Prioritizing COVID-19 interventions & individual donations

post by Ian David Moss, catherio · 2020-05-07T01:44:27.250Z · score: 11 (2 votes) · LW · GW · None comments

Contents

  I. Executive Summary & Recommendations
  II. Big picture: What's the bad thing that's happening? What could cause less of it to happen?
    First-order problem: a disease is spreading around, causing illness.
      What levers can we pull to make direct impacts less bad?
    Second-order problem: disruptions to people's lives and livelihoods
      What levers can we pull to make indirect impacts less bad?
  III. Prioritization: Which levers are likely "most effective" to pull on?
    Scale
    Neglectedness, and other properties of the ecosystem & organizations.
  IV. Specific giving opportunities
    Opportunities we plan to give to
    Promising, with some open questions
      Global poor
      Vaccines, diagnostics, and treatments
      Policy advice/knowledge/resources
      What about Johns Hopkins CHS?
    Also considered
    Opportunities we wish we'd found more of
  V. Addendum: Should you prioritize COVID response over other EA priorities?
None
No comments

NOTE: The COVID-19 funding landscape is evolving rapidly, and we are making updates to reflect new developments and revised donation recommendations. Please check the EA Forum version of this post [EA · GW] for the latest information.

Authors: Catherine Olsson [EA · GW] and Ian David Moss, with contributions from the collective members of the "Funding Rational Actors Promptly" Pandemic Endowment (FRAPPE).

At the beginning of April, a group of friends pulled together a messenger chat to discuss how to most effectively spend personal donation funds towards mitigating global suffering caused by COVID-19. We're about 20 people, with somewhere in the ballpark of $200k to donate. We gave out one chunk of ~$50k in mid-April, and are currently in the process of distributing a second round of funds.

A defining motivation of our group was to find time-sensitive and neglected bottlenecks to effective COVID response that could be eased with rapid funding or other supportive actions. Fast action can be an important source of philanthropic leverage in responses to the current pandemic, a factor that we did not see explored in depth in available analyses of COVID-related giving opportunities. Accordingly, we have summarized our research here in hopes that others can use it to inform their own giving.

This article is organized in two parts. The first shares our working framework for prioritizing interventions, which helped us get oriented in a fast-changing and otherwise confusing landscape.

In the second part, we enumerate specific giving opportunities (jump to section [LW · GW]) we have found that currently rate highly on this framework as of right now (early May 2020). One of these has a high minimum donation; if you would like to join a pooled gift, please reach out.

We've written this post primarily for the benefit of donors who have already decided to focus on COVID-19 for their own reasons. We haven't made it a priority to weigh the relative value of COVID-related donations as compared to other issues or causes, although we address this briefly at the end.

Some disclaimers: this research is being done and our donations are being made in a purely personal capacity, and none of us is acting as an employee, representative, or spokesperson of our employer or any other organization. Furthermore, because we don't have complete information on many opportunities and the situation is changing so rapidly, none of what follows should be treated as the final word on COVID-related giving opportunities. With that said, we tried hard to come to the best decisions we could in a short period of time using the resources we had, and we hope to update this post periodically as our perspective continues to evolve.

I. Executive Summary & Recommendations

When evaluating COVID-19 interventions for importance/scale, our intuition is to look for the following five "scale factors":

  1. Acting quickly, because widespread avoidable suffering is already taking place, because mitigation is more cost-effective when active case numbers are smaller, and because many potentially impactful interventions require lead time to set up.
  2. 🌍Concentrating benefits on the global poor, due to both disproportionate vulnerability and huge numbers.
  3. 😷Reducing the spread of the disease, even in populations where containment is not possible. Of particular interest is driving availability and adoption of cheaper mitigation strategies.
  4. 🔬Scientific research & development in support of any of the above facets of the problem, because a dollar spent on research can unlock orders of magnitude more benefit later. This includes treatments (including but not limited to vaccines) that substantially reduce the severity of the disease; diagnostics; and other areas.
  5. 📊Knowledge and advocacy to inform and motivate policy responses that are more likely to achieve desired outcomes from a global perspective.

For now, we are planning to pursue the following donation opportunities, as we believe they meet many of these criteria and have room for more funding:

In addition, we have identified a number of other organizations doing promising work that have the potential to emerge as top recommendations as we learn more about them and/or as their work develops.

II. Big picture: What's the bad thing that's happening? What could cause less of it to happen?

In this section we lay out the basic moving parts of the current crisis that one could intervene on to produce a better outcome. Parts of this section might be obvious to some readers; however, what is "obvious" to some people can be "surprising" to others, so we think it's worthwhile to just re-state the essential picture.

Two things are going on:

First-order problem: a disease is spreading around, causing illness.

The basic epidemiological picture is as follows:

1. Each person who has the disease infects some number of other people on average.

2. The disease at first spreads exponentially (R0 > 1) within populations of susceptible people who have contagious contact with each other.

3. If no measures are taken to bring R0 below 1 and there is no vaccine, the exponential spread begins to slow down in a population only when a large fraction of that population has been infected, such that the disease starts running out of susceptible hosts.

To get a feel for these dynamics, the simulator at https://ncase.me/covid-19/ is the best pedagogical resource we've seen so far.


Graphic by Kristen Tonga for FRAPPE

What levers can we pull to make direct impacts less bad?

1. Reduce the total number of people who get it. For example:

2. Reduce the amount of suffering per person who gets it. For example:

Second-order problem: disruptions to people's lives and livelihoods

In addition to the direct health impacts on people who get sick, there are indirect impacts. People who are sick or concerned about getting sick will not work, consume, travel, distribute goods, or participate in their communities at the same rate or in the same patterns as before. Additionally, mitigation strategies (such as lockdowns, test-and-trace programs, mandatory face coverings, or education campaigns) will further shape people's behaviors, as well as costing money. These altered patterns of activity and production, and direct and indirect financial costs, are already manifesting as job losses, food shortages, and other disruptions to people's lives and livelihoods. Particularly in poorer countries, the indirect effects of the disease could cause more harm than the disease itself, as they are not only harmful in their own right but worsen many existing social problems (other diseases, hunger, domestic violence, education, inequalities in access to essential services, etc.).

Some mitigation strategies are much more expensive than others, in money and in disruption. In the case of a previous pandemic, for example, one analysis concluded that "[early] contact tracing was estimated to be 4,363 times more cost-effective than school closures ($2,260 vs. $9,860,000 per death prevented)." While we can't assume that these ratios will necessarily hold for COVID, a similarly wide differential among the cost-effectiveness of different strategies would not be surprising.

Furthermore, some mitigation strategies take much more "setup time" than others (e.g. a school closure can be done immediately, but contact tracing cannot be started until tracers are trained and hired), and yet mitigations are best done when case numbers are low (which is true early in the course of disease spread, or after a successful period of suppression). This means that wealthier places can deploy expensive and disruptive methods early on to buy time to set up cheaper methods later while keeping case numbers low throughout the mitigation process; lower-income areas, by contrast, cannot afford to do so as easily.

What levers can we pull to make indirect impacts less bad?

III. Prioritization: Which levers are likely "most effective" to pull on?

When prioritizing interventions, the usual factors to consider from an effective altruist perspective are scale, neglectedness, and tractability. We think that time-sensitivity is another important factor in this case.

Scale

The above "napkin sketch" picture of what's going on yields some quick-and-dirty intuitions as to where the big "scale factors" are.

For one, exponential curves add orders of magnitude very quickly, so reducing the spread of the disease (especially in contexts where it can be done cheaply) is likely to be cost-effective. We emphasize that this is still the case even in communities that cannot avoid a high rate of infection. If a population has not been able to control the disease and cannot afford sustained lockdowns, and therefore may be on track to hit herd immunity before a vaccine is found, we originally entertained the hypothesis that it might not make a difference to the ultimate outcomes to slow the spread. However, we now understand that lowering R0 saves lives in all cases, because it both lowers the herd immunity saturation point, and reduces "overshoot" in which excess infections occur above the herd immunity level. These both correspond with vast numbers of lives saved.

Some strategies are orders of magnitude more cost-effective than others. We believe these cheaper strategies may include wearing masks in public, handwashing, contact tracing when case numbers are low, disease surveillance (i.e. finding undetected cases), and personal protective equipment (PPE) for healthcare workers. (see Juneau et al. preprint). We're excited about interventions that make cheap mitigation strategies more available, affordable, and accessible.

We also expect orders of magnitude could be found in substantially reducing the severity of the disease, through developing, manufacturing, and distributing highly-effective treatments.

Existing thinking about the role of leverage in cost-effectiveness can be applied here too. Borrowing from the framework from Open Philanthropy Project's blog post "GiveWell's Top Charities Are Increasingly Hard to Beat," the following sources of leverage are ways to add multipliers to impact-per-dollar:

We boil this down to five "scale factors": ⏰Acting quickly, 🌍Focusing on the global poor, 😷 Reducing the spread via cheaper strategies, 🔬Scientific research, and 📊Informing & coordinating policy.

Neglectedness, and other properties of the ecosystem & organizations.

We found that estimating neglectedness was critical to our understanding of opportunities, but more challenging than we expected because the landscape of other funders' attention is both difficult to track and evolving rapidly. For example, a simplistic view is that too much money is being spent on "coping with" the pandemic, as compared to "solving" it (see, e.g., the COVID-Zero messaging, and this tweet from Paul Romer). This lens might give the impression that vaccines are currently under-resourced; however, vaccines seem to have attracted a lot more attention compared to other treatments and have received billions of dollars in new investment just this week. The simplistic "coping" vs. "solving" lens also misses that many of the world's poorest people need support to literally survive pandemic-induced disruption, not just "cope" with it.

In addition to the usual ITN analysis, we came to realize that acting appropriately quickly is unusually important to an intervention's effectiveness on COVID-19 mitigation. This consideration is not usually called out explicitly in the ITN prioritization framework. In addition to the fact that exponential curves add orders of magnitude very quickly (as discussed above as a "scale factor"), there's the fact that fast-moving actors are rare; it's much more typical for decision-makers to respond slowly. This means that interventions that need to be undertaken quickly seem more likely to end up neglected, due to a lack of actors who can orient and act fast enough to do them. Slow-moving organizations might be appropriate to fund if the intervention is not especially time-sensitive, but if there is a narrow window of opportunity, and the opportunity passes, money spent on the attempt could be totally wasted. Finding an appropriate match between the time-sensitivity of interventions and the promptness of relevant actors is especially key. As a result, some donation opportunities that otherwise look good might not be effective due to a lack of urgency or readiness on the part of the specific potential recipients.

IV. Specific giving opportunities

We highlight first the top few overall giving opportunities we have found so far at this stage of our investigation. Then we outline other promising candidates by topic area. We emphasize that as an all-volunteer team we had limited time to identify and evaluate organizations, but still wanted to prioritize giving quickly to opportunities we felt we understood well enough.

Opportunities we plan to give to

Fast Grants. 🔬Scientific research, ⏰Acting quickly.

Development Media International. 😷Cheap mitigation, 🌍Global poor, ⏰Acting quickly.

Promising, with some open questions

Global poor

Africa CDC. 😷Cheap mitigation, 📊Informing & coordinating policy, 🌍Global poor, ⏰Acting quickly.

GiveDirectly (International). 🌍Global poor, ⏰Acting quickly.

GiveIndia. 🌍Global poor, ⏰Acting quickly.

Medecins Sans Frontieres. 😷Cheap mitigation, 🌍Global poor, ⏰Acting quickly.

Vaccines, diagnostics, and treatments

ACT Accelerator: 🔬Scientific research, 🌍Global poor, 📊Informing & coordinating policy.

Policy advice/knowledge/resources

Center for Global Development: 🌍Global poor, 📊Informing & coordinating policy.

COVID-END: 📊Informing & coordinating policy.

What about Johns Hopkins CHS?

The Johns Hopkins Center for Health Security has been the most frequently cited top donation opportunity for COVID-19 so far in EA community writeups such as the Founders Pledge COVID-19 Response Fund and SoGive's post on EA Forum [EA · GW]. Pandemic preparedness is core to the center's mission and it has been quite active in the current crisis, having launched a widely followed coronavirus case tracking map/database; it also publishes numerous white papers addressing questions of interest to policymakers and its experts have been prominently featured in the media. There are a few reasons why we have not prioritized it in our own research. For one thing, the profile of the center has risen considerably as a result of the pandemic and we are unsure what additional funding would make possible in the short term (notably, the center does not have an active appeal on its website and actually suggests other organizations for donations instead). In addition, the center's track record in providing high-quality advice does not appear to be unblemished; we noted that its experts explicitly recommended against wearing DIY masks in early March (a position reversed by the end of the month) and were not discouraging people from pressing ahead with travel plans as late as March 6, advice that may have led to costly decisions and missed opportunities during a period when infections were rapidly increasing.

Also considered

Other organizations we considered included Partners in Health, PATH, CARE, the Food and Agricultural Organization of the United Nations (FAO), EpidemicForecasting.org, IDInsight, Y-RISE, Center for Disaster Philanthropy, GiveDirectly (for its US-focused Project 100 campaign), the Emergent Fund's People's Bailout, National Domestic Workers Alliance, Meals on Wheels, Feeding America/World Central Kitchen, and United Way. While we have elected not to pursue these options at this time, we believe they are all doing relevant work and intend to track their activities periodically as bandwidth permits. Complete analysis of all organizations considered is available here.

Certain of these merit mention for donors with specific preferences or priorities that may be different from ours. For example, the Center for Disaster Philanthropy's COVID-19 Response Fund is a good option for donors who want to make a single donation to cover a wide range of interventions globally. For donors interested in helping vulnerable populations in the US, the Feeding America/World Central Kitchen partnership looks attractive on the basis of scale of impact and fast implementation. For US-focused donors with very strong social justice values, the Emergent Fund's People's Bailout could be an intriguing option.

If an organization doesn't appear anywhere in this blog post, it's possible we are not aware of it, or it's possible we were but decided not to investigate it in depth.

Opportunities we wish we'd found more of

Based on our analysis of the big picture, and the important levers we identified, we felt there were a number of "gaps" in the landscape that we wished we could support, but we could not quickly find as many strong organizations dedicated towards those efforts as we would've hoped to see. (Such organizations may exist, in which case we'd love to hear about them.)

V. Addendum: Should you prioritize COVID response over other EA priorities?

We've written this post primarily for the benefit of donors who have already decided to focus on COVID-19 for their own reasons. We have not made it a priority to analyze the relative value of COVID-related donations as compared to other issues or causes. This post should not be seen as taking any position for/against prioritizing COVID-19 over other issues or causes.

That said, we know this question is top of mind for many EAs, and we wanted to offer some brief thoughts on it here.

Evaluating the relative cost-effectiveness of "COVID-themed" vs. "non-COVID-themed" donations is more difficult than if these were wholly separable topics/areas. Instead, the effects of the pandemic itself are intertwined with both donors' actions and the work of the organizations they support in several ways. (This intertwinement seems less pronounced for cause areas such as AI x-risk, and more pronounced for cause areas such as global health and pandemic preparedness).Some of these entwined interactions point against prioritizing COVID:

Some interactions point in favor of prioritizing COVID:

These factors point to a complicated picture that we have not undertaken to disentangle here. Overall, while we would not go so far as to suggest that EAs redirect their giving away from effective charities they already support, we do feel there are strong reasons for EAs to consider COVID-specific giving. Moreover, the situation is so quickly evolving that there is not yet an established consensus about what is most effective to do, so we believe it is important that individual EAs take up the mantle of thinking carefully about what they think is best to do in the current unusual times, rather than exclusively deferring to the opinion of trusted voices in the EA world.

(For another EA Forum post which has a different take, see COVID-19 response as XRisk intervention [EA · GW])

None comments

Comments sorted by top scores.