Are all models wrong?

8 Sep, 2015 at 13:44 | Posted in Economics | 11 Comments

quote-all-models-are-wrong-but-some-are-useful-george-e-p-box-53-42-27If you say “All models are wrong” then the most important issue is to define the words. “All” is quite clear, “are” also is without much doubt. So, we are left with “models” and “wrong” …

The more interesting discussion is the one about the definition of truth. The philosopher Bertrand Russell has written something on truth about a century ago in his book The Problems of Philosophy (ch. XII):

“It will be seen that minds do not create truth or falsehood. They create beliefs, but when once the beliefs are created, the mind cannot make them true or false, except in the special case where they concern future things which are within the power of the person believing, such as catching trains. What makes a belief true is a fact, and this fact does not (except in exceptional cases) in any way involve the mind of the person who has the belief.”

Truth, Russell says, is correspondence with facts. If minds do not create truth or falsehood, but only beliefs, then I would argue that models are beliefs. They can be true when they correspond to the facts. So, there is hope! Models can be right after all! … A model is an abstraction, but as such it can be right. Of course, there is no proof that a model that has been right today will be right tomorrow, but that only makes economics an art.

Dirk Ehnts

Interesting reading that — once again — shows that being able to model and investigate a credible world, a world that somehow could be considered real or similar to the real world, is not the same as investigating the real world. Even though all models are false, since they simplify, they may still possibly serve our pursuit of truth. But then they cannot be unrealistic or false in any way. The falsehood or unrealisticness has to be qualified (in terms of resemblance, relevance, etc.). At the very least, the minimalist demand on models in terms of credibility has to give away to a stronger epistemic demand of appropriate similarity and plausibility.

The predominant strategy in ‘modern’ economics is to build models and make things happen in these “analogue-economy models” rather than engineering things happening in real economies. And as a rule the modelers consider their work done when they have been able to convince themselves that the model is valid. But — really — to have valid evidence is not enough. What economics needs is sound evidence. Why? Simply because the premises of a valid argument do not have to be true, but a sound argument, on the other hand, is not only valid, but builds on premises that are true. Aiming only for validity, without soundness, is setting the economics aspirations level too low for developing a realist and relevant science.


  1. ““All models are false because they simplify” is False.”

    Depends on what you mean by false.

  2. The problem isn’t the word “model” or the word “true” or the word “fact”.
    The problem is the word “science”.
    “Science” is not a synonym of the word “physics”. Newtonian physics is actually kind of an outlier among sciences. But we are living in the last days of the Enlightenment. Why do we imagine a “political spectrum”? Because Newton’s second most important book was “Opticks” and one of the most important commentaries on it was written by Voltaire. It is a horrible match for the reality of politics.
    Economists are in the same rut, but more so.
    Here is a true fact: biology is not reducible to physics. This fact should blow the mind of orthodox and heterodox alike.
    The way models work in physics is not the way they work in biology.
    Physics is an outlier.
    Please think about this, Lars. “All models are false because they simplify” is False. Until you understand this, there will be no progress.

  3. The happy end of the social science delusion
    Comment on ‘Are all models wrong?’
    For more than two thousand years now, physicists have seen falling leafs but until this day we have no scientific theory of this phenomenon. Instead, we have the theory of gravity, motion, ballistics, etc.
    The first thing to notice is that physicists are not concerned with reality as such but with certain aspects of reality. In other words, reality is narrowed down to those general aspects that can be dealt with effectively by applying tools which are already available or are developed specifically for the task at hand.
    What a scientist instinctively avoids are questions that cannot be answered in principle like: is God male or female? Exactly this type of question is what occupies non-scientists most of their time.
    Science is often accused of being narrow or over-simplistic. However, those who have tried to tackle the fullness of reality head-on have never produced more than colorful descriptions, interesting gossip, subjective interpretations, data-decorated hunches, or extensively interpolated individual histories. In the meantime, narrowly focused science has moved in an unbroken sequence of logical steps from the elementary Law of the Lever to the unobservable deep reality of quantum particles and quarks.
    Reality is not something given and simply to be looked upon — this is the delusion of naive empiricism — but is continuously redefined in the course of scientific research itself.
    While it is obvious that economics has much to do with human behavior it is a methodological mistake to primarily focus on this aspect of economic reality, just as it a mistake to focus on individual flying leafs in order to find out something general about falling bodies. No way leads from the observation of a myriad of falling leafs to the Universal Law of Falling Bodies. Likewise, no way leads from the observation of individual behavior to the systemic economic laws.
    The so-called social sciences have always been a misguided endeavor. The reason is simple. “By having a vague theory it is possible to get either result. … It is usually said when this is pointed out, ‘When you are dealing with psychological matters things can’t be defined so precisely’. Yes, but then you cannot claim to know anything about it.” (Feynman, 1992, p. 159)
    Certain knowledge about human behavior is hard to come by, to say the least, and to circumvent this fundamental difficulty by postulating constrained optimization is of breathtaking naivety — or worse.
    It has not been such a good idea to build economics on a green-cheese assumption about human behavior. Therefore, “… if we wish to place economic science upon a solid basis, we must make it completely independent of psychological assumptions and philosophical hypotheses.” (Slutzky, quoted in Mirowski, 1995, p. 362) see also (Hudík, 2011)
    This is exactly the opposite of what may be called Hume’s methodological delusion: “And as the science of man is the only solid foundation for the other sciences, so the only solid foundation we can give to this science itself must be laid on experience and observation.” (Hume, 2012, Introduction)
    The results of Hume’s program are as to be expected: “…there has been no progress in developing laws of human behavior for the last twenty-five hundred years.” (Hausman, 1992, p. 320)
    The science of man, or what we call social sciences, has no solid foundation. As a matter of principle, there is no such thing as a social science. Therefore, the first methodological task is to take economics out of what Feynman aptly called cargo cult science.
    The new definition of economics is objective-structural-systemic and entirely free of any behavioral connotations: economics is the science which studies how the monetary economy works. Thus, the popular yet forever pointless second-guessing of human motives and behavior can be left to psychology, sociology, political sciences, philosophy, mythology, literature or science fiction.
    Economics is a failed science and the ultimate cause is given with this definition: “It is a touchstone of accepted economics that all explanations must run in terms of the actions and reactions of individuals.“ (Arrow, 1994, p. 1)
    Accepted economics is scientifically unacceptable.
    Egmont Kakarot-Handtke
    Arrow, K. J. (1994). Methodological Individualism and Social Knowledge.
    American Economic Review, Papers and Proceedings, 84(2): 1–9. URL http:
    Feynman, R. P. (1992). The Character of Physical Law. London: Penguin.
    Hausman, D. M. (1992). The Inexact and Separate Science of Economics. Cambridge: Cambridge University Press.
    Hudík, M. (2011). Why Economics is Not a Science of Behaviour. Journal of
    Economic Methodology, 18(2): 147–162.
    Hume, D. (2012). A Treatise of Human Nature. Project Gutenberg EBook. URL
    Mirowski, P. (1995). More Heat than Light. Cambridge: Cambridge University

  4. “If you want to have rock solid logical entailment, don’t do social sciences!”

    I entirely accept what you say about inference to the best explanation but there are still problems around the notions of facts, beliefs and the nature of reality. I guess all we can do is find a consensus then be prepared to be corrected.

  5. I keep meaning to read Box’s original, but haven’t got around to it. But from the Russell/Keynes school of thinking it is clear that no empirical model can be known to be true to reality. Russell’s observation that the mind creates beliefs also needs unpacking. Russell is pointing out that they are only beliefs and not necessarily true to reality. Hence all beliefs are rather weak things, not like the classical ‘belief in God’.

    As Pontus points out, the notion that the solar system was geocentric was true to the dominant world view of the time, but we never could have known that it was true to reality. No Physics can be known to be true, but it can be useful in Box’s sense.

    It follows that no theory can be sound in Lars’ sense, but at least we can test the theory, and have some grounds for supposing that an intended application is in similar circumstances to those of the tests. (Think 2005-date.)

    • Dave, I think we have to be careful here since the way scientists justify and argue are quite different from what mathematicians do. In science we have to accept that the warrant and/or justification we come up with when relying on data/empirical facts isn’t deductive. Contrary to deductive proofs, our justifications always come in degrees and are more or less strong. This is inevitable. It is the prize we have to pay when doing non-entailing impliative science using induction & abduction. Coming up with sound evidence is not equivalent to deductive entailment. As I see it we can never come up with anything better than inference to the best explanation — looking for what causes things, we search for that particular cause that best explains them.

      • Lars, If we take Russell and Keynes seriously, then ‘inference to the best explanation’ can be a dangerous notion. In almost any empirical situation there will be a range of explanations that explain the evidence satisfactorily (at least, according to Jack Good’s development of Keynes). To most of us, most of the time, this may seem unimportant. But it seems to me that sometimes (as in 2005-date economies) it can be important, since what we now regard as ‘the best’ may turn out to be very bad.

    • ” As I see it we can never come up with anything better than inference to the best explanation.”

      While this might be the best we can do, it is still problematical. Who decides what is the best explanation? For instance, is the heliocentric theory the best explanation for the observed behaviour of the celetial bodies? It seems to me a reasonable explanation can be constructed around geocentricism.

      • I understand your frustration, but you seem to be looking for what I would call an exaggerated aspiration level. If you want to have rock solid logical entailment, don’t do social sciences! Of cause there is no absolute deductive security when reasoning in social sciences. As I have been arguing over and over again on this blog, that’s the reason we have to argue and justify the conclusions we draw and the assumptions we make. But make no mistake — no matter how hard you try to justify and warrant your assumptions/models/theories you can never in any logical meaning PROVE them!

  6. “Truth, Russell says, is correspondence with facts.”

    How do we define “fact”?

    700 years ago it was believed that the solar system was geocentric.

    Now the “facts” tell us it is heliocentric.

    • Henry, I agree. The sentence is circular.

Sorry, the comment form is closed at this time.

Blog at
Entries and Comments feeds.