Apt quote from Diane Coyle, who makes the point that a model is just a model and as such it is dependent on the assumptions used to construct it. But the quote doesn't go far enough.
Issues other than formal consistency arise when a model is interpreted semantically with respect to the world it models when that world is assumed to be real and therefore experience becomes a test of the semantic interpretation of the model. All models are necessarily simplifications. Replicas to scale are highly isomorphic representations, but differ in size from what they model.
But most models are more abstract. Their representational power lies in their isomorphism with what is modeled, but the degree of correspondence is much less exactly than replicas. Often this is not an issue, since either a higher degree is not required of the model with respect to its purpose, or the subject matter doesn't permit it. This is accomplished by loosening the assumptions.
The greater the abstraction, the more questions arise about how much weight assumptions are being asked to bear, and if they are being pushed too far.
Although we all knew at some level that the rational choice assumption was being made to bear too much weight, very few economists openly challenged its everyday use in justifying public policy decisions. Very few of us put this weight on it in our own work. But not all that many economists challenged its pervasive use in the public policy world…Well, now we know how that turned out, and it is not over even yet — over seven years later, with untold consequences, socially, politically, financially, and economically.
This is comparable to using wrong assumptions in engineering in building or bridge construction, or designing dams, bulwarks or levies, with the consequence that the structure collapses owing to unplanned for stress resulting in huge loss of life and property, and a setback in the area lasting years.
When a model is interpreted semantically as representative of a world, its isomorphism can only be decided based on comparing aspects of the model (hypotheses derived from it as theorems) to what it purports to model, based on experience. This is to say that no amount of analysis of a model that is claimed to be representational can justify the reliability of the model as a representational one. For that, comparing the model to what is modeled is required. Much of the scientific method is about doing this rigorously.
Of course, it is also a requisite that modeling be formally consistent, but formal consistency implies nothing about what's outside the symbolism of the model. Formal consistency implies that if the logic, and math is a brach of logic, is valid and the premises are correct, then the theorems that follow are true as hypotheses in a semantic interpretation. But this is no guaranteed of the truth of the hypotheses semantically. It is just an assertion that follows from the logic. It is claim that stands in need of testing. A logical pedigree is never an empirical warrant independently.
However, this is not the end of it. If it were, then a failure of the model could be passed off inadvertent, with the excuse that no one could have known before testing, or that conditions were more severe than could reasonably be expected. No fault. No blame.
But that is a often a bogus excuse. Analysis can reveal likely weak points in a model where the assumptions are being relied on to bear too much stress, just like in engineering. Engineers don't have to build a bridge to test it, for example. They use design principles that have already been found reliable for good reasons. In most cases, this is trivial where conditions are well-known and the capability of materials and procedure is well established. But in many cases, there is ambiguity or even the presence of unknowns, for example in putting a person on the surface of the moon for the first time. Clearly, those involved, the engineers especially, were not cavalier about this.
In economics, however, this doesn't seem to be the case, where being cavalier — appearing certain in the face of uncertainty — is expected of Very Serious People. Worse, non-economists have not only noticed weak assumptions, but warned in cases where economists were putting too much weight on assumptions in model construction, not only from common experience and common sense, which might be suspect, but also from work in related sciences such as psychology and sociology. Where the FBI saw massive fraud taking place in the real estate bubble as it was building, Alan Greenspan, the nation's "chief economist," saw "froth," assuming that financial experts, corporate management, and the market itself know better.
This failure would just be embarrassing to the field if macroeconomics were not used as a policy science and relied on extensively in policy formulation. Then mistakes become disastrous, and they may even led to catastrophes resulting in huge losses that can never be recaptured, or lives that can never be set right. It's grossly irresponsible — not to mention highly inefficient, when economics is supposed to assist individuals, firms and societies in avoiding or at least reducing inefficiency, vitiating a chief purpose of the discipline.
The financial crisis followed by the non-recovery, as well as the crisis in Euroland, should have put economists on alert to recheck their assumptions and those in charge of policy formulation and financial regulation to review their assumptions also. Instead, many have just doubled down. Science or ideology? Whatever, it's more than laziness.
Lars P. Syll’s Blog
Modern macroeconomics — an intellectually lazy ideologyLars P. Syll | Professor, Malmo University
No comments:
Post a Comment