Rule of thumb on coefficient error

Check out money velocity, it tells you how many data points are available.  Cancel out time and determine how many points of history you get per coefficient calculation.

Here we have M2 velocity.  From 1990 to 2010, lots of points.  Then after the crash, we have 40% fewer points.  Looking short term at this series, it recently dropped to new lows.

The Fed, the BEA and BLS are blind.  Just consider the sudden appearance of the Chinese home buyer, took us by a bit of surprise.  We had no stable coefficients for a sudden in rush from China, we had to do a large three quarter revision

Right now international pricing does not work, onshore and offshore yuan values differ by 8%.  Global trade has halted.

How do we fix the model?
Make the economy a statistical machine, it always attempts to improve the coefficients, reduce the error band.  Error bands are loops in the probability graph, the economy will run the loop to price something, its mandatory, Pricing is energy consuming, and the loops necessary to keep inventories from running off the rails.