Thoughts on Freakonomics

A number of people have asked me what I think of the bestselling book "Freakonomics" written by Steven Levitt and Stephen Dubner. On the front of the book, there is a glowing blurb by me which would suggest that I love it. On the other hand, chapter four of Freakonomics is devoted to the question of why crime dropped so dramatically in America—and particularly New York—in the 1990’s, and in that chapter Dubner and Levitt reach a very different conclusion than I do in "The Tipping Point." In fact, "Freakonomics" specially singles out for ridicule the theory of broken windows, which I suggest in the Tipping Point played a big role in New York City’s recovery. So what gives? Why do I love a book so much, if it contradicts my own book? Have I renounced the theories I put forward in the Tipping Point?

I have two answers. The first—obvious—point is that it is not necessary to agree with everything you read in a book to like that book. I have a number of problems with several chapters in Freakonomics, because I find the way in which economists approach problems occasionally frustrating. That being said, it’s very difficult to read Freakonomics and not find yourself saying "wow" every five minutes. I loved it.

Now for the long answer: what do I think of the substance of their crime argument? Is the Broken Windows theory central to the question of whether crime dropped, or isn’t it?

The Freakonomics argument starts off very much like the argument I make in The Tipping Point. The startling decline in crime in major American cities in the mid-1990’s is a mystery. No one predicted it. Everyone thought that high crime rates were a permanent feature of urban life. And the standard arguments to explain why crime falls don’t seem to work in this case. Levitt and Dubner go through all the usual explanations for crime decreases—a booming economy, decline in the crack trade, innovative policing strategies, tougher gun laws, aging of the population—and find only two that they think really matter. Putting more police on the street, they say, which happened in major cities all over the country in the early 1990’s, was a major factor. So were the soaring numbers of young men put away in prison in that same period. But neither of those two factors, they argue, are sufficient to explain the full magnitude of the crime drop. There has to be something else—and their candidate for the missing explanation is the legalization of abortion.

Levitt’s argument (and for simplicity’s sake, I’ll refer to the argument from now on as Levitt’s) goes something like this (and keep in mind that I’m grossly simplifying it here). The huge declines in urban violent crime rates happen, more or less, eighteen years after the passage of Roe v. Wade. States that legalized abortion earlier than the Supreme Court ruling saw their violent crime rates fall earlier. When you look at falling crime rates, the reductions in violent behavior are almost all concentrated in the generation born after the legalization of abortion, not before. People undergo abortions, in other words, for a reason: because they are poor, or don’t want a child, or live in an environment where it is hard to raise children. An unwanted child has a higher chance, when he or she grows up, of becoming a criminal. By removing a large number of unwanted children, legalized abortion ended up lowering the crime rate. Levitt makes it clear that he’s not passing judgment on this. He’s not pro-abortion, as a result of this observation. He’s just explaining the way he thinks the world works. He also stresses—and this is because even more important—that he doesn’t think that crime fell in major American cities solely because of abortion. He thinks abortion is simply one of several factors—albeit a significant one—in the crime drop.

Is Levitt right that legalizing abortion has played a role in lowering violent crime rates? Levitt has a few critics, and he’s dealt with them pretty effectively, I think. (Check out Freakonomics.com) There are some other technical critiques of his work from fellow economists, that, I have to confess, I can’t follow. My own response is chiefly that I find the argument incomplete.. For instance, the biggest drop in fertility in the U.S. came with the advent of the Pill in the mid-1960’s. The Pill allowed lots of women who would otherwise have become pregnant not to become pregnant because they were poor, or didn’t want a child, or lived in an environment where it was hard to raise children. But the fertility drop caused by the Pill didn’t lead to a decrease in crime eighteen years later. In fact, that generation saw a massive increase in crime. The advent of abortion in the early 1970’s, meanwhile, caused a far, far smaller drop in U.S. fertility but—Levitt argues—that drop is consistent with a fall in crime. In other words, the unwanted children whose births were prevented by the Pill would not have gone to become criminals. But unwanted children whose births were prevented by abortion would have gone on become criminals. Why is this? I can think of some hypotheses. But they are just that: hypotheses. I would have been a lot happier with Freakonomics if the crime chapter had been twice as long—and spent more time explaining just what is so peculiar, in terms of crime rates, about births prevented by abortion.

But that’s a quibble. In the course of making his argument for the importance of abortion Levitt is also pretty dismissive of other, alternate, theories—especially the theory that I spend a lot of time on the Tipping Point, namely the broken windows idea.

It’s here, though, where I think Levitt’s argument is a bit unfair. Levitt concludes that there are three factors that matter the most in the crime drop—abortion, high rates of imprisonment of young men, and increased number of police officers. The last of these three factors he glosses over pretty quickly. But I think that’s a mistake, because what is increased police presence? Well, having more police on the streets than before means that law enforcement can be more aggressive and pro-active. It means officers can do a lot better job getting guns off the streets. It means that they can be much more vigilant than before. It means that they have the time and resources to start cracking down on the kinds of seemingly minor "lifestyle" crimes than might have gone ignored before. The kinds of things that I argue were so important in responding a civil environment in New York State—the crackdowns on graffiti and public urination and panhandling and turnstile jumping in the subway system—are all the kinds of things that police departments can do when they have more officers on the streets. In Freakonomics, Levitt pretends he has refuted the Broken Windows explanation. He hasn’t at all. In fact, to the extent that he concedes the huge role played by the expansion of police departments in the 1990’s, he tacitly supports the Broken Windows theory.

So why is he so anxious to discredit Broken Windows? One—understandable—explanation is that he makes his own argument more compelling by dismissing all other arguments. (I know all about this tactic. I do it all the time). But a deeper explanation, I think, has to do with the difference between the perspective of economics and the perspective of psychology. Levitt is very interested in the root causes of behavior, in the kinds of incentives and circumstances that fundamentally shape the way human beings act. That’s the kind of thing that economists—particularly behavioral economists—think a lot about. And rightly so: who we are and how we behave is a product of forces and influences rooted in the histories and traditions and laws of the societies in which we belong.

But there’s a second dimension to crime, and that is the immediate contextual influences on human behavior. If you talk to a police officer (or a psychologist) they’ll tell you what a "typical" murder looks like. It’s two men, drunk at a bar. They get in a fight. They step outside. One pulls a gun in anger and kills the other. You can prevent that homicide by creating a population of people who are less likely to get drunk and angry in bars. You can also prevent that homicide by decreasing the likelihood of either of those drunken men having a gun. Police-work is concerned, necessarily, with this kind of immediate influence on behavior, and one of the things that having lots more police did was to make it possible to reduce the number of guns on the street that could end as a cause for a homicide. Drunken young men still fight in bars in New York City. But now they fight with fists—which are a lot safer.

Freakonomics is a book about deeply rooted influences on behavior, because it’s a book written by an economist. The Tipping Point is a book, by contrast, about the kinds of things that law enforcement types—and psychologists—worry about, because it was written by someone who is obsessed with psychology. I prefer to think of Freakonomics not as contradicting my argument in Tipping Point, but as completing it.

One final point (just to complicate things even further). Since Tipping Point has come out, there have been a number of economists who have looked specifically at broken windows—and tried to test the theory directly. Some have found support for it. Others—particularly Bernard Harcourt at the University of Chicago—find it wanting. If you crave a rigorous critique of broken windows, read Harcourt. He’s every bit as smart as Levitt.

Comments

[Ian D-B]

I can tell this blog isn't for those in a rush.

[Devraj]

Freco is a really good spicy book. It does give some insights but is largly aimed at the mass market and the marketing people have succesfully positioned it. The serious reader will and always find flaws in such kind of writeing. Nevertheless this doesnt matter or what is inside the book, but whether its a market leader or not.
Without a good study of economic principles reading such a book can creat a mind block in weak minded readers. These weak readers will analyze things in a wrong way and come to wrong conclusions. I thing this book should be sold in Pharmacys and with a prescription from an economics professor.

[Robby]

Personally I never understood the big deal with Freakonomics. Levitt is a well respected economist but the books was neither the first, nor the best book of economic analysis for the Everyman. Steven Landsburg's "The Armchair Economist" was much more interesting. David Friedman is another author who has some entertaining economic analysis. As a previous poster mentioned, Freakonomics, while decent, was really the beneficiary of some tremendous marketing.

[Ed]

Whose blog is this, anyway?

[Jeremy]

I'm with Ed.

Steve Sailer and a few others could spare all of us by limiting their l-e-n-g-t-h-y diatribes to their own blogs.

We'll read you if you've something interesting to say - don't piggyback off someone else's forum - it's cheap, annoying, and far less likely to get you read.

[Dan Zarrella]

the only wow moments I had with freakonomics were when I found myself saying "wow, these guys spent a lot of time dismissing truthiness, over simplification and logical fallacies only to fill a book with them"
loved your stuff tho, malcolm.

[zhoule]

Would I be remiss to ask that instead of Gladwell wondering about Freakeco, what does he think about Think, a rebuttal to Blink, and Blank, a parody?

[KGam]

So if Dr. Levitt is correct about the legalization of abortion having a direct effect on the crime rate, should we expect to see an increase in crime in South Dakota? I guess we should check back in 18 years....

[GP]

I've certainly been entertained and enlightened by books and papers written by both Gladwell and Levitt. When I first read The Tipping Point, the Broken Windows theory appealed quite strongly to me. I'd been studying in Singapore for a few years, and it really started to make sense to me that the Singapore government's freakishly severe fines for seemingly harmless acts such as spitting or walking on public park grass might have accounted for the low crime rate in singapore.

Then came Freakonomics, which I daresay is one of the best books on the applications of economics to everyday life's problems. I found Freakonomics alarmingly similar to The Tipping Point and Blink in their radical views of the way the world really works. The argument that legalized abortion is a significant reason of the sharp crime drop in the early 90s certainly is a valid one. After reading the book, I found myself so interested in the issue that I went on to read the paper Levitt wrote with Donohue and in all fairness that was a great paper (even though some holes have been pointed out by other economists). But I do agree with Gladwell that it's a little bit strange that Levitt only touched very briefly on what evidence he had to discount the Broken Windows theory.

Anyway I think, as an explanation to the dramatic crime drop in the 90s, both the Broken Windows and Legalized Abortion arguments have certain merits. The difference is that the Broken Windows theory might be a little bit easier to swallow. After all, no one will want to admit that abortion, to a certain extent, serves to lower crime rate.

[Jos Bleau]

The poor and minority women who abort are the ones who are striving to enter the middle class, or at least better their condition. Those who have no such dreams seem to have lots of babies. It's their kids that cause the most crime problems.

What if it wasn't the criminals that were aborted but the bottom part of the working class?

So the 'Roe effect' wouldn't be a lowering of crime but an increase in Hispanic immigration - to do the minimum wage jobs that have now no takers.

[Theophilus Punk]

Haven't read Freako. The argument reminds me of the 10-minute tempest that was stirred up around William Bennet (sp?) a few months ago.

Does Levitt account for which socio-economic, ethnic, etc., groups have sought more abortions or less since Roe v. Wade?

[jzb]

Let's suppose that the FREAKONOMICS argument is for the most part correct. Wouldn't this in some way contradict (or rather, complete) THE NURTURE ASSUMPTION argument that Malcolm is so fond of: that peers are more influential in a child's development than parenting? Can't we say that while there is little effect of either "good" or "bad" parenting, horrible parenting (i.e. having an unwanted child) can have a great effect on a child's development?

[zp]

I haven't read the book, but I've heard and read a number of summaries of the argument, particularly the abortion-crime argument. I've been a little puzzled about the authors would choose abortion statistics (who, when), which are difficult to come by and unreliable, on which to pin an argument. How can we argue if no one really knows who gets abortions and when since its all so stigmatized . . . for both doctors and patients. I think folks in health professions would argue that in many cases, both abortions and the pill (and other reliable forms of birth control) are most accessible to women who have good, consistent access to medical care.

[zp]

whoops. second sentence should read

I've been a little puzzled about WHY the authors

[Ernunnos]

Actually, that is not the typical anatomy of a murder. (See the Bureau of Justice homicide statistics.) Most murders are committed in disputes over drugs or drug territory, between young soldiers in organized crime. It's yet another power-law situation. A very few violent individuals are responsible for most of the murders. Incarcerating members of this population makes a disproportionate difference in murder rates.

[ars]

Malcolm, this post is an interesting comparison point between yours and Levitt's books, but I find myself slightly incredulous that you actually "loved" Freakonomics or found yourself saying "wow" every five minutes. Perhaps this is truly how you read, which, I suppose, is why you get quoted on the dust jackets of bestsellers more often than I do.

As you may already have surmised, I didn't love Freakonomics or find myself saying "wow" every five minutes, except maybe in the context of "wow, am I underwhelmed." And given your considerable erudition and the quality of your writing, I have to suspect you had some of the same problems with the book that I did.

Not that I expect you to trash a fellow traveler in the pop econ canon (although it sure would be fun if you did), but please don't pull punches. Generally speaking, I think this blog is great.

[brad]

Ethologists (scientists who study animal behavior from an evolutionary perspective) have a useful construct for distinguishing immediate causes of behavior from the underlying ones: the distinction of "proximate" versus "ultimate." When you ask "why does a certain behavior happen?" you can answer that in two ways: by looking for ultimate causation, such as why natural and social selection perpetuate a tendency for this behavior; or by looking for proximate causation, which is the equivalent of the two men fighting in a bar and one pulling a pistol. Lots of disagreements about causation boil down to people confusing these two concepts or seeing them as mutually exclusive when in fact both explanations can be correct. I've heard people arguing about whether puppies really "play" or whether they are in fact training for adult activities such as hunting. In fact both are correct. There's no reason to believe that puppies don't experience a similar kind of enjoyment that we get from play, just as there's a very strong argument that play evolved in the first place because young animals that played ended up being more successful hunters as adults.

Anyway, this separation of ultimate, or underlying, causes from proximate, or immediate causes is a useful tool for investigating behavior and it seems that some of the disagreement you cite might be related to the different perspectives of economists versus psychologists.

[Damien]

I've read both the TP and Freako, and both books definitely resonate strongly with me. On reading Malcolm's piece here, and the response form the Freako blog, I see one question which could perhaps help move the discussion forward.

What was the SocioEconomic background of those who used the Pill from the early sixties? Freako puts forward the idea that it was "potential criminals" (without passing judgement) who were being aborted. Surely the need for such high rates of abortion implies that they mothers in these backgrounds were not using the Pill, or that the people who were taking the Pill were less likely to have children who would be considered disadvantaged or prone to crime?

I'm in Ireland, so I don't know, but given the discussion of the US healthcare system, is it safe to assume that mothers on the Pill would have had to pay for it themselves without state support?

I ask these questions without trying to cast any aspersions on the Pill or on abortion, or those people who make use of either.

[Bill]

Damien raises an interesting point. Can we really equate The Pill with abortion? The Pill requires a certain amount of foresight, whereas for most people, abortion is seen as a last resort. I'm guessing that babies unborn because of the Pill are less likely to become criminals than babies unborn due to abortion.

Because of its up-front cost, there might be a demographic inconsistency between those that use the Pill and those that don't.

Just a thought…

[Abobtrader]

I agree with Robby that Landsberg's Armchair Economist is a better book. I greatly enjoyed Freakonomics but I think the marketers have over-hyped it. Either way, all books that encourage open-minded, out-of-the-box thinking are to be welcomed.

I have only just finished reading Freakonomics and was going to say that a good point that may refute the impact Broken Windows theory is that the crime rate fell simulatenously across the US...but I have just been over to the Freako blog (as directed by Damien above) and see they make the same point. Of course, this still leaves the Pill question open.

With respect to the mathematics behind the Freakonomics I am more pessimistic on the ability of econometrics to produce such clean results, and I thought Freakonomics didn't provide sufficient critique of the methods.

[chris]

Bill's comment that "babies unborn because of the Pill are less likely to become criminals than babies unborn due to abortion" lacks simple logic.

Unborn [i.e., non-existent] babies are not more or less likely to become anything.

[Mike D.]

Abortion vs. The Pill

A fan of both Levitt and Gladwell, I think both the abortion theory and Broken Windows theory have some merit and are not mutually exclusive. MG uses the availability of female contraception, i.e. the Pill, to suggest that if decreased unwantedness leads to lower crime, in should have happened years sooner.

The problem with that logic is universal access. Even with the existence of the Pill, three things need to happen for effectiveness:

1) A doctor's visit
2) Money to pay for the monthly prescription
3) No side effects & perfect usage

Lower-income families, from whom most future criminals come, do not at the outset or now have equal access to the Pill - in fact, access is dramatically different.

Roe v. Wade guaranteed universal access to post-conception birth control, which is a one-time expense not requiring vigilance on the part of the mother. This and not the Pill would have had a much greater effect on unwantedness on the criminal-producing cohort.

[Jason Hall]

I'm not sure that Levitt is tacitly endorsing the broken windows theory. Equating larger numbers of police on the street as being equivalent to the broken windows theory confuses cause with effect.

Maybe I am viewing this incorrectly, and it has been several years since I read 'The Tipping Point' - but I understood 'broken windows' to be an environmental approach to reducing crime, i.e. make the environment nicer, and people will be less prone to doing bad things. Is a larger police force a prerequisite to applying a 'broken windows' approach? I think there are alternatives. You could hire more street cleaners and maintenance staff to repair things more quickly. You could also install better public facilities to discourage petty crimes like turnstile jumping. Is there any evidence that doing these things will reduce the crime rate? I don't really know the answer - but, I still have a problem making the conceptual jump to 'more police' == 'broken windows'.

[brent toellner]

I've always found the "windows" theory to be interesting -- and would love to know more and have never taken the time. But coincidentally, I was thinking about this subject last night. I live in a "redeveloping" neighborhood -- one that was once beautiful, then became blighted, and is now making a comeback. So on my daily dog walk, I pass houses that are maticulously maintained, and ones that are not kept up at all.

What is interesting is that I there is a house on my block that always has tons of trash in the yard that is thrown there by passers-by. It's only the one house. Meanwhile, most of the other yards contain no trash in them.

While it is true that the people who live in these homes pick up the trash out of their yards better, I can honestly say that it will take me 5 years of picking up litter in my yard to get to the levels that currently exist in this one lot. This led me to thinking about this subject because I believe more people choose that place to throw out their trash because there is already trash there, vs places that are clean.

Interesting, because I think the correlation is there, and this just happened to come up on my daily walk.

[stay-at-home dad]

So, if a right leaning Supreme Court should overturn Roe, we would see an increase in crime 18 years afterwards. Interesting.