Teenagers Rational About Sex? (Author One-on-One: Tim Harford)
Dan, thanks for the kind comments about The Logic of Life. Anyone who read my Financial Times review of Predictably Irrational will know that the admiration is mutual.
Let me just make one small correction: I certainly don't say that high CEO pay is desirable; I think it grotesque. Half that chapter is devoted to explaining why--why can a system full of rational shareholders can deliver grotesque pay packets?
Now, let's discuss the central question: are human beings rational or irrational? Perhaps the best way to understand the different perspectives of the books is through an example, and since the sexual examples in both books have captured everybody's attention, I'll compare and contrast.
Dan wants to know whether our behavior changes when we're sexually aroused and more importantly, whether we are caught by surprise when it does. As always, his approach is to run an experiment, and he paints a picture that will forever be etched in my memory, of students masturbating over Saran-wrapped laptops, while answering multiple choice questions about whether they would use condoms, consider using date rape drugs, have sex with animals... (In case you're wondering, no, this kind of subject matter is not the only reason that Predictably Irrational has been a runaway success.)
What does he find? First, that his subjects give very different answers to these questions when in a calm frame of mind than they did when sexually excited. Second, that they did not predict the way their answers would change. Now, one could argue about this research. Maybe the Saran-wrapped laptop is not a good substitute for a real sexual encounter. Maybe the pornography did not change the subjects' preferences, but merely encouraged them to be more honest. But basically, I buy it. The research seems sound, and Dan's witty description of it is pretty unforgettable.
So, how is The Logic of Life different in the way it approaches this topic? I take a bird's eye view of the whole controversy over the so-called "teen oral sex epidemic"; I find out that, behind the hype, yes, teenagers are having more oral sex than they used to. But I also find that they are losing their virginities later and are more likely to use condoms. That isn't so much an oral sex epidemic as a safe sex epidemic. I argue, informally, that this would be a rational response of a typical teenager to increased education about pregnancy and particularly to the now near-universal knowledge of the threat of HIV/AIDS. (You can read a short article I wrote about all this in Slate. You can also see what Stephen Colbert made of it all here.)
Fine. But that's still a bit speculative. So then I look at some of the latest evidence from economics, which shows that state by state, when state legislatures introduce laws tightening up on the availability of abortion to teenagers, the rates of sexual infections in teenagers falls in those states, relative to that of adults. That shows that teens are switching to safer sex, or abstinence. This isn't some vague correlation (more conservative states = less risky sex) but a specific response from teenagers to laws that affected only teenagers.
"But that's rational!" spluttered one venerable journalist, when I told him about this. Well, yes--it seems so, doesn't it?
Okay--who's right? When you see both examples set against each other you realize that's not a smart question. Dan explains that our behavior changes when we're excited; I explain that teenagers have safer sex if a legal change increases the consequences of pregnancy. Obviously both those things could be true at the same time. I believe that both of them are true.
At the risk of making this seem like a big group hug in which everyone is right about everything, let me raise a point of disagreement. The Logic of Life leans on all kinds of evidence, everything from the kind of careful statistical work I describe above to the use of lab rats. Theory counts too, and so does the clever simulation work inspired by the likes of Thomas Schelling. Every type of evidence has its flaws. I think the weaknesses inherent in pure theory are obvious; so, too, are the weaknesses of the sort of data-detective work very common on modern economics. (I will let Dan say more about those weaknesses if he wants to.)
But the weaknesses of laboratory experiments are not always quite so evident, especially when they are described as compellingly as in Predictably Irrational. So, let me point them out. While laboratory experiments are great for creating controlled conditions, they also create artificial conditions. There are several examples of important clashes between what happens in the laboratory and what happens outside. We know, for example, that procurement managers systematically screw up when bidding in a laboratory auction, but they do much better job in the (apparently identical) real life auctions situations they face everyday.
The economist John List has tried to replicate some famous "predictably irrational" results from the laboratory; the results tend to evaporate in more realistic settings. In one example I describe in The Logic of Life, List shows that the "irrational" result (which is that people given an unexpected raise work much harder than they could get away with) only lasts for ninety minutes. A gratitude effect that lasts ninety minutes is not the basis on which to rewrite your company's human resources policy.
I'm not aware that any of Dan's experiments have been challenged in this way, and I was pleased to see that he often tried to carry out his work in realistic settings such as restaurants and bars. So this isn't an attack on his work--it's more of a question. Dan, how can we be confident that these experimental results hold up in real life? And what further work would you like to see, to make us more confident in them?
Your logical friend,
Tim
WhoDat on April 16, 2008 at 09:27 PM
Way to lay the homework on Dan!
It would be great if you took a single issue and discussed how you would approach answering the question...
Jim on April 17, 2008 at 09:03 AM
I would be interested to read how both authors interpret 'rationality' in economics. My own prejudice is that it is impossible to distinguish rationality from irrationality by observing people's behavior, yet that seems to be precisely what the authors are claiming. Would that be a fair characterization?
M. Simon on April 17, 2008 at 12:39 PM
I think one must differentiate between rule based behavior and rational behavior.
Figuring out when to sell stocks when you have no skin in the game may be more rational vs figuring out when to sell if you are invested.
The very word "invested" indicates a bias for holding on past when it is rational. So we know the rule. We know it is not rational.
Shannon Love on April 17, 2008 at 02:38 PM
We're often got in a trap between the need to simplify a problem enough that we model it mathematically and the need to include enough detail to make the model valid in the real world.
Unfortunately, we have a cultural perception that the more abstract an model the more true it is. We prefer geometry theorems based on abstract absolutely perfect straight lines to ones based on real world objects with messy edges. People who work in the real world soon realize that such abstractions are dangerous if misused. Mathematicians can assume the existence of ideal objects, engineers cant.
I think that most of the time when we say people are behaving irrationally we're actually saying we don't have enough information about the circumstance. Don't expect academics to adopt that viewpoint anytime soon.
Dan King on April 17, 2008 at 05:47 PM
Teenage prostitutes in Japan are acting rationally, or at least that is what is argued at http://anti-militant2.blogspot.com/#Hughes_2 . Another view about rational prostitution (though not really about teen agers, can be found at http://www.dankingbooks.com
DerekT on April 18, 2008 at 10:10 AM
Since Tim hasn't responded to the postings about rationality, could I quote from his book on his "simple" definition:
Rational people respond to incentives: when it becomes more costly to do something, they will tend to do it less (see his teenage sex comments above); when it becomes easier, cheaper, or more beneficial, they will tend to do more. In weighing up their choices they will bear in mind the overall constraints upon them: not just the costs and benefits of a specific choice, but their total budget. And they will also consider the future consequences of present choices.
I'm still not convinced by the idea that I'm rational in my choices (I believe in serendipity!). But I hope this debate will help convince me, one way or the other.
Colin Camerer on April 18, 2008 at 11:59 AM
Very useful exchange. A comment on the example from John List's field experiment on "gift exchange" (paying workers an unexpected bonus induces higher uncontracted effort). Note that in the first couple of hours of their six-hour field experiment there *is* positive reciprocity. That time span matches the typical length of the lab experiments. So rather than being evidence that lab results do not generalize to the field (as John interprets, and Tim suggests above) this is an excellent example of *good* generalization, if one compares the same time spans of the two experiments.
It is true that the reciprocity appears to wear off in the field experiment in the next four hours or so after the initial two hour period. The natural next question is whether a six hour lab experiment would show a similar decay to the field.
Another field experiment by
Bellemare and Shearer (http://ftp.iza.org/dp2696.pdf) shows substantial (cross-day) gift exchange effects in the field. Alexandre Mas at Berkeley has two wonderful empirical papers using a tire company strike and final-offer arbitration of police unions showing that there are substantial long-term effects that can be interpreted as reciprocity (the police clear fewer cases for a year after they lose in arbitration).
The reasonable response to all this is to ask which types of experimental results are likely to generalize and which are not. But note that the same question can be asked about generalizing from one empirical phenomenon to another. For example, studies of the macroeconomics of the Great Depression may not generalize to modern economic phenomena, or they might. Similarly, interesting field experiments on microcredit or corruption in Africa might generalize to American city politics, or they might not. The hope is that there are general principles which apply essentially everywhere (up to parametric variation such as local tastes) and that is what experimental economists like myself believe.
DerekT on April 19, 2008 at 10:04 AM
Re Colin (Camerer)'s comments, unfortunately the URL dp2696.pdf wasn't recognised by my server, so wasn't able to read (you make it sound of interest to me).
One question on your final para, which would apply to all economics ideas (whether lab-test-based or field), I suspect; and on which both Dan and Tim might like to comment, please. To what extent is economics theory as currently presented firmly rooted in a specific culture (Western, Oriental, or, as in your instance, African) - and thus, how transferable is it into other cultures?