August 26, 2004
REASON IS A MUG'S GAME:
IRRATIONAL MARKET BUBBLES: Are They "Post-Modern"? (Elliott Wave International's
Robert Folsom, Editor of Market Watch, August 25, 2004, Financial Sense Online)
Newsweek magazine recently ran an article with the headline "Mind Reading," and the subhead, "The new science of decision making. It's not as rational as you think."The piece rehashes a bit of year-old news about an experiment conducted by economists at Princeton University, known as the Ultimatum Game. The game has two players and goes like this:
"Subject A gets 10 dollar bills. He can choose to give any number of them to subject B, who can accept or reject the offer. If she accepts, they split the money as A proposed; if she rejects A's offer, both get nothing.... A makes the most money by offering one dollar to B, keeping nine for himself, and B should accept it, because one dollar is better than none."
The rational choice is to accept the dollar, yet the outcome was consistently irrational:
"People playing B who receive only one or two dollars overwhelmingly reject the offer. Economists have no better explanation than simple spite over feeling shortchanged. This becomes clear when people play the same game against a computer. They tend to accept whatever they're offered, because why feel insulted by a machine?"
As I said, this story was in the news more than a year ago. The only new twist in Newsweek had to do with a scanning gizmo that apparently spots which region of the brain produces the irrational choice. Yet what got me was Newsweek's claim that this research will "help understand some of the most vexing problems in postmodern society," such as "irrational market bubbles." Now, I'm a big fan of science 'n'all. I really am. But ... we're supposed to believe that market bubbles are "postmodern"? HEL-LO!! Calling all search engines! Tulip Mania? South Sea Bubble? The panics of 1837, 1857, 1873? And wasn't there a little episode in 1929?
Another question: Were market bubbles in the good old days something besides "irrational"?
The answer to the "irrational choice" paradox is so obvious that it would take a PHD in Economics to miss it. The game is not about choosing to receive a benefit, but about the valuation of relative self worth between two people. It is a social transaction, not a monetary transaction. Player A, in his choice of how much to offer player B, is making a statement about how he views his own social worth in comparison to B, with the expectation that B views the comparison similarly. Any offer to B that is less than $5 is a statement of social superiority, or dominance. An offer above $5 is a statement of social inferiority. Likewise, B's acceptance of an offer below $5 is an admission of inferiority, while a rejection of an offer of $5 is a statement of superiority. The "socially" rational offer by A is $5.
In a larger sense, market decisions are mainly a statement about beliefs which may or may not indicate statements of self worth, but which often are. Refusing to liquidate a losing position is often based on a belief that "I deserve to get my money back", not on the rational assessment of the likelihood of recovering it. Worldviews and expectations bias our ability to rationally calculate probabilities based on the available information, which often is contradictory. Bulls stay bullish in the face of bearish news, and Bears stay bearish in the face of bullish news. The human psyche requires a narrative, but market randomness offers none.
Posted by Robert Duquette at August 26, 2004 3:15 PMThe purely rational thing to do is shoot A and take all $10. That we don't is a function of our religious faith.
Posted by: oj at August 26, 2004 3:18 PMIn these sorts of economic games, it seems that the word "paradox" has the meaning of "the results are not what I wanted them to be."
Here's a variation I'd like to see tested-- B has to secretly write down what is the minimum amount it will accept before hearing A's offer. If that value is equal to or less than A's offer, they both get their shares, or they get nothing. By market strategy, A should write $1, while the social strategy says $5. (And a socialist would write $9, because A should be thankful just to get anything.)
Posted by: Raoul Ortega at August 26, 2004 3:29 PMOJ:
Bingo. I coincidentally read about this experiment just a few days ago and thought to myself that conceiving of rationality as narrowly as this experiment does is pretty shortsighted. It is quite rational to wish to punish another person's arrogance and deter him from such behavior in the future, even if you forgo some dough in the process.
The article mentions that the experiment was conducted at Princeton. Do any BroJudd readers know if Daniel Kahneman was responsible for it? If he won one-half of the 2002 economics Nobel for this, then Vernon Smith should have gotten the entire damn thing.
Posted by: Matt Murphy at August 26, 2004 3:33 PMDo the experiment with $100 bills and the results will be entirely different. For the reasons Robert suggests, this experiment will not scale.
I do wonder, though, whether people think about buying and selling stock in the market as a transaction with other people or as transaction with the market itself.
Posted by: David Cohen at August 26, 2004 3:35 PMAll good points (as usual). I'll attempt one more that makes the attempted correlation to a free market flawed:
If I understand the setup correctly, B can't lose. And A is simply playing with the "market's money."
The risk/reward trade-offs aren't remotely corelative.
Posted by: John Resnick at August 26, 2004 3:49 PMThis is an irrational game set up to determine rational behavior. What's particularly irrational about it is that there is no work to be done in exchange for the money. There is no value at issue here.
The rational thing to do, contra OJ, would be for A and B to collude and split all this free money coming down on high from Princeton. That they can't is an irrational rule imposed by Princeton.
Posted by: at August 26, 2004 4:00 PManonymous:
Bunk. Why should B share? Rationally speaking...
Posted by: oj at August 26, 2004 4:05 PMOJ: Sorry. It's me, EO.
Surely you aren't arguing that there are no ethical or cooperative components to rationality...
But, as with so many other things, the ethical dimension need not be reached. There are many unspoken practical matters that keep B's behavior in check. There is C, for example, who would beat B severely about the head and shoulders if A were cheated or injured. There are D, E, F, and the other members of B's community who would think it was pretty dumb of B to do some rash thing over a few dollars.
As an aside, why would A offer anything under your view of rationality? Why wouldn't A just pocket the $10 and run back to the eating club?
Posted by: EO at August 26, 2004 4:32 PMEO:
There are no ethical components to rationality.
A would.
Posted by: oj at August 26, 2004 4:39 PMOJ, shooting isn't an option. It's a controlled experiment. Nice try.
David is onto the right track. With $1 bills, B can blow off any offer that offends her and lose very little. If it were $100, she would probably swallow her pride and take it. But pride is a funny thing, the amount of the difference between A's take and her may just spark her indignation.
A has more to lose if he doesn't get B to go along. If B doesn't go along, he loses $900. He would have more incentive to ensure that the offer is accepted. I still think that $500 would be the optimal play.
John, when rapid appreciation in stock positions give people a large windfall, such as in 1999, people can get the feeling that they are playing with free money. That might explain a lot of the reckless trading we saw at the bubble top.
Posted by: Robert Duquette at August 26, 2004 5:33 PMra·tion·al ( P ) Pronunciation Key (rsh-nl)
adj.
Having or exercising the ability to reason.
Of sound mind; sane.
Consistent with or based on reason; logical: rational behavior. See Synonyms at logical.
Mathematics. Capable of being expressed as a quotient of integers.
Your definition is too narrow, much like this experiment. You are thinking one decision at a time. Yes, it may be rational to steal or kill or deceive for $10 but, later, a whole host of bad things will happen. Even if nothing bad happens, you'll eventually come to realize that life is no fun when everyone you transact with is dead, hates you, etc.
The most remedial application of game theory, which comes closer to approximating real human behavior because it is more complex (and human behavior is complex, with or without God), would yield my predicted result: $5 to A and $5 to B.
Posted by: EO at August 26, 2004 5:37 PMRobert:
Of course it's controlled--no one could afford the insurance to test pure reason.
Posted by: oj at August 26, 2004 6:19 PMPoint taken. Here is a restatement:
"You'll eventually come to realize that what appeared reasonable in the immediate moment turned out to have unforeseen and highly undesirable consequences in the medium and longer term that could have been avoided by applying a different policy that still would have made you somewhat-to-mostly satisfied at the immediate moment and would have avoided these new problems."
That we don't shoot A is not a function of religion but a function of our appreciation of these long-term consequences. And that my beloved and only slightly peculiar blogging host and cyberfriend, OJ, is neoconservatism in a nutshell. God is not reached.
Posted by: EO at August 26, 2004 8:20 PMThat's sort of what I was going to say, EO.
Rationality gives uncertain results unless you can predict the future accurately.
Posted by: Harry Eagar at August 27, 2004 3:57 PM