“Additionality” is central to claims of impact by development finance institutions (DFIs). At its core is the notion that DFIs are necessary to solve a market failure by providing capital, risk mitigation, or some other benefit to a market that is not delivering these services strictly through private actors. On a micro level, additionality implies a counterfactual: any specific DFI-financed project would not have happened as it happened without the support of the DFI. More broadly, this suggests that the publicly supported finance from the DFI should be additional to, not a substitute for, available private financing. These details matter because without additionality, any claim to development impact is irrelevant and, perhaps worse, suggests scarce public resources are being mis-deployed. A joint multilateral development bank framework claims that “additionality is a threshold condition, sine qua non for considering a potential intervention.”

That means the World Bank’s International Finance Corporation (IFC), Britain’s CDC Group, Norfund, Swedfund, the Dutch FMO, France’s Proparco, Germany’s DEG, the new US International Development Finance Corporation (DFC), FinDev Canada, and many other similar agencies all ride on the notion that they can deliver additionality. For obvious reasons, this has ignited fierce debate and much hand-wringing about what exactly constitutes additionality, how do we know when it is real, and how can we measure it. To which the answer is “we usually can’t, but that is OK as long as we follow some simple rules.”

What is additionality?

A narrow interpretation is provided by Melina Heinrich of the Donor Committee for Enterprise Development. She suggests that to prove additionality, an agency must establish that the company is (a) unable to self-finance the project; or (b) does not have the knowledge or skills to implement the project activities alone; and/or (c) is unwilling to implement the project without support. Further, the agency should establish that (d) the firm cannot obtain this finance, knowledge, or support from a private provider and that (e) no other private firm could deliver the project without support.

Is the capital additional or the development benefits?

The multilateral development bank (MDB) harmonized framework extends considerably beyond “supports a project that would not have happened otherwise.” This more expansive definition claims that DFI involvement can be additional if it provides financing not available in the market, risk mitigation, improved project design, better development outcomes, or if it adds environmental, social, and governance standards. (This “extended definition” apparently covers a lot of what DFIs do: In the 1996–2007 period, for instance, the World Bank Group’s Independent Evaluation Group estimated, using project staff and documents, that in only 27 percent of cases was IFC involvement essential for a project to go ahead.^[1]) By this formulation, additionality is not a binary “project vs. no project” view but rather “does DFI involvement make the project better from a development perspective?” 

The measurement conundrum

Choosing between definitions of additionality affects claims around development impact. In the narrow definition, a project might, for instance, claim impact by estimating the number of jobs created by the project or the amount of capital “mobilized.” If the project never would have materialized without the intervention, then the DFI can plausibly claim credit for any (and arguably, even all) subsequent impact. For the broader definition, “impact” isn’t all jobs or capital but rather the number of jobs created by the project or capital mobilized or other features compared to the counterfactual of the project without the DFI’s investment.

From a development perspective, the expansive definition of additionality may make the most sense, especially if DFIs consider potential negative externalities that their involvement helped to avoid. Take, for example, a large highway that might displace a vulnerable residential neighborhood. A DFI might get involved, not to provide the base capital or enable the project to proceed, but rather to raise the project’s reach and/or elevate its standards. Without the DFI, the highway might still get built. But with the DFI, it might add feeder roads to an otherwise isolated community. And, perhaps more importantly, the DFI’s involvement might insist on competitive bidding for the construction contract and best practice for relocating and compensating affected people. This would be additional development impact (both adding positive and avoiding negative) but not strictly from a financial market perspective. (Incidentally, this is also an argument for MDBs to get back into projects that can have big negative externalities if done badly, like large hydropower projects.) 

It is difficult to prove “narrow additionality”—the counterfactual that a project would not have gone ahead without DFI support. Anecdotal evidence can sometimes be persuasive: for example, showing that a market (e.g., Liberia) or sector (e.g., rural agriculture) has a dearth of private capital or that similar projects are failing to attract private capital. But DFIs have no approach to demonstrating narrow additionality that is rigorous. A number of authors have attempted to measure the impact of MDB and DFI investments ex-post by seeing if they are associated with an increase in total country or country-sector investment, but Carter et al. suggest the approaches used do not allow for causal statements of impact on investment levels.^[2] Related to that, Carter et al. raise questions about standard mobilization calculations, concluding that the “methodologies are not so much exercises in assessing additionality... as in asserting it.”

If proving additionality by the narrow definition is difficult, the measurement obstacles for calculating development impact under the broad definition of additionality can be equally considerable. DFIs could attempt to track and publish which of their environmental, social or governance safeguard requirements were not already being met before sponsors came to them and were imposed as part of loan conditions—although this would be subject to gaming. (And it is equally possible that overly burdensome safeguards could prevent projects from happening.) Proving counterfactuals for the impact of a specific project (ex ante) with or (ex post) without DFI support is likely impossible with any precision. Using either definition of additionality, measurement of development impact is an unresolved and potentially unresolvable challenge.

The subsidy question

 Additionality concerns are particularly large with projects that encompass direct or implied subsidies. Client firms will always want more favorable financing terms, and so will have a strong incentive to claim that subsidies are necessary for a project to go ahead. DFI staff seeking to deliver projects have an incentive to agree. But by enabling financing that is more attractive than would be available from the market, subsidies increase the risk that DFIs crowd out rather than crowd in private finance. Under those circumstances, subsidies would reduce the probability that a project is narrowly additional. They would simply convert public resources to private rents by padding profits for projects that would have gone ahead anyway.

Fortunately, we need not worry too much about proving additionality or providing exact measures of development impact on every project if development finance institutions operate under incentives designed to ensure they do not crowd out private capital or fund projects likely to have large negative development impacts. We suggest three simple approaches:

1. Impose a “Do no harm” test. Have strict safeguards around projects with potentially significant social, environmental or external economic effects. In many cases, this does not require banning any involvement, but ensuring projects limit harm and suitably compensate affected populations, for example. Ensuring such safeguard conditions on projects can in and of itself be a source of DFI additionality.
2. Harness market signals upfront. As suggested by Carter, the best approach is for DFIs to adopt procedures to ensure narrow additionality before deals are signed. The simplest mechanism to do this is an open competition, such as a subsidy minimization auction. Where an auction isn’t possible, DFIs can make sure that projects which do not require public support voluntarily go elsewhere by offering less attractive terms than commercial banks and other private investors. This could involve a combination of pricing at market rates and adding transparency, environmental, and/or social conditionalities, for example. 
3. Use transparent benchmarks. While it may be hard to prove additionality in individual projects, we know that, as a rule, it is harder to attract private finance to agricultural projects in low-income countries than power projects in upper-middle-income countries. A DFI could set up a traffic light system that shows how resources are being deployed across different counties and sectors that will illustrate how much it is focusing on projects that may be more likely to be additional. For instance, how many projects in very poor environments (e.g., low-income countries like Liberia) are there relative to upper-middle-income markets (e.g., Brazil). This approach could also apply to sectors depending on capital availability and development impact (i.e., agriculture vs. tourism).

We can be confident that any DFI that does all three is providing value and is likely far more “additional” than one stuck in a measurement morass.