Skip to content

Taking Absurd Theories Seriously: Economics and the Case of Rational Addiction Theories

Ole Rogeberg writes:

Saw your comments on rational addiction – thought you might like to know that some economists think the “theory” is pretty silly as well. It’s worse than you think: They assume people smoke cigarettes, shoot up heroin etc. at increasing rates because they’ve planned out their future consumption paths and found that to be the optimal way to adjust their “addiction stocks” in the way maximizing discounted, lifetime utility. To quote Becker and Murphy’s original article: “[I]n our model, both present and future behavior are part of a consistent, maximizing plan.”

Yeah, right

Here’s Ole’s article, “Taking Absurd Theories Seriously: Economics and the Case of Rational Addiction Theories,” which begins:

Rational addiction theories illustrate how absurd choice theories in economics get taken seriously as possibly true explanations and tools for welfare analysis despite being poorly interpreted, empirically unfalsifiable, and based on wildly inaccurate assumptions selectively justified by ad-hoc stories. The lack of transparency introduced by poorly anchored mathematical models, the psychological persuasiveness of stories, and the way the profession neglects relevant issues are suggested as explanations for how what we perhaps should see as displays of technical skill and ingenuity are allowed to blur the lines between science and games.

I agree, and I’d also add that this problem isn’t unique to economics. Political science and statistics also have lots of silly models that seem to have a life of their own.


  1. dk says:

    Rogeberg thinks the problem is bound up w/ the stupefying quality of mathematical models in economics. Perhaps. But I'd say he is underestimating the motivating effect of ideology, values, worldviews, etc. on cognition. He suggests that Becker et al. couldn't really believe their own theory–they are engaged in a "theory game." He obviously just doesn't know very many Chicago School economists! They really believe what they are able to "prove" with their models — that markets are always wealth- and autonomy-maximizing — and they believe the models because the conclusions are so congenial to their ideology (I have a mathematical model that proves this, by the way). The readers who are persuaded by the the theorists' models are in the same boat. And of course this same point applies to those who produce or are persuaded by mathematical models or by highly technical empirical analyses of one sort or another in political science and other disciplines. What's more, those who *disbelieve* the "absurd" theories are also motivated by their values, too. I don't think this means that mathematical models or empirics are "pointless" or "value driven" etc. On the contrary, I think it means that it is psychologically naive to blame the "impenetrability of mathematical arguments as compared to verbal arguments" (Rogeberg, p. 264) for acceptance of "absurd" theories; indeed, it is naive, or at least insufficiently reflective about one's own predispositions, to call the theories "absurd"….

  2. Andy says:

    It's very easy to roll one's eyes about how the rational addiction model assumes people make plans with perfect foresight, but it's much harder to argue when that model fits some aspects of the data that no one else considered and that other "irrational" models don't speak to. In particular, people respond TODAY to expected price changes in cigarettes in the FUTURE. Further, they do it in pretty close to the way the Becker-Murphy model predicts.

    This was a novel insight, and I think lends support to the notion that people, even addicts, can make plans for both today and tomorrow and that an aspect of addiction can be called "rational". The fact that the modeling requires perfect planning is generally considered a mathematical convenience and arguing over that is missing the point, at least in my opinion.

  3. Thorfinn says:

    Well, there are plenty of empirical studies on this. Some find, for instance, that higher taxes reduce consumption, even before the law takes effect. The Irish bar ban example is another intervention that radically reduced smoking. There's even a book; Addiction: A Disorder of Choice, that takes this idea seriously, but from a purely psychological and medical point of view. The model may or may not be true, but it's being evaluated (as Becker and Murphy called for) from an empirical rather than purely theory point of view.

    One can easily turn the table. What evidence do you have that there are no elements of choice or rationality in the decision to use drugs? If you don't, than you're just as guilty of using implicit and hand-waving assumptions as Becker and Murphy.

  4. Dave says:

    Thomas Schelling also did work on addiction. Here is Tim Harford on Schelling:

    For Schelling, the addict was neither perfectly rational nor irrational and helpless – he was a rational being at war with himself, who could deploy strategies to help him win that war.

    … It was many decades before Becker and Murphy formulated their hypothesis but Schelling says “I learned then that they don’t know what they’re talking about.”

  5. Andrew Gelman says:

    Andy, Thorfinn: As I wrote in my original blog entry (linked to above), there's certainly rationality within the context of addiction (e.g., wanting to get a good price on cigarettes), but "rational addiction" seems to miss the point. Presumably different people are addicted in different ways. Some people are definitely addicted in the real sense that they want to quit but they can't, perhaps others are addicted rationally (whatever that means). I could imagine fitting some sort of mixture model or varying-parameter model. I could imagine some sort of rational addiction model as a null hypothesis or straw man. I can't imagine it as a serious model of smoking behavior.

    To put it another way, of course it makes sense that smokers will respond to potential price increases, and this is a perfectly reasonable thing for economists to study. Contrary to Thorlinn's implication I never wrote that "there are no elements of choice or rationality in the decision to use drugs." My complaint is when rationality is viewed as all-emcompassing. (Recall Ziliak and McCloskey's statement that "cigarette smoking may be addictive. [emphasis added].") Rationality is a mode of thinking and behavior. It's an important part of life, and it's perfectly reasonable to include rational behavior in your model without letting it take over.

  6. What is "absurd" about rational addiction theories? I realize that the common understanding of the words rational and addiction make "rational addiction" sound like an oxymoron, but when these terms are defined precisely, as they are in the theory, I don't see anything absurd.

    Nor do I believe this framework forces researchers to oppose government regulations and taxes.

  7. Its like George Box said, all models are false, but some are useful.

    So the important question is, is the model of rational addiction useful? I would argue that has been (and I'm a sociologist!). But if someone wants to make the case that it distracted attention from better models, I remain open to persuasion. Note, depending on what the goals of your research are, I'm certain that there are better models.
    Until people lay out:
    1) the goals for using a particular model,
    2) alternative models

    I don't think discussion can be very fruitful.

  8. Andrew Gelman says:

    Michael: I agree that models with big holes in them can still be useful. In my own field, I think that so-called spatial voting models (where individuals live on a two-dimensional space of policy preferences) are useful, even though they are obvious oversimplifications. But I do think that Ole Rogeberg makes a good case about rational addiction models in his article linked to in the above entry.

  9. Martin says:

    The paper of Becker & Murphy is at :

  10. Andy says:

    I think I'm responding negatively to the claims that the Becker-Murphy model rational addiction is a tautological or contentless theory. Rogeberg's explicit claim is that it is not "empirically falsifiable".

    As I believe my previous example showed, I think this is wrong-headed. The rational addiction model has clear testable predictions about how addicts, both new and old, respond to current and future prices. A reading of the original paper supports this claim.

    Yes, economists and others often go too far in using rationality in a tautological sense, but I don't believe the rational addiction model is one of those cases. It's easy to attack because the claim that there are some people who want to quit but can't is a strong narrative case. But to quote the Becker-Murphy paper, although addicts are often unhappy and depressed "[they] would be even more unhappy if they were prevented from consuming the addictive goods."

  11. Ed says:

    A better way of putting it would be rational consumption of addictive drugs.

    The whole point of addiction is that you feel compelled to keep doing something regardless of whether its rational for you. So "rational addiction" is a bad term. But you can have reasons, sometimes even good reasons, for using drugs DESPITE their being addictive.

    This might come up for people with terminal illnesses. In the past, doctors were reluctant to prescribe effective pain relievers because the people with the illnesses might become addicted to them. But is that really a relevant consideration for someone who will likely be dead in a couple of years?

    You can consider the possibility of being addicted as part of the "cost" of using a drug that might have certain benefits. Now you can talk about externalities, part of the cost of addiction is borne by people other than the drug user (but most of it is a real cost to the drug user).

    Imagine if economists talked about people rationally getting into car accidents. No, people drive, sometimes too fast, and run the risk of getting into car accidents. I think behind alot of the confusion is that people poorly understand risk.

  12. @Ed, I guess we can argue for using different terminology in economics, but we're unlikely to have much luck. The way economists talk, "rational car accidents" makes sense… Perhaps an individual, or a country, could reduce the risk of car accidents by lowering the speed limit. We might consider the reduction in risk to be too small to justify slowing transportation.

  13. Ole Rogeberg says:

    @Michael Bishop, I completely agree that you cannot judge a theory before you specify the goals for using a theory. If you read the introduction to my article (linked to by Andrew above), you’ll see that I specifically criticize claims that the theories explain addictions (in the sense of describing a causal mechanism that could plausibly be behind the observed phenomenon) or are valid for welfare analysis  (i.e., allows us to make credible statements about how the welfare of real addicts will be affected by policy). These claims are frequently made (lots of quotes in footnotes 1 and 2) – but these claims are wrong.

    If someone based a theory on the assumption that bouncing rubber balls are blind to their environment but aim to minimize unpleasant internal pressure, it may well be that this theory could be adjusted to predict very well how it bounced. That doesn’t mean we can make meaningful statements about the welfare of rubber balls. It would be an as if theory that predicts well despite being based on false assumptions.

    @Andy, I guess I shouldn’t have said that the rational addiction theories are absurd – they might even predict certain things pretty well – I should have said they are absurd if taken as potentially true explanations for why people are addicted and if taken as useful ways of predicting the welfare of addicts under various policies. And predicting well is not that difficult given that you can alter the number of stocks and their dynamics to get constant, rising, cyclical and chaotic “rational addiction” consumption patterns, including (if you want) endogenous quitting of the cold turkey as well as gradual kind.

    A flexible modeling framework + assumptions loosely and selectively tied to anecdotes and data + empirical regularities that you can mould your dynamics to = as if theories that can predict (certain things) well yet give us no reasons to believe their proposed mechanisms or implied “welfare results”

    @dk, Whether these "welfare results" support liberalization or drug control, btw, is irrelevant in this regard – the method can do both (just add hyperbolic discounting, cue triggered cravings or some other behavioral economics ingredient to the stew to get the latter).