When Should Donors Mitigate Private-Sector Risk?

In the last two weeks, three large bilateral donors, a foundation and an international organisation have all approached me about their plans to use loan guarantees and other debt and equity instruments, alongside grants, to encourage more private investment in developing countries or in global public goods. Togther with the recent announcement of the Global Health Investment Fund, about which I have considerable doubts, this got me wondering when it is a good idea for donors to subsidise private sector risks.

The objective of supporting private investment is often sound; but I've managed to convince myself that if you are going to provide a public subsidy to the private sector it is nearly always better to amplify the returns than to reduce the risks. I suspect aid agencies are being pushed into suboptimal financial instruments by a combination of myopia and dodgy accounting rules rather than good economic analysis.

So I sent an email round some of my CGD colleagues - generally a good way to find out quickly if I'm talking rubbish. That triggered a lively internal discussion. Here is where we've got to so far.

There are many good examples of where it makes sense for an aid agency to provide a subsidy to the private sector for something which otherwise does not make commercial sense, such as developing medicines for developing countries or paying for a coffee processing plant. Since the problem for the private sector is that the risk and return do not add up to a viable investment, donors either have to mitigate the risk or increase the return, or both. The questions is, which is better?

In a hypothetical efficient market, an investor ought to be indifferent between an increase in return and a reduction in risk. And in theory, the expected cost to donors is the same either way. That suggests it doesn't matter which we do, and a loan guarantee ought to be just as good as a subsidizing the return, for example through an Advance Market Commitments for new products or Development Impact Bonds for the delivery of services,

But in the real world donors should usually increase the returns to the private sector rather than reduce the commercial risk. The intuition behind this is straightforward: the market failure is nearly always that the private returns are less than the social returns, not that the investor perceives too great a risk. So the optimal solution is to increase the returns, not to try to reduce the risk.  There are at least five reasons for this:

1. Moral hazard – these guarantees don’t just change the cost of risk, they change the risk profile and that distorts how risks are managed. We involve the private sector to manage risks and innovation, so it is weird to add a distortionary guarantee which insulates them from the risk we want them to manage.
2. Deadweight cost – it is much harder to to calibrate loan guarantees to the desired social outcomes whereas topping up the returns can be much more selective and precise.
3. Investor selection - different types of investors choose high-risk-high-return projects from those who choose low-risk-low-return projects, and it is likely to be in the public interest to attract more of the former into these socially advantageous investments.
4. Contestability - a subsidy which increases returns when the outcomes are achieved need not require the subsidy provider to pick winners in advance and so it can unleash competition among investors and firms to produce the goods and services; whereas a loan guarantee has to be provided in advance so somebody has to choose who gets it.
5. Reputational benefits - a loan guarantee hits the aid budget when it goes wrong, whereas a subsidy for returns scores when things go well. I know which spending I'd rather defend to Parliament.

These arguments are all set out in a bit more detail in a note I am writing in response to an informal request from one of the organisations which is thinking about this. The note will also explain why aid agencies might tend to make the wrong choice. Here is the current draft - it is a Google doc, but I don't think you need a Google login to access it. I'm working on this now (which is why I want feedback) so don't be alarmed if it is changing before your very eyes. You might prefer to wait for the final, more polished version which we will publish in due course. These are all thoughts-in-progress, so I'd welcome any comments or reactions, either to this blog post or to the draft note so far.