(1) One factor limiting our funding for THL and other groups is how much of their budget we’re both comfortable with OP being.
How often is this the limiting factor? If quite often, then I agree funging would be small.
(2) Room for more funding / neglect is only one consideration in our grant sizing for groups, including THL
I think my point holds as long as all the considerations boil down to you setting a given target funding in $.
(3) Our grant sizing for big grantees like THL is quite coarse: we only consider it once every three years and are unlikely to be swayed by small fluctuations in their funding.
In general, I do not think the consideration above matters much. The situation seems analogous to one eating 100 g less chicken leading in expectation to a decrease in chicken production by something close to 100 g (a little lower because the elasticity is lower than 1), despite the very low probability of eating less 100 g of chicken affecting the number of batches of chicken (which I guess respect at least tens of kg). Likewise for funding? The probability of my donation affecting the funding THL receives from Open Philanthropy could be low, but in expectation the decrease in funding could still be meaningful. To illustrate, if you only adjust your funding to THL in intervals of 100 k$, and set the target funding to THL in $ (instead of as a fraction of THL’s total expenditure), a donation of 1 k$ to THL would have something like a 1 % (= 1⁄100) chance of decreasing your funding by 100 k$ (given a uniform prior about how far away you are from updating you target funding), so the donation would in expectation decrease your funding by 1 k$.
Second, will donating to THL mean that OP gives less to farm animal welfare? Answer: almost certainly not.
I appreciate a donation of 1 k$ is super unlikely to change your animal welfare funding, but this does not necessarily imply the expected change in your animal welfare funding caused by a small donation is neglegible. For example, if you update your targer animal welfare funding in intervals of 100 M$, a 1 k$ donation to THL could have something like a 0.001 % (= 0.001/100) chance of updating your funding by 100 M$ (given a uniform prior about how far away you are from updating you target funding), thus decreasing your animal welfare funding in expectation by 1 k$. Am I missing something?
Thanks Vasco. On (1) and (2), I think that the grant sizing process is messier than it may seem. So the portion of a group’s budget we can be is often a major factor, but not necessarily the limiting one. And I don’t think our considerations all boil down to us setting a given target revenue for a group, in large part because we don’t want to create a perverse incentive for other funders to not fund groups we do and for our grantees to not fundraise.
On (3), I agree there’s some chance that in aggregate your donation will flip a group into a different funding category. I just think it’s quite rare, because the ideal revenue level for a group is not our only consideration in funding levels. See also the point above about us explicitly trying to avoid gaming other funders or groups’ fundraisers.
On the final point, I think you’re wrong to assume that if funding for farm animal welfare increased by $100M then there’s a 100% chance our program’s funding would decline by $100M (which to be clear is more than our program’s budget). I think reduced neglect could influence Open Phil leadership to allocate less funding to a cause area. But I think the odds it did so are much below 100% and the amount it would do so by is far less than the increased funding in the space (here $100M).
On the final point, I think you’re wrong to assume that if funding for farm animal welfare increased by $100M then there’s a 100% chance our program’s funding would decline by $100M (which to be clear is more than our program’s budget). I think reduced neglect could influence Open Phil leadership to allocate less funding to a cause area. But I think the odds it did so are much below 100% and the amount it would do so by is far less than the increased funding in the space (here $100M).
I gave a bad example because 100 M$ is a significant fraction of the amount granted in farm animal welfare over the number of years respecting the budget allocation. I also assumed an elasticity of 1, but I can see something like 0.5 would be more reasonable. So my corrected statement would be something like a new animal welfare donor granting 10 M$ in a similar way to Open Phil (i.e. not just an increase in 10 M$ of funding, which may be poorly allocated) would decrease Open Phil funding in expectation by 5 M$. However, I see your replies to points 1 to 3 would also apply, such that the elasticity may be closer to 1, and therefore one would not need to worry about Open Phil decreasing funding to animal welfare.
Yeah that makes sense. I think you’re right that it’s plausible that new funding could decrease Open Phil funding in the space. I just think it’s low odds, and would only be to a much lower extent than the size of new funding.
Thanks, Lewis!
How often is this the limiting factor? If quite often, then I agree funging would be small.
I think my point holds as long as all the considerations boil down to you setting a given target funding in $.
In general, I do not think the consideration above matters much. The situation seems analogous to one eating 100 g less chicken leading in expectation to a decrease in chicken production by something close to 100 g (a little lower because the elasticity is lower than 1), despite the very low probability of eating less 100 g of chicken affecting the number of batches of chicken (which I guess respect at least tens of kg). Likewise for funding? The probability of my donation affecting the funding THL receives from Open Philanthropy could be low, but in expectation the decrease in funding could still be meaningful. To illustrate, if you only adjust your funding to THL in intervals of 100 k$, and set the target funding to THL in $ (instead of as a fraction of THL’s total expenditure), a donation of 1 k$ to THL would have something like a 1 % (= 1⁄100) chance of decreasing your funding by 100 k$ (given a uniform prior about how far away you are from updating you target funding), so the donation would in expectation decrease your funding by 1 k$.
I appreciate a donation of 1 k$ is super unlikely to change your animal welfare funding, but this does not necessarily imply the expected change in your animal welfare funding caused by a small donation is neglegible. For example, if you update your targer animal welfare funding in intervals of 100 M$, a 1 k$ donation to THL could have something like a 0.001 % (= 0.001/100) chance of updating your funding by 100 M$ (given a uniform prior about how far away you are from updating you target funding), thus decreasing your animal welfare funding in expectation by 1 k$. Am I missing something?
Thanks Vasco. On (1) and (2), I think that the grant sizing process is messier than it may seem. So the portion of a group’s budget we can be is often a major factor, but not necessarily the limiting one. And I don’t think our considerations all boil down to us setting a given target revenue for a group, in large part because we don’t want to create a perverse incentive for other funders to not fund groups we do and for our grantees to not fundraise.
On (3), I agree there’s some chance that in aggregate your donation will flip a group into a different funding category. I just think it’s quite rare, because the ideal revenue level for a group is not our only consideration in funding levels. See also the point above about us explicitly trying to avoid gaming other funders or groups’ fundraisers.
On the final point, I think you’re wrong to assume that if funding for farm animal welfare increased by $100M then there’s a 100% chance our program’s funding would decline by $100M (which to be clear is more than our program’s budget). I think reduced neglect could influence Open Phil leadership to allocate less funding to a cause area. But I think the odds it did so are much below 100% and the amount it would do so by is far less than the increased funding in the space (here $100M).
Thanks, Lewis!
I gave a bad example because 100 M$ is a significant fraction of the amount granted in farm animal welfare over the number of years respecting the budget allocation. I also assumed an elasticity of 1, but I can see something like 0.5 would be more reasonable. So my corrected statement would be something like a new animal welfare donor granting 10 M$ in a similar way to Open Phil (i.e. not just an increase in 10 M$ of funding, which may be poorly allocated) would decrease Open Phil funding in expectation by 5 M$. However, I see your replies to points 1 to 3 would also apply, such that the elasticity may be closer to 1, and therefore one would not need to worry about Open Phil decreasing funding to animal welfare.
Yeah that makes sense. I think you’re right that it’s plausible that new funding could decrease Open Phil funding in the space. I just think it’s low odds, and would only be to a much lower extent than the size of new funding.