The GiveWell Blog

Thoughts on “Moonshine or the kids?”

Nicholas Kristof’s recent column argues that

if the poorest families spent as much money educating their children as they do on wine, cigarettes and prostitutes, their children’s prospects would be transformed. Much suffering is caused not only by low incomes, but also by shortsighted private spending decisions by heads of households.

This argument has provoked a lot of strong reactions, for and against. I feel it is helpful to separate three separate questions the article raises:

  1. Is selfish/bad spending by the poor ever a problem?
  2. Is selfish/bad spending by the poor a major/widespread/leading problem?
  3. Is selfish/bad spending by the poor a promising target for aid? (As we’ve written before, we think there is a vital – and often underrecognized – distinction between “biggest problem” and “best target for aid.”)

With this separation in mind, I think it becomes fairly clear that Mr. Kristof has told an interesting story that points to an interesting hypothesis (perhaps worthy of investigation), but that he’s a long way from making a case for action.

I think the distinction between “interesting story/hypothesis” and “good case for action” is also chronically underrecognized in the world of giving.

1. Is selfish/bad spending by the poor ever a problem?

I’m confident that it is. I didn’t personally observe the two “irresponsible fathers” Mr. Kristof talks about, and I don’t know whether he represented them fairly, but I doubt anyone would try to argue that there are no irresponsible fathers among the over 2 billion people living on under the equivalent of $2/day.

People in our own society aren’t immune to selfishness and shortsightedness; I don’t see why we would expect the poor to be.

2. Is selfish/bad spending by the poor a major/widespread/leading problem?

Most of Mr. Kristof’s column is anecdotal, focusing on two men in one village. His only reference to more systematically collected, possibly representative data is to “The Economic Lives of the Poor” (Banerjee and Duflo 2006) (PDF).

We are familiar with this work and have published our own summary of its implications. I don’t feel that Mr. Kristof gives a fair picture of it. It’s one thing to find specific fathers who could be sending their kids to school if they drank less, but when looking at broader data a lot more care is needed. Much of what looks at first glance like “irresponsible spending” could in fact be benign and reasonable. (For example, when looking at the fact that the poor spend money on “festivals,” I have the same reaction as Aid Watch, which asks, “Is it really such a big surprise that the poor also want recreation? That the poor have a life? Including some of the same vices that the rich have?”)

I think the details of the flaws of Mr. Kristof’s interpretation are well covered by this Wronging Rights post.

None of this disproves his claim. It seems like we simply don’t know the size of the problem he’s describing (and, importantly, where and when it’s a problem).

3. Is selfish/bad spending by the poor a promising target for aid?

Even if selfish/bad spending were shown to be a dominant and widespread problem, there would still be the question of whether aid can reasonably expect to do anything about it, and how our prospects compare to our prospects of (for example) vaccinating more children.

Perhaps because he is aware of this, Mr. Kristof doesn’t suggest heavy-handed interventions. In fact, he suggests expanding microsavings and “giv[ing] women more control over purse strings and more legal title to assets” – both goals that have much to recommend them regardless of whether selfish/bad spending is a problem at all.

Bottom line

I credit Mr. Kristof for saying what many may be afraid to. It is possible that many of the poor’s problems come down to selfish/bad spending.

However, the information we have – or at least the information Mr. Kristof presents – isn’t enough to make the argument any more than a hypothesis. And the last thing we need is another instance of pouring resources into a hypothesis as if it’s already been proven.

For those who find his column compelling, the appropriate response is more investigation, not action.


  • Tim Ogden on May 27, 2010 at 3:51 pm said:

    While I agree with most of the post, I think you’re missing some things on the third question.

    First, we need to reframe the question. Do the poor express a desire to build assets and do at least some of them express frustration with their ability to build assets because of temptations of various kinds (incl. temptation goods)?

    The answer to that is unequivocally, “Yes” based on lots of high quality research including but not limited to Duflo and Banerjee’s paper and Portfolios of the Poor.

    Then the next question is, “Are there low cost interventions that have proven efficacy at helping those who want to build assets and avoid temptation do so?”

    Again here the answer is unequivocally, “Yes.” See for instance Karlan on commitment savings, Karlan on Commitment contracts to quit smoking, Dupas on Savings accounts, Duflo and Banerjee on timing of fertilizer sales and Linden on timing of conditional cash transfers based on school attendance.

    In each of these cases the costs are minimal and the evidence is strong that these interventions help the poor invest in building assets be that financial, physical or human capital.

    Now there is a final question of relative efficacy of these interventions versus others. I have a bias towards these types of interventions because they are at least neutral and often positive in terms of reinforcing the agency of the poor.

  • Holden on May 28, 2010 at 10:36 am said:

    Tim, good point; we agree that there is a pretty good evidence base for the kinds of interventions you describe and we are positive on them too.

    These sorts of programs seem to apply to “shortsighted” spending though not so much to “selfish” spending.

Comments are closed.