The CBO is already off by 20%, a mere one month after publishing

Here we have the CBO 20 year projection. Below we have the actual amount a month past the dotted line called actual. The one below has already risen about one percent faster than the CBO predicts. I compound that over the 20 years and, the CBO is 20% off, already, on its 20 year timeline. In other words, economists who use this as a predictor are simply statistical frauds.

 
Why does the CBO do this? They are required by law to assume variation is linear with Congressional elections. Congress gets elected precisely on two year intervals, with no variation in election days. So Congress counts by base one, 1,1,1,1... one number for each election.  Since Congress has no method to correct itself with aperiodicity, then Congress wants linearity in variation.  They certainly do not want the CBO intimating some symmetrical uncertainty in decisions since each Congress member has no choice but to sit out his or her term until the next fixed election day.

The central banker, on the other hand, is expected to make at least moves, up and down, over its adjustment period.  The Fed will make counter moves to offset some prior change in rates.  They have two degrees of freedom, one more than Congress, and we can expect variation in rates to be square symmetric, a Bell shaped variation.

This is the type of analysis we want mathematicians to use when modelling the economy. Find out what the degrees of freedom would be, then break the aggregate variation into a recursive polynomial based on exponents being the degrees of freedom. California and Texas governments triangulate, three degrees of freedom. The system, in total, I would think, computes to a multiple Poisson distributions, the queue sizes being the degrees of freedom.

Mark Thoma has a pointer.

Stationary processes - The Leisure of the Theory Class

The author has a paper coming up in which the subject is stationary processes when order can be exchanged.  This is where we want to go.  The entity managing a Poisson queue can exchange order among the queue members, this is an agent re-ordering the delivery schedule from inventory. It is the way in which spectrum is shaped, the method by which Jerry Brown or Apple can alter the variation to keep the degrees of freedom it has, or expand them.

The abstract from Nabil Al-Najjar, Eran Shmaya:

A Bayesian agent learns about the structure of a stationary process from observing past outcomes. We prove that his predictions about the near future become approximately those he would have made if he knew the long run empirical frequencies of the process. 

This is Stiglitz, screening. Jason Smith, information passing, Kling, PSST; the new economics. And it is the theory of counting, from the vacuum to the quasar.   We are almost here, do not give up, this is all great stuff.