Can the Federal Reserve Stop Being the World’s Biggest Savings & Loan?

Published in Housing Finance International Journal. Also published in HCWE.

The Federal Reserve’s huge monetization of residential real estate mortgages is one of the radical developments in the history of American housing finance, central banking, money creation, and credit inflation.

Through this investment, the Federal Reserve has become a new element in the massive interventions of the U.S. government in housing finance. Through Fannie Mae, Freddie Mac and Ginnie Mae, the U.S. government guarantees about $8 trillion of mortgage debt, which represents about 70% of the country’s total mortgage loans. In addition, we now find that mortgage funding, cheap mortgage interest rates, and inflated house prices have become dependent on the Federal Reserve. Getting mortgages on the central bank’s balance sheet was a fateful step.

To summarize the story: Into the American financial system strode the Federal Reserve, its pockets filled with newly printed dollars and its printer handy to create more. It bought up the biggest pile of mortgage assets than anybody ever owned. It bought even more Treasury securities, too, accumulating $5.8 trillion of them, so that its balance sheet has reached the memorable total of $8.9 trillion. This is a development not imagined by anyone beforehand, including the Fed itself.

As of March 2022, the Fed owns $2.7 trillion of mortgage securities – in other words, about 23% of all the mortgages in the country are on the central bank’s balance sheet. Moreover, mortgages have become 30% of the Fed’s own inflated total assets. These are truly remarkable numbers, which would certainly have astonished the founders of the Federal Reserve System. I enjoy imagining the architects of the Fed in financial Valhalla, in a lively but puzzled discussion of how their creation of 1913 turned into a giant mortgage funder. Nothing could have been further from their intentions.

Since March 2021, one year ago, the Fed has added $557 billion to its mortgage portfolio, increasing the balance by $46 billion a month on average. It had to buy a lot more than that in order first to replace the repayments and prepayments of the underlying mortgages and then increase the outstanding holdings. The Fed has been the reliable Big Bid in the market, buying mortgages with one hand and printing money to make the purchases with the other.

During this time, the giant U.S. housing sector, representing about $38 trillion in current market value of houses, has experienced an amazing price inflation. U.S. house prices shot up 17% in 2021, as measured by the median sales price, and by 18.8%, as measured by the Case-Shiller national house price index. In January 2022, average house prices rose at the rate of 17%. House prices are far over the 2006 peak which was reached during the infamous Housing Bubble (which was followed by six years of falling prices).

Faced with the rapid escalation in house prices, the Fed unbelievably kept on stimulating the housing market by buying ever more mortgage securities. It thus further inflated the asset price bubble, subsidizing mortgages and distorting house and land prices.

By acquiring $2.7 trillion in mortgages on its balance sheet, the Federal Reserve has made itself far and away the biggest savings & loan institution in the world. Like the savings & loans of old, the Fed owns very long-term fixed-rate assets, and it neither marks its investments to market in its financial statements nor hedges it’s extremely large interest rate risk.

Like the saving & loans of the 1970s, it has loaded up on 15 and 30-year fixed rate mortgages, just in time to experience a period of threateningly high inflation.

Will the Federal Reserve withdraw from being a giant savings & loan? On March 16, 2022, the Fed announced that it “expects to begin reducing its holdings of Treasury securities and agency debt securities and mortgage-backed securities.” This move is very late. The specific decision may be made, the Fed said, “at a coming meeting.” For now, it will keep buying to replace the runoff of the current portfolio. However, shrinkage of the investments in the future seems intended and may be “faster than last time.” How far will such shrinkage, especially of the massive mortgage portfolio, go?

The Fed’s $2.7 trillion in mortgage securities is double the level of March 2020, but more to the point, infinitely greater than the zero of 2006.

From 1913 to 2006, the amount of mortgages the Federal Reserve owned was always zero. That defined “normal.” What is normal now? Should the Fed’s mortgage portfolio be permanent or temporary? Can the mortgage market now even imagine a Fed which owns zero mortgages? Can the Fed itself imagine that?

Should the Federal Reserve’s mortgage portfolio not just shrink some, but go back to zero?

In my opinion, it should indeed go back to zero.

In the beginning of its mortgage buying, this was clearly the Federal Reserve’s intent. As Chairman Bernanke testified to Congress about this bond and mortgage buying program (or “QE”) in 2011:

“What we are doing here is a temporary measure which will be reversed so that at the end of the process the money supply will be normalized, the amount of the Fed’s balance sheet will be normalized, and there will be no permanent increase, either in the money outstanding, or in the Fed’s balance sheet.”

Obviously, that was bad forecast, but that does not mean the intentions were not sincere at the time. “I genuinely believed that it was a temporary program and that our balance sheet would go back [to normal]” said Elizabeth Duke, a Fed Governor from 2008 to 2013.

The recent book, Lords of Easy Money, relates that:

“As far back as 2010, the Fed was debating how to normalize. Credible arguments were made that the process would be completed by 2015, meaning that the Fed would have sold off the assets it purchased through quantitative easing.”

A formal presentation to the Federal Open Market Committee in 2012 projected that the Fed’s investments “would expand quickly during 2013 and then level off at around $3.5 trillion after the Fed was done buying bonds. After that, the balance sheet would start to shrink, gradually, as the Fed sold off all the assets it bought, falling to under $2 trillion by 2019.” Obviously, that didn’t happen.

Early on, the Fed realized that if it did sell assets, and interest rates had normalized to higher levels, and the prices of bonds and mortgages had therefore fallen, it could be realizing significant losses on its sales. Of course, the Fed has made no sales as yet.

The yield of the 10-year U.S. Treasury note touched 2.38% on March 22 as I write, and the 30-year fixed rate mortgage rate reached 4.62%. These rates are high compared to recent experience, but they are still very low, historically speaking, especially compared to the current inflation. Historically, more typical rates would be 4% or more for the 10-year Treasury, and 5% to 6% for mortgages, or more. The interest rate on 30-year mortgage loans was never less than 5% from the mid1960s to 2008.

If the Fed stops suppressing mortgage interest rates and stops being the Big Bid for mortgages, how much higher could those interest rates go from their present, still historically very low level? When the mortgage interest rates rise, how quickly will the runaway house price inflation end?

This leads to an interesting question about the Fed itself: As interest rates rise, how big will the losses on the Fed’s vast mortgage portfolio, and on its even vaster portfolio of long-term Treasury bonds, be? Could the losses be big enough to make the Fed insolvent on a mark-to-market basis?

The duration of the Fed’s $2.7 trillion mortgage portfolio is estimated at about 5 years. That means that for each 1% that mortgage interest rates rise, the portfolio loses 5% of its market value. So a 1% rise would be a loss in market value of about $135 billion, and a 2% rise in mortgage rates, a loss of value of $270 billion.

With the $5.8 trillion of Treasury securities on the Fed’s balance sheet, it owns about 24% of all Treasury debt in the hands of the public and a lot of very long bonds. I tried to estimate the duration of the portfolio and came up with about 5 years. Then on a 1% increase in rates, the market value loss to the Fed will be $285 billion and on a 2% rise, $570 billion.

Add the mortgage results and the Treasury portfolio results together and you get these market value losses in round numbers: for a 1% rise in rates, over $400 billion; for a 2% rise, over $800 billion.

Now compare that to the net worth of the consolidated Federal Reserve System, which is $41 billion. A capital of $41 billion is subject to a potential market value loss of $400 billion or $800 billion. It is easy to imagine that the Fed may become insolvent on a mark-to-market basis, just as the savings & loans were in the 1980s.

But does mark to market insolvency matter if you are a fiat currency central bank? Most economists say No. If the Fed were mark-to-market insolvent, probably nothing would actually happen. You would still keep accepting its notes in payments. The Fed would keep accounting for its securities at par value plus unamortized premium, and the unrealized loss, however massive, would not touch the balance sheet or the capital account.

But suppose the Fed actually does sell mortgages and bonds into a rising rate market, and takes realized, cash losses. Suppose they bought a mortgage security for the price of 103, and sell it for 98, for a realized loss of 5, and that 5 is gone forever. Wouldn’t that loss have to hit the capital account, and if such losses were big enough, force the Fed’s balance sheet to report a negative capital – that is, technical insolvency?

It may surprise you to learn, as it surprised me to learn, that the answer is that this realized, cash loss would not affect the Fed’s reported capital at all. No matter how big the realized losses were, the Fed’s reported capital would be exactly the same as before. That ought to be impossible, so how is it possible?

It is because in 2011, while considering how it might one day sell the mortgages and bonds it was accumulating, the Fed logically realized that if interest rates returned to more normal levels, it would be selling at a loss. What to do? It decided to change its accounting. The Federal Reserve has the advantage of setting its own accounting standards. It decided to account for any realized losses on the sale of securities not as a reduction in retained earnings and capital but as an intangible asset! This was clever, perhaps, but hardly upright accounting. I would certainly hate to be the CFO of an SEC-regulated corporation who tried that one.

But here is a great irony: This is precisely what the old savings & loans, facing insolvency, did in the early 1980s. So the new, biggest savings and loan in the world, potentially faced with the same problem as the old ones, came up with the same idea.

Will the Federal Reserve really get out of its “temporary” mortgage program now? Will it ever get back to the formerly normal zero? And if it ever sells mortgage securities, what losses will it realize?