The GiveWell Blog

December 2019 open thread

Our goal with hosting quarterly open threads is to give blog readers an opportunity to publicly raise comments or questions about GiveWell or related topics (in the comments section below). As always, you’re also welcome to email us at info@givewell.org or to request a call with GiveWell staff if you have feedback or questions you’d prefer to discuss privately. We’ll try to respond promptly to questions or comments.

You can view our September 2019 open thread here.

Comments

  • J garcia on December 14, 2019 at 5:36 pm said:

    I have a question about fundraising.
    When a given charity spends 1 extra dollar in advertising, promotion or fundraising, that increases their donations by some amount. Presumably, a part of it is from people who would not have donated otherwise, and some is diverted from other charities. Are there any estimates of this cannibalization effect? Is it likely to be important? Is this something people worry about?

  • Catherine (GiveWell) on December 16, 2019 at 5:30 pm said:

    Hi J,

    Thanks for your question!

    Our mission is to direct donations to effective giving opportunities. We think that reallocating a donation from a less-effective to a more-effective cause—as a result of marketing or other reasons—is generally a positive thing, if it enables a donor to accomplish more good in the world. (Though if you are a major, repeat donor to another organization, we would advise communicating with that group before shifting your giving, to help with their planning.)

    We know that some of GiveWell’s donors are giving to charity for the first time, some are increasing their total donations to support our top charities, and others are reallocating from their previous gifts to our top charities. We published some data on this in 2015: see “How has GiveWell changed donors’ giving” here.

  • In allocating grants for 3Q 19, you determined that HKI’s program was expected to be 28 times as cost-effective as cash transfers (“28x cash”). The numbers for other top charities, other than GiveDirectly, also seem to be at a large multiple of GiveDirectly’s unconditional cash transfer program.

    In light of the order of magnitude difference, do you plan to reconsider whether the unconditional cash transfer programs should be the informal baseline for cost-effectiveness for a top charity? Stated a different way, if you had the other seven top charities and were evaluating GiveDirectly today, would it make the list — or has it effectively been grandfathered in?

    I could see the case for creating a new list in between top and standout charities. I feel people are prone to assume that charities within a category have similar effectiveness, which doesn’t seem borne out by the more detailed analysis for those who dig into it. So maybe top = current criteria + significantly more cost effective than cash; high-performing = same but about as effective as cash, standout = current definition?

  • Catherine (GiveWell) on December 26, 2019 at 10:12 am said:

    Hi Jason,

    Thanks for your question! You’re absolutely right that we don’t think all of our top charities have equally pressing or cost-effective funding needs. A major part of our research process is prioritizing among our recommended charities so we can direct funding to the places that need it most.

    Over the years, we’ve seen that having two lists—top and standout charities—has led to some confusion about the differences between the lists. I’d guess that adding a third list would cause additional confusion. It would also be hard to know which list to allocate charities to. We break out charities’ cost-effectiveness by country or project, so it would be challenging for us to rank a whole charity as being at the top tier or “high-performing” tier when we might, for example, estimate that one of its funding needs (such as expansion into country A) is 80x as cost-effective as cash transfers, while another (such as continuation of the program in country B) is 5x cash transfers.

    Rather than creating new lists of charities, I’d guess that we should work to communicate more effectively about our relative ranking of our top charities’ funding needs. We’ve aimed to make our prioritization transparent on our website (see, for example, here). We have also more highly emphasized the option for donors to give to “Grants to recommended charities at GiveWell’s discretion” in recent years. Donors who choose this option can be confident we’ll allocate their gift to the top charity or charities that we believe can use it most cost-effectively at the time it is granted, no matter the overall threshold for our top charities list.

    I’d also note that part of the reason GiveDirectly’s cost-effectiveness estimate is currently much lower than others is because of parameters we’re highly uncertain about, such as moral weights, and the way we model how different outcomes (e.g. health, wealth) interact. This reduces our true best guess of the difference between GiveDirectly and our other top charities, relative to our estimates.

    Finally, not all GiveWell staff members agree about the threshold we should apply for top charities. You’ve hit on an important question, and one that we discuss internally, as well!

  • Oscar Delaney on December 27, 2019 at 7:11 am said:

    I see in your cost-effectiveness analysis that Evidence Action’s Deworm the World initiative is significantly more cost effective than any other charity listed, specifically it is five times better than the HKI VAS program (~60.6/12.6), which is where my donation went (from the Q3 discretionary funds allocation). My understanding is that you chose not to give to DtW because it’s funding needs had already been met for the next few years. However if this intervention is so effective, surely they could expand to more countries or somehow scale up their work with further donations? It seems strange to me that DtW couldn’t find a good way to use more money. I mean even if they have already picked the lowest hanging fruit in terms of countries and interventions I would have thought their next best wouldn’t be five times worse.
    Sorry if I have missed the explanation for this elsewhere.

  • Catherine (GiveWell) on January 2, 2020 at 1:47 pm said:

    Hi Oscar,

    You’re right that we have not prioritized directing funding to Evidence Action’s Deworm the World Initiative, even though its current work is very cost-effective. This is in part due to our understanding that funding directed to Deworm the World Initiative will largely be used to build the organization’s reserves, rather than being used to directly implement or scale programs in the next three years. We note in our review: “We expect Deworm the World to have opportunities to spend $3.5 million more than we expect it to receive over the next three years. Funding beyond this level would allow Deworm the World to build its reserves.” Deworm the World currently holds more than $3.5 million in reserves.

    You ask, “However if this intervention is so effective, surely they could expand to more countries or somehow scale up their work with further donations?” Deworm the World has chosen to prioritize strengthening these reserves over additional expansion due to its view that longer funding runways and reserves will ensure stability for its program and contribute to its long-term success. We generally defer to charities’ spending priorities when deciding how to prioritize allocating funding (though we may incorporate our views on how well they prioritize spending into our qualitative assessments), and so we aren’t prioritizing sending funding to Deworm the World at this time, even though we believe its work is highly cost-effective.

    In addition, our estimate of Deworm the World’s current cost-effectiveness is driven in part by its work in India and Kenya, its first two programs. The cost-effectiveness of deworming programs can vary widely, due largely to differences in rates of worm infections in different locations as well as how much funding governments contribute to the programs. We model the deworming programs that Deworm the World has started in Nigeria and Pakistan as significantly less cost-effective than its work overall (where India receives a lot of weight) and than our top charities Helen Keller International’s vitamin A supplementation program, Malaria Consortium’s seasonal malaria chemoprevention program, and Against Malaria Foundation. This suggests that if Deworm the World were to expand further (which it is not prioritizing doing), its next priorities would likely be less cost-effective than the programs we are recommending on the margin.

    I hope this helps to clarify our approach!

  • Colin Rust on January 5, 2020 at 5:24 pm said:

    Catherine writes:

    [P]art of the reason GiveDirectly’s cost-effectiveness estimate is currently much lower than others is because of parameters we’re highly uncertain about, such as moral weights, and the way we model how different outcomes (e.g. health, wealth) interact. This reduces our true best guess of the difference between GiveDirectly and our other top charities, relative to our estimates.

    I don’t understand. The first sentence is about uncertainty. The second sentence is about expected values. How does that follow? The dramatic uncertainty in the estimates might well reduce your confidence that e.g. AMF is better than GD; is that what you mean? Otherwise, I don’t see whey it would have a big impact on your best guess of the ratio of their cost effectiveness. (It is true that your best guess of the ratio is technically not the same thing as the ratio of your best guesses, but I don’t think that’s a material issue here.)

  • Colin Rust on January 5, 2020 at 6:04 pm said:

    Give Directly seems like a thoughtful, well-run, efficient charity. So I don’t expect you could find another charity with a similar model that is materially more cost effective. But I do wonder, have you looked at other models in the cash transfer
    space?

    In particular, remittances are a massive transfer — estimated by the World Bank at $529 billion in 2018 — to individuals in low- and middle-income countries. So even modest improvements to the efficiency and/or convenience of remittances are potentially really high impact.

    My immediate prompt for asking about interventions around remittances was this crazy story. Apparently for remittances from US to Nigeria, for some time it was a better deal to buy a gift card, sell it for bitcoin then convert to local currency than to just use Western Union. That’s no longer cheaper, but many people still use this method especially for smaller amounts because Western Union turns out to be less convenient than this byzantine run-around! To my mind this really suggests how inefficient and inconvenient — and therefore potentially ripe for disruption — this huge sector is (or at least can be). Of course it’s not obvious whether the right player would be a charity, a regular for-profit or something in between (double bottom line, social enterprise). Anyway, I’d be interested if you have any thoughts in this space.

  • Catherine (GiveWell) on January 8, 2020 at 12:35 pm said:

    Hi Colin,

    On our cost-effectiveness estimates:

    I’m happy to clarify. Our top charities list breaks down into two categories: (1) charities that increase incomes or consumption and (2) charities that avert deaths. We’re more confident in the modeled differences between interventions within a category than those across categories.

    Counterintuitively, even if each of our cost-effectiveness estimates is individually unbiased, we would expect the option that looks better on our cost-effectiveness analysis (charities that avert deaths) to be overestimated because of a phenomenon called the optimizer’s curse. A paper discussing that phenomenon is here and a simpler explanation is here.

    If the error terms in those estimates are uncorrelated (which is more likely if e.g. they depend on different moral weights), we’d expect the optimizer’s curse to be a worse problem—and thus that the true difference between GiveDirectly and other top charities (which we think are more cost-effective) to be smaller than what we estimate.

    On remittances:

    We haven’t done research on remittances so far, although we think we should likely have this on our long list of interventions to consider. We are aware of Wave, an organization that aims to streamline remittances. Sorry to not have more to add at this time!

New Comment

Your email address will not be published. Required fields are marked *

*