Hey everyone,
I have a question for you that could inform an EA related research topic: What do you believe to be the most important reasons to support multiple different interventions? If you donate to a single fund that itself supports multiple interventions, what reasoning should apply to the fund’s allocation decisions? This question is relevant as I am looking for a mathematical model to describe an individual’s satisfaction with an allocation that may differ from the preferred allocation of resources.
Here are some reasons I could come up with:
Maximizing utility in the presence of diminishing returns: Individuals attempt to maximize impact according to their own worldview. As interventions receive more funding they encounter diminishing returns of investment, which is why another intervention may yield better impact per resource spent.
Reducing risk: Individuals may be willing to fund an intervention with lower expected impact in order to reduce risk (such as reducing variance or guaranteeing to do at least some amount of good).
Composing Individuals of Subagents: Individuals may be composed of different subagents. Letting them “vote” on a donation allocation may result in interventions receiving funding.
Some completely different reason I could not think of yet
Why am I asking?
As I will be working on donor coordination problems in my thesis over the next couple of months, understanding the different kinds of motivations to donate to multiple interventions just became a lot more important. In a donor coordination problem, individuals have different opinions on the optimal resource allocation over different interventions. By coordinating their donation efforts they can increase overall satisfaction with the outcome.
Naturally, in such a setting, most donors will end up with a donation allocation that differs from their preferred result. The literature considers many different evaluation metrics to measure the satisfaction of individual donors, each of which does not capture the full list of reasons why one might wish to have a diversified donation allocation in the first place. A surprising fraction of papers assume some sort of distance function between the desired allocation and the observed allocation (such as distance you may know from linear algebra or Leontief utilities that focus on the intervention you consider to be underfunded by the biggest factor relative to your ideal allocation). While this mathematically intuitive, I find it difficult to justify from the individuals perspective.
In case you diversify because of multiple reasons, ranking their importance in practice would also help greatly as mathematical models become more and more complicated the more nuance you are trying to capture with them.
(In case anyone is interested: The papers I looked at so far can be found here, here, here and here though I do not expect anyone to go into that much detail. Any quick take on the reasons for diversification is greatly appreciated :) )
Relevant reading: https://forum.effectivealtruism.org/posts/x2iT45T5ci3ea9yKW/dialogue-on-donation-splitting
Thank you for sharing!
(Simply giving this link was really valuable already. I will put my main takeaway points from this post here, but I do not expect a response in case you do not want to participate in this discussion again as it appears you already did so 3 years ago :’D.)
To me, it appears as if the authors mainly agree that diversification (within one cause area) is motivated by an attempt to maximize utility, though they disagree on the degree to which diminishing returns of investment (and therefore the role of diversification) actually matter in practice. I briefly want to point out that even though this idea is obvious from an EA perspective the literature on donor coordination problems does a very poor job in capturing this intent.
I agree that donation funds help reduce coordination problems. However, assuming there is more than one grantmaker, this just shifts the “burden of aggregating different opinions” from the general public into the organization since grantmakers still need some mechanism to reconcile their differing beliefs. That said, I don’t know enough about typical grantmaking processes to judge whether grantmakers differ significantly enough in their individual assessments for this to matter in practice.