Proposal for Forecasting Givewell-Charity Impact-Metrics

By Patrick Gruban 🔸 @ 2022-04-13T10:21 (+28)

As GiveWell has stopped sharing estimates on the future impact of donations to individual charities, I propose to use forecasting as a replacement. However, the resolutions will not apply to individual donations, so that I would be interested in ideas for alternative estimations.

Last fall, GiveWell discontinued its impact calculator. On their website, they write:

We’ve decided to use impact estimates based on grants that GiveWell has made from our Maximum Impact Fund or has recommended to Open Philanthropy. By using only past grants, we can share more concrete estimates of impact than if we’re trying to guess what the future impact of a donation will be.

Backward-looking estimates don’t fit well into our impact calculator design, which offered donors an estimate of what their donation will accomplish in the future.

They still report backwards-looking estimates, although only for grants made from the Maximum Impact Fund and recommendations to the Open Philanthropy Project.

Different EA organisations use impact calculations to communicate individuals' impact through their donations. The GWWC calculator "How Rich Am I" gives an estimate on bednets distributed, schistosomiasis treatments, and lives saved per dollar amount. EA Australia has an impact calculator (disclaimer: I was involved in writing the underlying software), Ayuda Efectiva uses the GiveWell numbers for their impact calculations, and One For The World communicates them. These are only a few examples.

In most cases, they can seem like forward-looking projections for the money donated now. At the same time, they seemingly rely on GiveWell estimates that are backwards-looking and do not include individual donations. This could lead to donors finding out in the future that their donations' impact is different than expected and losing trust in recommendations from EA recommender websites. 

An easy solution is to change the website's wording to reflect the changed approach of GiveWell or to discontinue the use of impact calculators or the communication of impact for individual donations altogether.

However, there might be cases where new donors can be motivated more easily with impact estimates than without. I propose not using backwards-facing impact numbers but estimates of current ones. As GiveWell is increasing the funding opportunities for the Maximum Impact Fund, they are lowering their funding bar. This will also affect individual donors' impact.

Given the knowledge about forecasting in the EA community, I would be interested to see ideas on how the impact of current donations for GiveWell's top charities could be projected. I assume the most straightforward questions would be around numbers that GiveWell will retroactively publish concerning the impact of the Maximum Impact Fund: There will be a clear resolution if GiveWell continues its publications.

For donations on the individual level, GiveWell does not calculate backwards-looking estimates. They are stating:

We exclude the funding [for donations that donors make to an organization we recommend, but that aren’t directed to a specific funding opportunity] because we don’t have estimates of the cost-effectiveness of those donations. We only calculate the cost-effectiveness for specific funding opportunities. Funding directed to our top charities could be either more or less cost-effective than the funding directed to specific funding gaps. Rather than guessing at a number and using it for our calculations, we’ve excluded this funding.

So making forecasts for these numbers won't lead to a resolution by GiveWell. 

I'm hoping for suggestions in the comments on how to get to forward-looking estimates on the individual level. If there are ideas worth setting up discussions for, I'm glad to facilitate them.