A Comment On Mathematical Models In Economics

This article is a followup comment on a good article by Nick Edmonds,  "On The Role Of Models", in which he discusses the use of mathematical models in economics. I agree with what he has to say, but I just want to add a comment on how I approach this topic.

I would first note that I am fairly skeptical about mainstream economic models, in particular DSGE models. This is not just an aversion to the mathematics, or the lack of realism of the underlying assumptions. (For example, Steve Keen is a well known critic of the realism of the underlying assumptions of "neoclassical" economics.)  Instead, even if we take as given the model assumptions, there are still difficulties with how the models are constructed and interpreted. This is a large subject, which will take time for me to cover, but I give as examples of such analysis:

• The problems associated with applying the logic of single-good economic model to real economies.
• The pursuit of Quantitative Easing (QE) by central banks, even though mainstream modelling assumptions should imply that QE is ineffective.
• Nick Rowe argues that the standard New Keynesian models assume full employment in this article on Worthwhile Canadian initiative. (Rowe: "They have been teaching their students to just assume the economy eventually approaches full employment, even though there is absolutely nothing in the model to say it should.")

Nick Edmonds follows in the tradition of Wynne Godley in using Stock-Flow Consistent (SFC) models as teaching models ("didactic models").  I have not yet had a chance to unveil my own SFC models (under development), but the idea is that we can learn about the potential dynamics of an economy, even if we do not attempt to fit the model to actual economic data. In other words, we can use it to learn about what is possible, but not to make forecasts for upcoming quarters.

I think this is useful, but it is clear that we have to compare the model dynamics to real world data. As an example, take the hyperinflation model of Vincent Cate. (If you follow the link, the model dynamics are available online.) I have only looked at the model quickly, but it appears to me that it would be possible to tweak the model so that hyperinflations never occur.

How to decide which modelling assumption is correct? If these are teaching models, what exactly are we teaching? We need to compare the implied dynamics of the model to real world data. In the case of the hyperinflation model, I expect that the problems will show up in the assumed model for the velocity of money, and the following frictions to the idealized behaviour:

• Wages in modern economies are not indexed to the price level at a high frequency. Wage contracts are typically fixed one year at a time. This limits the possibility of consumption to match an assumed rise in the CPI level.
• Taxes are imposed on nominal wages, and tax rates rise as wages increase. This will create an increasing fiscal drag.

These factors should show up in the historical data, and thus it will be possible to validate whether the teaching model is a reasonable approximation to reality. But the key is that even if we do not attempt to fit the model parameters to real data, we still need to check whether the dynamics make sense.

(c) Brian Romanchuk 2013