For posterity (esp. because I don't trust Facebook as a repository of knowledge), quoting in full the original FB post that riceissa linked:
For three different foundations: Open Philanthropy Project, Bill & Melinda Gates Foundation, and Laura and John Arnold Foundations, I calculated that the total money granted per hour of staff time is approximately $1000 - $3000. This includes all staff time (obtained by taking number of people on staff and multiplying by 2000 hours for a year, then comparing with annual grants).
Is there a reasonable argument that foundations would generally have this ratio of money granted to staff time? For instance, if we break down the cost into direct grant investigation cost + cost of time spent getting familiar with the domain and evaluating strategy, etc., are we bound to arrive at a comparable figure?
One foundation that has a much higher ratio of money granted to staff time in recent years is Atlantic Philanthropies, but they are in spend-down mode right now and I don't have a good picture of their overall spend trajectory and employee counts yet.
Open Philanthropy Project:
Grants in 2016: $50 to $100 million
Staff at year-end: ~20 (+ some shared operational staff with GiveWell)
Laura and John Arnold Foundation
Grants in 2015: $185 million
Staff in 2016: ~50 listed on their site
Bill & Melinda Gates Foundation
Grants: ~$4.2 billion
Staff: ~1500
I can't say that surprises me too much. Cost correlates pretty strongly to the complexity of a project, and complex projects should take more staff evaluation.
You might also be seeing offsetting effects - bigger budgets may have some efficiencies in evaluating good proposals; they may also attract more low-quality proposals that must be rejected. It could wind up being about even, efficiency wise, over a fairly large range.
[Edit: original source has been found. The way I framed this question was off, since I'd mis-remembered it as "amount that a given grantmaker grants per year" rather than "grants per hour of staff time". Updated the title]
I recall reading an article once that claimed that, when examining many small and large foundations, it turned out that there was a maximum amount that a given grantmaker typically gave out. And as organizations scaled to give out more money, this amount stayed surprisingly fixed, with a higher overhead ratio than you might have expected.
(i.e. when an org gives out a million a year, it has N grantmakers, and when it gives out 100 million a year, it typically has 100N grantmakers).
I don't remember the number, or the methodology that determined it. Curious if anyone can remember the article. (It might have been from OpenPhil's blog, or it might have been some random news site).
I vaguely remember the number "3" being involved, possibly $300k, or $3 million.
The takeaway I remember was something like "you might naively think you can scale up an organization and then give away money more efficiently, but weird forces seem to limit that."
Does this sound familiar to anyone?