Does it Matter if Early Models are Flawed?

One of the central objections to this post was that ‘later’ models incorporate dynamism, and early ones are just introductory, meant to establish basic concepts. I’m not sure this is accurate.

Firstly, are models like supply-demand and IS/LM just introductory? It strikes me that demand-supply is regularly trotted out by academic economists as an argument against the minimum wage and other price controls, while probably the most famous economist in the world has been vehement that IS/LM is all that’s needed to understand our current situation.

Furthermore, ‘basic’ models are all that the overwhelming majority of people ever learn about economics, including many policymakers. If they are that flawed, the correct response isn’t to dismiss 99.9% of the population as ignorant and carry on, but to try and make the basics of economics resemble reality better.

However, neither of these are my main concern with this line of thinking. The problem is that economists regularly make comparisons to physics/engineering, where early models do not incorporate friction, planets are modeled as perfect spheres, etc. I must stress this: economists do not understand the difference between a good assumption and a bad one.

If basic economic models were taught like physics, we would be told the impact of each individual assumption on the analysis, and as things got more complex, we would relax each assumption, one by one. But this isn’t what happens. What happens is that the core of the analysis is kept intact while more epicycles ‘frictions’ are incorporated when reality doesn’t conform to the model. Steve Keen describes this as a ‘dog walking on its hind legs‘ – it can seem to work for a while, but look away and it will return to its former four legged glory.

Keen offers the example of Paul Krugman, attempting to model the crisis within a neoclassical framework:

In what follows, we begin by setting out a flexible-price endowment model in which “impatient” agents borrow from “patient” agents.

Note how the standard ‘banks as intermediaries between savers and borrowers’ model isn’t questioned – instead, it is protected as Krugman attempts to model the crisis ad hoc, a one off due to the ‘patience’ of Chinese people and ‘impatience’ of Americans. The standard neoclassical framework can then be reestablished once the paper is finished.

Why do economists do this? Well, once you start to relax assumptions, models like Arrow-Debreu simply collapse completely. If people are unable to predict the future and hence to model their consumption and production functions based on that, the model just has nothing to say. Thus, to protect the model, we have to ‘add’ rather than ‘take away’ criteria. If relaxing an assumption in engineering meant the entire model collapsed, it would be abandoned.

But, as I said on Daniel Kuehn’s blog, people have this weird cognitive dissonance with economics, a sort of ‘let’s not throw the baby out of bathwater’ mentality despite empirical or logical contradictions. And the funny thing is that the same people don’t apply it to other sciences/social sciences. For me, it feels like defenders of economics are engaged in lengthy special pleading.