« Place for comments on "Keynesian Parables of Thrift and Hoarding" | Main | External validity, fallacies of composition, and the Wawa multiplier »


Feed You can follow this conversation by subscribing to the comment feed for this post.

I'm torn. Tobacco's contribution to morbidity is direct and causal, while alcohol may be masking the mental health issues for which it is the self-medication of choice. Both substances are abused as "self medication," but while alcohol's "self-medicating" effects are overwhelmingly negative, tobacco's are overwhelmingly positive. Except for the slow death from horrible cancer.

Erik - I'm curious - what do you see as the overwhelmingly positive self-medicating effects of tobacco? Reduced anxiety/improved concentration (I've heard this anecdotally, but haven't found evidence). If this is really true, then perhaps e-cigarettes, which can deliver the nicotine without the tar, have a huge upside potential?

I wouldn't say that we know for sure that alcohol's self-medicating effects are overwhelmingly negative. At least, if you can point me to research, I'd be interested. Eric Crampton writes about the alleged "J curve" of ill-effects from alcohol - so abstainers (including probably a good number of ex- or probable alcoholics) have slightly worse than usual health outcomes, moderate drinkers have the best health incomes, and heavy drinkers have bad health outcomes. I self-mediate my shyness and social anxiety with alcohol. Is that so overwhelmingly negative?

No talk of cancer, it might upset certain people. We're going for sudden, massive coronary arrest here. Ideally after a day of canoeing, followed by a bottle of wine and steak cooked over the campfire.

Nice. I'd looked at comorbidity between alcohol and mental illness a bit more thoroughly here.

Co-morbidity is difficult to deal with. When somebody self-medicates for a pre-existing condition using substances that themselves can be harmful, it will be a mistake to attribute the total difference between that person and the average person to the substance use. The observed difference in outcomes between the user and the average non-user could be greater or smaller with substance use than it otherwise would have been without the substance use, but it requires thinking at margins and about interactions rather than just the usual dumb Cost-of-illness-study thing that just compares unemployment rates or income levels across users and non users and assumes that, but for use, people would have been the same.

Oh: also worth noting that there's some evidence that those with schizophrenia self-medicate with nicotine.

People ingest psychoactive substances for reasons. Ignoring those reasons and just claiming irrationality is unbecoming of economists.


Thanks for writing and for those comments.

"Ignoring those reasons and just claiming irrationality is unbecoming of economists."

Part of the Akerlof research agenda is to change the basic paradigm of economics. But there is that saying about babies and bathwater.

Frances: "We're going for sudden, massive coronary arrest here. Ideally after a day of canoeing, followed by a bottle of wine and steak cooked over the campfire." :-)

Are there any societies that are both economically successful and that abstain from alcohol? Does alcohol create trust? Because it is hard to run any economy without trust.

Nick: UAE?

At the risk of stating the obvious: there's also the consequences of regulation to consider. Prohibition and the War on Drugs(TM) weren't/aren't exactly shinning examples of policies that created a net increase in aggregate happiness. Except for those involved in organized crime.

Nick - basically the only oil-poor that's both rich and sober is Singapore. The Democratic Republic of Korea stands out as being a country that drinks a lot, despite having fairly low income. Otherwise there's the kind of relationship between income and alcohol consumption that you'd expect to see if alcohol was a normal good (this same pattern holds true with Canadian micro data, too - people with higher incomes generally speaking drink more). But the alcohol-creates-trust story might have some truth in it.

Patrick - I don't think the discussion here is about the best way of preventing harm - I agree with you about prohibition not working terribly well (also there's a literature suggesting that the increase in the legal drinking age in the US in the 1980s led to an increase in marijuana use). The more interesting question to my mind is "how much of a case is there to be made for trying to minimize harm?" For example, having alcohol available in supermarkets will decrease the non-monetary costs of purchasing alcohol, so would be expected to cause people to drink more (also it's there, and people are impulsive and suggestible). Is this a good thing or a bad thing? It all depends upon how harmful alcohol is. And there's very little serious discussion of this.

Fair enough. There's no doubt lots of public health care money to be saved and productivity gains to be had by reducing alcohol consumption. But the road to hell is paved with good intentions. It doesn't take prohibition to create inadvertent subsidies to organized crime. I think we could easily end-up trading e.g. reduced cancers for billions more to be spent trying to stop bootlegging and smuggling. Or you could get another Beer Store.

It's sad to see such sloppy work. I think that economics done right is science. Economics has so many beautiful results, both theoretical and empirical. Akerlof and Shiller set a poor example here on that front. Another Nobel laureate, Richard Feynman had this to say about finding things out:

“Science is a way of trying not to fool yourself. The principle is that you must not fool yourself, and you are the easiest person to fool.”

Akerlof and Shiller should be far more concerned about fooling themselves than about the phools in society.

Sometimes I find even smart economists forget or never really understood how it is that we know things. When Shiller discussed his Nobel prizing winning work, he had this to say about Fama's need for a proper definition of a bubble:

"What is a bubble? You [Gene Fama] said nobody defines it. So I will define it. A speculative bubble is a fad. People get excited sometimes. Too excited... Prices start going up, they start talking, the newspapers start writing about it, more and more people pile in to a market and they push prices up more and it goes on for a while. eventually it breaks and the bubble bursts."

That's bunk. That's not a DEFINITION, that's a THEORY. Scientists have little time for circular reasoning. If Noble prize winning economists can't tell the difference between a definition and an explanation, well perhaps very little of the field is actually engaged in science.


A theory is a definition of something that may or may not exist. When scientists work with theories, they begin by finding definitions for them because only then can they use evidence to determine whether the theory is good or bad. When Shiller makes a definition like that he is saying "This is what I and many others think is a bubble". It's a definition of a word, or more precisely a defnition of a concept. A clear definition is useful because it encourages fruitful discussion. That does not make it any less of a theory.

That was a bit OT, sorry. On a more substantive point, I think you're absolutely right to criticize Shiller and Akerlofs reasoning. The reason being that no situation occurs in isolation. If we allow poor reasoning in some cases because we think the conclusion drawn from it is beneficial, people may begin to accept poor reasoning elsewhere with less benign results. Of course, if you only do it once, that probably wouldn't not happen but it would be giving people a nudge in the wrong direction.

"A theory is a definition of something that may or may not exist." This is not what a definition means. You can't argue with a definition. There is nothing to debate. A theory is an explanation for something - it uses definitions, but you can argue for or against a theory.


Definition: The kilogram is defined as the mass of an object called the international prototype kilogram made of a platinum alloy known as "Pt‑10Ir", which is 90% platinum and 10% iridium (by mass) and is machined into a right-circular cylinder (height = diameter) of 39.17 millimeters to minimize its surface area.

Theory: Two one kilogram masses separated by distance 1 meter will experience an attractive force of 6.674×10^−11 N.

Notice that we can debate the validity of the theory. Maybe the 1 kg objects experience that force and maybe they don't. But we can't debate the definition of the kilogram.

Your second example is not a theory, it's a prediction BASED ON a theory. To be able to think clearly about that theory you will need to define it first.

Of course you can debate the definition of a kilogram. As you may know, the IPR is slowly losing weight and many would prefer to replace the definition you're using with one that defines the kilo in terms of the Planck constant. Even if the definition is changed in this fashion it is perfectly possible that some future great discovery in physics will cause us to reconsider that definition (say, if someone finds that the Planck constant is actually false).

Or take the definition of the word 'definition' which is what we are discussing here. Clearly there are at least two different definitions of the word which is why we are debating it.

Hugo André,

Debating definitions is about not understanding what definitions are. That's fine in day-to-day life - people aren't careful with language and they don't need to be. But in science you have to be careful or you will fool yourself. I can define the kilogram anyway I want - some definitions will be poor because other physical effects might affect my attempt at a definition or it might be hard to compare two standard masses. But that's a technical issue, not one of definition. Same thing in how I define temperature scales, etc.

In Shiller's case, he "defines" a bubble by giving me an explanation - a THEORY - of price movements. He says that price goes up because of a fad. That's the same thing as saying two objects attract one another because matter enjoys company or because of Newton's law of gravity or because little green men live inside all material. But those are theories - explanations. Fama is asking for a DEFINITION of a bubble - something that will allow me to tell when this fad or whatever it might be is actually going on. Shiller's response in a scientific context is just silly. Fama is not being pedantic here.

If economists can't get definitions separated from theories, well Feynman might have been justified in his contempt for the social sciences.

It is possible, as you say, that Shiller misunderstood Fama. That Fama was asking for a definition that would allow us to empirically separate "bubbles" from sharp price movements caused by shifting fundamentals. In answering, Shiller defines the abstract idea of a bubble instead of giving the sort of empirically testable definition that Fama is (perhaps) asking for. In that case, either Shiller should have paid more attention to what his counterpart was saying or Fama should have explained his point more clearly. Or both.

That is a wholly different discussion from the one about definitions, however. Science requires definitions both for the theories and the kind of empirically testable patterns that you (and perhaps Fama) are asking for here.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad