The Lively and Logical Logic of Life

Boredom drives a lot of academic research.  After you’ve studied a subject for decades, it isn’t much fun to keep repeating the standard lessons, so you mischievously start looking for counter-examples and loopholes.  Unfortunately, when the mischievous academic talks to a broader audience, he often leaves the impression that the standard lessons are a waste of time.  Frankly, I think that a lot of recent popular economics books fall into this trap.

Tim Harford’s The Logic of Life is a welcome antidote.  Harford argues that the standard economic assumption of human rationality usually works.  In fact, it works in a lot of cases where you might think it doesn’t.

The best example in chapter 1 is condom use by Mexican prostitutes.  It’s easy to say "A prostitute would have to be a brain donor not to use a condom every time."  But Tim demurs.  By bargaining about condom use, instead of using every time, prostitutes raise their income by about 25%.  Still not worth it?  Think again:

In fact, the prostitutes know that while the risks are real, they are modest.  Only one in eight hundred Mexicans carries HIV, and even among prostitutes it afflicts just one in three hundred.  Even if a prostitute is unlucky enough that one of her unprotected jobs is with a man who is HIV positive, the risk that she will catch it is less than 2 percent if one of them is carrying some other sexual infection, and less than 1 percent otherwise…

As far as we can tell, the typical Morelian prostitute is acting as though she valued one extra year of healthy life at between fifteen thousand and fifty thousand dollars or up to five years’ income.

Tim may sound like a typical insensitive economist in this quote, but he’s firmly in the Alan Blinder "hard heads, soft hearts" tradition:

[A] rational world is not necessarily a wonderful one… Rational individuals can make choices that are bad news for others; risky sex is just a particularly clear example.  And when rational individuals face a miserable set of choices, as do the Morelian prostitutes, they cannot do better than pick the best of a bad lot.  We will not solve social problems if we pretend that they are caused only – or even mostly – by the mad, the stupid, and the morally degenerate.

As an academic, I’m tempted to immediately highlight a counter-example.  Morelian prostitutes value a year of healthy life at up to five times their annual income.  But what about Levitt and Dubner’s drug dealers who risk their necks for minimum wage?  Aren’t they irrational?

But for now, I’m going to resist the temptation to dwell on counter-examples.  You’ve got to learn to walk before you can learn to run.  And you’ve got to understand rational explanations for human behavior before you can understand irrational explanations.  The Logic of Life may well be the best introduction to the rational choice approach on the market.  Even better, it’s well-written enough to inspire even jaded academics to get back to basics.   Bravo, Tim.

Comments

Agree with you 100% Bryan ... Harford has written a great book that demonstrates to the reader the power of the economic way of thinking.

I believe the odds of a prostitute getting AIDS are far greater due to adverse selection. HIV positive males may look for reckless prostitutes, since for them the cost of unprotected sex is much smaller for them than for other males (the cost being the condom-free sex premium plus the discounted value of treating AIDS and shorter life span).

Meanwhile, there's also a moral hazard problem. HIV positive prostitutes may charge a smaller premium for unprotected sex.

Nonetheless, the game is still rational.

Pete – Sorry that your brother was caught out. A couple of people have raised eyebrows about the research on teenage sexual behavior. Perhaps it’s all about expectations. This stuff has all been discussed everywhere from Oprah to Atlantic Monthly to the op-ed page of the New York Times, usually with much hand-wringing and very few facts. Nobody seems to find that inappropriate. When an economist adds to the discussion, suddenly it is shocking.
In fact, “The Logic of Life† devotes only four pages to the subject, brings facts to the table and presents a rather optimistic view: because sex is getting riskier, teenagers are being more careful. Sure, I could have opened the book by talking about something else, but I think it’s a (rare) responsible discussion of an important topic.

The drug dealers who risk their necks for a very low return may seem irrational to people sitting up there collecting a paycheck. But if you ask who else will hire a ghetto kid with no education who speaks a lower-class version of English and who may have a record, you have a clearer view of his choices.

Then add that he gets to carry a gun for his own protection, whereas the fast food worker is a sitting duck for whatever robbers are in the neighborhood.

On top of that, the drug dealer may be too young to be legally employable, or why runaways end up either stealing or in prostitution.

As you said, the best of a miserable set of choices.

"As an academic, I'm tempted to immediately highlight a counter-example. Morelian prostitutes value a year of healthy life at up to five times their annual income. But what about Levitt and Dubner's drug dealers who risk their necks for minimum wage? Aren't they irrational?"

Another possibility for this behavior is that serving as a foot soldier in a gang can be viewed as a rational gamble that they'll end up at the top of the heap someday themselves, in the same way that lots of would-be actors, directors, dancers and so forth accept staggeringly poor working conditions and pay for the chance to be one of the lucky few that make it.

Alternatively, being a gang member may provide non-money compensation in the form of protection...

In response to Pat Mathews's excellent comment. He/she wrote:

The drug dealers who risk their necks for a very low return may seem irrational to people sitting up there collecting a paycheck. But if you ask who else will hire a ghetto kid with no education who speaks a lower-class version of English and who may have a record, you have a clearer view of his choices.

Then add that he gets to carry a gun for his own protection, whereas the fast food worker is a sitting duck for whatever robbers are in the neighborhood.

On top of that, the drug dealer may be too young to be legally employable, or why runaways end up either stealing or in prostitution.

Add to that the fact that the minimum wage may price him out of jobs with a more-certain, safer future, and it may be even more rational than thought.

A black friend and I agreed, during the O.J. Simpson trial, that if we had a son, we would rather he be a drug dealer than be a leech like Kato Kaelin. At least he'd learn how to do math.

Best,

David

Sorry, I don't see why we should regard the prostitute's behavior as "rational" unless you beg the question by regarding all economic transactions as rational.

By what standard is the behavior "rational?" Would you still regard it as rational if the premium were 10%, or 5%, instead of 25%?

So far, I have found the book to be immensely enjoyable. I am curious about the rates of STI infection offered in the first chapter. You state, as evidence that life as a Mexican prostitute is less than pleasant, that one in six prostitutes has a sexually transmitted infection; this seems incredibly low to me. One in five Americans has herpes (although I believe this figure includes oral herpes which may or may not be sexually transmitted) and, depending on whose figures you believe, as Americans age their chances of contracting HPV skyrocket. This leaves out assorted other STIs as syphillis, chlymidia, etc. If prostitutes have STI rates lower than the average sexually active American, that is something interesting indeed.

I think there's also a moral hazard issue in the BJE. Yes the kids are practicing a safer sex relative to other sexual activities. But does this prospect of a "safer sex" lead more kids to begin sexual activity at a younger age? I won't pretend to know what the ideal age to begin sexual activity is, but to say that kids are becoming safer I think you have to show that those kids choosing oral sex are choosing it over regular sex and not over no sex.

Tyler writes:
Yet another approach is to use rationality, not as a description of the person's behavior, but rather as a description of how the economist should proceed, namely by classifying behavior into the dual categories of preferences and constraints. That tends to be my own approach and I think it is found in Tim's book as well.

In this case, there seems to be no such thing as "irrational" behavior. There is only behavior we don't understand.
That's well and good, but Bernard/Enrique's points about falsifiability comes into play.

I guess you would follow up by saying that "X is rational" is not a very content-ful statement, it's more like a reminder to look for motives.
So, you can't prove or disprove "X is rational." What you can prove or disprove (and this is where the science comes in) is whether the individual really acts so as to maximize Q given constraints XYZ.

So, the name of the game here is, "who can come up with the most accurate model of human behavior."

"Just so stories" are a trap here, as they are in evolutionary biology. We are positing the underlying mechanism behind what we observe, but to be good scientists we need to make sure that our guess about the underlying mechanism gives us new hypotheses, which future observations can confirm or refute.

"...teenage fellatrices..." Ha!

Tyler Cowen wrote:
One notion of rationality is that the person made the right decision and chose the right margin. We don't have much evidence for that in these cases.

I'm a computer scientist, not an economist, but it occurs to me that it wouldn't necessarily be required for the prostitutes to choose a margin. Having decided to offer such a deal to customers, they could in theory learn over time what choice of margin maximises their profits.

Having decided to offer such a deal to customers, they could in theory learn over time what choice of margin maximises their profits.

How could they learn that? They might learn how to maximize revenues, but I don't see how they can learn to maximize profits when the cost is reduced expected lifespan.

Yet another approach is to use rationality, not as a description of the person's behavior, but rather as a description of how the economist should proceed, namely by classifying behavior into the dual categories of preferences and constraints. That tends to be my own approach and I think it is found in Tim's book as well.

I don't really understand this. Are you suggesting that economists should simply study behavior as the choices that get you the greatest utility subject to given constraints? If so, then "rational" becomes a meaningless term, doesn't it? (How do we determine preferences?) I suppose you could argue that a given set of behaviors is rational if it does not imply inconsistent preferences, but then no single behavior is rational in itself. Anyway, I suspect if you dug deep enough you would find very few individuals who are rational in this sense.

Bernard Yomtov wrote:
They might learn how to maximize revenues, but I don't see how they can learn to maximize profits when the cost is reduced expected lifespan.

If we are assuming they know the average risk per customer, I would have thought the calculation was no harder. If they do not know the average risk per customer (or have some equivalent notion of risk) it's unclear to me how they could be acting rationally in any sense. But perhaps I'm oversimplifying this far too much?

"Another notion is that the person had *some reason* for doing what he or she did; that approach strikes me as too weak to be useful."

I think there are a couple of ways to think about irrationality. If a person really does things without a reason or purpose, that is irrational. Another notion of irrationality is that the person merely did something for reasons I disagree with; that approach strikes *me* as too weak to be useful. The debate is whether or not human beings are fundamentally rational or irrational, and I would say they are fundamentally rational. There are exceptions, but relying on the availability heuristic here or confirmation bias there is not evidence of irrationality. Most people, most of the time, still pass Bryan's "gun to head" test with flying colors; i.e. raise the costs enough, they change their behavior. It may seem like cheating to take individuals' risk preferences, etc. off the table, but personally I don't see how. Preferences are fundamentally subjective, and cannot be used to define rationality in the economic sense.

I think his definition of rationality would have been better served by simply claiming that we attempt to carry out some kind of calculation

Actually I think his main, key point is that rationality does not require conscious calculation. The word "attempt" suggests conscious effort, so it may not be the best word for him to use here. As for the rest, maybe he should revise his statement so that there is less of a suggestion about how, precisely, the brain manages to optimize, but I don't think it affects his key point about it not necessarily being conscious.

Constant said:
The assertion that behavior is rational may be falsifiable in the sense that some particular conceivable behavior may simply not be analyzable into preferences and constraints....

If we view economics as asserting that all behavior is rational, then economics is falsifiable. If we view economics as a calculus for deriving inferences about rational behavior without asserting that all behavior is rational, then economics is not falsifiable but the claim that it is applicable in a particular way to this or that behavior is still falsifiable....

These points make sense, although I would like to prod things in a more pragmatic direction.

What does it really mean to be falsifiable? What does it mean to say that I cannot analyze a person's behavior into preferences/constraints? I can come up with some example where a person behaves in a "contradictory" way. But how do I know there is no lurking variable?

Let's say I like cheese on Sundays but not on Tuesdays. How do I decide that I cannot explain this in terms of preferences? In the most trivial logical sense, why can't I just "have a preference for Sunday cheese?" Or to look at things in a practical way, how could I possibly rule out all possible preference explanations?

Can you really falsify "rationality's applicability to some situation" without breaking determinism? What does it really mean for an action to be "analyzable in terms of preferences?" What is a preference? If it was a brain hiccup that made me behave a certain way at a certain moment, was that a transient preference mediated by a brain hiccup? If brain hiccups are what cause irrationality, how do I tell the difference between a brain hiccup and "legitimate, ordinary preferences"?

Or, if I only like cheese when I am in a good mood, does that mean that I am irrational (my behavior is not analyzable in terms of preferences)? Or does it mean my preferences change with my mood?

I think the answer to all these questions is "whatever set of answers and definitions tells the simplest, most usable story is probably the one we want." We graduated from geocentricity/epicycles to heliocentric elliptical orbits not because epicycles were refuted by the data, but because of pragmatic, even aesthetic considerations, and matters of simplicity.

So I would argue that we don't have a good definition of "irrationality" because we haven't identified some clear set of patterns that it would be useful to call "irrational." And if we try to say something is "irrational" if it "cannot be analyzed into preferences", I would say we need a clear definition of "preferences". If preferences can be whatever I want them to be, even unconscious things that happen in the brain, then our definitions are sounding vacuous again.

Of course, it is ALWAYS possible to ask whether or not a person's behavior is consistent with a particular model of how they act. But when the person's actions do not match the model, it merely tells us that our model is not accurate. "Rationality" is a statement of the existence of some model, not a property of a model.

In other words, the right formulation is:
       IsRational(Behavior) iff Exists(Model) SuchThat ModelPredictsBehavior(Model, Behavior)

as opposed to the wrong formulation which many like to use:
       IsRational(Behavior) iff ModelPredictsBehavior(MyFavoriteModel, Behavior)

Aron,

I too have just begun reading Beinhocker's book: "The Origin of Wealth." It provides evidence that full blown general equilibrium models have little connection to reality. Although I've only read 127 page of Beinhocker's book, I am getting the impression that he is ready to "throw the baby out with the bath water". What I mean is that he seems too eager to believe that the concept of "equilibrium" is a completely mistaken metaphor in economic in general (no matter whether it is a partial equilibrium model, or a simplified general equilibrium model, or a full blown general equilibrium model). But, as I said, I'm only on page 128, so I'll have to hold off a bit until I reach page 454. (I might mention that "Sugarscape", for example, was far less impressive to me than it seems to have been to Beinhocker as evidence repudiating the usefulness of the concept of equilibrium in economics.)

But, in light of the discussion above, I look forward to having a look at Tim Harford's book in the near future to POSSIBLY (remember I'm less than a third into "Origin") regain my balance after finishing Beinhocker.

Applying the David Friedman quote to this chapter: a common way of acting rationally is to simply copy what other people are doing in similar situations. If prostitutes & customers have come to a sensible equilibrium pricing schedule (and there's sufficient transparency) then it's sufficient to price in line with the market to be rational, without doing any cost-benefit analysis at all. And that schedule might be reached in the first place simply by aggregating some very noisy estimates of the cost-benefit tradeoff.

Obviously this can be biased, and its response to new information (say a spike in the HIV rate of customers) may not be incorporated very fast. But in general, the forces that make markets (mostly) efficient also make relying on a market price a good heuristic for being close to rational.

As for the BJE, there aren't so much market forces, but there is to some degree a shared analysis of the problem and thus a common consensus (or perhaps just the average solution) rates to be reasonably optimal.

I would guess that most of the computation that goes into rational decisions is actually outsourced to others in this way.

Some great comments and questions, and I am sitting in JFK airport, heavily jetlagged. I feel I need to offer some (brief, partial) responses before the discussion moves on.
Re: rationality. I am unabashedly pragmatic here. The mathematical definition of rationality is framed in terms of consistent choices that allow a mapping into a utility function. This is a useful concept for economics texts, but not for the general reader. I prefer to say that the preferences have to be intelligible (we not believe that the typical prostitute would prefer to catch an STD, for instance) and look for evidence of tradeoffs - accepting money in exchange for a higher risk of infection. Also, I look for consistency: no endowment effects or hyperbolic discounting, to use the technical terms.
This is a fairly relaxed standard of rationality but still a strong statement when applied to eg teenage sexual behavior, dating and divorce, office politics and so on. It is also falsifiable, and (in chapter two) sometimes falsified.
With the prostitutes, we have an additional point of reference: we can compute their implicit valuation of their own health, relative to income (a year of healthy life is valued at 2-5 years income, if I recall correctly) and compare that valuation to those derived from other sources, such as surveys and the risk premium paid to workers in hazardous jobs. The numbers match up very well, which suggests some sort of consistency across people in very different circumstances, although not an absolute standard of rationality.

Now, a question. In the research I describe on Mexican prostitutes - which was led by the highly respected economist Paul Gertler - we see that sex workers from Morelia often use condoms and demand a decent premium for putting them to one side.
In the new Levitt/Venkatesh paper on prostitution in Chicago - which I have skimmed but do not have here with me - I recall that Chicago sex workers use condoms more rarely and demand very little premium for unsafe sex.
This, to me, seems to be a genuine puzzle. The Chicago sex market seems to be far more of a niche, wheras in Mexico most prostitutes are paid simply for plain vanilla sex. Does that suggest any resolution of the puzzle? I suspect, also, that Chicago prostitutes are in a much more desperate situation relative to the society around them. Prostitutes in Morelia earn a premium relative to other women. But this is wild speculation - does anyone have something more informed to offer?

One interesting thing about the definition of "rational" that Tim proposes is that we have no idea if it is conducive to happiness. Phenomena like hyperbolic discounting probably reflect important facts about the brain (e.g. "When I am thinking about Potentially-Imminent-Thing X, I get anxious for it and can't wait to have it"). Is it wise or unwise to contradict these impulses in the name of "rationality?" Do we suffer a hit to our endorphins each time we do contradict our desires in this way? Maybe after training, we could make ourselves avoid this endorphin effect.

Of course, we circumvent endorphin-rush desires all the time in the name of planning, but I wonder whether some degree of hyperbolic discounting is not intrinsic to successfully pursuing desire.

Of course, no one ever said happiness and rationality were the same thing. But consider a silly thought experiment. Think of your brain as part of the "external world": it is an object that you have some, but not complete, control over.
Now imagine that the "endorphin rush" is treated as endogenous to the choice problem. Human "souls" choose option A or B, where one of the factors under consideration is how each choice will affect the endorphins in their brain.
Suddenly, it looks quite rational for a human soul to apply hyperbolic discounting!

With the prostitutes, we have an additional point of reference: we can compute their implicit valuation of their own health, relative to income (a year of healthy life is valued at 2-5 years income, if I recall correctly) and compare that valuation to those derived from other sources, such as surveys and the risk premium paid to workers in hazardous jobs. The numbers match up very well, which suggests some sort of consistency across people in very different circumstances, although not an absolute standard of rationality.

Maybe, or maybe not. I don't doubt you can do a calculation. I wonder how much confidence one can have in the figures going into the calculation. Is the risk of infection 1% or 2%? Is the rate of HIV infection among unprotected customers 1 in 800 or 1 in 200? (Isn't an HIV positive customer much more likely to seek unprotected sex?) How many customers does a prostitute have in a year? How many unprotected? The choice of these values, which can hardly be determined with much precision, is going to have an enormous effect on the outcome. How does the model take account of the fact that some prostitutes are already HIV-positive?

And of course there is the question as to whether the pricing represents decisions by prostitutes or their pimps.

I was also curious about the study that showed people with a relative with AIDS changed their sexual behavior. That seemed a strange example to me. When risks are well-understood (certainly the case by '92), but behavior changes in response to a relative's sickness, that is somehow evidence of rationality. Hmm. If I were an economist, I'd study the role of emotional connection in choice. Maybe this is an area where the neuroeconomists can contribute something useful.

msi,

You should familiarize yourself with Kahneman and Tversky, who address these kinds of issues. The changed sexual behavior you discuss is a good example of the importance of the salience of information - its prominence. It is the difference between reading that Toyotas are very reliable, for instance, and having a good friend make a point of how little trouble he has had with his Toyota. The latter type of information tends to carry more weight in decison-making than it should, because it is more salient.

meter,

To be more explicit, you probably mean their "net benefit" not their "benefit".

I'm sorry, but I thought a downward sloping demand curve was a sufficient condition for rationality, and that seems to be what Harford is saying as well. If the cost of sex goes up, prostitutes, and teenagers, will supply a smaller quantity of sex. If the prostitutes(not the teenagers) are compensated well enough, the cost effectively decreases and they will supply more. Any eloquent language and extravagant excuses do not invalidate the law of demand. Just as planes and helicopters do not disprove gravity, they only do what they do in the context of gravity. As they fly, gravity still acts upon them. So it is with the law of demand.

Comments for this post are closed