[Owen Abroad] Is budget support less fungible than project aid?

Is budget support less fungible than project aid? 5 May, 2011 Something about which I am not particularly bothered is the possibility that, if we give aid to a country to contribute to its school system or health clinics, the effect may be that the government chooses to spend less of its own money on these services, and instead to spend more of its money on something else which did not catch our eye, such as infrastructure or even defence. (This is a concern usually known in development circles as fungibility of aid - though see the pedantic footnote.) One reason that 'fungibility' doesn't bother me much is that I think that it is a good thing for a government to be able to choose (and to be accountable for) its own spending priorities. To govern is to choose, so if we want to encourage the emergence of capable, accountable and responsive states we should think twice before trying to limit the ability of a government to make choices over its spending priorities. I also think there is a reasonable chance that they are better informed about their priorities than we are. But suppose you don't agree with me, and you think that donors should try to ensure that when they give aid, the overall effect is to increase spending on the thing for which they have given the aid. You have broadly two choices. Option 1: you can give your aid as a project. You use your own procurement system, pay the contractor directly and check that the clinic has been built, or that the teacher has been trained. So you know that your money has built the school. (Right?) Option 2: you can give your aid through the government, either through the finance ministry or through the relevant line ministry. They pay for the work to be done, and you can still check that a clinic has been built. It may look at first like a no-brainer: the project seems to give you more certainty that your money has been used as you intended. But that superficial view is pretty much exactly wrong. There may be good reasons for preferring project aid in particular circumstances, but the 'fungibility' of budget support is not among them. Under option 1 the government is still able to make its own budget allocations. So if you build a clinic the government can, if it wishes, reduce health spending to offset your project spending dollar for dollar. Now it can spend that money on defence, or the President's palace, or any number of other things. When you ask for assurances from the finance ministry that it will not spend the budget on line items of which you disapprove, it will probably point out that none of your money is involved and that you should mind your own business. There is no way to compare actual health spending with what it might have been without your project, so you can never determine what impact you have had on the government's budget allocations. Under option 2, by contrast, the government budget is at least partly your business. You might employ staff to understand the budget allocations and to discuss them with the finance ministry and with line ministries. Before making your grant you seek assurances about the future trajectory of spending you think is important. You might obtain assurances that spending on social services will continue to rise from one year to the next, if that is what you think should happen. If the government decides to spend money in ways you think unwise, or worse, you have some standing to have a conversation with the finance ministry and to ask it to reconsider, or even to make your aid conditional on the overall budget allocations (for example, donors attempted to constrain the growth of Uganda's military spending in this way). Here is an example of how our normal assumptions about fungibility can mislead us. A USAID assessment of Mozambique alleges that, "more than $100 million of donor funds were used in 2001 to bail out the failed privatization of the Commercial Bank of Mozambique (BCM)". Let's think about what this means. In what sense were 'donor funds' used to bail out BCM, rather than the government's own revenues? This assertion seems to require us to speculate that if the donors had not given aid to the government, the government would not have bailed out BCM. But is there any reason for thinking that if there had been no aid at all, the government would have felt obliged to spend its own resources on health and education first, instead of the bank bail out? Or would it have bailed out the bank just the same and provided fewer schools and clinics? And if all the donors had run health and education projects themselves, instead of giving aid to the government, what would the government have done with its budget savings in these areas? Would it not have used the money for the bank bail out? The (explicit) claim that donor money was used for the bail out, and the (implicit) claim that this could happen because the aid was provided through the government budget rather than as stand alone projects, both seem doubtful. The bank bail out would probably have happened anyway: in which case the effect of aid was that there was more provision of health and education services than there would otherwise have been. If so, then in ordinary language we would say that donor funds were used for heath and education (not for the bail out) because that's the difference the aid made. Furthermore, in this particular case, the fact that the donors were mainly giving budget support almost certainly resulted in higher spending on social services than if they had been giving only project aid. The donors giving budget support had robust discussion with the government of Mozambique about its plans to bail out the banks, and thereby perhaps limited the resources used for the bail out. If the donors had been giving all their aid through projects, they probably would have not been able to have that conversation at all. If budget support gives the government less room to reallocate its spending because donors have more influence, this suggests that aid provided through the budget is less 'fungible' than aid provided as projects. Whether or not you agree with these judgments about what might have happened in Mozambique, the example shows that our talk about 'fungibility' is somewhere between meaningless and irrelevant. We ought to be concerned about whether the clinics or schools got built. We can observe these outputs equally well whether we give project aid or budget support. We get into a mess when we start to speculate about what would have happened in an alternative universe in which we did not give aid, or in which we gave aid in some other way, which is what concerns about 'fungibility' invite us to do. We can't predict whether those outputs would have been built without us, irrespective of whether we give project aid or budget support. Nor can we predict what would have happened to the rest of government spending whichever way we choose to give our aid. Any statement about 'fungibility' requires us to compare the real world with some hypothetical world in which we did not give aid or in which we gave aid in some other way. Whatever we do to improve public financial management in the real world does not solve the problem that we can't audit the hypothetical world. Even if we had perfect information about the real world - which is what we try to approximate by running our own projects - that wouldn't tell us more about the alternative world, so it wouldn't reduce the uncertainty about 'fungibility'. If you are concerned about fungibility, greater use of stand alone projects is not a rational response. If we are concerned about how the rest of government spending is allocated, the best way to have some influence is to give some of our aid through the government. Then at least donors can have a conversation with the government. (This isn't merely a theoretical point. It actually happens. The US tends to be less influential over government budgets and public financial management in developing countries than other donors because it normally provides project aid.) There are countries in which public financial management systems are incomplete and weak. In those environments we are inclined to be especially careful that the aid we give is used for the purposes we intend. It is common to assume that providing aid as a project makes it less fungible, and so less susceptible to being used to finance spending which is not consistent with our priorities. But this is nonsense. Governments can use the fiscal space created by aid equally well whether you give aid as a project or through government systems; when you give project aid you shut your eyes to the problem, but it doesn't go away. Pedants' corner Though this issue is usually known as fungibility of aid, to be pedantic the correct label is liquidity of aid since the concern is that aid may substitute for a different asset - namely domestic revenues - rather than another unit of the same asset. See the post online, read comments by others, or add your comments. Unsubscribe | Subscribe | RSS