What is the impact of aid on overall health spending?

April 23, 2010

     By Duncan Green     

makes aid complicated. Where
does the money go?

The Lancet has put the cat among the aid pigeons with its recent piece on the arcane, but important issue of ‘aid fungibility’. This claims that for every $1 given in health aid, the recipient government shifts between 43 cents and $1.14 of their own spending to other priorities. (If the aid goes to NGOs, by contrast, government health spending appears to increase.)

The Lancet paper falls into a familiar pattern. Number crunchers assemble a data set, however flaky (eg these guys had to impute nearly half the data points for low income countries, because the data were missing). Then they crunch it to find some correlations between A and B. They then argue that A causes B, but being number crunchers, give only passing attention, if any, to why/how that causation might occur. They then lob it into the public arena and stand back. A similar process followed publication of a paper by Xavier Sala-i-Martin  arguing that poverty was falling fast in Africa, and of course, more or less everything ever written by Paul Collier.

Frequently, the ensuing debate doesn’t reflect very well on anyone involved, with people imputing each other’s methods, motives (‘the Gates Foundation/World Bank/some other hate figure funded it, so you can’t possibly believe it!’) and belittling their academic credentials (or entire discipline). 

This time around, however, there have been some interesting exchanges. Owen Barder accepts the results at face value and thinks that actually, they could be a sign of progress- ‘yes, that’s what country ownership looks like’.  If we believe that effective states are essential to development, then surely they should be encouraged to take the difficult decisions on resource allocations.

Well yes, but the health advocates don’t like that idea much. As Laura Freschi on Aid Watch remarks:

‘The problem is that aid agencies have long used the argument that earmarking aid for a specific project or sector is a credible way to force recalcitrant recipient country priorities into line with donor priorities—to coerce bad governments into making good decisions.

If governments that don’t prioritize their people’s welfare respond to an influx of aid money by simply shifting their existing resources around to circumvent donor priorities (and we don’t know what is happening to the resources shifted away from health—they could be going to private jets and presidential palaces, or to education, infrastructure, or loan repayments, or really anything at all), then the aid agency argument for project aid falls apart.’

causation v correlation cartoonOthers question the quality of the research on several important methodological aspects (see Aid Thoughts for a discussion). Some think it has confounded correlation with causality.  As David Roodman at CGD argues “It is really hard for cross-country statistical analyses to prove what causes what. Maybe aid is indeed reducing spending by the receiving governments. Maybe governments with lower domestic health spending attract more health aid (reverse causality). Maybe third factors simultaneously affect both variables.” Some point out that given the well-known volatility of aid, it is hardly surprising if governments seek to smooth out health spending when they get an aid windfall, rather than use it to train loads of new medical staff only for the money to run out in a couple of years’ time.

Other critics point towards a usual suspect that is strikingly absent from the Lancet paper – the IMF. They argue that the IMF has historically pursued reduced public spending, increased privatization (including of health), and wage and other budget ceilings.  More particularly, it has argued that increased donor funding, even for health can be inflationary. I don’t quite get the logic of this– that argument would work if for every dollar in aid to health, the government reduced other spending, whether on health or other things, but it doesn’t explain the diversion of money between ministries.

As usual, I end up stuck somewhere between the various camps. Let’s assume the findings are fairly robust (I haven’t the stats to check, although the authors themselves admit there are huge holes in the data). Surely it is not right for donors to say ‘we are going to pump PIIS0140673610602334.gr3.lrgmoney into your health service, and you’re not allowed to spend a single dollar of your existing health spending on anything else, ever’. That does not build effective states. On the other hand, it does matter how much is spent elsewhere and on what (water and education are one thing, presidential palaces quite another).  

A compromise could be for donors and governments (who often negotiate an overall agreement on health spending anyway) to agree a floor or overall target for government spending (eg African governments have publicly committed to spending 15% of their budgets no health – why not stick to that?). Malawi offers a possible model – its SWAp agreement with donors had a clause specifying that the government had to increase its own health spending, and its health spending has rocketed compared to its neighbours, (see map – Malawi is the blue one on the bottom right hand side, surrounded by red. Blue indicates the highest rate of growth in health spending).

Above that level, governments would be free to consider rival demands on scarce resources and make the best decisions possible. Do aid donors really think they are better placed to decide between, say, clean water and new clinics?

The findings that funding to NGOs is associated with extra government health spending definitely deserves study – is it because NGOs invest in things like community health organizations which then generate demand on public health systems? Or because NGOs flock in greatest numbers to countries where health is already a government priority? Or less positively (as the authors speculate), because NGOs inflate local wages and prices by poaching staff and increasing demand for drugs and other materials? And before NGO readers declare victory, please remember what happens when too much health spending is channeled through NGOs – a health system that is chaotic, fragmented, and frequently ignores the poorest. I remember talking to doctors in Bolivia a while back, when 600 different healthcare providers were in charge of the country’s health system– no-one knew what on earth was going on.

Update: Aid Watch has another critique of the methodology and ‘mirages of fungibility’ by David Roodman, plus more links

April 23, 2010
Duncan Green