The GiveWell Blog

Donor impact vs. donor attribution (or, does your $120 really buy a sheep?)

One of the questions we struggle with a lot is the question of what impact a donation has – i.e., what happens because of your donation that wouldn’t have happened otherwise? In other words, what do you get for your dollar?

It’s a tricky question, especially for relatively small donations going to relatively large organizations. A future post will discuss a couple of approaches we’ve taken to answering this question; for now I want to focus on an approach we haven’t taken, that of literally attributing each donation to a “set of purchased items.”

A good example is this GlobalGiving project, which promises to educate 1 woman for every $45 you give. Organizations like GlobalGiving, Kiva, DonorsChoose and Heifer International make similar offers.

We’ve never been sure what to make of projects like these. For example, I see two interpretations of the “teach 4200 women” project:

  1. The pitch is literally true. For every $45 that comes in, one woman will be educated. Therefore, if the project is underfunded by $45, one qualified and interested woman will be turned away from the class.
  2. Funds are fungible. The organization conducting the class (Women for Women International, a large international aid organization) has already drawn up a full budget for the project, and will fund it with “general” funds if necessary. Therefore, for every $45 you give, the organization moves $45 of its “general” funds back into the “unrestricted” pool, meaning that your $45 is effectively an unrestricted donation to the organization running the class.

I’m guessing that #1 is the one that donors subconsciously picture when they are drawn to the “tangibility” of a project. It’s the scenario under which a donation is literally matched to a discrete person or item. Yet if #1 were the case, this would seem a horribly inefficient way to run projects – spending all the overhead to pay staff, set up the class, etc., and leaving one woman out because of a $45 shortfall. (Or in the case of Kiva, setting up a micro-bank and then turning away a screened, qualified borrower because of a missing $850.)

Yet if #2 is correct, your donation is really an unrestricted donation to a large organization; to understand your impact, you need to understand the entire organization.

GlobalGiving’s FAQ suggests a mix of the two scenarios: sometimes a funding shortfall is covered by other funding, and sometimes a project is scaled back or canceled when funding falls far short. As a side note, I would guess that when a project is scaled back, the “cost per person” generally increases (since the program’s overhead can’t fall fully in proportion to people served).

Other organizations are explicit that #2 is the case (see small print on this “buy a sheep” page).

All in all, I’m skeptical of any claim that says “your $1000 buys X.” It’s a good way to make things feel tangible, but a donor truly trying to understand his or her impact should take a different approach. (Some possibilities will be discussed in a future post.)


  • Ron Noble on October 4, 2008 at 8:38 pm said:

    I sympathize with the dilemma raised in this post because I suffered it myself. After initially considering giving to Population Services International (based on learning about them from GiveWell), I realized I was giving a drop in the bucket to a huge organization. I decided to go ahead and give to PSI, because I’m convinced that the expected utility of my gift, per dollar, is higher than it would be with most things I could give to.

    In hesitating to give to PSI, I was succumbing (temporarily) to what is called “the certainty effect.” The idea is that people over-value certain, or apparently certain, outcomes when making decisions.

    Another problem with the certainty effect is that the “certainty” is usually an illusion. In the case of $45 to educate a woman, let’s assume 100% certainty my money will educate exactly one woman. Why am I willing to pay to educate one woman? Because I hope good things will come of that education. But they may or may not. She might get a job she wouldn’t have otherwise, resulting in her children being spared many nights of painful hunger. Or she might fail to do so, and the time away from other work actually hurts her children. So my “certainty,” in terms of doing some good with my one donation of $45 is an illusory—but I still might donate, because I think it is a good bet.

    If I donate to a large organization that educates women, I’m also gambling. It might turn out that my $45 doesn’t result in a single extra woman being educated. Or, it might result in 50 more being educated, if my $45 gives the organization just enough funds to open a school they otherwise couldn’t. The total funds spent divided by the number of women educated probably reasonably estimates the expected amount it takes to educate a woman, on average.

    I think many people will share the feelings I had when considering PSI and that I gather from this post the Holden and Elie have had. But that raises another troubling question: GiveWell wants to give the public accurate information about charities and the good they do. What would happen if GiveWell made a point of informing donors that there individual donation would only on average do X amount of good, and might do none at all? Might this not result in people (arguably irrationally) giving less? For each individual donor and donation, that might save them donating a small amount that ended up doing no good. But in the long run, the results of many donors being spared this possibility would almost certainly be bad.

  • Adrian on October 5, 2008 at 3:03 pm said:

    But hasn’t there been research showing that people prefer donations phrased as “help this person” rather than “help people like this one” and both more than “help these people over here” due to this certainty effect? It may make significant differences in fund raising if the offer is phrased with certainty — “buy this person a sheep for only $X” vs. “we need $XY to by Y people sheep.”

    On a completely different note, have you guys considered looking into open source pharmaceuticals?

  • Stephanie Fischer on October 7, 2008 at 1:47 pm said:

    Holden brings up some interesting issues. At GlobalGiving we pride ourselves in demystifying philanthropy – enabling donors to provide project-specific funding for great work to take place around the world.
    On the donor side, we offer “donation options,” which illustrate the impact specific dollar amounts will have on a particular project. Many donors find that donation options help them get a tangible sense of how their contributions will be used and are indicative of the “bang for the buck” of your donation (e.g. educate a girl for $45). There is some fungibility of donations, however NOT from project to general operating budgets. Each project’s budget includes the various costs of getting a project done, and the donation options show a sampling of these.

    In working with NGO leaders to identify the projects they wish to post and to assist them in their presentation, we encourage them to pick projects that are discrete, but where incremental funding can create incremental benefits. Project leaders can post projects that will only be executed with complete funding (e.g. a capital project), and only accept funds when the project is fully funded. However, most project leaders choose to accept incremental funding for their projects. GlobalGiving disburses funding on a monthly basis, and projects accept funds for these specific projects, which can include project staffing and overhead expenses, but NOT organizational overhead/general administration.

    Typically, Project Leaders use a variety of channels to raise money for their projects. When you donate to a project through GlobalGiving, your donation will always be directed to that project and its defined activities. However, depending on the overall amount of funds projects raise, both through GlobalGiving and from other sources, Project Leaders may pursue a combination of different activities, allowing the project to deliver a complete suite of relevant products and services to its constituents.

    (For example, a project provides school meals and uniforms for 25 students in Kigali, Rwanda for a year. Donation options for this project include $20 for a month of meals and $30 for a school uniform. Even though the project is not fully funded, if donors have already given 25 uniforms for the year, incremental funds raised will go towards ensuring a full year’s worth of meals for the 25 students – even if a donor chooses the $30 school uniform donation option. What we can ensure is that students will be educated in Kigali when the project finishes raising funds, and provided with meals and uniforms.)

    Sorry for the long-winded comment, but it’s a complex subject with many nuances – hopefully we’ve been able to provide insight into how we work. Thanks for initiating this dialogue.

  • Holden on October 10, 2008 at 7:49 pm said:

    Ron: you raise a valid concern about whether publishing true information about charity would (irrationally) lower giving. We’ve addressed the issue before here.

    Adrian: by “looking into open source pharmaceuticals,” do you mean initiatives like this one? (Note that our current focus is on direct aid, and we are excluding research funding for the time being.)

    Stephanie: thanks for the comments. I agree with you that the #1 and #2 I presented above aren’t the only two possibilities, and that there is often likely a middle ground where an individual donation has some impact on the probability of a project’s being carried out (and its size), though that impact can’t be captured as simply as “$45 educates a woman.” To be clear, I think GlobalGiving’s FAQ is appropriately transparent on this point, but that many donors nonetheless have a simplified picture in their heads.

Comments are closed.