For one, yes, the funders would have invested their money elsewhere. So, the org will only have a positive Enterprise Impact if it is more cost-effective than the funder’s alternative. I think the ‘generous-to-GiveWell option’ is more extreme than it might appear at first glance. It’s not only assuming that the funders would otherwise donate in line with GiveWell (GW). It’s also assuming that they are somehow suckered into donating to this less effective org, despite being GW donors.
Yes, I think the “generous-to-Givewell” model should be seen as the right bookend on defensible models on available data, just like I see GIF’s current model as the left bookend on defensible models. I think it’s plausible that $1 to GIF has either higher or lower impact than $1 to GiveWell.
As for the counterfactual impact that other funders would have, I would expect funders savvy enough and impact-motivated enough to give to GIF-supported projects to be a cut above the norm in effectiveness (although full-GiveWell effectiveness is a stretch as you note). Also, the later-round funders could plausibly make decisions while disregarding prior funding as sunk costs, if they concluded that the relevant project was going to go under otherwise. This could be because they are thinking in a one-off fashion or because they don’t think their fund/no fund decision will affect the future decisions of early-stage funders like GIF.
Although I like your model at a quick glance, I think it’s going to be challenging to come up with input numbers we can have a lot of confidence in. If there’s relatively low overlap between the GiveWell-style donor base and the GIF-style donor base, it may not be worthwhile to invest heavily enough in that analysis to provide a confidence interval that doesn’t include equality.
Also, GiveWell’s diminishing returns curve is fairly smooth, fairly stable over time, and fairly easy to calculate—most of its portfolio is in a few interventions, and marginal funding mostly extends one of those interventions to a new region/country. GIF’s impact model seems much more hits-based, so I’d expect diminishing returns to kick in more forcefully. Indeed, my very-low-confidence guess is that GIF is more effective at lower funding levels, but that the advantage switches to GiveWell at some inflection point. All that is to say that we’d probably need to invest resources into continuously updating the relevant inputs for the counterfactual impact forumula.
Yes, I think the “generous-to-Givewell” model should be seen as the right bookend on defensible models on available data, just like I see GIF’s current model as the left bookend on defensible models. I think it’s plausible that $1 to GIF has either higher or lower impact than $1 to GiveWell.
As for the counterfactual impact that other funders would have, I would expect funders savvy enough and impact-motivated enough to give to GIF-supported projects to be a cut above the norm in effectiveness (although full-GiveWell effectiveness is a stretch as you note). Also, the later-round funders could plausibly make decisions while disregarding prior funding as sunk costs, if they concluded that the relevant project was going to go under otherwise. This could be because they are thinking in a one-off fashion or because they don’t think their fund/no fund decision will affect the future decisions of early-stage funders like GIF.
Although I like your model at a quick glance, I think it’s going to be challenging to come up with input numbers we can have a lot of confidence in. If there’s relatively low overlap between the GiveWell-style donor base and the GIF-style donor base, it may not be worthwhile to invest heavily enough in that analysis to provide a confidence interval that doesn’t include equality.
Also, GiveWell’s diminishing returns curve is fairly smooth, fairly stable over time, and fairly easy to calculate—most of its portfolio is in a few interventions, and marginal funding mostly extends one of those interventions to a new region/country. GIF’s impact model seems much more hits-based, so I’d expect diminishing returns to kick in more forcefully. Indeed, my very-low-confidence guess is that GIF is more effective at lower funding levels, but that the advantage switches to GiveWell at some inflection point. All that is to say that we’d probably need to invest resources into continuously updating the relevant inputs for the counterfactual impact forumula.