Can aid create incentives for politicians in developing countries?

The question of whether and how donors might try to create incentives for politicians in developing countries lurks behind many of the debates about how to give foreign assistance.  It has come up for me twice in recent days:

  • First, I was talking on Thursday evening with a donor agency official in Tanzania, who was explaining to me that her agency gives budget support in order to exercise policy influence with the government of Tanzania
  • Second, in the context of Cash on Delivery aid, several people have argued that the whole concept is flawed because it relies on the idea that a promise of more aid will create incentives for developing country governments, and this is unlikely to be true.

I think it is worth establishing two points about the use of aid to create incentives. First, donors are almost never able to use aid to create incentives that work for developing countries, and they probably should not try.  Second, I am in favour of piloting Cash on Delivery aid, not because I think it will create stronger incentives in developing countries, but rather because I think it will create better incentives for donors (and it might at the margin increase the accountability of developing countries to domestic stakeholders).

Why donors cannot create effective incentives through aid

It sounds straightforward to say that if a donor offers a lot of money for a developing country to change a policy or achieve a particular outcome, this should, at the margin, have some effect on incentives within the developing country and so increase the likelihood that this policy change will occur or the outcome be achieved.  People respond to incentives, right?

Yet there is a mountain of empirical evidence that aid conditionality does not work. (Check the references at the bottom of this post for details.)  There are two good reasons why donors cannot create effective incentives for developing countries.

First, donors do not make credible threats. Most donors have an organizational imperative to get the money out of the door.  Aid agency staff want to sustain their bureaucratic status by managing large budgets.  They do not, in practice, withdraw aid when developing countries do not comply with conditions.  Officials in the governments of developing countries know that donors rarely stop aid – indeed, it is much more likely that aid will not be disbursed because of donor maladministration. The Government of Kenya was able to include the same promise to reform agricultural policy in five successive World Bank loan agreements over fifteen years: the condition was never met, and the aid was always disbursed. If the threat is not credible, it creates no incentive.  Svensson (2000) set out the argument and the evidence in detail.

Second, the incentive created by aid is not strong enough to influence decisions of government officials.   Aid is used to deliver services within the developing country, and that has only a very indirect impact on the interests of a minister or senior official.  After all, most aid does not end up in the minister’s personal bank account.  Of course, other things being equal, most ministers and officials want to see their fellow citizens better off.  They might also calculate that more aid which improves service delivery will make them more popular and so more likely to hold on to office.  But these benefits are often much less direct and immediate than the political costs to them of reform.  A politician is not going to take on vested interests and risk being ousted from office just to increase the number of wells being dug in a rural area.  In other words, people do respond to incentives, but aid is not as big an incentive as you might think.

Consider the choices made by ministers in industrialized countries.  They too want to improve the economy, increase spending and improve public services for their citizens, for the same reasons as politicians in developing countries.  But they are constrained from implementing sensible economic reforms, or raising more money in taxes, by powerful vested interests. Responding rationally to the incentives they face, they choose carefully which vested interests they are willing to confront at any given time; and that means compromising on the services they can deliver and the amount of economic growth they can bring about.  The prospect of faster growth and better services does not lead them to make suicide runs at reforming agricultural subsidies, removing trade tariffs or increased taxes on the super-rich.  Developing country ministers are no different.  Rational ministers will not risk being ejected from office just to bring more aid to the country.

This is why conditionality based on cutting aid does not work, and it is why it is a great mistake for development agencies to allow their success to be measured in terms of the amount of policy reform they have brought about.  Aid works because it pays for essential services, not because it brings about policy change.

I am not sure that industrialised countries could not create meaningful incentives for politicians in developing countries: but I believe it would be much more effective to impose travel restrictions and visa bans, to uninvite Ministers from key conferences, or to sequestrate money accumulated in Swiss bank accounts, than to cut off aid.  These approaches would have the added advantage that they wouldn’t leave the poor to starve.

Incentives and Cash on Delivery

If I don’t believe that developing countries will respond very much, if at all, to aid-based incentives, why do I support Cash on Delivery aid?

Some have argued that this is a show stopping argument against Cash on Delivery.  Tom Harrison says that “the biggest issue with ‘cash on delivery’ is that it assumes that where governments do not provide key public services it is because they lack the incentive to do so rather than because they lack the capacity to do so”.  I agree with him that it would be a mistake to think that we can create incentives for developing countries to deliver key services; but I disagree that this is the assumption on which Cash on Delivery is based. And in a well-argued paper, Ngaire Woods and Paolo de Renzio criticize the idea of Cash on Delivery Aid, partly on the basis that “external actors may have limited leverage on domestic political realities and accountability”.  I entirely agree with them about this.

I don’t believe that most developing countries need incentives to deliver better education for their children, better health care or access to safe water.  I think in many cases that what they need is money.   The problem is that donors won’t give that money without a whole rash of conditions, milestones, benchmarks, policy dialogues, missions, evaluations and reports.  Donors are forced to impose all that paraphernalia because they need to demonstrate to their taxpayers that the money has achieved something. It makes aid costly for developing countries, and highly unpredictable. I believe that Cash on Delivery can cut through all that: by providing money on the basis of results, donors can put more money into countries that are willing and able to deliver more and better services, without imposing hassle on either donor or recipient.

The promise of Cash on Delivery might – just might – nuance the political incentives in developing countries a little.  If the media, parliamentarians and civil society know that cash is available for any country that delivers better outcomes, that may help them to put pressure on their government to do a better job.  The government will not be able to hide behind the old excuses of lack of money or the pernicious impact of donor conditions and foreign interference. They will have to explain to their own citizens why they have not been able to do more.

Furthermore, it is possible that I’m wrong, and that in some cases Cash on Delivery aid will provide a modest incentive for some countries.  Some of my friends think it might help, and I’ve never heard anyone make a convincing case that it would do any harm.


I think a reasonable position to take is:

  1. Donors cannot, in practice, create incentives for developing countries through the promise of aid. That is why aid conditionality does not work.
  2. It is absurd to measure the success of aid by the policy change it brings about. We should measure aid’s success by the services which are provided.
  3. Cash on Delivery does not depend on the assumption that developing countries need incentives to provide key services.  There are good reasons for testing the idea even if you do not believe that incentives are needed or would work.


Killick, T., (1998), Aid and the Political Economy of Policy Change, Routledge, London and New York.

Collier, Paul. (1997). “The Failure of Conditionality.” In Catherine Gwin and Joan Nelson, eds., Perspectives on Aid and Development. Policy Essay 22. Washington, D.C.: Overseas Development Council.

Devarajan, Shantayanan, David Dollar, and Torgny Holmgren. (2001). Aid and Reform in Africa. Washington, D.C.: World Bank.

Nelson, J.M., (1996), “Promoting Policy Reforms: The Twilight of Conditionality?”, World Development, Volume 25 Number 9, September.

Svensson, Jakob, (2000). “When is foreign aid policy credible? Aid dependence and conditionality,” Journal of Development Economics

13 thoughts on “Can aid create incentives for politicians in developing countries?”

  1. Excellent points, Owen.

    If we look at COD Aid primarily as an alternative mechanism for transferring foreign aid, then it really makes sense to give it a try. Nancy and I made this point in the book “Cash on Delivery” by discussing the approach in relation to trends in foreign aid. However, most people tend to think of COD aid in the context of payment for performance (P4P) and other results based financing approaches where the emphasis is on the incentive for a particular service provider or individual.

    The need for shaking up the foreign aid financing mechanisms has been really obvious to me in our recent trips to Ethiopia, Liberia and Malawi. People from aid agencies and development banks will assert that programs are more successful when the country “owns” the program but at the same time claim the country doesn’t have “capacity,” or a strategy, or the right implementation plan to succeed with a COD Aid agreement. When you look at the big picture, I think the most dramatic reforms have come from developing countries themselves, not from conditionalities or technical assistance. It’s about time we recognized that and tried something different.

  2. Great post but a couple of objections: ‘Aid is used to deliver services within the developing country’… In many countries this is only true for certain off-system aid instruments. I prefer the logic in your earlier post on fungibility. In most heavily aid dependent African countries, ministries of finance operate a rule of thumb whereby recurrent spending (delivering govt services) is restricted to within the domestic resource envelope. As well as being a prudent macro policy, this helps to protect them from the huge unpredicability of aid, whether because of administrative cock-ups or because the donors try to use it to exert ‘policy influence’ – it means even a sustained pull-out is not going to affect services too badly. This doesnt hold in really aid dependent post conflict settings (e.g. rwanda), but does in e.g. uganda and mozambique and could be added to your list of why aid conditionality exerts weak incentives – aid, including DBS, is often not really financing recurrent spending and hence services. This is NOT a rationale for reverting to more heavily earmarked aid modalities which deny this coping mechanism but remain inherently unpredicable (PEPFAR, GFATM etc) – quite the opposite.

    Second, Ngaire Woods and Paolo de Renzio have already expained pretty convincingly why COD is more ‘innovationitis’ than a useful tool for development:
    Again, I prefer the reasoning in another one of your posts (on global health intiatives). We need to at least apply the same level of scepticism to ‘innovations’ as to the current orthodoxy. GAVI’s ISS performance payments in Uganda are a spectacular example of what can go wrong when shiny new innovations ignore local political incentives.

    Its much more messy and prosaic, and cannot so easily be packaged for an advocacy campaign, but why not focus more on lobbying Governments to allow donors such as USAID and the World Bank to adopt less harmful and more flexible practices & templates for the aid they already deliver and to moderate their expectations regarding what can be done with it? The devil is very often residing in the country-level detail. Maybe this agenda needs an acronym before it will fly?

  3. Very good points, but I think you are essentially wrong.

    The current donor practices in a planned and negotiated system. (contrary to a bottom up trial and error system) consist of a strong (unintendet) incentive system for the political leaders. I start with a few, but others will be able to find more:
    – budget aid and IMF loans support the government in charge, and democratic conditions are seldom enforced. The lesson learned by governments is to keep the World Bank on board, and power will be long term.
    – The incentive is strong to create the facade of popular consultations (NGOs, village meetings, etc.) which is higher in esteem by donors than the messiness of real life politics with often incompetent and inexperieced politiecians.
    – Ignoring technical advice is safe, as long as the higer level meetings go smoothly. The stakes are so high that the aid decisions are essentially political, and as we all know, politics is about personal relations. As one Belgian politician said to the press after a state visit to Mobutu Zaire: I love this country, its people, its leaders.

    However, in general, I do agree that the intended incentives of the donors don’t materialise. One reason for this is when a donor says he gives budget aid in order to influence policy, the donor proposed changes to the policy are often intrusive, a ragbag of important and irrelevant issues and based on not too much in depth knowledge of the ins and outs. As we learned a long time ago: if you have too many objectives, the result ends up being the same as having no objective at all.

  4. This is not the first time that de Renzio and Woods (dR&W 2006) has been cited incorrectly as a critique against COD Aid. In fact, the paper was commissioned by CGD as an input to designing a COD Aid agreement. We asked for criticism so we could assess and improve the proposal. This is exactly what we did in the book, Cash on Delivery, which incorporates and addresses the points voiced by dR&W.

    De Renzio and Woods’ paper, despite its criticism, actually supports using COD Aid. The paper’s analysis and criticism is aimed at identifying the “pre-conditions” that the authors believe will make a COD Aid approach successful. They name three: (1) that all potential distortions in the performance reward mechanism be addressed, (2) that COD Aid should replace rather than complement existing aid modalities, and (3) that it should be tried in places where political culture and domestic accountability mechanisms are operative. We addressed issue #1 in the book by designing a careful reward structure for achieving universal primary completion. Issue #2 essentially supports COD Aid and simply argues that its effectiveness will be diminished by other aid flows. This can be read as an argument for reducing traditional aid rather than for not trying COD Aid. Issue #3 says that COD Aid will function better in countries with better domestic accountability systems; it doesn’t address whether – in countries with weak domestic accountability – transaction intensive aid in budget support or traditional projects does any better than a COD Aid approach that at least requires evidence that outcomes or outputs close to those outcomes were achieved. In fact, if you follow Owen’s reasoning, Issue #3 may be quite irrelevant to the arguments about trying COD Aid since the incentive effect for the country is secondary to the incentive effect it creates in the aid-providing institutions.

    By the way, I’m puzzled by Sam’s comment. COD Aid is exactly the solution to the problems Sam describes in his final paragraph. It focuses on a single objective, e.g. universal primary completion, and the donor refrains from imposing conditions or approaches or strategies. So, Sam, why do you disagree with Owen?

  5. I disagree with the main statement of Owen: “First, donors are almost never able to use aid to create incentives that work for developing countries, and they probably should not try.” Donors do create strong incentives that work for developing countries, however, I must admit it are most of the time not the incentives they intend to create.

    I quote Duncan on Haiti: “There is no apolitical option: A disaster of this magnitude is also a political shock. New actors will emerge, old ones will decline, politics will shift.[….] Donors won’t solve Haiti’s problems (which of course predate the earthquake), Haitians will. But the way reconstruction is designed could help or hinder efforts to tackle poor governance, mass unemployment, inequality and crime.

    Aid will always be an element in the local power plays. However, donors routinely consider the government as their main partner to be strengthened, strengthening them in relation to the other local institutions and branches of the state. In their choices, they also strengthen some departments, like womens affairs, more than others, like police.

    I have no established opinion on cash on delivery and I am still open for convincing, but I have a gut feeling it is one more fad, with little advantage compared to a lot of interventions that are evidence based and are still too seldom supported. As pre-financing by the recipient is necessary, there is a high level of complication. The reward is also too far away from the result, and mostly going to someone else, (see fungibility).Also, we all know donors can and will move the goalposts during the execution, e.g. when a law on homosexuality comes up. It might be for the department striving to get the agreed results more like a lottery than a system. Now, lotteries are in general more successful than aid schemes.

  6. Hi Owen,

    I agree with you that aid conditionality has been a failure (and am skeptical too of the latest generation of performance-based aid modalities) but this doesn’t mean aid has no positive incentives at all on developing country governance. On the contrary, I think aid (and donor activity more broadly) creates a whole panoply of incentives for developing countries, some of which can be extremely valuable.

    I would challenge both your arguments for why donors cannot create effective incentives.

    Your first argument (that donors don’t make credible threats) implicitly assumes that good policy is against recipient governments’ interests; why else would incentives need to involve some sort of punishment to be effective? This relies on a crude caricature of developing country governments. As you yourself say later, “other things being equal, most ministers and officials want to see their fellow citizens better off.” What donors have to recognize is that those same ministers and officials have other interests too and the good results they seek are not always easy to bring about. But this doesn’t mean donors need to employ punishments to have a positive incentive effect.

    Your second argument (that aid cannot make big enough incentives to influence behavior) is contrary to the experience of many people who have worked at the interface of developing country governments and donors and seen the powerful, positive incentives aid can have. Aid can buy donors regular access to policymakers from which donors can exert positive incentives in various ways: by reminding policymakers of their own pledges, by equipping them with the technical case to defend good policies, or simply by appealing to policymakers’ good faith.

    I think donors should aspire to influencing policy reform, but adopting a modest approach, in recognition that donors’ behavior may affect recipient governments in a myriad different ways, not all of which can be understood. This is especially true when one considers the complexity of the decision-making process and the multitude of different actors – including far too many donors! Experience suggests that when donors attempt to influence policy with excessive hubris, their efforts usually backfire. You’re right that creating effective incentives is a complex task; the growing body of research on the political economy of policy reform should help shed light on this.

    Owen replies: Laurence, thanks.

    On the first point, I think you are interpreting the term “credible threat” too literally. The point is not about punishments vs rewards, but inconsistency of donor behaviour. If donors do not keep their positive promises, or carry out their threats, there is no incentive effect. You can’t erase this point by saying that not all incentives are punishments, though that’s certainly true.

    On your second point, I myself have worked many times at the interface of donors and governments, in many contexts, and I don’t agree with your observation. You often hear donor staff in country justifying their existence with arguments like the ones you give (reminding policymakers of their own pledges, equipping them with the technical case to defend good policies, appealing to their good faith) and some others (backing reformers, building confidence). But all the empirical evidence, including 40 years of careful evaluations of technical assistance programmes which make all these claims, show that it is self delusion. It doesn’t work for the reasons I give in my post: when the policy meets any kind of resistance, none of these effects is strong enough.

    But I agree with you completely about the need to avoid excessive (or indeed any) hubris.


  7. Pingback: Magnets all the way up « Waylaid Dialectic

  8. Pingback: Sunday Africa Blog Roundup: Mali, Sudan, Somalia, and African Football « Sahel Blog

  9. Pingback: Cash on Delivery & the Last Mile « Bottom Up Thinking

  10. Pingback: Incentives, results and bureaucracy in aid

  11. Pingback: A New Approach to Aid: Cash on Delivery | Development Policy blog

  12. Pingback: Isomorphic mimicry | Development Policy blog

  13. Pingback: Magnets all the way up « Waylaid Dialectic

Leave a Reply

Your email address will not be published. Required fields are marked *