A great idea in principle. But watch out! The devil’s in the details, at least from an equity perspective.

For starters, there’s the problem that unless the focus is on additional births, rather than on all births, the principal beneficiaries are women would would have delivered anyway. These are overwhelmingly concentrated among upper-income women almost everywhere: in the average developing country, for example, the attended delivery rate is well over twice as high among women in the top economic quintile than among women in the bottom economic quintile. (In Africa, the ratio is over three to one.) Please note that this is not simply a hypothetical problem: in both of the two known cases of delivery incentives where there are adequate records to determine just who benefited (Ghana, Nepal), the overwhelming majority of benefit accrued to upper groups.

So the first thing that’s needed is to reward additional attended deliveries, rather than all deliveries. But while necessary, this isn’t likely to be sufficient. For unless one limits the rewards to births among poor women, there’s a strong incentive for program managers to reach out to the woment who are most readily available; and these are also likely to be concentrated among better-off, more educated groups rather than among distant excluded ones.

So the second need is for a focus on additional deliveries not to all women, but to poor women. Is this feasible operationally, and can it be monitored? In each case, the response is “yes” — or at least “probably.”

Operationally, a focus on poor women is straightforward in the several Latin American countries with well-established systems for identifying poor people through a proxy means test. In such countries, rewards could be given on the basis of the number of women served who qualify in terms of that test. Elsewhere, as in Africa, where such systems do not exist and administrative systems are far weaker, identification of poor individuals might well be problematic; but a viable, less demanding alternative is available. This is simply to focus on rural women. Because poor women tend to have far more children than do better-off women, and because poor women tend to be heavily concentrated in rural areas, an effort rewarding additional deliveries among rural women would almost inevitably deliver benefits disproportionately to poor women. Also, the great majority of poor rural women in Africa — over 80% on average — attend at least one antenatal session; and such poor women usually constitute the majority of the ones coming for ANC service in a rural facility. This makes them easy to reach — through, for example, distributing vouchers for a delivery payment to all women coming in to rural factilities.

As for monitoring, a viable system already exists in the form of household surveys undertaken regularly in most poor countries through the Demographic and Health Survey (DHS) program. The reports on these surveys include quintile-specific attended delivery rates with standard errors that are adequately small for most purposes. One could use them to estimate the attended delivery rate for, say, the poorest one or poor quintiles, extrapolate to estimate the increase in the number of births in those quintiles — and those quintiles alone — during the period between the current DHS and the last one, and pay $25 for each on those. (To be sure, the DHS is usually undertaken only every five years or so in a country. If this is thought to be too long an interval, similar information is available through the UNICEF MICS surveys undertaken in many countries once every three years. Or if even shorter intervals are desired, it shouldn’t be hopelessly expensive to organize a yearly survey limited to asking only about household assets and whether the last delivery in the household was attended or not.)

Best, Dave G.

The COD approach is certainly attractive. It recognizes the host country as the final duty bearer and, in principle, holds the aid recipient accountable for a final meaningful outcome, without dictating what approaches are to be taken. It is also encouraging that there are new signals of donor interest in reducing maternal deaths, which – whether happening at a rate of 350,000 per year or 500,000 – are scandalously high, given that the overwhelming majority are in principle preventable. But choice of skilled-birth-attendant-at-delivery as the indicator for such an incentive scheme would be a big mistake. I would like to echo the point made by April Harding; a focus on where a woman delivers or the supposed qualifications of the health worker attending her tells us nothing about the actual content and quality of care given, and it is these which will determine whether or not the service is effective in reducing risk to her and her newborn. Progress in reducing maternal mortality has been impeded by incentivizing the contact between health worker and beneficiary or the location where these services are delivered rather than the contact. Use of such indicators as benchmarks of overall program performance has pushed program managers towards strategies of training thousands of ‘skilled birth attendants’, or upgrading health facilities as ‘emergency obstetrical care sites’. Neither of these approaches are necessarily bad but they are by no means the only strategies available. Furthermore, use of indicators that only measure deployment of a strategy rather than provision of the actual interventions which save women’s lives (e.g. surgery for ruptured uterus, magnesium sulphate for eclamptic seizures) pushes attention away from where it needs to be. Use of skilled-birth-attendant-at-delivery has the appeal of seeming to be very simple. But as Einstein has warned us, “Everything should be made as simple as possible, but not simpler.”

I don’t really understand why there is such a fuss about CODAID. Is it really a “new” approach?

1. Linking general budget support and results. The European Commission has developped the “variable tranche budget support” where governements receive a “tranche” of budget support when they reach a target in social sectors (i.e. skilled attendance rate, or secondary school completion rate). There is a direct link between results achieved by the country and the amount of budget support received. MOF and line minitries have – in theory – an incentive to succeed.

After 8 years of implementation, one of the lessons learnt is that incentives at governement level make insufficient change on the frontlines, at provider level (health facility or school). Governements have trouble in reforming social sectors, increase efficiency, implement efficient information systems and independent control mechanisms. Incentivising the ministry of finance does not mean schools and health facilities are incentivised.

2. In order to incentivise the providers – and ensure a change for populations – governements and donors of many african countries (DRC, Burundi, Rwanda, CAR, Cameroun, and many more are preparing similar schemes) have decided to provide cash to health facilities in direct relation with their results (quantity and quality). In the region, it’s called performance based financing or PBF. The funding is provided by an independent fundholder that transfers ressources to the facility based on its performance: for example 1$ for a fully immunized child, 8$ for a birth attended by a skilled health worker, 25$ for a cured TB patient, 50$ for a girl that achieves the primary education test, 15$ for a boy. Usualy, this funding mechanisms shows impressive results in terms of utilization of services.

In central africa, those schemes have been in place for years, and channel multi-million dollars of funding from donors and governements. In Burundi and Rwanda, they are part of the national health financing system.

based on my experience, my opinion is : (i) incentivising the governement (in fact the ministry of finance) is not sufficient, you need to ensure that the incentive goes to schools and hospitals (ii) incentivising providers means that governements have to implement specific systems for that: independent and accountable fundholders, efficient information system and audit mechanisms (split of functions). That is where the issue and difficulty is. (iii) The funding of those schemes will come both from Aid and governements. Therefore it is not really straightforward to focus on aid only (what CODAID does).