« Bargaining theory | Main | Why is Tide so popular? »
The Law of Unintended Consequences
Dubner and Levitt have an article in the NYTimes with three examples of the law of unintended consequences, the Americans with Disabilities Act made it more costly to hire people with disabilities and reduced their employment, ancient Jewish sabbatical law intended to help the poor has made them worse off, and the endangered species act has resulted in habitat destruction.
In light of this Andrew Gelman asks a deep question, What kind of law is the "law of unintended consequences?"
The law of unintended consequences is what happens when a simple system tries to regulate a complex system. The political system is simple, it operates with limited information (rational ignorance), short time horizons, low feedback, and poor and misaligned incentives. Society in contrast is a complex, evolving, high-feedback, incentive-driven system. When a simple system tries to regulate a complex system you often get unintended consequences.
Unintended consequences are not restricted to government regulation of society but can also happen when government tries to regulate other complex systems such as the ecosystem (e.g. fire prevention policy that reduces forest diversity and increases mass fires, dam building that destroys wet lands and makes floods more likely etc.) Unintended consequences can even happen in the attempted regulation of complex physical systems (here is a classic example involving turbulence).
The fact that unintended consequences of government regulation are usually (but not always or necessarily) negative is not an accident. A regulation requiring apartments to have air-conditioning, for example, pushes the rental contract against the landlord and in favor of the tenant but the landlord can easily push back by raising the rent and in so doing will create a situation where both the landlord and tenant are worse off.
More generally, when regulation pushes against incentives, incentives tend to push back creating unintended consequences. Not all regulation pushes against incentives, some regulations try to change incentives but incentives are complex and constraints change so even incentive-driven regulations can have unintended consequences.
Does the law of unintended consequences mean that the government should never try to regulate complex systems? No, of course not, but it does mean that regulators should be humble (no trying to remake man and society) and the hurdle for regulation should be high.
Posted by Alex Tabarrok on January 24, 2008 at 07:47 AM in Economics | Permalink
Comments
I don't want to hijack this thread, but this is as good a place to ask as any: has anyone written a book/article about what the consequences, intended or unintended, would be of returning to a world of 5% tax rates and far fewer agencies? For instance, what would happen if the labor department, ag, social security, ADA, etc. were dismantled. That seems like a crucial thought experiment, but I've never seen anybody consider it.
Posted by: bjk at Jan 24, 2008 8:35:57 AM
The Tacoma Narrows failure was due to resonance, not turbulence.
Posted by: Dave at Jan 24, 2008 8:39:58 AM
The simple fact is that almost all of the time markets make the right decisions and consequently government interference with markets can't avoid making things worse. Put simply, markets tell the truth and governments lie. Hence the ironical effects of government regulation.
Posted by: critic at Jan 24, 2008 8:49:05 AM
Critic, I might rephrase that with something from Marginal Revolution and one of its friends earlier this month: markets make unfortunate decisions all the time, but governments have even worse records. Markets usually do well and rarely do worse than the leading competitor, hence the last paragraph of the post.
Posted by: Zubon at Jan 24, 2008 9:03:07 AM
Yes, the percentage of disabled people holding jobs has fallen since enactment of the ADA, but that's not necessarily because lawsuit-wary employers are reluctant to hire the disabled. An alternative explanation is that more and more healthy but lazy people are falsely claiming to be disabled in order to get Social Security disability and avoid having to work. I don't know if there's any way to prove this. One possibility is to see if easy-to-fake, hard-to-disprove disabilities (psychiatric, back pain, maybe others) have increased as a proportion of all disabilities, though even a significant increase wouldn't necessarily be conclusive.
Posted by: Peter at Jan 24, 2008 9:22:42 AM
The general point about unintended consequences is of course important, but there's some good reason to think that Levitt and Dubner might not really know what they are talking about in this case. See, for example, this:
http://michaeldorf.org/2008/01/unintended-consequences.html
This in turn ought to make us go very slowly in accepting that unintended consequences mean we should not try to make some change. Such things are hard to get right, and if you mostly just ignore a lot of factors and evidence (like Levitt and Dubner seem to have) you'll also get some pretty bad results.
Posted by: Matt at Jan 24, 2008 9:25:20 AM
Isn't this effect largely the result of looking at 'marginal' regulations, regulations whose use is most disputed? there are enormous amounts of regulation that pretty mcuh achieve there goal, but these are rarely discussed.
Take aircraft safety regulations and the FAA. Goal: keep aircraft safe and the public trust in aircraft high. Works like a charm. Industry likes it because it gives them clear safety targets, airlines like trust the public has, passengers of course end up paying for the safety measures, but without regulations they probably wouldn't dare to fly.
Is it perfect? No. There are constantly disputes, these get solved most the times but sometimes they linger on. Does it achieve the social optimum between cost and safety? Probably not, but few people are complaining. Could it be done without the government? Perhaps yes, an S&P type organization giving out AAA ratings to airliners might be possible. But that organization would just produce its own regulation, most likely very similar to the present one.
Of course, FAA regulation is hardly 'simple', it has lots of experts backing it up and good feedback channels. That's the whole point. Good regulation can be useful, even if isn't perfect. There's loads of regulation around us. Some of it made by the government, some of it by private organisations. Ever heard of ISO norms?
Posted by: GreatZamfir at Jan 24, 2008 9:32:05 AM
I disagree that the political system is simple. It is also a complex (adaptive) system with its own dynamics, heterogeneous agents, non-linear interactions, adaptation, positive and negative feedback etc... However, this only serves to increase not decrease the chance of unintended consequences. The resulting regulation or policy coming out of this complex political system is usually the simplistic thing that usually leads to the unintended consequences that make economist knowingly smile.
Posted by: GoodneesOfFit at Jan 24, 2008 9:33:02 AM
Isn't this effect largely the result of looking at 'marginal' regulations, regulations whose use is most disputed? there are enormous amounts of regulation that pretty mcuh achieve there goal, but these are rarely discussed.
Take aircraft safety regulations and the FAA. Goal: keep aircraft safe and the public trust in aircraft high. Works like a charm. Industry likes it because it gives them clear safety targets, airlines like trust the public has, passengers of course end up paying for the safety measures, but without regulations they probably wouldn't dare to fly.
Is it perfect? No. There are constantly disputes, these get solved most the times but sometimes they linger on. Does it achieve the social optimum between cost and safety? Probably not, but few people are complaining. Could it be done without the government? Perhaps yes, an S&P type organization giving out AAA ratings to airliners might be possible. But that organization would just produce its own regulation, most likely very similar to the present one.
Of course, FAA regulation is hardly 'simple', it has lots of experts backing it up and good feedback channels. That's the whole point. Good regulation can be useful, even if isn't perfect. There's loads of regulation around us. Some of it made by the government, some of it by private organisations. Ever heard of ISO norms?
Posted by: GreatZamfir at Jan 24, 2008 9:33:11 AM
I didn't mean to double post,must be some bug in the system?
Posted by: GreatZamfir at Jan 24, 2008 9:45:59 AM
When a simple system tries to regulate a complex system you often get unintended consequences.
Thank you, Alex, this is an excellent definition and I agree with everything you've written above. I wish more libertarians would realize that it applies to deregulation as well as new regulation. The law of unintended consequences is one reason I distrust the Fair Tax and abolishing the FDA (sorry, Alex!). I think the law requires that libertarians and de-regulators be more humble, too!
Posted by: MostlyAPragmatist at Jan 24, 2008 9:55:14 AM
Markets rule! Let's bring back slavery!
Posted by: ES32 at Jan 24, 2008 10:02:54 AM
this sounds similar to a black swan which are the highly improbable expected these consequences are kinda expected.
Lively Money
Posted by: Lively Money at Jan 24, 2008 10:07:07 AM
Peter:
"An alternative explanation is that more and more healthy but lazy people are falsely claiming to be disabled in order to get Social Security disability and avoid having to work."
A theory is based on evidence; an accepted theory is the best explanation of how things work based on evidence.
Where is the evidence to support your claim?
Posted by: George at Jan 24, 2008 10:07:36 AM
"I wish more libertarians would realize that it applies to deregulation as well as new regulation."
I don't follow this. Removing the regulation will restore the original complex system. Maybe there will be a period of adjustment as people get used to the return to the old regime... is that what you're saying?
Posted by: Cliff at Jan 24, 2008 10:09:07 AM
Critic: "The simple fact is that almost all of the time markets make the right decisions..."
Fun statement. Any proof? Like decisions made by cabals of the rich somehow defy the law of unintended consequences. Right.
Posted by: converse at Jan 24, 2008 10:20:22 AM
When a simple system tries to regulate a complex system you often get unintended consequences.
I don't see any evidence for this beyond the anecdotal. Lets try not to rely so much on an inferential distance that you assume your readers to have, but that make statements like this cry for real evidence and not just a collection of stories.
Posted by: rtsevo at Jan 24, 2008 10:27:02 AM
And who passed this disastrous unintended consequences law? GEORGE W BUSH AND THE REPUBLICANS
Surely examples of unintended consequences are just as numerous in unregulated/market interactions. I holiday somewhere to enjoy the native charm, my tourism contributes to the destruction of that traditional charm. and so-on.
It's the old argument about how extensive externalities are...
I don't buy the idea that a complex system necessarily leads to better outcomes than a simple one, and I also would suggest that the way political regulation is designed is a vital factor.
Posted by: Tim at Jan 24, 2008 10:36:55 AM
Have unintended consequences been studied for congestion pricing? Does anyone know if there are any studies showing examples?
Posted by: bronxilla at Jan 24, 2008 10:39:03 AM
I think the name "Law of Unintended Consequences" is a bit misleading. It's a bit like having a "Law of Low-Probability Events" which says they occur -- of course they occur, with frequency matching their low probability, by definition!
"When a simple system tries to regulate a complex system you often get unintended consequences" sounds nice. But what would an example be of a complex system regulating a simple one? In other words, aren't our regulations always much simpler than the real world? The same is true in the physical realm, gardeners try to control an ecosystem using a shovel.
I agree that the lesson from all these examples is for regulators (and gardeners) to be humble. We don't understand most things, so imagining how things will change if we take a certain action is no substitute for actually trying it in the real world. (In fact this is really what the scientific method amounts to.)
Posted by: improbable at Jan 24, 2008 11:07:28 AM
Superb, Alex!
Posted by: Matthew at Jan 24, 2008 11:07:37 AM
Political action is not amenable to this kind of analysis. Regulations are often imposed in an unserious manner, as self-sufficient gestures of political theater, with vastly less enforcement than they would require to be effective. Every actor who is affected by any regulation has an emotional impulse to sabotage it, which they will indulge if they think they can get away with it. This factor overwhelmingly dominates any incentive-based analysis.
Regulations should only be imposed when there is an absolute determination to enforce them and a thorough understanding of all the costs of doing so.
Posted by: Frank Wilhoit at Jan 24, 2008 11:11:03 AM
Cliff: Have you lived under such a low-tax, low-regulations system? Even so, would you have a perfect memory of it? There are many reasons such systems were abandoned, and a lot of such reasons will lurk invisible until someone attempts to (re)create such a system.
Posted by: Leo Petr at Jan 24, 2008 11:21:14 AM
I'm with Andrew Gelman on this one. To put it more strongly than he does: The "law" of unintended consequences is not a scientific law at all. It is a very important social sciences concept the knowledge of which should be compulsory for policy makers. But I have yet to see a falsifiable hypothesis with respect to unintended consequences which has empirical support. (If I read Alex correctly, he seems to be suggesting the following: "The more complex a regulated system is relative to the regulating system, the more unintended and unwanted consequences relative to intended consequences will be observed." Sounds hard to test.)
To the posters above who praise market solutions: Ever heard of externalities?
Posted by: LemmusLemmus at Jan 24, 2008 11:24:12 AM
When a simple system tries to regulate a complex system you often get unintended consequences
While true, I think this is inherent to the overall complexity of the combined system and not a result of a simple system interacting with a complex system. Further, I would expect more unintended consequences as the regulatory structure becomes more complex.
Posted by: Jody at Jan 24, 2008 11:24:29 AM
