The Global Innovation Fund (GIF) is a non-profit, impact-first investment fund headquartered in London that primarily works with mission-aligned development agencies (USAID, SIDA, Global Affairs Canada, UKAID). Through grants, loans and equity investments, they back innovations with the potential for social impact at a large scale, whether these are new technologies, business models, policy practices or behavioural insights. I've been told that so far their investments have been roughly 60% grants and 40% 'risk capital' (i.e., loans and equity).
Recently, they made a bold but little publicized projection in their 2022 Impact Report (page 18): "We project every dollar that GIF has invested to date will be three times as impactful as if that dollar had been spent on long-lasting, insecticide-treated bednets... This is three times higher than the impact per dollar of Givewell’s top-rated charities, including distribution of anti-malarial insecticide-treated bednets. By Givewell’s estimation, their top charities are 10 times as cost-effective as cash transfers." The report says they have invested $112m since 2015.
This is a short post to highlight GIF's projection to the EA community and to invite comments and reactions.
Here are a few initial points:
- It's exciting to see an organization with relatively traditional funders comparing its impact to GiveWell's top charities (as well as cash transfers).
- I would want to see more information on how they did their calculations before taking a view on their projection.
- In any case, based on my conversations with GIF, and what I've understood about their methodology, I think their projection should be taken seriously. I can see many ways it could be either an overestimate or an underestimate.
Let me introduce myself – I’m the Chief Economist of the Global Innovation Fund. Thanks Jonathan for the original post and others for your comments. I’d like to offer a few clarificatory responses on some of the issues raised.
Apples vs oranges
Yes, you could say this is a deliberate apple vs oranges comparison as an aid to constructing an efficient fruit salad :-) Apples: The philanthropist in search of impact can invest in a proven intervention, where the marginal impact of each additional dollar is well understood. Buying more bednets, for instance, will save lives and prevent illness. Oranges: the philanthropist could invest in a risky innovation which might fail – but if successful, might catalyze far-reaching impact.
I don’t know the philosophy underlying Givewell’s portfolio strategy of 75%“Top Charities” vs. 25% innovation, but it doesn’t necessarily mean that Top Charities have a higher expected return. In general terms, allocating a philanthropic portfolio between proven interventions vs. innovations involves considerations of risk. Innovations are inherently risky, so a philanthropist might well seek a balanced impact portfolio with ‘bond-like’ Top Charities (sure impact return) and ‘equity-like’ innovations (high but uncertain impact returns). GIF has a very risk-tolerant strategy, but that may not suit everyone.
Projecting long-term impact
Impact projection for innovations is undoubtedly difficult. Certainly these projections involve informed judgment on things like the probability of failure and speed at which the innovation diffuses. But our projections of impact are no more audacious than the valuation projections that underlie venture capital finance for start-ups and indeed often draw on the same set of assumptions.
Typically our impact projections are based on a rate of growth of the innovation’s take-up or influence (number of customers or beneficiaries), multiplied by our best estimate of impact per person reached. To get the impact per beneficiary, we review the literature and draw on RCTs where possible and appropriate. We adjust for the risk that the innovation fails, both during our investment and during the hoped-for subsequent scale-up. We regularly update these projections based on new evidence on impact, rate of uptake, risks surmounted or not. Many of our grants have built-in RCTs or other evidence generation that lets us update the impact per beneficiary parameter.
Additionality and contribution
As readers of this thread know, these are difficult things to estimate. GIF invests in early-stage innovations that arguably might fail or falter absent the funding round in which we participate. So we think the counterfactual of no impact is a defensible assumption. Regarding contribution, we allocate future impact in proportion to our participation in that funding round. We are eager to keep up with methodological developments in these issues.
Transparency
We will update our website with more information on our methodology. Complete transparency on the underlying calculations presents some issues, since some of the information is sensitive or confidential, especially for risk capital. I’m not aware of any investor that publishes projections of a for-profit investee’s impact – I’d be grateful for any references.
Yes, I think the "generous-to-Givewell" model should be seen as the right ... (read more)