http://bondtraderforum.com/phpBB3/viewtopic.php?f=2&t=49 Here is the Fed's balance sheet this week. http://www.federalreserve.gov/releases/h41/Current/ Code: Remaining maturity Within 15 16 days to 91 days to Over 1 year Over 5 years Over 10 All days 90 days 1 year to 5 years to 10 years years Mortgage-backed securities (4) Holdings 0 0 0 0 0 427,552 427,552 Weekly changes 0 0 0 0 0 - 3,928 - 3,928 4. Guaranteed by Fannie Mae, Freddie Mac, and Ginnie Mae. Current face value of the securities, which is the remaining principal balance of the underlying mortgages. Then the New York Fed notes for the week period they were a net buyer of 25B+ securities. http://www.newyorkfed.org/markets/mbs/index.html So refinances and mortgage cashflows actually more than completely offset the face value of what they are buying. There is a little paradox here, in that the lower the rates go, the refinancing demand causes the Fed to merely replace its existing portfolio with lower yielding and higher priced assets. Say if they buy mortgages at 3.5%, a large portion of that 'new buying' will merely roll over existing portfolio holdings, so for every dollar the Fed spends on new mortgage assets, they are not necessarily increasing the size of their balance sheet (in face value) by that same amount. On the other hand, there is an adequately sized secondary market to compensate for this "problem" of the Fed doling out money. If the Fed owned the entire mortgage market however (so bought all outstanding agency MBS in the many trillion), they would run into practical limitations to how much they could lend out. Every substantial move lower in rates would result in a refi round not raising the size of the Fed balance sheet, and as well would suck in all of the remaining 'qualified' buyers of homes at diminishing rates. I guess none of this is a practical problem, because you probably wouldn't see it kick in until the Fed had acquired many trillions in assets. ie, if the Fed bought 3% market rate mortgages, it would increase housing demand considerably from today. But all of this has a worthwhile talking point: The Fed was a buyer of 25B of mortgages for that weekly period yet their asset holdings (outstanding money supply created by that QE problem) decreased by almost 4B. Refis explain a bit of that, principal paydown explains the rest. It gives the Fed ammo to juice up its purchase rate, I think... to... 50B/week... 25B is simply not enough, and the proof is in the fact that not only rates rose substantially, but -fundamentally- that outstanding money supply from this agency MBS program FELL while the Fed was a net buyer of 25B. It might also give support to the fact the Fed may be onto something with the self neutralizing nature of providing these timed loans to the market. If buying 25B of mortgages a week results in a aggregate -reduction- of money supply, that is somewhat negative for the hyperinflation crowd.