Wanted: a mechanism for additionality.
Additionality is the thorn in the side of Development Finance Institutions (DFIs). It means: making an investment happen that would not have otherwise. The problem is that DFIs cannot provide rigorous evidence that their investments are additional.
Rigorous quantitative evidence based on investment data requires some credible method of estimating the counterfactual (what would have happened otherwise). And that requires something like a randomised control trial, or a ‘natural experiment’, or that mythical beast of econometrics a ‘valid instrumental variable’. And, to my knowledge at least, none of that is feasible in the world of DFIs. And without such tricks, we are unable to distinguish correlation from causation.
DFIs have no shortage of anecdotal evidence. They can all talk convincingly about projects where their support was vital. But correlations and anecdotes are not admitted as “rigorous evidence” by academics, who then report that evidence does not really exist, leading civil society to conclude that there is very little evidence of additionality. And so DFIs are continually haunted by their inability to prove they actually do what they say they do.
DFIs can hope their critics will become more realistic about what counts as evidence, but there is another possibility: rather than trying to demonstrate additionality after the fact, DFIs could adopt procedures to ensure additionality before the deals are done.
One of the fundamental reasons why DFIs struggle to convince sceptics is that when they negotiate, project sponsors (entrepreneurs) will not show their hand: whether they have willing private backers in the background is private information. To the extent that DFIs offer terms and conditions that are more favourable than private investors would, then the project sponsor has an incentive to pretend that the project will not be viable without the help of the DFI. If you add into the mix the possibility that DFI investment officers have an incentive to play along with this pretence, because they get rewarded for deal flow, it is easy to see where worries about crowding-out come from.
This is a live policy problem. Donors are pouring money into their DFIs, and the degree of concessionality is widely expected to rise. Industry insiders are watching nervously to see how the World Bank’s IFC will use the new $2.5bn private sector window in IDA18.
The good news is that whole field of economics — known as mechanism design — deals with problems of private information and how to overcome them. The bad news is that it is not obvious how these insights could be applied by DFIs. ODI has been working with the EBRD and leading mechanism scholar Professor Francesco Decarolis to explore the application of mechanism design to the problems faced by DFIs (and we will publish the resulting paper shortly) but preliminary results reveal two basic problems.
First, the investment business is dominated by decentralized bilateral bargaining, not formal competition between financiers. That limits the leverage of mechanism design to induce project sponsors to reveal their private information (whether they genuinely need public support).
Some sort of auction is the typical mechanism for inducing economic actors to compete against each other and, in so doing, reveal private information. In principle, we could ask DFIs to run something like a subsidy minimisation auction. Then if the auction reveals that the project could be financed at market rates (and we assume that DFIs are more onerous to deal with because of their higher environmental, social and governance standards) we can expect entrepreneurs with access to private finance to choose it, thus ensuring DFIs do not crowd-out private finance. And if the auction reveals that some form of concessional finance is required, it will be minimised, thus ensuring DFIs do not needlessly subsidise private profits. But no entrepreneur is going to voluntarily enter an auction for project finance designed to give them financing on the least favourable possible terms. So DFIs would need to enforce participation as a condition of accessing their finance. Project sponsors would not like that, and DFIs might find it hard to agree to, especially if the auction design is too crude, and incapable of capturing non-monetary dimensions of project finance that cannot easily be expressed as a price, which DFIs and entrepreneurs regard as important.
Alternatively, if a formal competitive process is not feasible, DFIs may still be able to set their terms and conditions so that projects that do not genuinely require public support voluntarily go elsewhere — this is known as ‘screening’ by mechanism designers. But — and here’s the second problem — that would probably require DFIs to offer less attractive terms than commercial banks and other private investors, and that is hard to reconcile with most DFIs’ mandates to promote investment. It would also require DFIs to know what ‘market rates’ look like, which may not be possible when every project is unique and capital markets are undeveloped.
But we haven’t given up hope. Mechanism design is one of the most active areas of research — and it is possible that innovative thinking could yet generate ideas that DFIs could use to solve the additionality problem. Perhaps, for example, DFI investments could be structured to bring expected returns up to market thresholds when returns are genuinely too low to be commercially viable, and hence crowd-in private investors, but also cap returns so that those project sponsors who secretly know the pay-off is likely to be larger than they are letting on will voluntarily go elsewhere.
So, whilst we finalise our initial foray into this field, we want to say: calling all mechanism designers. Get your thinking caps on!
