Tuesday, June 26, 2012

A brief note on dyanamic vs. static analysis

I'm reading some of the literature on Hayek's discussion of Ricardo effects to help write a much weaker (although hopefully completely adequate) couple paragraphs on the subject myself. It's tougher reading than, say, the Cantillon effects literature. I find it hard to follow verbal renditions of dynamic processes. But that's mostly what this consists of.

Anyway, one of the issue that gets raised with Ricardo effects is the difference between static and dynamic analysis. There are claims that Hayek was doing dynamic analysis and nobody understood him because they were all doing static analysis, and there are also claims that Hayek was doing naive static analysis and just calling it "dynamic" (this is where the verbal renditions can obscure what the model is really communicating). But in all of these discussions, there's always this underlying insinuation that static analyses are weak and dynamic analyses are strong.

I don't really like this assumption. Maybe in specific fields or in answering specific questions a general aversion to statics is wise, but I don't like it as a general statement.

Think about engineering a bridge. If your primary concern is how much of a load it can carry, that's really a statics question. If your primary concern is how it can deal with shocks from weather or vibrations from the traffic, that's a dynamics question. In most cases, you're probably concerned with the statics question. Of course that doesn't mean the dynamics aren't important:

In other situations, the dynamics take precedence.

Just as in engineering, in economics the question of which we should prefer probably depends on the problem at hand, and we're inevitably going to have arguments about that. But it doesn't make sense to consider one approach inherently more naive than the other.


  1. I do not think you are right here. The static approach is necessarily less robust.
    The results of a static model are going to be the same as a dynamic one only if the modelled process converges to some stable state. Therefore, by omitting dynamic interactions there could be mistaken things to say about economy, which - in real world - exists in historical time.

    1. Well maybe I'm explaining myself poorly because I certainly agree with the last sentence.

      Yes, if you have a dynamic model which generates the same static model in the steady state, the only reason not to go with the dynamic model is parsimony, I agree.

      But if you are modeling two different processes the fact that one model presents a dynamic picture and one a static picture does not inherently give the dynamic model an advantage. It all depends on what is most important to talk about. The nature of the process you're modeling is the very first thing to consider.

    2. The problem with much dynamic analysis is that people assume that it subsumes static analysis as a special case, but they don't check. They also tend to assume only one dynamic effect is taking place.


All anonymous comments will be deleted. Consistent pseudonyms are fine.