Monday, 29 September 2008

Procedure for generating macromodels from microfoundations

Macroeconomic models today are often generated from microfoundations, whereby an economy's behaviour is constructed from assumptions about the behaviour of individual consumers, companies, and countries. The approach tends to allow for a fuller inclusion of Keynesian and classical behaviour in the model than approaches which start from macroeconomic behavioural assumptions.

The modelling also is usually a bit more involved and so the published models are fairly simple with assumptions like identical consumers, one company, or two countries. I've been trying to improve my own humble modelling, and think that the following procedure is quite helpful for building models without immediately being overwhelmed by the simultaneity and interactions between the variables. It has its own simplifications built in, but gets the modeller going down the path.

The steps start by specifying a utility function for the demand side of one of the fundamental goods or services in the economy. The utility function should describe how the good and its trade-off payment interact - for example the trade-off between a good and wealth, or between production and labour, or between income and leisure. Simple linear utility functions like U(good) + U(wealth-payment) tend to be analytically amenable. The function is solved to get an equation connecting the two elements. A supply equation is then specified, which may be monopolistic for simplicity so that the supplier can choose the point on the demand curve, or may be competitive to get an upward slanting supply curve. The two equations for supply and demand may be solved explicitly to get the amount of the good purchased or solved implicitly or graphically. This is a solved model for the good/trade-off pairing.

Either the supply or demand curves may then be expanded to include another good/trade-off pairing, and the new model solved in the same way. The model can keep on expanding as required.

I have only applied this approach to a basic model (one good, one company, one consumer, one bank) to get the money and good demand in an inflationary environment, but it is really easy to use compared to everything else I have tried and looks vaguely like the models of the leading figures in the field. I haven't tried closing the model with simultaneous equations using labour demand.

A real scoop would be a full growth model built from microfoundations, as most major microfounded macromodels do not seem to be aimed at analysis of economic growth. I am not sure if it would lead to many more insights than basic growth models, however.

No comments: