We should be Scared, but not of what we’re Scared of

Be scared of him, not the Governator.

You should be afraid, just not of what you were afraid of at first...

Amidst the brouhaha over the Lancet article on aid fungibility, there’s one underlying question that seems not to have been addressed explicitly: what is the basis on which we expect funding (aid and domestic resources) to be allocated? I work in this field, and this basic problem has only been addressed with any quality in one place.

The comment on this topic is of sharply variable quality. Most of the voices critical of this evident of ‘fungibility’ deserve to be dismissed out of hand. Many suggest that by reducing own-funding to health the Governments concerned are committing a grave sin, based on absolutely no evidence on what the marginal return of the money moved was in the health sector compared to other sectors. These are single issue activists who lack the will or capacity to think more broadly than their specialism, and as Owen Barder has said, we should scorn them. Two contributions raise specific issues about resource allocation that we should explore further.

Not surprisingly, Owen’s is one of them, and his is the only contribution I would class as unreservedly useful, indeed excellent. He makes the most important point that we must hold in mind: the Lancet article actually does not address fungibility of aid. Aid would be fungible if the exact aid dollars that were earmarked for health could be used for education, transport or even private jets. What the Lancet article shows is that aid money in a sector can free up resources that the Government was always able to spend anywhere it wanted in a new area. That money was already fungible – it has simply been moved from one place to another in response to non-fungible funds. This is of crucial importance to this debate, because it means we can dismiss the question of fungibility altogether. What we are really talking about is resource allocation procedures governing the always-fungible Government resources.

Laura Freschi at Aid Watch makes the other contribution we should pay attention to, though I’m not uncritical of it. She says that donors use project support and earmarked funding to try and ‘force’ recipient Governments to use their resources ‘well’ by which she means ‘as the donors think they should’. This assumes that Governments will not change their own fungible resource allocation after the introduction of new, non-fungible resources by donors, so all aid money is purely additional to the sector it appears in. She says that if it doesn’t do this, then the argument for using earmarked funds disappears. This contribution is important, because it now introduces into the discussion the intentions of donor resource allocation and structure decisions, though I argue below this is an incomplete understanding of how it actually works.

So, how should resources for development be allocated, and how does the reality depart from this norm? Ideally, our resource allocation procedure would be entirely rational. Imagine a world in which no distinction is made between aid and local funds, and all money is fungible, the resource allocation procedure of Government should be straightforward. Looking at all the funds the Government has, it allocates a certain amount to core running costs (salaries, electricity bills and such) and then distributes the balance based on an analysis of the key constraints to the development of the country. If they have problems in health, education, infrastructure and private sector development, the rational resource allocation procedure would then address the constraining factors in each area. Given resources are scarce, we allocate them by looking at where the marginal benefit of each dollar is highest. If after we’ve spent $20 in health, we find that the marginal benefit in education is higher, we switch our attention there, and so on.

What we don’t do is just pour money in one sector until all the problems it faces disappear: doing this is counterintuitive, since after a point, each dollar would have a bigger impact in a different sector.

Now let’s complicate the picture with our first slice of reality. The donors provide some new funds, but they earmark them, so they are non-fungible. Say, originally, we had $100, and we gave $25 to each of the sectors I listed above. Now a donor comes in and gives us $25 that must be spent in health. Our decision of how to allocate the $100 has changed. We have $125, of which $25 is pre-committed to health. If our constraints and marginal benefit analysis tells us that marginal benefits are maximised by having equal funding in each area, then the new best approach is to use the $25 from the donor, and then increase this by $6.25 until we have $31.25 in this sector. Meanwhile, from our original $100 that was always fungible, we allocate $31.25 to each of the other three sectors. As such, we have now maximised the benefit of our spending by reducing our own health allocation, but increasing it overall.

Such is our ideal world. In reality, though neither Governments nor donors behave in this way. Take Governments first. Governments are political machines, and the incentives governing them are political. They use resources to strengthen patronage networks through the differential allocation of development projects, and even simple handouts. They also choose politically prestigious projects, such as fancy new buildings, roads or subsidies. Finally, they have a genuinely developmental incentive, since a better economy and country wins votes, too, though questions of ideology and politics can determine what kind of ‘better’ country this means.

These three incentives work in different ways, and mean that they don’t simply allocate resources according to the best marginal benefit. Instead, they pursue a number of objectives simultaneously, and this lends budgeting and resource allocation a chaotic feel. I’ve interacted with budgets in several African countries, and don’t think in any is it a fully or even largely rational allocation process. It’s very much a balancing act, with the real decisions taken far away from the technocrats (I’m not saying it’s a world different in the West, by the way).

This complicates things greatly: we are dealing with a mix of good intentions, bad ones and mixed ones. But it gets even more complicated when the donors’ own mixed up incentives. We saw that Freschi argued that earmarking was an attempt to fix Government allocations, one that was failing. This is only a small part of the story. In fact, earmarking has probably got more to do with the areas where a donor country office finds it acceptable to spend vis-à-vis its paymasters, be they Parliament, electorate or a Board. Most bilaterals spend in Health, HIV/AIDS, Water and Sanitation, and Education because these are easy sells – far more so than the unromantic realms of tax collection, private sector development and infrastructure development.

What’s more, when donors earmark they also don’t generally expect to peg the Governments’ domestic expenditures. When they do, they make this explicit. In Malawi, for example, under the conditions of the Health Sector Wide Approach the Government were compelled to maintain real funding levels to the health sector or donor funding would be reduced. This is easy enough to apply. If it isn’t, earmarking is probably motivated by a desire on the part of donors to protect a minimum expenditure in a given sector such as health (the level of their earmarked aid) rather than to influence Government resource allocation decisions – i.e. it assumes some level of reallocation of local resources.

As this makes clear, the donor’s incentives are not clean and simple to unbundled either: they gravitate towards certain sectors because they are more popular, and they design their aid modalities package based on a range of assumptions of Government behaviour.

What this leaves us with is a far knottier and scarier problem than fungibility of aid. It’s ultimately a question of the incentives facing both donor and Government agencies. These incentives are complicated, working in many ways at the same time, with results that are sometimes positive and sometimes negative. Solving this has nothing to do with fungibility questions. We need to improve resource allocation decisions by Governments by helping realign their incentives. Transparency and financial management is part of this, but it goes much deeper: the political settlements in most countries are a major part of the problem. Electorates captured in a web of patronage won’t want a more rational resource allocation procedure, they will wish to continue to reap the rewards they get. The ones in opposition search not for fairness, but their time in the sun. This is what needs to change.

On the donor side, marketization of aid may improve much, but ultimately, the decisions that lie behind developed country Governments giving aid at all are political. Not all uses of this money are equal, because of electoral pressures and political objectives. A better educated electorate and a clearer demarcation between aid and realpolitik-based state-building will go further in addressing these unhappy incentives.

What’s good about this whole controversy is that we can go beyond the shallow thinking that characterised many first responses and start thinking about what really needs change.

13 thoughts on “We should be Scared, but not of what we’re Scared of

  1. Justin Kraus

    April 15, 2010 at 2:06pm

    Two things frighten me about those who defend government reallocations away from sectors that receive donor funding.
    1. I am skeptical that even an “ideal” rational marginal benefit analysis can exist. Sounds pretty “planner-ish” to me. But even if it can, as you point out, politics (good, bad, and otherwise) generally runs the show anyway. So why should I believe people who justify such allocations in the name of “rational” marginal benefit analysis?
    2. I have a moral beef, please dissuade me of it. If I give you money to improve a project because you need help and because of that you remove some of your own money to finance something else, my gut instinct is to be rather miffed, why shouldn’t I be?

  2. Ranil Dissanayake

    April 15, 2010 at 3:02pm

    Justin – looking at these in turn:

    1) it’s not planner-ish, and even if it was, that wouldn’t bother me. Earmarking donor funds is far more planner-ish – if everything was unearmarked, and left for the market to allocate between sectors, the same result would obtain.

    Secondly, I never justified the reallocation based on marginal benefit analysis. I pointed out that this is the ideal, and that it doesn’t happen for either donors or for Governments. Instead of worrying about fungibility, since it’s going to happen (if you don’t want it to affect government resources now, cut aid to 0. Otherwise it will happen. No way around this, none at all – this is just realism), we should be addressing the incentives underlying both donors and governments that take them away from the ideal. This will mitigate the issue in the long run.

    2) If I have $8 for health, and need $12, and you give me $12, why should your over-generosity mean I have to spend it all where I don’t need it. What if I need $8 for education, too?

    Government’s reallocating suggests that either

    a) they’re using money for stuff they shouldn’t use money for (in which case we should be advocating against this regardless of any aid provided – they’re either stealing money or buying frivolities with it). This has nothing to do with aid money. They will do this with or without aid money and the problem is to do with their incentives – why they misuse funds wilfully.

    or b) we’ve given them more than they need in an area, so they want to use it elsewhere. I have no problem with this. If I give a friend a birthday present they don’t like I’m not the type who is bothered by them returning it and getting something new. I normally give them ‘gift receipts’ to allow them to do so if its possible. And I certainly won’t take offence if aid I provide through tax or whatever gets used to help people in a different way to how I originally envisaged it. This is all about the flexibility to search for the best solution. If you use easterly’s planners/searchers thing, then earmarking and getting miffed about aid being used for something else is stifling the searching.

    The challenge is to tell the circumstance apart, whcih is why we need good audit etc.

  3. Justin Kraus

    April 15, 2010 at 3:28pm

    1. Maybe.
    2. a. I agree. b. why are we being overgenerous? That sounds like a bad idea. To use your example, I would hope that before I gave you $12 I would have asked how much you actually need.

  4. Ranil Dissanayake

    April 15, 2010 at 3:36pm

    Unfortunately, that’s exactly the problem: donors and Government don’t communicate all that well, and certainly NGOs and donors don’t either (at a central level). So loads of donors pile in to fix one problem and step on each other and wind up giving more than needed, which is why Government needs to be stronger at coordinating them.

    On the other hand, if donors are adamant that they want to give in Health (and many donors are, which is why some sectors are so massively funded) then the Government’s incentives are clear: accept more than we need, and reallocate our own funding to service those areas where donors are unwilling to fund.

  5. Justin Kraus

    April 15, 2010 at 3:49pm

    I agree. But, not to be moralistic or naive, if I did give you $12 and it turned out you only needed $8 and you didn’t give me the $4 back, I’d be miffed. Why don’t governments do the same? Also, if you did give me back the $4, I’d be much more willing to help you out the next time you needed money. But maybe thats just me. Thanks for the feedback.

  6. Sam Gardner

    April 15, 2010 at 10:05pm

    You are very much right on most of the issues, but one aspect is out of bounds: The allocation of resources over a government budget is NOT an economical but a political process. How much art should a country support compared to sports and education? There is only a political answer to this. However, political processes never lead to optimal results. Technocracy, populism, most -isms, lead to suboptimal resource allocation, especially for the poor. As we all now, democracy is the least bad option for leading this resource allocation process.
    Another area that should be highlighted is that if both partners respect each other and there is depth of commitment from the donor, there is no development problem by a donor earmarking aid to a sector and giving their full support in that area (advocacy, capacity building, attention to progress, etc.), and the government reallocating according to their needs. There is only a public relations issue. The added value of donor agencies is not only in the transfer of money, but also in the commitment and support on top of the money.

  7. tulip

    April 19, 2010 at 5:03pm

    Aid would be fungible if the exact aid dollars that were earmarked for health could be used for education, transport or even private jets. What the Lancet article shows is that aid money in a sector can free up resources that the Government was always able to spend anywhere it wanted in a new area. That money was already fungible – it has simply been moved from one place to another in response to non-fungible funds. This is of crucial importance to this debate, because it means we can dismiss the question of fungibility altogether. What we are really talking about is resource allocation procedures governing the always-fungible Government resources.

    umm … I don’t want to descend into semantics, but I don’t think aid fungibility means the exact dollars given for one thing are spent on the other, the whole idea is simply that you give say $100m with the intention of it being spent on health, but spending on health does not rise, relative to the counterfactual, by $100m. Of course this works by aid “freeing up” money the government was always able to spend as it liked, as you put it, to be spent elsewhere, but that is what fungibility means.

    Exact earmarked dollars has nothing to do with it

  8. Ranil Dissanayake

    April 20, 2010 at 7:02am

    Sorry Tulip, I think you’re wrong on this one. Fungibility of *aid* must refer to the aid money itself, since donors have and should have no direct control over how Governments spend their own funds. They do, however, have control over their own funds.

    These definitions are quite standard for those working in aid effectiveness. This is why we describe general budget support as fungible aid, sector budget support as semi-fungible aid, and project support as non-fungible aid.

    Your definitions would imply that there is no such thing as non-fungible aid, and the Paris and Accra conferences wasted a significant amount of time talking about fungibility for no reason.

  9. Luis Enrique

    April 23, 2010 at 11:49am


    I’m with Tulip. For the sake of argument, imagine somebody carefully kept a record of all the serial numbers on the dollar bills generated by domestic revenues and the dollars donated by aid donors intended to be spent on certain things, so we could see if aid was fungible in your terms. But imagine somebody gets into the vaults and mixes up the dollars! Do you really think that, holding expenditures constant, changing exactly which dollars are spent on what would amount to changing the fungibility of aid?

    I think if you read all the lit. on fungibility (starting with Feyzioglu et al and working forward) you will see that “standard definitions” are about counterfactuals. If a recipient government would have spent $100m on health in the absence of aid, and is given an additional $100m to be spent on health, if actually aid spending turns out to be only $150m because the govt cut its own revenue allocations, then aid is said to be fungible regardless of “exactly which dollars” make up that $150m. The point is, you give $100m for health, but health spending only increases by $50m. Think about “treatment” and response. Read the Roodman blog post about this Lancet study, it’s quite clear this is how he sees fungibility.

    You are quite right that this definition means that in theory there is no such thing as non-fungible aid. Check the fiscal response literature (Morrissey at al). And in practice there is no such thing as non-fungible aid either unless donors have some way of constraining the expenditure choice of recipients. This is why the result “aid is fungible” is so unsurprising … of course it is.

    It is an empirical question to what extent in practice aid is fungible, because it’s an empirical question what constraints donors are able to impose on recipient budgeting. So researchers and conference delegates are not wasting their time discussing an empirical question, nor by discussing questions like whether we should care about fungibility, or what measures could be used to constrain fungibility.

  10. Luis Enrique

    April 23, 2010 at 12:06pm

    Actually, the question is often ill-posed and I think I have repeated the error in my “treatment – response” analogy, above.

    If you give somebody £10 to spend on food, but their total spending on food only rises by £5, this does not mean that you have found, empirically, that money is “only 50% fungible”. Money is 100% fungible, and the recipient can spend your money anyway they choose, even if you march them into a supermarket and watch them to buy £10 worth of food, they can simply reduce their own food spending, and in this example it just so happens they choose to increase their total food spending by £5. Money only becomes less than 100% fungible if you somehow manage to impose some constraints on not only how your money is spent, but on how the recipients own money is spent.

    I think fungibility is not measured by asking “we gave $X to be spent on Y, by how many $ did spending on Y rise?”, but by what extent actual spending deviates from what spending would have looked like if $X was given to be spent in any way the recipient wishes, because of course the recipient may wish to spent some of the $X on Y, anyway. This is why it’s such a pig of an econometric problem, and why economists expend largely unconvincing efforts trying to construct theoretical counterfactuals in order to get a proper measure of fungibility.

  11. insomnia cures

    May 23, 2010 at 6:29pm

    Super-Duper site! I am loving it!! Will come back again – taking you feeds also, Appreciation.

Comments are closed.