Thursday, August 7, 2014

Lets Matching liquidity and rates, and discover if the Fed is ripping off the taxpayer



I compute the real curve by matching rates and liquidity under two assumptions. One, the real rate is the Treasruy rate of one year, and two the fake rates are set by the interest on overnight reserves.
 

Constructing the real yield curve:

Assume we do not have the interest on excess reserves. Looking at the two year bill at about .5%, compounded at five years one gets the ten year rate about 2.5%.  A good match.  Look at the one year bill at .09, compound that to two years and get about .18, no match for a two year bill. That is because the one year bill seems to be trading much more frequently than its term indicates, it is acting like 1/3 year cash, so .09 compounded six time comes to about the two year rate.

But wait!
How do I know these are real numbers since the interest earned on excess reserves is set by fiat, at .25?  I don't, so lets make the other assumption. The real yearly rate is the one year bill at .09%, then that makes the two to ten year rates way to high by half, they are set to the IOER. Which is it? It is hard to tell because compounding from the overnight rate to the yearly rate is fraught with potential error. But I am still guess that the two and ten year are too high by half.  The government is likely overpaying on  its debt.

That means the Fed has deliberately inverted the curve for the past six years and the Fed owes the government another $700 billion os so..


 My method:

This is my little short cut for getting the -pLog(p) right and finding the Shannon equivalent curve.  That Shannon equivalent curve gives you the actual terms that make the bond network trades occur with a minimum  of redundant trades. The simple approximation: -pLog(p) = (1/n)*[(1+i)^n], these should all be close for the set of N terms and less than one. The approximation is I assume an infinite sequence and just use frequency rather compute number of occurrences in a finite set of, say 20 years. I use n as a relative change, say from 2 to 10, mostly.  This is real back of the envelope, and more common sense then exact math.

Minimal redundancy is normal, and should be the method used naturally in all the industrial sectors. The Banking fractional reserve ratios work when they are matched to the inventory cycles of the collateral against which lending is made.
In general, the economy always tends to create the Shannon ccondition across all sectors, but inefficiencies abound and the result is localized pricing bubbles the cruise the economy. Our economy has a pricing bubble that takes the same path and returns back to the same place, synchronized to the eight year DC recession cycle. 1) Business expansion, then 2) housing bubble, then 3) stock market bubble, then 4) loaning to DC to cover its expenses, then 5) recession; wash, rinse and repeat. Time is actually a fake variable, the real variable is the number of things in the queue that can be processed before DC returns to the borrowing market.

No comments: