Unrealistic Mental Models 6

In the previous post (Three Types of Models 5), we discussed three types of models. The first type is based purely on patterns in observations, and does not attempt to go beyond what can be seen. This is an “observational” or Baconian model. The second type attempt to look through the surface and discover the hidden structures of reality which generate the observations we see. The best approach to this type of models has been developed by Roy Bhaskar, so we can call it a critical realist model or a Bhaskarian model. The third type of model creates depth and structures in our minds which create the patterns we see in the observations. The question whether our mental structures match reality is considered irrelevant. These may be called Kantian, or mental models.  The models of modern Economics are largely Kantian, while Econometric models are largely Baconian. The key defect of both of these approaches is that they GIVE UP on the idea of finding the truth. Max Weber’s ideas about methodology played an important role in this abandonment of the search for truth, but it would take us too far away from our current concerns to discuss this in any detail. Briefly, Weber thought that heterogeneity of human motives made “explanation” of social realities via “truth” impossibly complex. Instead, he argued that we should settle for a weaker concept based on “ideal-types” – deliberately over-simplified models of behavior which create a match to observed aggregated patterns of outcomes. We now discuss the disastrous consequences of this abandonment of truth in greater detail.

There is a famous article of Milton Friedman on methodology in economic theory, which recommends the abandonment of truth: “Truly important and significant hypotheses will be found to have “assumptions” that are wildly inaccurate descriptive representations of reality, and, in general, the more significant the theory, the more unrealistic the assumptions.” For more details, see Friedman’s Methodology: A Stake Through the Heart of Reason.”  What Friedman expresses, in inaccurate language, is the idea that an assumed structure of reality which is a mental model designed to match observations, need not match the true hidden structures of reality. All that matters is that observable implications of the model match the our observed data. This idea is called “saving the appearances”. For example, if we imagine that there is a heavenly sphere surrounding the earth and the moon is pasted on that sphere. Motions of the moon occur because of the rotations of the sphere. According the idea of “saving appearances”, as long as the observed motion of the moon matches the predictions of our model, we need not be concerned with whether or not the heavenly sphere actually exists.

This is fundamental methodological mistake at the heart of economics: the idea that we can make up any crazy model we like. As long as our models produce a match to the observations, it does not matter if we make wildly inaccurate assumptions. This has led to DSGE models, currently the dominant macroeconomic models, which have been held responsible for the fact that the profession of economists as a whole was blindsided by the Global Financial Crisis. Economists make completely unrealistic assumptions without any discomfort, because of Friedman’s idea that “wildly inaccurate” assumptions will lead to truly important and significant hypothesis. In a previous portion of this article, we documented the fact that economists are not bothered by conflicts between their models and reality. Below we provide quotes which document the crazy models that now dominate economics because of adherence to Friedman’s Folly: the crazier the assumptions, the better the model.

Keynes: The classical theorists resemble Euclidean geometers in a non-Euclidean world who, discovering that in experience straight lines apparently parallel often meet, rebuke the lines for not keeping straight as the only remedy for the unfortunate collisions which are occurring. Yet, in truth, there is no remedy except to throw over the axiom of parallels and to work out a non-Euclidean geometry. Something similar is required today in economics. (GT)

Solow: Suppose someone sits down where you are sitting right now and announces to me that he is Napoleon Bonaparte. The last thing I want to do with him is to get involved in a technical discussion of cavalry tactics at the battle of Austerlitz. If I do that, I’m getting tacitly drawn into the game that he is Napoleon. Now, Bob Lucas and Tom Sargent like nothing better than to get drawn into technical discussions, because then you have tacitly gone along with their fundamental assumptions; your attention is attracted away from the basic weakness of the whole story. Since I find that fundamental framework ludicrous, I respond by treating it as ludicrous — that is, by laughing at it — so as not to fall into the trap of taking it seriously and passing on to matters of technique.

Narayana Kocherlakota: Minneapolis Federal Reserve President (2010-2015), “Toy Models”, July 14 2016  “The starting premise for serious models is that there is a well-established body of macroeconomic theory… My own view is that, after the highly surprising nature of the data flow over the past ten years, this basic premise of “serious” modeling is wrong: we simply do not have a settled successful theory of the macroeconomy.”

Olivier Blanchard IMF Chief Economist (2010-2015), “Do DSGE Models Have a Future?”, August 2016  “DSGE models have come to play a dominant role in macroeconomic research. Some see them as the sign that macroeconomics has become a mature science, organized around a microfounded common core. Others see them as a dangerous dead end…”  and “There are many reasons to dislike current DSGE models. First: They are based on unappealing assumptions. Not just simplifying assumptions, as any model must, but assumptions profoundly at odds with what we know about consumers and firms.”

All of these authors are expressing the same complaint, in different forms. Structures of our Mental Models have no match to the true Structures of External Reality. The only job mental models have to do is to produce a match to the observed data. Whether or not mental models are realistic has no bearing on whether or not they are good models. There is complete lack of concern about whether our mental models make assumptions which are realistic. The mystery of how models based on false assumptions can help us “understand” and “explain” the real world has been the subject of a long and complex methodological debate. For example, a leading methodologist, Mary Morgan, writes that “Despite the ubiquity of modelling in modern economics, it is not easy to say how this way of doing science works. Scientific models are not self-evident things, and it is not obvious how such research objects are made, nor how a scientist reasons with them, nor to what purpose.” In the “Explanation Paradox”, Julian Reiss writes that it is widely accepted that: (1) economic models are false; (2) economic models are nevertheless explanatory; and (3) only true accounts explain. A whole subsequent issue of the Journal of Economic Methodology is devoted to the attempt to EXPLAIN how all THREE of Reiss’ premises can be true. Alexandrova and Northcott – philosopher-outsiders – point out the obvious: economic models do not explain. However, this simple explanation falls on deaf ears; economists are too much addicted to meaningless mathematical models to realize that these models are mental structures which are “castles hanging in the air, having no contact with reality”. For full references and discussion on these issues, see  Tony Lawson’s Beyond Deductivism 

NEXT: We plan to discuss how econometric models are mostly observational models, while economic models are mental models. However, differentiating econometric models from real structural models requires a discussion of the hidden and unobservable causal structures which generate the observable data. This is done in a sequence of five posts on Simpson’s Paradox. These posts provide a leisurely pedagogical introduction to the topic, and also relate the paradox to causal structures, something not available in the conventional statistical literature on the subject. After covering Simpson’s Paradox, and some related materials on causality and econometric regression models, we will come back to the present topic – the nature of economic and econometric models, and why those models create results conflicting with easily observable reality.

1 comment
  1. Algorithmic comments are as follows:

    1. In my views, heterodox economists sometimes misunderstand mainstream economists, consequently criticizing too much. To my understanding, many mainstream economists have had being struggling for the demerits of their writings for long times. The real problem is: no way of improvement. Complaining should always be quite easy, but, how to do next?

    2. The defects of economics reflect the nonexistence both of unified economics and unified social science. If one agrees with the nonexistences, is it astonishing that economics witnesses so many failures? Up to date, it is alleged that the methodology of social sciences has not been desirably established yet.

    3. The Algorithm Framework Theory (AFT) is exactly designed to constitute the whole principles of economics and social science, and therefore to settle such methodological problems as above.

    4. In addition to the comments on the previous post, I’d like to stress here that all the three types of models are not essentially different. Both Baconian models and Reality models are thoughtful or Kantian, resulted from information-processing (or computations). No nude fact available, no nude reality available. The differences lie only at the technical aspects such as processing depth, processing skills, quality of results, etc. “Reality” is only a belief, a certainty, a hypothesis, an analytical strategy, or a relatively more reliable or “better” imagination.

    5. Any theoretical premise (or assumption) cannot be certified completely, so it is more or less “unrealistic”. Human thinking is fragmented (or discrete). Sometimes the inferences based on “reliable” premises cannot go ahead, then the premises may be revised into “unrealistic” and the following inferences will thus go smoothly to a desirable destination. Otherwise Einstein’s Relativity Theory may not be found out. The mistake of Friedman’s idea, in my opinion, is not his preference to establish unrealistic assumptions, but that he forgot the following necessary steps, that is, the steps of verifying the assumptions with realities, or of coordinating the assumptions with other assumptions or theories, and then of deepening the researches. Friedman stopped, just showing off his “creative” idea, so it is quite misleading. In fact, when an unrealistic assumption is “successfully” established, it implies that a big theoretical crisis happens, and the following re-examination of the theories relevant must be done. It is not a moment of celebrating for theorists, but of mourning.

    6. However, Algorithmically speaking, conflicts are more or less inevitable. Theories expand without (noticeable) inconsistencies but slowly, or expand quicker but with some internal conflicts. This is a trade-off, subject to weighing up from time to time.

    7. Fortunately, above all of the demerits, there is a high-order consistency, which AFT reveals, and I have repeat many times before. Time will make everything different. Viewing the following paper is welcome. Thanks! https://goingdigital2019.weaconferences.net/papers/how-could-the-cognitive-revolution-happen-to-economics-an-introduction-to-the-algorithm-framework-theory/

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: