CALCULATEDRISK
By Bill McBride
This is a technical note from housing economist Tom Lawler.
Some Thoughts on Quantitative Easing and Quantitative Tightening
It has now been well over a year since the Federal Reserve began its “quantitative tightening.” The overall size of the Fed’s balance sheet is down by about $1.4 trillion from its peak, with the bulk of that decline coming from 3/22/23 to 2/28/23 ($1.167 trillion). Some analysts have argued that this “QT” represents a reversal from the previous (and massive) “QE” (quantitative easing) the Fed undertook in earlier years, and as such monetary policy is significantly “tighter” than the level of the Federal funds rate. In one very important perspective, however, that is absolutely not the case.
To understand why I say this, one must understand what the “QE” policy entailed. Under QE the Federal Reserve purchased large quantities of long-term fixed rate assets (mainly Treasuries and agency MBS), “funded” by creating very short term bank reserves/deposits at the Federal Reserve. These reserves represent liabilities of the government, just as Treasuries (and for all intents and purposes agency MBS) represent liabilities of the government. In essence, then, what QE did was significantly alter the maturity profile of government liabilities to the private sector, replacing longer-term liabilities with shorter-term liabilities. One of the purposes of QE was to put downward pressure on longer term interest rates (including mortgage rates), which the Fed viewed as stimulus needed, especially when the Federal funds rate was at its “lower bound.”
A “reversal” of QE, then, would involve the Fed selling longer-term government obligations and reducing shorter-term government obligations (bank reserves/deposits). But that is not what has happened under QT. The Fed has sold virtually no agency MBS and virtually no longer-term Treasury securities. Rather, the Fed has not replaced agency MBS principal pay downs with new MBS purchases and has not fully replaced maturing Treasury securities with new Treasury securities (though the Fed has still purchased Treasury securities.) As a result, most of the decline in the Fed’s balance sheet has involved a decline in very short-term (maturing) government obligations owned by the Fed, combined with a decline in very short-term Fed obligations (bank reserves) to the private sector.
(In terms of Federal Reserve Treasury holdings, from 2/22/23 to 2/21/24 the Fed sold virtually no longer-term securities, while at the same time it increased its holdings of Treasury Notes/Bonds with a maturity of 2 years of more by $167 billion, with a weighted average maturity as of 2/21/24 of 8.09 years. It also increased its holdings of long-term (10-year and 30-year) TIPS by about 5.7 billion.)
If the Federal Reserve’s QT policy had been to “reverse” QE, it would have involved selling longer-term Treasury securities and agency MBS – thus increasing the amount of long-term government obligations held by the private sector – while at the same time reducing very short-term bank deposits/reserves at the Fed – thus reducing the amount of very short-term government obligations held by the private sector. Just as QE appeared to put downward pressure on long-term US interest rates (though estimates of by how much vary considerably), a “full reversal” QT policy would have put upward pressure on long-term US interest rates (which so far has not been the case.)
Focusing now first just on Treasuries, below is a table showing end-of-year marketable Treasuries securities outstanding and the weighted average maturity of marketable Treasury securities outstanding, as well as marketable Treasuries securities held by the Fed and the WAM of Fed Treasury securities. (Note: Fed holdings are the last Wednesday of December, and not year-end, as the data are reported weekly. Also, the totals exclude “inflation compensation” from TIPS.)
Also calculated are (1) marketable Treasury securities held by the private sector along with the WAM; and (2) marketable Treasury securities plus Fed bank/reserves/deposits held by the public “created” solely by Fed Treasury purchases. (Other factors also affected Fed reserves, but I am only focusing here on the impact of Fed Treasury purchases.)
There are a few things worth noting. First, of course, marketable Treasury debt outstanding has soared, both in absolute terms and relative to the size of the economy. Marketable TDO was over 94% of Q4 GDP at the end 2023, up from 30.7% at the end of 2007.
Second, the average maturity of TDO has increased significantly (with the exception of 2020).
Third, the net effect of Fed Treasury purchases and reserves created by Treasury purchases on Government Obligations owned by the private sector has been to reduce the average maturity of such obligations over 1.5 years, which is pretty sizable, and certainly suggests that Fed activity resulted in lower longer term interest rates than would have been the case in the absence of Fed activity.
Now let’s look at MBS. Below is a table showing Agency MBS outstanding Agency MBS held by the public, and Agency MBS held by the private sector. (Again, the Fed holdings are as of the last Wednesday in December).
The Fed’s share of total Agency MBS outstanding hit a year-end high of 32% at the end of 2021 and was still above 27% at the end of last year. Again, Fed MBS purchases essentially involved taking longer-maturity fixed-rate MBS held by the public and “replacing” them with extremely short-term bank reserves/deposits at the Fed, thus reducing NOT the amount of government obligations held by the private sector, but significantly reducing the maturity/duration of government obligations held by the private sector.
Computing the impact of Fed MBS purchases on the maturity of government obligations held by the private sector is more challenging for MBS than for Treasuries, as the expected weighted average maturity/duration of MBS are heavily dependent on where interest rates are relative to the interest rates on the mortgages backing MBS. For the end of 2023, the “expected” duration of MBS outstanding was about 5.55 years and the expected weighted average maturity was about 9 years (according to the Bloomberg MBS Index. The expected durations/WAM of Fed MBS holdings at the end of last year were longer than that for the MBS market as a whole, as the Fed’s share of very low-coupon 30-year MBS is much higher than is its share of the MBS market as a whole, and are probably around 6 years/9 years, respectively.
Using those durations/WAMS, at the end of 2023 the Fed’s holdings of agency MBS has “converted” about $9 Trillion of privately-held government obligations (including, of course, the very short term bank reserves/deposits “created” from the Fed’s MBS purchases from a duration of about 5.55 years to a duration of about 3.91 years, and a weighted average maturity of from 8 years to about 5.5 years.
Below is a table showing the impact of the weighted average maturity of privately held government obligations (only including reserves created by Fed purchases of Treasuries and Agency MBS) from the Fed’s holdings of Treasuries and Agency MBS. Also shown are projections though 2026 assuming (1) median projections of Treasury borrowings from the Treasury Borrowing Advisory Committee January report; (2) Federal Reserve Treasury holdings projections from the same report (which assumes total holdings will decline more slowly this year and not decline much in 2005 and not at all in 2006); (3) reinvestment of maturing Treasuries sufficient to hit target holdings levels at an average maturity of 3.5 years; and (4) continued runoff of Agency MBS at the same pace.
Note that the impact of Fed Treasury/MBS holdings on the average maturity of private sector general obligation holdings was to reduce the average maturity of private sector general obligations by a whopping 1.75 years. That “gap” should move down to about 1.2 years in 2026.
Under these realistic scenarios (1) the private sector will have to “absorb” a lot of Treasuries and a modest amount of MBS; (2) the average maturity of debt the private sector will have to absorb will be longer, though not by a massive amount; and (3) the likely impact on the so-called “term-premium” – the extent to which long-term Treasuries exceed expected future short-term rates – will probably be “modest” but noticeable. I’ll have more later on this subject, but the “best guess” right now is the term premium on the 10-year Treasury will probably gradually increase from today’s “guess” of about 0 to about 50 bp by the end of 2026.