, programs that pay for results need to recognize that the payment purchases a result. Once the result is verified and paid for, the funder has no more claim on determining how the money is spent. As one of our colleagues recently noted, countries pay Saudi Arabia for delivering oil all the time without introducing safeguards on how that country spends its revenues. After purchasing something which is good for the planet – reduced emissions – why would we insist on telling tropical forest countries how to spend their money? The recipient country should decide how the funds will be used, as Brazil does with Norwegian results payments which are channeled to the Amazon Fund.
GCF can see the importance of these rules by looking at other efforts to pay for reduced deforestation. Norways’ agreements with tropical forest countries are among the best. By focusing on a few countries that can produce results that are globally important, Norway has provided amounts that are large enough to have impact at the national level, to be visible and to bolster political willingness. Norway’s agreements with Brazil and Guyana respect those countries’ autonomy over deforestation policies, but its later agreements with Indonesia, Liberia and Peru have incorporated preparatory phases that infringe or impose rigid approaches on domestic policymaking, tie funding to specific strategies and actions, and delay the disbursement of results payments.
The Forest Carbon Partnership Facility (FCPF) Carbon Fund is also trying to pay for reduced greenhouse gas emissions but has been slow to move ahead. One difficulty is that money in the FCPF Readiness Fund is spread thinly across 47 participating countries, some of which contribute only marginally to global greenhouse gas emissions. Fortunately, the Carbon Fund itself plans to pilot payment for results in fewer countries. But to become eligible for these payments, countries must present strategies to reduce deforestation, procedures for complying with environmental and social safeguards, and detailed activity plans. And these hurdles are just the start. Once selected into the Carbon Fund pipeline, it may still take years before a country actually receives a performance payment. After all, the Carbon Fund has been operating since 2011 and only 11 countries have completed the first of 8 steps needed to begin receiving results payments.
The GCF Board has an opportunity to learn from other programs and implement an ambitious program to pay for reduced greenhouse gas emissions from deforestation. It can make this pay for result approach effective by focusing its effort on countries that can deliver more averted tons of carbon emission, rewarding countries that develop and succeed with the strategies they choose (without up-front clearance from outsiders), and recognizing verified emission reductions as the deliverable for its payments. The GCF’s objective and potential contribution are too important for it to take timid steps or ignore these basic lessons from the past.
We’d like to thank Jonah Busch and Frances Seymour for their valuable comments and corrections.
What do you picture when you hear ‘humanitarian aid’?
Most of us imagine temporary shelter to help people get back on their feet after a natural disaster, or food supplies and clothing to help refugees for a few weeks after they’ve fled a conflict. Those images are increasingly out of touch with what’s happening on the ground.
Emergencies are more protracted: a girl born in Kenya’s Dadaab refugee camp the year it was founded would be well over 20 years old today and could still be living there. Last year, two-thirds of official humanitarian assistance went to long-term recipients. That’s putting humanitarian budgets under strain. According to the UN, the number of people in need of help has doubled in the last decade; humanitarian aid grew by less than half.
There’s a growing consensus that humanitarian cash transfers can help to bridge the widening gap between needs and resources, empowering people affected by disaster and using local markets to deliver the goods and services we previously thought only aid agencies could provide.
What we know about humanitarian cash transfers
The Cash Atlas, a platform to help track how and where cash transfers are being deployed, can add some context to our understanding of the current state of cash transfers. The data are self-reported by agencies and so not a comprehensive record. Nevertheless, focusing on the years 2005 to 2014, the Cash Atlas includes nearly 800 programs implemented by 49 different organisations.
We combine information about transfers with financing data from OCHA’s financial tracking service, convert figures to today’s prices, and allocate budgets equally over each project’s duration to estimate spending per year. The resulting data represents $2.8 billion worth of cash transfer programming. (Click here to download the data appendix, .do file, and a note on methodology).
What did we learn?
1. Humanitarian needs are outstripping funding
Humanitarian agencies set out the financing they need to tackle crises using the humanitarian programme cycle (previously called the consolidated appeals process). Adding what agencies asserted they needed and comparing it to what donors funded paints a stark picture of a growing deficit between needs and resources. (Click off ‘Funding’ and ‘Requirements’ to zoom in on this growing shortfall).
It’s plausible that agencies overstate their needs. But with a deficit of over $7.3 billion last year, dissembling by humanitarians, if present, probably isn’t the main story.
2. Agencies are scaling up ‘cash-type’ transfers, but most come with conditions
Cash transfers are the Rorschach test of humanitarian aid— agencies see what they want in it. It can mean putting $60 a month on a debit card so that a family can decide what to buy (and where to get it cheaply), but is also used to describe everything from vouchers for work to cash for training.
We translate the Cash Atlas’ tags into something more intuitive by calling a transfer conditional if it requires the recipient to do something (‘cash for training’, ‘cash for work’, and ‘conditional grants’), unconditional if it’s described as such (‘unconditional grant’), and a voucher if people can only spend it in specific ways (‘vouchers’, or ‘vouchers for work’).
Breaking the data down like this shows that organisations have mainly used programs that come with strings attached for beneficiaries. However, this buries an interesting lead: unconditional transfers are growing quickly, reaching about one-fifth of the total estimated budget for cash programs last year.
3. Cash is still a tiny share of humanitarian aid
What does this rise in cash programming look like in context? A reasonable comparator is how much donors spend on humanitarian aid more broadly (appeals, together with various kinds of emergency funding and bilateral aid).
OCHA’s data indicate that donors spent $23 billion on humanitarian aid last year. On a purely fiscal basis, then, cash-type interventions remain the unloved country cousins of humanitarian response, accounting for less than $1 billion of this. (Zoom in by clicking on the legend to see how cash programming has grown over time but remains a very small share humanitarian spending).
4. Most programs haven’t been (only) about transferring cash.
Our data show very few examples of ‘pure form’ programs: a budget line that only pays for beneficiaries to receive modest payments and lets them decide how to spend them. Leaving out reports that don’t flag which of these types of budget it was or which ‘modality’ was used leaves 368 programs. Of these, just 22 — less than six in every hundred — fit this description.
Type of transferUnconditional ConditionalVoucherType of programCash only225154Cash / in-kind889657
A more inclusive definition could include programs that gave unconditional grants irrespective of whether their budgets included in-kind aid. That brings the total up to 110, just under one-third of the total.
5. Most budgets haven’t been (only) about transferring cash.
A head count of programs based only on whether they include in-kind aid is instructive, but isn’t weighted by how much agencies actually spend. It might be, for example, that a small number of unrestricted transfers actually represents a very large share of humanitarian cash programming.
The 368 programs about which we know both the type of budget and how the cash transfer was made represent $750 million (of the $2.8 billion) worth of programs included in the Cash Atlas. Of this, less than 5 percent, nearly $37 million, is for ‘cash only’ programs that delivered unconditional transfers.
However restrictive the definition, and regardless of whether we’re looking at the number of programs or weighting them by spending on each, it seems that agencies haven’t comprehensively moved to large-scale, unconditional programs, or separated those programs from traditional, in-kind aid.
6. Most beneficiaries don’t (only) experience cash.
Though it’s easy to focus on top-line budgets, what ultimately matters is the number of people a program benefits. Combining information about the total budgets with the number of ‘beneficiaries’ that agencies purport to reach picks up some interesting trends.
Most importantly, this shows that larger budgets aren’t coupled to coverage. The size of the spheres indicates that voucher-based programs reach the most people, but have consistently received a smaller share of total budgets. Similarly, in the last few years fewer people benefited from conditional than unconditional programs (1.48 million vs. 1.54 million), but the graph shows that conditional transfers had larger budgets: $234 million more in 2014.
What we don’t know, but should
7. Budgets per person vary a lot, and we don’t really know what that means.
Comparing budgets to the number of beneficiaries begs the question of what this looks like on a per-person basis. Since our data are in current (2015) terms, we get a ‘budget per capita’ by adding up cash programming budgets over the last decade and dividing by the number of beneficiaries that aid agencies report assisting.
The concept of ‘budgets per capita’ highlights a fundamental gap in what we know about HCTs.
People who are skeptical about ‘cash for training’ can point out how expensive those programs are. But organizations delivering training programs might point to the same number and argue that, no, this actually shows much value they transfer to beneficiaries. In short, budgets aren’t the same as costs. If these different modes of transfer cost the same to make, ‘cash for training’ might well pass on the most to those we’re focused on helping.
But even if we took these numbers at face value, and even if we assumed (without basis) that they cost the same, we still couldn’t make a value judgment about what donors should focus on. Transferring $60 dollars to a family, for example, might help them get what they most need by making tough choices between heating, school fees, or clothing. That could ultimately be much more ‘valuable’ than other interventions that cost as much, or much more
* * *
CGD has been working with the Overseas Development Institute to look into humanitarian cash transfers, supporting a High-Level Panel on Humanitarian Cash Transfers chaired by my colleague Owen Barder. The panel will launch its report on September 14. Its findings reflect decades of hard-won, front-line experience, opportunities arising from new ways to transfer money safely and securely, and insights from a fast-growing evidence base.
Data-driven initiatives like the Cash Atlas are an important start to understanding the role that HCTs can play. But we won’t be able to get a fix on the value added by different kinds of aid without better information, not only about budgets but also on costs and, more importantly, what those we have a duty to assist actually need, want, and can use.
This, ultimately, is why the high-level panel has been convened: to help us weigh the evidence that we have on cash transfers, and to chart a path forward for humanitarian aid that strikes an informed balance between the thorny yet critical issues of cost, value, and growing need.