The GiveWell Blog

Eat a peanut, save a child!!

What would you do to make the world a better place?

Would you …

Of course you would. And of course you can’t.

If people buy from (or look at ads for) companies based on their donations to charity, I promise the prices will rise by however much they’re donating. When sponsors give to charity for every home run the White Sox hit, they have a range in mind – and you can bet that if they end up owing way more than expected, somehow this is going to come out in next year’s pledge (even if it’s through their insurance company’s quote, etc., etc. …)

I’ve debated the specifics of various schemes along these lines before. Right now, I want to drop my usual preference for concreteness, step back from all the mechanics, and just be as general as I can be.

If someone tells you you can make a difference without either giving up anything valuable or doing anything useful, they are wrong.

When trying to figure out which schemes work and which don’t, it seems that’s about all you need to know.

As for the notion that these kinds of schemes “raise awareness” … man, I’m sick of hearing about “raising awareness.” If you’re not, just – read this.

I’m surprised by how many otherwise intelligent people get pulled in by promises of saving the world by yawning. Sure, they’re tempting, and it can be very complicated and confusing to figure out precisely why they don’t work … but the fundamental problem couldn’t be more obvious. Please don’t fall for this stuff. That’s all.


  • Holden on October 31, 2007 at 8:21 am said:

    For those of you interested in more detailed analysis of how these kinds of things usually work, I’m verbatim-posting this exchange between me and Elie over gchat.

    Elie: OK. I was thinking about your blogpost
    Its obvious that no cmpy is doing one of these programs out of the goodness of their heart
    but, it also seems that if a progarm exists, there’s no reason not to donate your taco (except that you could prob use that time/energy better).
    Holden: no
    it’s all complex and game theoretic
    Elie: Basically, the reason this seems silly to me is that i dont value extra money givne to charity much
    Holden: that’s why i didn’t go into it
    here’s the thing
    Elie: Well, i don
    Holden: if i say i’ll give $100 for you to scratch your armpit
    what i am really doing is paying for your eyeballs
    /your attention, and the prospect of your purchasing something else from me
    like the guy who held the vote on HR #756 ball
    Elie: yea, ok
    Holden: so
    if you try to manipulate me by taking my $ and not buying my product
    all that does is affect the ROI for these kinds of projects
    which will eventually get priced in
    sure there could be little fluctuations, but they go in either direction because when i do this i try to anticipate how many people will “game” the system
    so more “gaming” does not lead to more charity in the long run
    just like how the White Sox hitting more HR’s means more given to charity that year, but moving Comiskey to Denver would not result in more given to charity over the next decade
    nor would signing A-Rod
    nor would anything, really, in expectation
    Elie: if i assume that cmpys price things perfectly, then i want to clik each time i hear it to try to make them give more now.
    Holden: but if lots of people think like that, dude, then the pricing will change for future charity projects
    Elie: well, of course, it’s true in the long run. cmpy’s make profit; they don’t give charity (excpet to enhance profit)
    Holden: companies will donate less per click, etc.
    your other option of course is not to game the system, is to actually prefer their products, but then you’re just donating in an extremely annoying and sideways way
    Elie: or, they might start advertising another way
    Holden: yeah, whatever
    Elie: i want them to advertise by giving charity
    Holden: one way or another it all washes out because they have a certain amt they’re willing tos p end
    Elie: but, it depends where they spend it. I’d rather they spend advertising budget on charity, not NYTimes.
    Holden: so what are you going to do about it?
    buy products from companies that use that form of advertising, instead of from companies whose products you prefer?
    then you’re donating
    Elie: talk up how great these programs are so more comanies do it
    Holden: THAT might work if you successfully fool them. but that’s difficult. doing the click (the part that’s not value-added or difficult) will not work because it will actually worsen their cost:benefit
    it’s so complicated, that’s why i say you have to start with the principle I put in purple. just like if someone sends you a fancy math equation for a get rich quick scheme …
    you won’t always be able to figure out hte math
    but it will be easier armed with the fundamental principle, before you go in, that this scheme is a ripoff
    Elie: dont agree. it’s not necessarily a ripoff. i agree that someone is trying to rip me off. But f them. im going to get them to give to charity, nad if that reprices charity ads in the future, so be it.
    Holden: yeah it does
    there’s no expected benefit
    just some fluctuations
    Elie: but, there’s something very certain about my clicking now. i dont know who’s going to do what in the future. i click now, and money goes to cute kids
    Holden: your manipulation could lead to more going to charity but it could also lead to less. the distribution isn’t necessarily even.
    no, they price that in though.
    this is my point.
    it might not be equally likely that you cause it to increase or decrease. increase might be more likely.
    cause they try to put in a “manipulation premium” before they even start.
    but in the event that a lot of people behave as you do – more than expected – they will OVER compensate for this next time by a LOT
    to deal with the risk on their end
    so you might hav ea 99% chance of giving an extra $1 to charity and a 1% chance of being the straw that breaks the camel’s back and causes $100 less to be given next time around
    but there is no “certainty benefit”
    it’s 99% vs. 1%, not certainty vs. uncertainty
    the size of the cost is guaranteed to compensate for its low likelihood
    Elie: the things i dont get are: 1) seems positive for a cmpy to advertise in charities rather than nytimes and 2) given that there aren’t real distinctions between, say, t-shirts, im happy if the GAP gets ppl to give a little money to Africa
    Holden: if there really weren’t distinctions in t-shirts, GAP wouldn’t do the (RED) thing, it would just cut its price by 1c
    it is positive for a company to advertise via charity instead of NYT … but that’s the company’s move
    Elie: the distrinciton is perception. what status the clothing provides. So, GAP markets that giving to AFrica provides cache, that gets ppl to give more
    Holden: how effective it is is how effective it is. you can’t manipulate the copany into doing it … except by donating.
    Elie: that’s true
    Holden: so someting like (RED) might be a net benefit for the world … but you are not accomplishing anything by falling into line.
    Elie: well, that’s a clearer statement
    Holden: except to the extent you give something up (quality of shirt), in which case you may as well donate
    Elie: im happy about RED because it gets people to donate
    but, i wouldn’t buy RED (unless i want hte cache), because id just donate
    Holden: right
    Elie: well, this makes more sense to me now
    Holden: yeah. i was just thinking about when we got my mom a RED ipod.
    we didn’t give a flying fig what color it was, so it seemed good to make it RED.
    but again, because we didn’t give a fig … that means the RED didn’t actually make us more likely to buy an ipod
    so to the extent that more people are in our camp, RED just costs apple $ without giving it benefits
    which means it’ll do less next time around
    Elie: right, i understand this

  • michael vassar on October 31, 2007 at 10:51 am said:

    The whole scenario is rather directly analogous to the non-vegetarian who argues that the animals he buys in the store are already dead so he isn’t killing them or causing them to live in factory farms when he buys them in the store. Obviously he is ignoring the expected economic consequences of his choice to buy meat, which signals demand to the stores and companies etc.

    If (RED) spends their money efficiently, which seems somewhat unlikely a-priori, then by buying their t-shirts you are not just donating to them but making it convenient (and fashionable?) for you to spend your days as a living billboard advertising them, and hence, by the hypothetical, advertising efficient giving.

  • Sean Stannard-Stockton on October 31, 2007 at 10:54 am said:

    I agree with you in theory, but think things work out a bit differently in practice.

    Economic theory says that the more of their earnings a company pays out as a dividend, the lower the growth rate will be. The idea is that they will have less money to invest in growth opportunities. But some research shows that higher dividend payouts are correlated with higher growth rates. It seems that when management is deprived of some of the earnings, they are more choosy about the growth opportunities they pursue.

    Similarly, some research shows that companies who give a high percentage of earnings to charity actually have better performing stock prices. My guess is that charitable giving is not an expense like rent, that it actually produces positive returns to the company.

    Anyway, I think that 1) the “cost” of the programs you describe, may in fact generate positive ROIs for the company and not force them to raise prices to offset the costs, 2) to believe that there is no “free lunch” you must believe in efficient markets. I believe that so little work has been done in this area that there is likely to be a lot of low hanging “free lunches” that will result in consumers, businesses and charities all gains. A zero sum analysis such as the one you put forward only makes sense if there are no inefficiencies available to be squeezed out.

  • michael vassar on October 31, 2007 at 9:53 pm said:

    Sean: The claim was not that the company looses money with such programs, but rather that they intend to make money with such programs so if you try to game the system by making the programs into money-losers they will (in expectation, like the expectation of an extra .001 cows dying from a hamburger or whatever) discontinue them.

  • Holden on November 1, 2007 at 9:39 am said:

    I’m wary of getting too bogged down in the details here, because we have so much else to do and because I firmly believe that focusing on the details/mechanics – instead of the big picture principle in purple – is a mistake. But here’s quick shot at it.

    We need to distinguish between:

    1. The impact of the program itself (launching “Scratch your armpit, save a child”).
    2. The impact of your activities that DO involve giving up something valuable or doing something useful. These include publicization/promotion/telling friends about the charity.
    3. The impact of your activities that DON’T involve giving up something valuable or doing something useful – i.e., scratching your armpit.

    #1 has value: the RED Campaign has resulted in more money going to the Global Fund, because it was a decision by companies to use charity as advertising rather than other channels. Charity can be an effective form of advertising.

    #2 has value, although it seems that in most (not all) cases, once you isolate this value you realize there is a better way to carry it out. The advantage of talking to your friends about a cause, rather than armpit-scratching, is the same as the advantage of donating to charity rather than buying a more expensive charity-subsidizing product: you get to choose the cause you support.

    (#2 is what Michael’s first comment addresses – the value in publicizing a cause.)

    #3 is the one that doesn’t have value, unless (as Sean suggests, and perhaps as Michael is suggesting in part of his comment) there is a systematic inefficiency – i.e., that every time a company runs a promotion like this, they repeatedly underestimate participation, and don’t adjust for it the next time around.

    There is simply no reason to think this inefficiency would exist. I’m not claiming markets are perfectly efficient – I’m claiming that we don’t know what direction any inefficiency runs. It’s clear that something like the “I’m” campaign is launched with full knowledge that people will try to “manipulate” it by typing “I’m.” So do they overestimate this risk, and give less than they would otherwise, or underestimate it? When well-meaning Sean types “I’m” 4000 times and puts them on the hook for more than they thought, do they under-react to this next time around, or over-react? I say we don’t know, and therefore it’s appropriate to assume zero effect (in expectation, though the distribution is wide and complex). Sean says he does know: they underreact. I don’t see where that comes from.

    Conclusion: my decision to pay $1000 for everyone who scratches their armpit is a donation, and possibly a way of getting attention. Your decision to tell your friends about it is valuable. Your decision to scratch your armpit is not. Scratching your armpit 3000x is, in fact, a waste of your time.

    As a side point, Sean, I am very skeptical of the financial studies you point to. It’s always dangerous to try to analyze an intelligent system that responds to analysis, like stock prices (and like the charity-incentive systems we’re discussing). It’s especially dangerous when there are thousands of conceivable patterns you could examine, and only the ones that show a one-in-a-thousand chance are publicized and celebrated. I’m more comfortable with my “we don’t know which direction the inefficiency is” analysis unless I see unbelievably overwhelming evidence in the opposite direction.

  • Richard Potter on November 24, 2007 at 5:55 pm said:

    Whoa! I read the blog entry and thought I’d make a comment. Then I read the comments above and forgot what I was going to comment about! Talk about getting bogged down in the details.

    So I went back to the blog entry and remembered. Here is the statement that made me sit up and say “YEAH!”:

    “If someone tells you you can make a difference without either giving up anything valuable or doing anything useful, they are wrong.”

    True statement. But not complete. People need to receive the message that giving time and money benefits the donor as well. Hedonism (the doctrine holding that behavior is motivated by the desire for pleasure and the avoidance of pain) is not a bad thing. Research indicates that giving stimulates the pleasure centers of the brain much like sex and drugs (and IMHO rock & roll). But because the concept is so counterintuitive, not enough people are doing it. Giving, that is. There’s plenty of sex and drugs and R&R.

    I love your blog and your web site and am recommending GiveWell to people in my circle of influence. If you find a way to promote the hedonistic aspects of philanthropy it will be even better!

  • This analysis ignores the extent to which people’s ability to give to charity is limited by willpower rather than money, i.e. some people maybe have $10 to give, but their brain won’t let them give it, because losing money feels “icky”, but they can spend an hour clicking links (valued at, say $15 in terms of the amount they could earn in that hour) for charity.

    No, they can’t just work another hour and donate the $15. They’d have to keep it, because losing money feels icky.

    Fascinating post, though.

  • Holden on December 2, 2010 at 4:11 am said:

    Roko, I agree that some people may be more inclined to volunteer $15 worth of labor than to give $15. However, this post wasn’t intended to apply to all donations of labor, but only to actions that are clearly not creating any kind of value. I think the point stands regarding those actions.

Comments are closed.