I noticed that the EA Funds often disburse large amounts of money at a time. For example, all the grants paid out by the EA Long-Term Future Fund in September 2020 were worth $5,000 or more. As a more extreme example, the Global Health and Development Fund tends to give out more than $100,000 at a time. In either case, an additional small donation of $5 seems like it would have no effect on what the EA Fund does, whereas an additional $5 donation to the Against Malaria Foundation would directly purchase one net. What is the expected marginal impact of donating a small amount such as $5 to an EA Fund?
You can think about this in terms of Shapley values, rather than counterfactual/marginal values, and you'll get an answer which is, in my opinion, less confused.
In particular, suppose that 2001 people donate $5 to an EA fund, and the fund gives out a grant of $100k. Then I claim that you should think of the impact of the $5 as closer to $100k/2001 (~the Shapley value), rather than as 0 (the counterfactual value of each donation, because without one $5 donation the $100k would have still gone through.)
In other words, you might want to think about each of the $5000 donors as coming together to enable one big $100k donation and sharing its impact, rather than emphasizing that their counterfactual/marginal impact is lower.