The Federal Reserve spent three and a half years shrinking its balance sheet, then quietly stopped. Runoff ended on December 1, 2025. Since early January 2026 the System Open Market Account has been buying Treasury bills again, close to $250 billion of them by the start of July, and total assets stood at $6.74 trillion in the first week of September 2026, higher than a year earlier.
A central bank buying securities while gold trades in the thousands invites a confident story: the Fed prints, the metal rises. The record is messier than that, and the mess is the useful part. If you hold bullion or coins, it helps to know where the link between the balance sheet and the metal is real and where it falls apart.
What the balance sheet actually is
On the asset side the Fed holds Treasury securities and agency mortgage-backed securities, currently about $4.55 trillion and $1.91 trillion respectively. On the liability side sit the things those assets are funded with, mainly physical currency and the reserve balances that commercial banks keep at the Fed.
Quantitative easing is the Fed buying securities in the open market and paying for them by creating reserves. Quantitative tightening, or QT, is the slower reverse: the Fed lets bonds mature and declines to reinvest some of the proceeds, subject to monthly caps. Nothing is sold, and nothing is shredded. The portfolio simply runs off.
The scale changed twice in living memory. Before the 2008 crisis the balance sheet was roughly a tenth of what it later became, by the Congressional Research Service’s reckoning. Three rounds of QE lifted it to about $4.5 trillion by the end of QE3 in October 2014. A first attempt at runoff brought it down to roughly $3.8 trillion by August 2019, and then the pandemic response took it to a peak of $8.96 trillion in April 2022.
Quantitative easing and gold prices: the 2008 to 2015 record
The first half of the story fits the popular version well. Gold fell to about $765 an ounce in November 2008, then climbed for three straight years through QE1 and QE2, reaching its record in September 2011. Anyone who bought the relationship then had every reason to feel vindicated.
The second half is where it breaks. In December 2012 the Fed raised QE3 to $85 billion of purchases a month, the largest sustained pace it had ever run. Gold fell 30 percent across 2013 in dollar terms. On the London benchmark the metal lost 25.4 percent in three months, breaking below $1,535 on April 12 and through $1,400 by April 15. QE3 did not actually stop until October 2014, and gold kept sliding until it bottomed a little above $1,000 in December 2015, the month the Fed finally raised interest rates off zero.
Read that sequence slowly. The most aggressive stretch of quantitative easing in American history coincided with gold’s worst calendar year in a generation. What turned the market was not the volume of purchases but the growing expectation that they would end.
What QT did, and what gold did anyway
The second experiment was just as unkind to the simple model. From June 2022 the Fed capped reinvestments, starting at $30 billion a month in Treasuries and $17.5 billion in agency securities and doubling both after three months. By the time runoff ended, total securities holdings had fallen by more than $2.2 trillion, split roughly $1.6 trillion in Treasuries and $600 billion in mortgage-backed securities.
Over that same window gold went on a long run of record highs, moving above $5,000 an ounce in January 2026 and gaining more than 25 percent in the year to early September 2026. The balance sheet shrank by an amount larger than its entire pre-2008 size, and the metal rose anyway. It has since come off that peak, trading near $4,400 in early September, roughly a fifth below the late-January high.
Why the connection is looser than it sounds
Several things get in the way of a clean relationship between quantitative easing and gold prices.
Markets trade expectations rather than settlements. By the time the Fed’s desk is buying, the announcement has been priced for weeks. That is why 2013 hurt so much: the flows were still enormous, but the expected future flow had turned.
Reserves are also not spending money. They sit in accounts at the Fed and can only move between banks. QE changes the composition of what the private sector holds, swapping bonds for reserves, which is a different thing from putting cash in anyone’s pocket. The clearer transmission runs through yields and the dollar, which is why real interest rates and the exchange value of the dollar have tracked gold more reliably than any balance sheet line ever has.
And the Fed is only one buyer among many. Sustained official-sector demand, the kind covered in our look at central bank gold buying, has done a good deal of the work in recent years without any reference to what Washington’s own portfolio was doing. Policy rates matter too, though not always in the direction people expect, as the record on what gold does when the Fed cuts shows.
Plumbing is not stimulus
This distinction matters for reading the current headlines correctly. On December 10, 2025 the FOMC began purchasing shorter-term Treasury securities, and the Fed has been explicit about why: to maintain an ample supply of reserves, accommodating trend growth in demand for its liabilities and normal seasonal swings. Reserve balances reached about $3.1 trillion by July 1, 2026. The purchases are bills, not long bonds, and they are not framed as an easing of policy.
There is a good reason the Fed treats reserve scarcity as an operational hazard. In September 2019, after the first runoff, reserves hit roughly $1.34 trillion, their lowest level since 2012. Overnight repo rates spiked above 5 percent on September 17, well above the target range. The Fed responded with $75 billion of overnight operations and, by October 11, standing bill purchases of about $60 billion a month. That was the money-market plumbing backing up, not a stimulus program.
Whether a balance sheet that grows with currency demand and bank reserves eventually matters for the metal is a fair question, and reasonable people answer it differently. What the last eighteen years suggest is that it is a slow background variable rather than a trading signal.
What a metal holder can reasonably take from this
Treat the balance sheet as context, not as a trigger. The headline total tells you very little on its own, since the same number can reflect emergency purchases, routine reserve maintenance, or currency in circulation growing because people carry more cash. If you want to follow it, the more informative series are reserve balances and short-term funding rates, both of which say something about whether the system is tight.
It is also worth remembering how much else sits between a policy decision and what a coin is worth. Metal value moves with spot, which you can follow on our markets page, but the premium on a given piece answers to mintages, condition, and collector demand that have nothing to do with the Fed. A pre-1933 gold piece and a current American Gold Eagle can move quite differently in the same week even though the metal inside them is priced identically.
The balance sheet is one variable in a crowded field that also includes real yields, the dollar, official-sector demand, and the occasional shock covered in our piece on tail risks in the gold market. Holding several of those in mind at once is less satisfying than a single explanation, but it has the advantage of matching what actually happened. The rules of gold we keep coming back to are mostly about patience and about not mistaking a loud number for an important one.
