The Federal Reserve loses billions on mortgages
Published in Housing Finance International.
Can house prices achieve a soft landing?
The largest holder by far of residential mortgages in the United States, a country often supposed to favor private markets, is none other than the central bank, the Federal Reserve. The Fed owns the remarkable sum of $2.5 trillion (that’s “trillion” with a T) in mortgage-backed securities (MBS).[1] That is more than one-quarter of the entire federal agency residential MBS market, the dominant U.S. source of housing finance. The Fed has mainly invested this enormous amount in the typical American 30-year fixed rate, freely prepayable, mortgage, an instrument of notorious interest rate risk. This risk infamously sank the savings and loan industry in the 1980s, and in 2023 was a principal factor in sinking Silicon Valley Bank, the second largest bank failure in U.S. history at the time of its collapse.
In this Fed MBS investment program, the entire risk of the mortgage is absorbed by the government: the credit risk of the Fed’s MBS investments is taken by other government organizations (Fannie Mae, Freddie Mac, Ginnie Mae), but the Fed itself bears 100% of the interest rate risk. How has the Fed managed this risk as interest rates have dramatically risen since 2022? Not well.
Faced with the runaway inflation of 2021- 22, the Fed pushed short-term interest rates rapidly up from close to zero to 5 ½% after it belatedly stopped buying MBS. Interest rates on 30-year mortgage loans have more than doubled from less than 3% to more than 7%. The resulting mark-to market on the Fed’s MBS portfolio is a loss of $394 billion as of June 2023, or more than 9 times the Fed’s total capital of $42 billion. Striking numbers![2]
The Fed’s huge purchases of mortgage securities began as an emergency, “temporary” action in the crisis of 2008, accelerated again in the Covid financial crisis of 2020, and lasted longer than needed—until March 2022. These central bank purchases drove the interest rates on 30-year fixed rate mortgage loans down to under 3% in late 2020 and 2021-- as low as 2.7% in late 2020—not much return for a 30-year risk! —and the yields on MBS can run 1% or so lower than the mortgage loan rates. The Fed bought a lot of MBS at the bottom in MBS yields—that is, it bought a lot at the market top in MBS prices which its own buying created.
In a leveraged balance sheet like the Fed’s, mark to market losses move into operating results through negative interest spreads. The Fed now has a $2.5 trillion MBS portfolio yielding a little over 2%, which to finance, it pays its depositors and repurchase agreement lenders over 5%. So this massive investment is now running at an interest rate spread of about negative 3%.
One hardly needs to say that lending at 2% and borrowing at 5% is a losing proposition. The $2.5 trillion MBS investment at a negative 3% spread is costing the Fed (and the Treasury and the taxpayers) about $75 billion a year. When added together with the rest of the Fed’s operations, principally the profits from issuing currency and losses from investing in long-term Treasury securities, the Fed accumulated from September 2022 through the end of August 2023 net losses of the notable sum of $95 billion.[3]
The losses will clearly continue for the rest of 2023, by which time the Fed will have lost on operations about three times its capital, and on into the future if interest rates stay at their current historically normal levels. Since the Fed’s profits mostly go to the Treasury, its losses mean that the U.S. Treasury and the taxpayers also lose. The many banks which also bought MBS and hold fixed-rate mortgage loans acquired at 3%, also have very large mark-to-market losses and negative spreads on them. These banks, like the failed Silicon Valley Bank, could correctly claim that they were only doing the same thing the Federal Reserve was.
From the point of view of mortgage borrowers, those who borrowed at 3% or less for 30 years obviously got a great and very attractive deal. A current borrower at 7.2% has to pay 2.5 times as much in monthly interest to buy the same house at the same price as a lucky 2.7% borrower did. On a house in the median price range costing $400,000, that is $1,500 more a month. Naturally, this much more expensive mortgage financing puts a downward pressure on house prices, but it also gives former borrowers a strong motivation to stay in their current houses and keep them off the market, in order to preserve the large financial advantage of the great 30-year mortgage rate the Fed created.
The astronomical house price increases of the second great American house price bubble of the 21st century are over. National average house prices, according to the S&P CoreLogic Case-Shiller Index, fell 1.2% yearover-year in June 2023. Over the same period, U.S. consumer price inflation was 3%, so in inflation-adjusted (real) terms, house prices fell 4.2%. That is a far cry from the bubble’s annualized increases in nominal terms which reached 18% and more in 2021 and early 2022. However, U.S. house prices have not fallen as much as many, including me, expected.
Serious reductions in house prices have indeed occurred in overpriced cities in the U.S. West. These year-over-year decreases reported by Case-Shiller as of June 2023 include the following. In real terms, the drops are in double digits.
House Price Changes Year-Over-Year June 2023
| Nominal | Real
San Francisco | -9.7% | -12.7%
Seattle | -8.8% | -11.8%
Las Vegas | -8.2% | -11.2%
Phoenix | -7.5% | -10.5%
These drops are, however, from very high peaks, and prices in these areas remain historically high.
The two metropolitan areas with the largest price increases in the June Case-Shiller report are the Midwestern cities of Chicago, at +4.2%, and Cleveland, at +4.1%, being in real terms, 1.2% and 1.1%, respectively, which are similar to the long-term growth rate in real house prices.
The AEI Housing Center finds that in July 2023, national average house prices increased year-over year by 3.5%, or by +0.3% real, after falling year-over-year in real terms for eight months in a row. Can falling real prices becoming more or less flat make for a “soft landing” of U.S. average house prices? Maybe--considering that 7% is in a historically normal range of mortgage interest rates. It only seems so high to us because of the previous suppression of mortgage rates to abnormally low levels by the Fed.
Still, many observers wonder why U.S. house prices have not fallen more, given the large reduction in buying power for most people from the sharply increased cost of mortgage loans.
Two measures which have dropped like a rock are the volume of home sales and of mortgage originations. Sales of existing houses were down 40% in July 2023 compared to the pre-pandemic July 2019[4] . The volume of interest rate commitments for new mortgage loans is down about 30%- 40% from 2019, and the volume of mortgage refinancings (“refis” in U.S. parlance) simply to lower the interest rate, is correspondingly down 95%.[5] The mortgage banking and brokerage businesses are experiencing heavy stress and their own painful recession with the disappearance of business volumes. Their total staffs have been cut by 43,000 employees or about 11% over the last year.[6]
How can house prices stay high while sales volume shrivels?
One factor is that while sales of existing houses are way down, so is the inventory of houses being put up for sale. The July 2023 inventory of houses for sale was down 36% from July 2019, which helps support house prices.[7]
A popular and plausible theory is that this low inventory at least in part reflects the desire of the lucky mortgage borrowers at 3% or less to hang on to their great mortgages by simply staying in their current houses. As a typical article explains, “Borrowing costs have not had the expected effect of cooling house prices [because] most US homeowners…have in effect been trapped in their properties by [the] low rates.” So the Fed’s manipulation of mortgage interest rates to exceptionally low levels created an unexpected factor to later restrict the supply of houses for sale and help hold house prices up. In addition, when working from home, the borrowers don’t have to change houses to change jobs, which supports the ability to hang on to the old mortgage.
These factors help slow prepayments of 30-year mortgage loans, extend the duration of the Fed’s MBS portfolio, and increase the Fed’s mark-to-market and operating losses.
It doesn’t help right now, but as a long-term strategy the Federal Reserve needs to return the amount of MBS in its balance sheet to what it always was from its founding in 1913 to 2008--namely zero.
_________
1 Federal Reserve H.4.1 Release, August 31, 2023, Section 5.
2 Federal Reserve Banks Combined Quarterly Financial Report, June 30, 2023.
3 Federal Reserve H.4.1 Release, August 31, 2023, Section 6.
4 AEI Housing Center, Home Price Appreciation Index report, July 2023.
5 AEI Housing Center, Housing Market Indicators report, August 1, 2023.
6 “Nonbank mortgage jobs undergo a seasonal dip,” National Mortgage News, September 1, 2023.
7 AEI Housing Center, Home Price Appreciation Index report, July 2023.
8 “Investors keep close eye on US mortgage rates,” Financial Times, September 3, 2023.