Let’s keep it strictly C.O.D.

The UK’s Conservative Party has just released a policy paper on international development. This ‘green paper’ is basically a discussion of the party’s agenda for international development – the sort of reforms they would enact if they took control of the government. Since the Conservatives are widely expected to take power sometime next year, the green paper has received a large amount of scrutiny.

One of the policies embraced in the green paper is the new(ish) aid modality Cash on Delivery, which was thought up by the folks at the Centre for Global Development a few years ago. The basic description of COD aid (yes, it already has its own acronym) can be found on their website:

Under “cash on delivery” aid, donors would commit ex ante to pay a specific amount for a specific measure of progress. In education, for example, donors could promise to pay $100 for each additional child who completes primary school and takes a standardized competency test.

A credible baseline survey would be conducted, the country would publish completion numbers and test scores, and then the donor would pay for an independent audit to verify the numbers. The payment would be made upon a successful audit. Payments would be “cash on delivery” – made only after measurable progress, only for as much as is verifiably achieved, and without prescribing the policy or means to achieve progress.

The payment for the results would then be fully fungible – the recipient government would be allowed to do anything they wanted with it (although the reality is that there will likely be some limits on this). COD aid is initiatlly being targeted at the education sector, likely because the outputs are, relative to most outcomes in this field, easier to measure. The CGD has been working on this concept for several years, writing discussion papers and concept notes – the background information can be found here, and a full-fledged FAQ section.

There are a couple of things about COD aid that I find quite promising:

  • The donors don’t get involved in policymaking; they just pay for the results = no more development training wheels!
  • There would be no attempt to tell the government what they should do with the payout, again another point for
  • It would represent a way of thinking about aid that, for a chance, is impact centered.

However, there are a number of things about the scheme I’m not as confident about – some of my concerns are purely theoretical (and so are likely wrong) – some are observational:

  • The burden of the task - Much of the literature on incentives and the public sector has come to the same conclusion: designing incentive contracts for public institutions is not easy, and most of the time low-powered incentives prevail. Part of the reason is that outputs are usually hard to measure. One would argue that COD, as it’s currently being presented, avoids this problem by very carefully measuring output. However, as Duncan Green pointed out in his recent post on the Green Paper, there are plenty of reasons school results and attendance could worsen (or get better) that are totally out of the control of the education authority. The less control they have, the greater the risk burden they carry (Nancy Birsall responded to that concern here). Agents that are forced to face too much risk might opt to just not play the game – they’ll make little or no effort to affect the outcome. While incentives might be useful in the short-run, a distant, difficult target that requires unprecedented effort might just be too much for the average ministry. For good examples of public incentive schemes falling short of their the desired impact, see Heckman’s work on the JTPA or Burgess on Jobcentre Plus.
  • A numbers game – Development is a tricky business – on one hand, we want to know that our intervention has a measurable impact.  On the other hand, we should always be concerned about turning the business into a stats game (readers familiar with The Wire will know the pitfalls of the pursuit of stats). One always worries if quality is being abandoned for the sake of quality. To be fair, CGD has repeatedly addressed this issue – their hope is also that very strict evaluation will deter attempts to game the numbers.
  • Donors are still playing with sticks and carrots – COD still carries with it that uneasy premise that still makes me wince: it is our job (as donors) to incentivise recipient governments to do the right things for their people – i.e. we know the way to salvation, if only these bloody governments would listen to us. Again, to be fair, this is no worse than the way aid has historically been handled – it’s just a bit patronising. It could be the right way to approach things – I tend to believe that we should be less concerned about getting governments to treat their people properly because we’ll give them money for it and more concerned with getting governments to treat their people properly because they have a natural, endogenous incentive to do so.

All these things said, I’m certain that the Center is just as worried about these same issues. They aren’t blindly pushing this new modality as an instant cure to the woes of ineffective aid – they’re approaching it cautiously, slowly building on the discussion year, and rolling out a pilot programme to see how successful it really can be. That’s the right approach – yet sometimes great-sounding but untested ideas can be quickly adopted and converted into policy. My worry is that the Conservative party, eager to distinguish its new development policy, will take up the idea and run with it before the Center finishes making up its mind whether it’s really a good idea or not!

3 thoughts on “Let’s keep it strictly C.O.D.

  1. Ranil Dissanayake

    July 20, 2009 at 8:06am

    Interesting analysis. I made a similar point about target chasing in your post about the health intervention (information sharing).

    One further point to note, though, is that the COD approach can still be quite distortionary in terms of policy. It’s very similar to the current Performance Assessment Framework structure for budget support payments. At present donors measure progress on about 15-40 indicators spread across a number of sectors, and based on progress against these indicators, they decide how much GBS to provide in the following financial year. Currently, about 2/3rds of donors use a ‘general assessment’ allowing for a great deal of discussion and analysis of the grey areas as well as the actual performance on the indicators before deciding how much money should disbursed the next year, while a few, notably the EU, explicitly link money to the achievement of certain indicators. This is pretty much how the COD would work, but with different kinds of indicators. The money disbursed is fully fungible, being GBS.

    Where the distortions come into play is in the balance of COD or PAF indicators. In Malawi, for e.g., until the most recent PAF which included the World Bank as a stakeholder, there was almost no mention of Trade and PSD, while Education and Health made up about 70% of the indicators.

    Developing country Government chase money – that’s an observation most of us working aid would agree with, I think – so they wind up spending a great deal of time trying to achieve the indicators (or fiddle the figures of them) that will give them the money next time out. As a result, far more money and effort is spent on areas where donors link indicators than in other areas. And unfortunately, it appears that donors use indicators in areas which are often not the Government’s central interests.

    Talk of aligning indicators to the NAtional Development Strategy is a red herring – the indicators are aligned, but the balance of them isn’t. so, donors might take all of the indicators from the NDS on Education, Health and Governance, and only 5% of the indicators on PSD and agriculture.

  2. Matt

    July 20, 2009 at 5:43pm

    Matt

    GAVI already does exactly this: there’s some debate around about the extent to which it creates perverse incentives and whether the performance payment is full compensation for all the costs it gives rise to in the health system, starting with
    doi:10.1016/S0140-6736(06)69337-9

    Matt

Comments are closed.