Hide table of contents

The first post of this series provides a case for policy-based poverty interventions in the US as a cost-effective means of improving global health. 

In this post, I’ll discuss how we might build a prioritization framework for this kind of work, and how it might differ from other EA causes. I’ll also add some thoughts on the pros and cons of EA investing in this area.

Prioritization Framework


We can evaluate impact in the same way as we might for health interventions in Low Income Countries. I personally believe some partiality is fine and as I am an American, I am particularly concerned about poverty in America. However, I respect those who want to assess causes completely impartially, and I think this work can hold up to that standard.


Getting legislation passed is a low-certainty endeavor. I think this uncertainty is likely on-par with medical research, but I haven’t worked in advocacy and so it’s hard for me to say. To make these problems tractable, EA would need to bring in significant legislative expertise - ie, lobbyists. The term grosses people out, but lobbyists are typically former staffers who understand how to word legislation, estimate its cost, and politically maneuver legislation into reality. That’s an extremely niche skill that requires practical experience. I’m quite interested in the question of whether EA should build out a lobbying function (but I will save it for another post).

For prioritization, I think the major factors in tractability can be broken down into:

  • How much does the program cost (it’s harder to pass more expensive programs)
  • Is there popular opposition to the program from constituents, making it difficult for a legislator to support?
  • Is there moneyed opposition from corporations or organizations who will spend adversarially to prevent the program?

For actual tractability of solving the problem, I believe most programs have high implementation tractability. Most of the tractability challenge is political.


Typically, EA operates on the assumption of diminishing marginal returns - the more you add people and funding, the less valuable the addition. In the case of policy, I think the opposite can (but isn’t always) true.

I think in many cases, policy requires a critical mass of people who care about an issue, who will vote on, publicly talk about, or show up at protests to convince legislators. Legislation is highly momentum-driven based on the issue of the day, and momentum represents increasing marginal returns. Additionally, legislators care about what their constituents want, and being able to ally with a major interest group or voting block like a union or Medicaid recipients strongly increases the likelihood of success.

However, there are certain issues that are not neglected but are still highly risky because they have very strong opposition. If your goal is opposed by a group with deep pockets and / or loud feelings, there’s both a spending race and a lower likelihood of success, making ROI plummet. So there may be higher certainty on neglected issues, especially smaller and cheaper interventions. Holden Karnofsky calls this out. This is captured in the tractability section.

Pros and Cons of this Cause Area

  • Pro: given that many in EA are Americans, some may feel heightened responsibility towards the poor in our own country, since we share a government and infrastructure that we all benefit or are failed by.
    • This is especially relevant when we consider that many in EA likely benefit from tax “handouts”: employer-paid healthcare (which is untaxed as income, even though it is effectively income) and tax-privileged retirement accounts, which are effectively unusable for those living paycheck-to-paycheck. It’s not unreasonable to look at my tax refund and see that 20% of that money would have been spent on welfare programs like SNAP.
  • Pro: in metro areas like the Bay Area, Seattle, New York, and DC, EA probably represents a relatively well-organized and well-funded constituent group that could have legitimate influence with congresspeople and state legislators.
    • EA is probably too small to represent a meaningful voting block: 7k members vs 80k for the Democratic Socialists of America (DSA) vs 4.3 Million for the National Rifle Association (NRA)
    • However, it is extremely well-funded (1B to Givewell vs $230k for DSA vs $200M to the NRA) and well-coordinated.
    • This offers an opportunity for coalition-building: since these initiatives are meant to help large numbers of people, it should be easy to partner with grassroots movements that represent larger voting blocks but are more resource-constrained. These groups are also more likely to have the necessary advocacy experience that we need.
  • Con: the economy is extremely complicated, and it’s not always obvious to me (or even between expert economists) how anti-poverty interventions like health care, minimum wage, or welfare benefits will impact inflation, employment, or economic growth. I personally am quite left and think that more social services are a moral imperative, but it might make sense to focus on interventions where there’s less disagreement about economic effects.
  • Con: Policy advocacy is an unpredictable domain and there’s high uncertainty of success in passing a law (eg 5%?), compared to public health interventions like vaccination (50%?). EA generally has a high tolerance for uncertainty, but I think discounting for uncertainty is fair.
  • Con: Many people external to the movement think EA is shady and has too much power, and building closer relationships with legislators could increase that perception. It could also enable any bad actors in the movement to do a lot of damage, in the case of corruption.
  • Pro: A lot of EA work has re-oriented towards working with the US government, because of its power to both regulate and direct massive funds. Therefore, this work could be very much in line with general EA strategy, and focusing on improving conditions for Americans could help build more popular support for the movement within the US.
  • Pro: because this topic is of significant interest to US researchers, there is significant existing research (though not much in the experimental realm) and expertise to guide effective policy and / or grantmaking.

Other considerations:

  • This area is evaluated on a near-termist basis. I actually think you can make an argument that a healthier, less impoverished American populace has major implications for long-termist concerns (ie, electing presidents who will preserve democracy and avoid war), but that’s not the analysis above.
  • EA has historically avoided marking itself with a political label, for fear of alienating whichever side it didn’t choose. In my experience EAs skew libertarian, and this expansion of social services definitely not aligned with that bent.
    • However, I think EA has really struggled in political work because it has not built coalitions with other political movements that have experience with getting out the vote, power-mapping, and building popular support. This work is an opportunity to begin that engagement and learn from other movements.

Thanks for reading! Would be curious to hear other pros and cons of this research, and your feedback on the framework. In the final post, I’ll go through some possible areas of research, and discuss how we might build the ability to intervene.

Sorted by Click to highlight new comments since:

Hi Abbey,

I think you are going to have a very hard time convincing EAs that this should be a core, or even peripheral, EA cause area. 

In your previous piece, you cite Matt Desmond's estimate that 5.4 million Americans live in extreme poverty. I think this is probably an overestimate, but taking it at face value, that's less than 1% of all the people who live in extreme poverty globally. Accordingly, if global poverty is your top priority, the impartial altruism principle implies that the extreme poor in America  should receive less than 1% of the attention you devote to it. Exceeding that 1% attention budget is going to be a very high burden of proof to meet, given our strong shared commitments to on-the-margin thinking (there are many, many social programs and organizations devoted to the welfare of the poor in America) and impartiality. 

On a different note; if you want to work on reducing poverty in America, what's to be gained by applying an EA label to it or convincing others to do so? My 2c: it's perfectly well and good to work on causes that motivate you and where your work is plausibly +EV, labels be damned.

Hey Seth, thanks for your thoughts! I agree it's pretty uncommon as an area of excitement for EAs, and I think it's because people have the (correct) intuition that interventions are much more expensive in the US. What I wanted to point out was that the problem can be framed differently, and that the broad EA intuition might be wrong here.

I'm not sure I agree with the attention budgeting point. Givewell and OpenPhil seem to look at funding interventions on the margin (eg, what is the return for this particular intervention) and plenty of those interventions are quite small-scale, so I think these interventions are in line with others they research. Indeed, Givewell's interest in Policy advocacy in developing nations suggests an interest in developing this muscle. Holden Karnofsky has taken US Policy intervention seriously enough to deeply research the topic, and make incarceration in America (1M people) a cause area for OpenPhil. From what I've seen, he focused more on preventative health interventions more than poverty interventions in his research.

I think this topic is overlooked in the community, and I wanted to draw some attention to it. I'd like to find a way to donate to cost-effective charities for reducing poverty in America, and part of my goal in writing this post was to research and develop some ideas about what those might be.

I think a fair amount of EAs also feel compelled to donate locally, but they designate those donations as "fuzzies." I don't think this has to be the case, and it would be great if more local donations went to truly cost-effective initiatives.

Anyway thanks for your thoughts, I appreciate you pointing out the relative scale and agree that poverty outside the US is much larger! I also care deeply about global health outside of the US and currently donate there.

Executive summary: The post makes the case for EA investment in US policy interventions to alleviate poverty as a highly cost-effective way to improve health, then lays out a framework to assess tractability, neglectedness, and other considerations in determining priority.

Key points:

  1. Impact can be evaluated similarly to global health interventions using cost-effectiveness. There is also an argument for partiality towards one's own country.
  2. Tractability depends on cost, popular opposition, and moneyed opposition. Implementation tractability is generally high if legislation passes.
  3. Neglectedness varies - some issues need critical momentum to pass legislation. Highly opposed issues are riskier despite neglectedness.
  4. Pros include sense of responsibility for poor Americans and being a well-funded, coordinated group that could influence legislators.
  5. Cons include economic uncertainty, low policy success rates, and increased perception of EA having too much power.
  6. This cause aligns with recent EA government work and could help build US popularity. Existing research expertise also helps guide policy.



This comment was auto-generated by the EA Forum Team. Feel free to point out issues with this summary by replying to the comment, and contact us if you have feedback.

I would try to improve the political system which is upstream of the dysfunctional politics which is upstream of bad policy that leads to needless policy in the US. I think you'll get better ROI be cause the dividends will affect so many other areas EAs care about.

Curated and popular this week
Relevant opportunities