Humble Pie, Guaranteed

Our fragile egos notwithstanding, one of the privileges of working at CGD is learning from our mistakes. We dug into some humble pie last week when Megan Rapp, Africa regional team lead for the Development Credit Authority, set the record straight about a USAID-backed credit guarantee.

We argued in a recent paper that guarantees always have some expected cost, and that expected cost could be spent more effectively if we linked the subsidy to success rather than failure.

We pointed to a USAID-backed guarantee in Kenya as an illustration. The guarantee covers loans made by Kenya Commercial Bank to clinics to buy new medical equipment exported by GE (things like MRI machines are pricey, so it makes sense to get a loan to buy them).

We said two things about it.

It’s distortionary

Guarantees are financial seatbelts that protect the holder from the costs of getting things wrong. And just as seatbelts tend to cause riskier driving, Kenya Commercial Bank makes loans to riskier clients because USAID’s guarantee offsets those risks.

It’s difficult to “see” this because we can’t compare a lender’s book with loans it would have made without the guarantee: we don’t have a control group or counterfactual. When we do, like in a nice paper using data from German banks, we see that guarantees induce worse lending to riskier clients.

It’s tied

In this Kenyan case, it looked as if USAID was paying for a guarantee so that Kenyan clinics could buy American exports. That would effectively have been a form of tied aid, since American taxpayers would have been subsidising exports from a US firm.

Megan rightly flagged that we got this second part wrong. In fact, this deal was one of a handful in DCA’s portfolio in which the firm itself transfers money to the Treasury — GE, rather than taxpayers, is on the hook if loans go sour. We’ve edited the description of this guarantee in our paper and added a footnote to explain the change.

More generally, Megan pointed out, DCA’s $4.3 billion portfolio is distinct in America’s aid ecosystem. It uses guarantees to pursue development outcomes, rather than support US exports (like ExIm) or American-owned firms (like OPIC, which can only work with companies that are at least one-fourth US-owned).

Along with its unique mandate, DCA practices good fiscal policy by lodging capital against potential losses. We think this is good: as we say in our paper, in the absence of this kind of accounting, guarantees can become a way for policymakers to spend freely and kick the fiscal can down the road.

How much to provision against losses is a tough question: DCA’s argued that they’re required to use an overly conservative model of potential losses, having historically locked away 10 times what’s been called in defaults. Of course, this could also be symptomatic of partnering with local lenders who skim cream by only lending to the very best clients, or of DCA or USAID being overly conservative in setting their target populations.  Or it might be that widespread defaults are relatively rare, but expensive when they happen, and they are rightly priced but haven’t yet occurred.  In any case, making guarantees reflect the expected cost to the taxpayer, and budgeting for them appropriately, at the time they are issued is smart public policy that should be widely adopted.

Not tied but “tied”?

This is a relatively novel deal, moving GE financing through USAID’s pipes to facilitate targeted lending in Kenya. It’s attractive because it fulfills DCA’s mission without locking away financing. In an era of stretched budgets, this model might become much more popular.

Yet it’s important that we remember that financing isn’t the only scarce resource — bureaucratic bandwidth is finite. If DCA were to greatly increase guarantees backed by firms to support their own exports, it would tilt the organisation’s portfolio towards export promotion, rather than its mandate of expanding access to credit for development impact. To the extent those exports were from American firms, the realignment would recreate some of development finance’s bad old days.

In reality, of course, we’re a long way away from that. This was an interesting pilot for DCA that should be written up and learnt from, not the organisation’s new normal. (Of course, many other governments do use development finance to support their own exports, including through export credits and guarantees).

As development agencies tackle the new agenda of working with firms to deliver on the ambitious Sustainable Development Goals, however, we should keep in mind that the risks of capture, distortion, or crowding out also apply to the agencies themselves.