Wednesday, January 26, 2005

Free Riders

Dr. Dan, a frequent reader of this blog, with whom the Hatcher just recently broke bread, did some research that was recently featured in an issue of the Economist. He and his co-author set up four person games, where players were given some amount of chips that they could eventually trade in for money. Each player has the option to throw her chips (anonymously) into a common pool or horde them; the chips thrown into the common pool are doubled, and distributed evenly among the four. If everyone throws into the common pool, each doubles his money; but there is an incentive not to throw into the pool, because if you are the only one who does, you end up with only half the chips you began with.

It is common in economics to treat the players as rational agents who seek to choose a best response to the anticipated behavior of others playing the game; in contrast, biologists who have used game theory to theorize about evolution, generally treat a given species as being equivalent to a pre-defined strategy that they do not deviate from, no matter what strategies are "adopted" by competing species. Which beckons the question - are humans best described as rational agents who adapt their strategies as they learn what other players in the game are doing, or are we just born as a strategy we stick to? Dr. Dan's results suggest we are born with a strategy, and we stick to it. We don't adapt to the behavior of others.

We are either free riders (20 percent in the study) who horde their own chips, co-operators (13 percent) who give up their chips to the common pool, or recipricators (63 percent) who mimic the behavior of others in the game. Of the participants in the study, Dr. Dan was able to unambiguously determine which of these three categories the subjects fell into with the exception of only a handful of people. He and his co-author, having learned the strategy type of each player, could then mix and match different players in new games, the results of which they could predict quite accurately.

The interesting result is that the average payoff to the different strategy types (presumably across many games matched in different combinations of the different types of players) was the same - i.e. no strategy type led to better results than any other. That would be consistent with the persistence of the three different types - if one strategy were dominant, people who are defined by it would be "preferred" by evolution, and the other types would eventually die out because they don't make enough money to convince anyone to mate with them. (Which means that somehow I managed to pick a strategy outside of these three for many many years, but finally righted my course). There is some evidence that cavemen actually played this game to divide up the cavewomen (there is even an episode of the Flintstones where Fred loses Wilma to Joe Rockhead in just such a game, but the Great Kazoo sets it right for Fred).

Here are a few questions for Dr. Dan, if he would be so kind to respond in the comments section:

1) Were there gender differences in the distribution of strategies? If so, who had more free riders? Remember that if you say women, you'll suffer the fate of Lawrence Summers, President of Harvard, who will go to his grave apologizing for spouting common sense. But if you say men, the same will happen to you - the key point is that men and women have to be the same, otherwise you will be a controversial right-wing nut like myself, your tenure will be revoked, and you'll be left only with Ideas Hatched as a venue to publish your work. I'll give you a cut of my voluntary subscriptions, but I think you can guess what that buys you.

2) Did the players that you couldn't characterize do better or worse than the those with a clear strategy? Did it appear that they were just random idiots, or did they look like they were trying to learn and adapt to what others in the game were doing?

3) I seem to recall evidence suggesting that economists and economics students were more likely to be free-riders, precisely because they had been taught the benefits that can accrue to such behavior. This would seem to suggest that even if we are born with some innate strategy, we can take a course from Professor Vic, for example, and switch to become free riders. Any thoughts?

1 Comments:

Anonymous Anonymous said...

Great post! At least until I found out that I had to do something. So being a good academic I followed usual protocol and tried to find someone else to do the work for me.... not really (or at least not successfully), but actually we do find that professors behave that way in experiments, on average. We've got a bunch of different lab experiments where we can measure people's trust and trustworthiness, and it turns out that university econ faculty are usually less trusting and less trustworthy than undergrads. It's not just a faculty effect, because econ grad students score about the same as the econ faculty. Not sure what this says about us...well, it's only averages...

You're right that there is no good way to answer questions about gender effects...in fact it doesn't even work to claim to be uninterested in the question. I know...I tried it during a seminar (different paper) at a universty not long ago. A person in the audience asked about gender effects, I said I wasn't interested, and he became very annoyed that I would not be interested in such an important topic and stormed out of the room... oh well... Anyway, different studies find different things about gender. One result out there is that women can be more cooperative with men than with other women, so it's not just a gender effect but probably something more complicated going on... subjects didn't know who they were matched with in our experiment, and we didn't see any obvious gender differenes.

I'd like to run an experiment on people who blog, they must be highly cooperative, contribute to the public good...

Thanks for the post! - Dan

8:02 PM  

Post a Comment

<< Home

Sign up for my Notify List and get email when I update!

email:
powered by
NotifyList.com