Underpants games

One of our Liberia studies is tracking 1,500 ex-combatants over two years. In the beginning, these particular ex-coms were illicitly mining diamonds or gold, occupying a rubber plantation, or making other mischief. Half are now in an agricultural training program that will also set them up with land and tools.

We are surveying both groups, and we have two staff doing qualitative interviews full time. That seems like enough data for one project. But when behavioral economist Julian Jamison came to town to work on a street youth project with me, I persuaded him to design some behavioral games to measure risk and time preferences.

The problem: the agricultural school didn’t want us giving out any cash (or quickly cashed-in rewards, like phone credit). They tried sweets and toiletries, but nothing seemed to elicit widespread excitement or desire.

Then star research assistants Rebecca Littman and Camelia Dureng stumbled upon the miracle good that everyone wants: underwear.

Turns out, everyone in the world wants more than one or two pair of underwear. It’s not as universal as cash, but it seemed to do the trick.

Here’s one of the ensuing risk games. (“Lucky ticket” is Liberian for lottery.)


“Make your decision: Briefs/panties OR lucky ticket.” Sounds like a paper title to me.

5 thoughts on “Underpants games

  1. Why no cash? I’ll reserve judgement until after the answer, but it smells like condescension. These people are doing you a service, I hope the argument against paying them like you would anyone else i.e. in cash, is a pretty good one.

  2. That was the NGO partner’s decision. In their defense, there has been a drug and transactional sex problem on the campus, and they were fearful of exacerbating that. Boarding schools definitely tend to the paternalistic side.

  3. This is great. I wonder if something like this might not actually work better than using cash. The problem with money is that people use it for all different reasons, and it would be hard to control for all the potential behaviors that might not represent risk preferences. For example, maybe non-ex-coms who come home with a pile of money they won in a game will find relatives asking for a share, whereas ex-coms have fewer relatives around and face fewer such claims. Then you’d see ex-coms appearing to be more risk-loving, when in fact the result had nothing to do with risk. Similarly, poor people in Liberia are liable to be faced with all kinds of short-term circumstances that might affect their behavior in a game where cash was the reward but do not really reflect their underlying risk preferences (i.e. what if I’m risk averse but I need a lot of money right now for some reason?).

    With underpants, though, you don’t really have those problems. As long as you control for wealth/income, it seems like you are probably getting a pretty good idea of underlying risk preferences, and probably a better one that you would if used cash.

  4. “transactional sex” is this simply a euphemism for prostitution or is there a substantive difference? The only paper I found on it defined it “as the exchange of gifts or money for sex.” which sounds pretty darn near identical to prostitution.
    Also one hates to be nitpicky (and as one who has done research in Africa I know that conditions are always far from ideal) but if the NGO’s motivation is to control the spending habits of those in the school (which is as you say paternalistic, and I add just plain wrong) then do they really think that your handing out of underwear instead of cash is going to be effective in this regard? Especially if these people have drug/sex addictions I would bet you that they will find very creative ways of turning those tighty-whiteys into drugs and “gifts” for sex.
    wrong + ineffective = very bad idea