Must We Choose a Paradigm?

Simon Wren-Lewis makes a plea for a more open attitude to behavioural economics in macro.  Overall, Wren-Lewis supports the idea that macro models should be microfounded, but believes that the micro foundations used are outdated and ignore inconvenient facts about how people really behave.

To me, this is merely symptomatic of a more general problem in economics.  This is the tendency to focus entirely on one particular paradigm or modelling style, and reject any approach that might contradict it.  This seems to me a dangerous attitude.

Before I go any further I should say something about my own approach.  Those familiar with this blog will know that I often resort to models to develop my own understanding and to illustrate what I think are interesting points.  They will also probably have noticed that many of my models are of a particular type, with a heavy focus on social accounting and institutional structure. 

However, the reason for this is not that I think this is the only way to model, but is rather more basic.  The reason for the absence of New 'Keynesian' style modelling in this blog is mainly that I'm not very good at it.  I never formally studied it and although I have enough of a grasp to be able to understand other people's work on this, I do not feel fluent enough to present ideas within that framework myself.  So I stick to what I know.

But that does not mean that I think DSGE modelling is wrong.  Actually, I don't think any modelling is right or wrong.  As I have said before, all any model does is illustrate how certain conclusions flow from certain assumptions.  It never actually tell us how things really are, because the assumptions always represent a gross simplification of the real world. 

This means, I think, that there is a lot to be learned from taking into account the results of different models.  Some models may be better than others, but all have their weaknesses.  Using different approaches helps us identify those weaknesses and understand how to interpret our results accordingly.

So what happens if two different models give us seemingly contradictory results?  Does this not mean we have to choose to simply reject one of them?  Absolutely not.  In fact, I think this provides us with an excellent opportunity to identify how crucial certain assumptions are and to give us greater insight into the strengths and limitations of both approaches.