Models as Fairytales

One of the meta-economics discussions we had in our Micro class this past term was about how we should view economic theory. Is it an attempt, as in science, to discern the rules by which the real world is governed? Is it a tool to predict the future? Is it trial-and-error? To what extent do the assumptions we make have to match reality?

Our prof told us he thinks of economic models as fables. As in a fable, the assumptions can be all wrong (ants can't speak to grasshoppers!). But, as with a good fable, a good model provides you with insight into the world. For a model to have value it must simplify the world, and part of that simplification will by necessity do harm to 'reality'. This view of models (which is basically what economic theory consists of) comes from Rubinstein. His Microeconomic Lecture Notes reflect on this in the introduction. Scroll down to that section and read the first two pages.

I'm still trying to decide where I stand on this issue. In some ways I think of the 'fables' interpretation of models as something of a cop-out. It's easy for a theoretical economist to say that models shouldn't be criticized for having unrealistic assumptions, but what if you want to apply your models to data? What if you want to make recommendations about how society should be run? You have to think more seriously about what the consequences of models of one type or another are. What is the result of assuming homogeneous agents? Does it matter that people don't always behave 'rationally'?

Obviously on some level it does matter. But still we must simplify. So where is simplification harmful? I tend to think of this as an empirical question, but there are some important things we cannot test. Ask me again in a couple years, I will hopefully have a better answer then.