Wednesday, May 25, 2011

Gift goats (HINT: Look them in the mouth!)

Caltech economist Jean Ensminger has spent many years studying African economies from inside and out. I met her when I visited Caltech a couple years ago, and she told me one really memorable story.

The story concerns a goat restocking program; donations were supposed to buy goats for poor villagers, to build their herd up to a self-sustaining size. Instead of going directly there, though, the money first went to a board of village members, who were to use their local knowledge to purchase the goats. And here's the breakdown of what they actually did:

Most of the money went straight into their pockets.

Most of the remaining money went into the pockets of high-ranking village members who were not on the board.

The remaining money did go to the poor villagers. In particular it went to buying them sick (and thus discounted) goats...

...which then got the original goats sick...

...and then there were no goats.


If I remember correctly, she estimates that upwards of 90% of charitable donations going into Africa end up in the wrong hands. But please check your condemnation at the door; these "privileged" board members are still the second-farthest thing from rich fatcats, and what would you have them do with their windfall? We can whine all day about how people should behave, or we can accept how they do behave and design institutions that take proper account of that behavior...


  1. One of my favorite quotes of all time, and one I continually remind myself of, is this one from Bucky Fuller:

    Don't fight forces, use them.

  2. Exogen(e)ous CombusionMay 26, 2011 at 12:10 AM

    Every people gets the government it deserves.

    The only more wasteful thing than to pour useless monies into corrupt governments whose people won't overthrow them is to use it to fund economists who *think* they can outsmart anyone.

    One great anecdote:
    The Enron guys weren't at all afraid of "regulatory fixes." They knew they were smarter than any politician or economist that could put down a fixed set of hoops for them to jump through. (And they were!)

    When the smartest guys you know in academia use sixteen variable, fifteen parameter models, recognize they are like cavemen with tools compared to entrepreneurs with catapults. Engineers don't need to know why something works to use it (we were firing cannonballs accurately centuries before we understood why they kept moving).

  3. Or our favorite quote Hayek quote,

    "The curious task of economics is to demonstrate to men how little they really know about what they imagine they can design."

    But there's no problem here with including in the optimization problem the fact that the institution designer has limited information/processing ability. In this case, for example, the institution *at worst* could be "just don't give goat money to the village." Since they could just do that, which is better than the previous policy which actually did harm. On the other hand, they could also try to design something even better. Perhaps they could decide to buy goats themselves, and give them directly to the poor villagers. Obviously there are nonnegative returns to Ensminger going in and actually observing what happens to funds, obviously she will come out of that with a *better* idea (in expectation at least) of how to design a mechanism than before. In the worst case, you could imagine iterating the process, trying a bunch of things and observing which one works the best, and going with that.

    It is one thing to assert that we can figure out an optimal mechanism. It is quite another thing to assert that we can do nothing, which is tantamount to saying that we generally cannot tell better from worse, or even that in fact we systematically tend to make things worse.

    A realistic understanding of one's understanding is necessary though. You could argue that economists tend to put too much faith in their overly simplistic models, and that it can systematically lead them astray if they treat the world like it is too absent the complexities that their models aren't accounting for.

  4. Exogen(e)ous CombusionMay 26, 2011 at 3:20 PM

    Improving "local" efficiency, as I shall argue, increases "global" inefficiency.

    "It is one thing to assert that we can figure out an optimal mechanism. It is quite another thing to assert that we can do nothing, which is tantamount to saying that we generally cannot tell better from worse, or even that in fact we systematically tend to make things worse. "

    I argue that whenever we try to "improve policy" given that it exists, we do make it systematically worse as a rule.

    We're in a world where a bunch of stationary bandits steal from us and pour our money down the drain (say, in the production of whiskey, which they enjoy). Read: the realities of political economy.

    Making their still more efficient inevitably leads to more effort being put into banditry.

    Better to make taking money from me and buying corrupt people goats with it more inefficient, rather than more efficient.

  5. And thus it is written: thank god taxation is inefficient, or what would stop them from taxing 100%?

    It is a good point, but I feel it is also too sweeping. It may be often the case, but it hardly holds in generality. For example, on the particular topic of *goat allocation*, i.e. the topic my comment was referring to, it need not be that improving the system will make matters worse. If the money comes from donors who think their money is being handled well, then they *start out* by giving too much, and getting them more for their money's worth is a good thing.

    Anyway, I am happy to accept the theoretical point of your criticism. Without seeing a whole lot of evidence that doesn't reek of selection, I will *not* accept the general empirical claim that almost always, increased local efficiency is globally undesirable. In any case, I will continue to make similar theoretical points as above, and you can interpret their *empirical validity* as conditional on the actual empirical validity of your own theoretical point.