Mathematically ridiculous; the economic DSGE model

Economists, some of them, use something called a Dynamic Stochastic Equilibrium model to predict the activity of multitudes of households and firms.  That model relies on something called Bayesian conditional inference. Thus, a household may decide to buy a car if house prices are stable in the neighborhood.  A business might expand its operation if the town is growing by 5%. The idea is that each choice is conditional upon some event in the economy.  The model is correct if the number of events possible in the economy, arranged in their possible sequences, form a Bell shaped curve.  But you need enough of a set of disorderly events in your economy to make this curve smooth, otherwise you calculations are for shit and worthless.  The curve must be smooth. How many sequences of vents to we have?

Bell Shaped Curve

That means, some sequence of events will probably happen as indicated by its coordinate on theX axis, at the left. The probability of that action being the coordinate on the Y axis.

Unemployment Rate

Take a look at the unemployment rate for the last 60 years. Ask yourself how many motions are there in that curve? Well, starting with 1980, I see basically one major motion, up and down about every eight years. Then I see a mid course correction every so often that lasts a year. That mid course correction contains about three jiggles, more or less.  So in total, this unemployment rate demonstrates the households and firms engaging in about 5 or 6 choices that are conditions upon the economy at large.  Every other motion they engage in is basically business as usual, about 90% of the time.

Those few degrees of freedom is not a Bell curve, folks, those DSGE equations are not going to work.  This is an economy dominated by four large agents, jerking everyone else around.

Better Models:

The correct model seems to be Public Choice, though I cannot be sure.  Young economists should look carefully at what I am saying, then go look at what other models do, Public Choice, and queuing models, mainly. This model is definitely the third choice. I suspect the best model whould be a queuing model with two or three lines of firms or households getting goodies from a few agents.

Can Stochastic Equilibrium be fixed?

Sure, make a stochastic version of the Hyperbolic differential equation. Get some help from the mathematicians, and the physicists will share the cost. Where the physicists have quarks, the economists will have the California, Texas and Florida economies.  Where the physicists have electrons, the economists will have medium sized firms across the USA. You are going to estimate the differential coefficients for these large movers. Once you get a few principals components modelled, then recursively gang them up into complex graphs.