The sector-wide approach (SWAp) has been a prominent way to operationalise principles for aid effectiveness. A health SWAp is a formal agreement between an aid-recipient government and its donors. The agreement commits both to a government-led sector-wide health strategy, and asks donors to harmonise processes and rely on recipient-government financial management. More than 30 aid-recipient countries have entered into health SWAp agreements. Our new paper in the Journal for International Development shows that using SWAps to coordinate donors and donor contributions has paid dividends, delivering substantial improvements in population health.

This finding is important because the uptake of aid effectiveness principles remains incomplete. One recent Devpolicy blog suggests that aid fragmentation remains a problem. When aid is fragmented, the bureaucratic effort needed to manage numerous donors and their reporting requirements may not be proportionate to the funding provided. Our paper suggests that health SWAps have reduced these inefficiencies, concluding that health SWAps have likely achieved health gains by freeing up domestic resources (health expenditure and health officials’ time).

Criticisms of fragmented aid are nothing new. Buse and Walt presented a compelling critique in 1997 of the inefficiencies and harms from proliferating health aid (and proliferating donors). The Paris Declaration saw development partners agree to do better – by coordinating their contributions, reducing fragmentation, and giving greater ownership of aid management and allocation to recipient governments. Health SWAps embody these principles.

Over the past 10 years, support for SWAps and the aid effectiveness agenda has waned. Shorten et al. noted this shift – a “diluting commitment and attention to alignment and harmonisation” – at the Fourth Aid Effectiveness Forum in Busan in 2011. One important reason for this has been the lack of robust and compelling evidence demonstrating impact.

Some case studies suggest SWAps have increased health aid and made improvements in service delivery and population health. The problem with case studies though is we can’t know if these improvements occurred because of SWAp implementation, or if they would have happened anyway under business as usual. We need a counterfactual.

Measuring impact is not completely straightforward: countries that have implemented health SWAps don’t look like the ‘average’ aid recipient country. They tend to be poorer in terms of GDP per capita and have worse health outcomes than other health aid-recipient countries. A SWAp also requires buy-in from the aid-recipient government as well as their key donors, and donor support for SWAps appears to have made SWAp implementation more likely in some regions. So, the likelihood of SWAp varies across countries and it’s not appropriate to compare SWAp-implementing countries with countries that are very unlikely to ever implement a SWAp. To control for this selection bias, we identified factors that predict SWAp implementation and then made sure our control countries were well matched on characteristics like health profile (life expectancy, infant mortality, HIV), GDP per capita, population size, health aid profile (health aid levels, number of donors), government effectiveness and geographic region.

Countries that have and haven’t implemented a SWAp might still differ on important but unobservable determinants of funding flows or health outcomes, so we took the further step of controlling for baseline differences in outcomes using a difference-in-differences approach. We don’t have a perfectly controlled experiment, but these steps get us much closer to an ‘apples and apples’ comparison.

So, what do we know about the impacts of health SWAps? First, SWAp implementation can have a significant impact on aid flows. Specifically, SWAps increased the levels of aid given as ‘general sector-support’ by an average of 360% (granted, from a very small base). SWAps also facilitated reallocations of aid contributions, specifically away from HIV and maternal/child health. A plausible explanation for this was that SWAps increased recipient control over health aid and that, in the view of recipient governments, too much aid had been directed towards HIV and maternal/child health relative to other priority areas.

Second, SWAp implementation can leverage more domestic health spending. Where SWAps are in place, about 70c per dollar of health aid effectively stays in health. This is in contrast to just 10c to 30c in non-SWAp settings (see here and here). Most likely, these gains occur through contractual obligations embedded in SWAps and increased confidence in health sector financial management. This is important because donors are understandably concerned that health aid has been used as a replacement for domestic health expenditure in some settings.

Third, the effects of SWAp implementation haven’t always been positive. For the poorest implementing countries (GDP less than $1 per day in 1990), SWAp implementation led to a nearly 30% reduction in levels of health aid. These reductions may reflect a lack of confidence in the ability of poorest countries to appropriately manage less constrained health aid.

Finally, if the primary aim of health aid is to improve health, our new research suggests that the impact of SWAps has been overwhelmingly positive. SWAps have facilitated, on average, a 5-8% reduction in infant mortality, even in poorest implementing countries. Our analysis suggests a likely pathway for this effect is indeed the reducing of the significant burden that comes with managing numerous donors outside of a SWAp.

So where to from here? In recent years, a lack of compelling evidence of impact has reduced donor interest in health SWAps and eroded faith in the Paris Principles for aid effectiveness. Our analysis shows that, given time to mature, health SWAps can (and have) achieved significant gains across a range of important donor- and recipient-relevant outcomes.

There remains some uncertainty (hard to avoid in real world policy impact evaluations), and there’s a lot more to learn about the pathways by which SWAps have achieved some of these identified impacts. But, in an increasingly austere aid environment, these findings should give hope and confidence that efforts to engage with health SWAps appear very much worthwhile.

Read the full paper ‘The impact of health sector-wide approaches on aid effectiveness and infant mortality’ in the Journal of International Development here.