What do you think?
Rate this book


Unknown Binding
First published July 6, 1995
Many of the models in the literature are not general equilibrium models in my sense. Of those that are, most are intermediate in scope: broader than examples, but much narrower than the full general equilibrium model. They are narrower, not for carefully spelled out economic reasons, but for reasons of convenience. I don't know what to do with models like that, especially when the designer says he imposed restrictions to simplify the model or to make it more likely that conventional data will lead us to reject it. The full general equilibrium model is about as simple as a model can be: we need only a few equations to describe it, and each is easy to understand. The restrictions usually strike me as extreme. When we reject a restricted version of the general equilibrium model, we are not rejecting the general equilibrium model itself. So why bother "testing" the restricted version? If we reject it, we will just create another version (pg. 4).
"Estimation" suggests a Bayesian approach to data, while "testing" suggests a classical approach. I prefer estimation, since I think researchers who want to test often choose models that are more specific than the economics require. They talk of adding restrictions for no reason other than making their models easier to reject (pg. 97).
I think it's better to "estimate" a model than to test it. ... Best of all, though, is to "explore" a model. This means creating many specific examples of a general model, where each one explains a single stylized fact or perhaps a few features of the world. It means using some of these examples to elucidate microeconomic evidence. ...
"Exploring general equilibrium," then, means generating examples and specific versions of the full general equilibrium model, to explain stylized facts and other data. It means changing examples quickly when they don't fit the facts. It means avoiding formal testing, or even estimation of heavily restricted models. When a general equilibrium model and another kind of model both fit the facts, it means favoring the general equilibrium model (pgs. 4, 5).