Cash on Delivery: A new approach to aid

This article is part of the Aid Open Forum, a series of blogs to encourage debate, and the emergence of new ideas for the aid program. In this article, Richard Curtain looks at a new aid instrument that is creating a buzz in Europe and the US.

Cash on Delivery Aid (COD aid), a proposal from the Center for Global Development, promises to fundamentally change how aid is delivered, by only paying for verified results. What is this new approach to aid and could it help Australia to improve aid effectiveness in Timor-Leste and the Pacific?

Overcoming dependence and limited impact

Giving aid isn’t easy. Traditional aid programs can create unintended incentives for rent-seeking rather than results. Australia’s aid program operates in countries which already have a heavy dependence on aid. While a large aid program can provide much needed funding for service delivery, it can also be a major contributor to the resource curse: providing windfall gains which create and sustain a political class of rent seekers. This can distort the allocation of aid and the choice of activities that it funds.

Partly for this reason, traditional aid programs often have had disappointing impact on development outcomes. The Asian Development Bank in its strategy document ADB’s Pacific Approach 2010-2014 acknowledges that many of its efforts to support policy reform in Pacific countries have had little impact. The main problem is the lack of political demand in the recipient country for change. Elinor Ostrom, the 2009 Nobel Prize winner in Economics, notes that the impact of aid on the recipient country is often shallow and not sustained. This is caused by the large number of stakeholders involved with large differences in power and capacity, resulting in a viper’s nest of perverse incentives. A new approach is needed which provides positive incentives to both the donor and the recipient to deliver real and lasting outcomes (see Owen Barder’s post on this).

Cash on Delivery Aid aims to address these problems by refocusing aid delivery on achieving results. Its major innovation is in reversing the roles of the donor and the aid recipient to overcome the latter’s tendency to a passive response in the face of the demands of the donor. This is done by moving the responsibility for service delivery from the donor to the party receiving the aid.

What is COD aid and how does it work?

Cash on Delivery Aid links final payments to the recipient’s achievement of specific outcomes. The operating principle is that the recipient government can run the program in whatever way it sees fit, without interference of any kind from donors.

At the core is a contract between donor and aid recipient that stipulates a fixed payment for each unit of confirmed progress toward an agreed-upon goal. Once the contract is struck, the donor takes a hands-off approach. Payment is made only after progress toward the goal is verified by a third party. The contract, the amount of progress made, and the eligible payments are made public.

The proposal comes from the Washington-based Center for Global Development (CGD), a small but influential think tank. Nancy Birdsall, the Center’s President, and her coauthors have also written a book on the concept with a specific application to primary schooling. (Also see Nicholas Kristof’s article in the New York Times on A New Approach to Aid for a concise summary of the idea. Another more extended summary is to be found at Oxfam International’s Duncan Green blog entitled Cash on Delivery – worth a try?)

What are the benefits?

The potential benefits for both the recipient countries and AusAID’s domestic constituency are huge in terms of the focus on delivering sustainable outcomes. COD aid addresses the legacy of aid dependence by giving governments a positive incentive to build up their capacity to deliver improved services. As sustainable outcomes are achieved, the payments will taper off, avoiding any surprises about when the external funding will cease. Australian taxpayers will have a ready means of assessing the value of their investment in aid programs.

COD aid can also improve the effectiveness of other aid. For example, the success of technical assistance depends on how outside expertise is used by the recipient government. If it is used to prop up weak capacity, it is likely to do more harm than good. If it is a part of a government’s clear plan to lift capacity to achieve specific ends, it can play a valuable role. A delivery agency with a clear results focus and responsibility for paying for inputs is much better placed to know what external expertise it needs and for how long.

Working in fragile states?

These potential benefits from COD aid are especially important in Australia’s aid programs in fragile states. The challenge for any donor in these settings is to ensure that its aid is having a long term beneficial effect and is not perpetuating weak institutions. COD aid addresses this challenge of aid to fragile states.

However, where initial capacity to deliver services is weak, a hybrid model of COD aid will be required. This will need to be based on working out agreed steps for handing over responsibility for achieving outcomes and who funds these steps. The essence of the new approach, however, is the donor’s focus on giving ever increasing responsibility to the delivery agency over a specific time frame. A fragile state setting will also require a detailed joint situation analysis to identify all the potential obstacles to ensure that these are addressed.

Who is trialling COD aid?

The new UK government is planning to trial COD Aid. Andrew Mitchell, the UK’s international development minister, has stated his government’s preference for results-based aid:

We’re also fundamentally redesigning our aid programmes so that they build in rigorous evaluation processes from day one. The focus will be on outputs and outcomes rather than inputs. In these difficult, economic times donors have a double duty, a responsibility to achieve maximum value for money: not just results but results at the lowest possible cost. With this in mind, we want to test the concept of cash on delivery aid that’s been mooted by the Centre for Global Development.

And interestingly, DFID has not used fragile states as an excuse to avoid trialling the concept as much of its aid now focuses on fragile states.

A new approach to Australian aid?

Cash on Delivery Aid is one aid instrument that deserves serious consideration as part of the Australian aid program. In addition to the benefits outlined above, it would also free up stretched staff from day-to-day management tasks, that with the doubling of aid threaten to overwhelm AusAID. Mark Baird has highlighted this problem in his review of evaluations of AusAID’s service delivery programs for the poor and his contribution to this blog.

All this will require, on AusAID’s part, a fundamental shift from the narrow form of accountability that now dominates its aid delivery system. This often takes the form of the donor ensuring that required processes are followed.  Mark Baird has also noted how an expanded aid program will require a new skills set within AusAID – one that goes beyond the risk adverse micro-management of projects. It will also require a different whole-of-government operating mode, to produce  country-specific, high-level bilateral agreements and third-party independent monitoring.

AusAID, in its response to the current review of aid effectiveness, has an opportunity to propose new ways of delivering aid. Any suggestions, however, need to show how they  overcome the perverse incentives generated by the existing system of aid delivery. In particular, the challenge for AusAID is to shift away from its close monitoring of inputs and outputs at the expense of outcomes. Included in this new approach must be a way of giving aid recipients real responsibility for the continued sustainability for aid funded initiatives.

Richard Curtain is a Melbourne-based, public policy consultant, who has spent 18 months in Timor-Leste in 2008 and 2009, working on projects funded by USAID, UNICEF and AusAID. His current work for two major multilateral agencies in the region relates to Timor-Leste and to pacific island countries.

image_pdfDownload PDF

Richard Curtain

Dr Richard Curtain is a Research Fellow with the Development Policy Centre.


  • This is a very interesting and timely idea. I wonder if and how we could translate it in a humanitarian context? Perhaps not at the onset but later in the emergency cycle? The sooner the better to avoid the wait and see attitude of beneficiaries, what they call “attentisme” in Democratic Republic of Congo.

  • Good to see the responses to my posting. I am just back from a week in Dili, Timor-Leste, conducting interviews for a workshop on a related issue – how to provide demand-focused technical training. Informal feedback from colleagues I spoke with is that big changes are still needed there in how aid is delivered.
    The use of a hybrid system of COD in a post-conflict setting will require careful analysis and innovation. Both parties will also need to negotiate changes if it is not working as expected. Better and more independent evaluation tools will be needed to provide feedback. However, the longer-term benefit must always be kept in mind – the full acceptance by the recipient government of the funding and implementation of the intervention. This is a key test of aid effectiveness.
    The change in the mode of aid delivery that COD encompasses is akin to the fundamental reforms to labour exchange services in Australia over a decade ago. Australia led the world in these reforms, reviewed favourably by the OECD in its 2001 report Innovations in Labour Market Policies: the Australian Way.
    The switch from the direct delivery of services to job seekers through the Commonwealth Employment Service (CES) to outsourcing these services to a network of for-profit and non-profit organisations was a huge change. The new focus was on payment for outcomes – jobseekers in jobs, training or education more quickly.
    Switching from a centralised funder-driven model of service delivery to a ‘steering not rowing’ focus was difficult. It took a number of attempts to finetune the new arms-length system. Above all, it required within the funding department a change of mindset from service deliverer to service designer based on high-level conceptual skills It also took alot of ‘learning by doing’ about what worked and what did not to design tenders that worked more effectively over time.
    These reforms were an outgrowth of a period of experimentation in public policy in Australia. Is the Australian Public Service up to the challenge in the current era?

  • COD Aid deserves a go!

    Coffee connoisseurs can get their cappuccino, café latte, or the like in nearly every little town on the planet. And while the price of a café latte differs considerably across locations, the quality – if my tastebuds are any guide – does not. The same can hardly be said of a classroom-experience in the local primary school. And the reasons may rest with incentives – the café thrives on serving good coffee; the primary school may not.

    Lots of funds have been spent training teachers, equipping schools, and improving the curriculum to improve literacy and numeracy in the developing world. This, using my café latte example, is akin to increasing the quantity of beans, the size of the coffee grinder, and possibly the dispenser to improve the quality of coffee produced. Absurd you might think. But this is exactly the strategy being pursued by donors such as Australia in Papua New Guinea.

    A lot is made of the poor quality of teachers for the poor educational outcomes in PNG. However, I find it hard to believe that the level of skills required for delivering a simple lesson in arithmetic or English is any more demanding than those for making a good café latte. Sure – they are very different skills, but which of the two is more demanding is debatable.

    Good coffee gets served cash on delivery. (BTW, no one has yet given me coffee on credit – but it may just be me!) So why not try the same for primary schooling? I have suggested a small pilot in PNG to test the effectiveness of COD in increasing the quantity and quality of throughput from primary schools. And it is only through carefully designed experiments such as these that we will know if COD works.

    It is time to roll up our sleeves and give COD a go.

  • The Center for Global Development have a really good FAQ on COD Aid that provides some answers to the questions raised in the previous comment. The full FAQ can be found at

    How can the country be expected to make progress if it does not receive payments until it has already achieved some of the goals?

    Countries that enter the COD Aid contract are not starting from a blank slate. They already have their own programs financed with domestic and foreign resources. The COD Aid funds provide an incentive to make existing technical assistance and educational investments “pay off” in improved outcomes. Once the COD Aid payments start flowing, these funds can be used to sustain and further accelerate progress.

    Wouldn’t there be a big lag between investment and the payoff to the country?

    This depends on the appropriate time frame. Most development programs have to be thinking of 5 to 10-year horizons, and any program aimed at making substantial changes in institutions or behaviors also has to consider long time frames. The lag in payments is likely to make the contract seem unattractive to short-sighted leaders, but very attractive to those who have longer-term visions. This is, then, another attraction of this approach since it is most likely to be taken up by the same leaders who are most likely to use it well.

    What if a country doesn’t meet the goal?

    The COD Aid contract is not an “all-or-nothing” form of assistance. A country would be paid a set amount for each additional unit of progress. So, rather than “passing” or “failing,” the recipient is rewarded in proportion to the amount of progress achieved.

    Governments only control a fraction of the factors that affect outcomes. How does COD Aid account for this?

    While it is true that governments do not have full control of all the factors that affect outcomes, there would be no reason to provide aid if funders didn’t believe that governments can influence those outcomes. The outcomes chosen for COD Aid agreements should be outcomes that funders care about and over which governments can have significant impact. The funder may end up paying for progress that is not directly attributable to government action but we view this ‘risk’ as more acceptable than the ‘risk’ of paying for inputs that may have no effect on outcomes at all.

    What if a drought or other unforeseen shock keeps the government from achieving progress?

    Sharp changes in international interest rates or terms of trade and natural disasters such as droughts can interfere with development programs. The COD Aid contract should contain contingencies for adverse shocks that interfere with measuring and verifying the outcome indicator. However, COD Aid funds should never be disbursed except against progress measures. If emergency programs are necessary to assist a government, they should use a different aid mechanism rather than compromise the credibility of the COD Aid agreement.

    Is lack of money really the problem? What about insufficient demand, political obstacles, or weak technical capacity and coordination?

    Lack of money is not necessarily the problem. The funds disbursed under COD Aid provide an incentive for the country to undertake the right diagnosis and design effective strategies, whether or not those strategies require additional funding. For example, if the only obstacle is political will to enact a particular law or enforce specific management practices, COD Aid would provide leaders with an incentive to take those measures. In such a case, the funds that the government receives for successfully improving educational outcomes could then be applied to other public priorities. In most cases, however, some funding is still necessary though not sufficient for improving outcomes – whether those funds go to additional training, more staff, infrastructure, supplies, research, information systems, administrative reforms, supply incentives, or demand incentives.

  • Interesting idea, appears to raise a number of implementation issues, though I haven’t studied in depth and may be missing something

    1. Cash flow. Results aren’t instant. There are often long time frames to achievement of results and the recipient is likely to be at an inherent disadvantage in obtaining funds. Possibly this would need further loan aid funding simply to cover the cash flow, more agreements, more administrative complexity. And the longer the time frames the greater this cost.

    2. Impact of failure to achieve results. What happens when the aid recipient fails to achieve outcomes at all? Examples are legion, and the more difficult to achieve results, the greater the likelihood of failure. Using the example above, does the recipient default on the loan? Insure against failure (a further overhead)? [Also that nice issue about what is results tied in, do we measure something hard that resembles what we’re really looking for, or something easy to measure (and achieve) that isn’t]

    3. Horse before cart? Kinda related to point 1 above. Where aid is intended to build capacity this may generate change, which may improve performance, with risks in both connections. Donors hate this, they want to see aid deliver results in terms of performance. But commonly the capacity fix (e.g. the old chestnut training) is needed as the hoped for catalyst for change (e.g. new techniques) to improve performance outcomes. But if the aid funding is tied to performance, the process can’t start.

Leave a Comment