On September 16 the Federal Reserve raised interest rates for the first time since July 2023. The vote was 12–0, the new range is 3.75 to 4 percent, and the statement ended with a sentence central banks seldom put so flatly: “The Committee will deliver price stability.” The minutes, published three weeks later, added that most officials thought another increase would likely be appropriate by the end of the year. Gold fell 8.5 percent in September. For anyone who learned the standard rule, that gold pays no interest and so suffers when interest rates rise, the month looked like the rule at work.
The rule is half right, and the half it gets wrong is expensive. The Fed has completed three rate-hiking cycles this century. Gold rose through two of them and finished the third slightly higher than it began. What separated the good cycles from the bad one was not the number of increases or how high rates went. It was a different interest rate, one the Fed does not set. That rate is rising sharply now, which makes the history worth knowing in some detail.
Three cycles, three results
2004 to 2006. Starting in June 2004, the Fed raised rates at seventeen consecutive meetings, from 1 percent to 5.25 percent. Gold was about $393 an ounce on the day of the first increase and about $587 on the day of the last, a gain of 49 percent. Along the way it never fell more than a couple of percent below where it started.
2015 to 2018. The Fed raised rates nine times in three years, from near zero to a range of 2.25 to 2.5 percent. Gold had been falling for four years before the first increase, partly in anticipation of tighter policy. It hit its low, about $1,051, the day after that first hike in December 2015. By the last one it was 16 percent higher.
2022 to 2023. This was the fastest tightening in four decades, with eleven increases in sixteen months taking rates from near zero to a range of 5.25 to 5.5 percent. Gold fell 15 percent in under eight months, to about $1,631 in November 2022. Then it recovered all of that and ended the cycle 3 percent above where it began.
Key Data
2004–06: 17 hikes, 1.00% to 5.25%. Gold up 49% ($393 to $587). 10-year real yield up 0.5 points (2.10% to 2.61%).
2015–18: 9 hikes, 0–0.25% to 2.25–2.50%. Gold up 16% ($1,078 to $1,252). Real yield up 0.1 points (0.84% to 0.97%).
2022–23: 11 hikes, 0–0.25% to 5.25–5.50%. Gold up 3% ($1,909 to $1,970) after falling 15% to its low. Real yield up 2.1 points (−0.61% to 1.50%).
2026 so far: 1 hike, to 3.75–4.00%. Gold down 4.4% for the year to the end of September. Real yield up 1.0 point since January (1.94% to 2.93%).
Gold: COMEX futures closing prices on the days of the first and last hike; the 2026 figure is the World Gold Council’s. Real yield: the US Treasury’s 10-year real yield curve rate.
The rate that mattered
The federal funds rate is the overnight rate the Fed sets. The rate that matters most to gold is the real yield on long-term government bonds, meaning what a ten-year Treasury pays after inflation. The market prices it directly through inflation-protected bonds. It measures the return an investor gives up by owning a metal that pays nothing, over the kind of horizon on which people actually hold gold.
The two rates do not have to move together, and in the first two cycles they did not. Between 2004 and 2006 the Fed lifted its policy rate by more than four percentage points, and the ten-year real yield rose by half a point. Alan Greenspan, the Fed chairman at the time, called the failure of long-term rates to follow short-term ones a “conundrum.” From 2015 to 2018 the real yield rose by about a tenth of a point over the whole cycle. In both cases the cost of holding gold barely changed while the Fed was busy, which left gold free to respond to everything else. In the first cycle that was a broad commodity boom, and in the second a deeply depressed starting price.
The 2022 cycle was different. When the Fed began, the ten-year real yield was below zero, at minus 0.6 percent. Less than eight months later it was 1.7 percent, a swing of more than two points, and gold fell 15 percent over the same stretch. That is the textbook rule working as advertised. The textbook would not have predicted the recovery. Real yields stayed high through 2023 and gold still returned to its starting level, because central banks were buying more than a thousand tonnes a year, by the World Gold Council’s estimates, for reasons that had nothing to do with yield. We examined that break in the old relationship in an earlier piece on real yields.
The cycle everyone remembers
There is one episode in which rate hikes did crush gold, and it still shapes how people think about the subject. In the late 1970s the Fed raised rates repeatedly but never by as much as inflation was rising, so the real cost of money stayed low or negative. Gold climbed from under $140 an ounce at the start of 1977 to $850 in January 1980.
Then Paul Volcker’s Fed pushed its policy rate to about 20 percent while inflation fell from nearly 15 percent to under 4. Real interest rates went from negative to the highest levels of the post-war era. By June 1982 gold was below $300, a fall of about 65 percent. An investor who bought at that peak waited some 27 years to break even in dollars, a story we tell in Gold’s First Real All-Time High Since 1980.
Set beside the three recent cycles, the Volcker years make the pattern clear.
Where 2026 sits
On the Fed’s own rate, this cycle has barely begun. The policy range of 3.75 to 4 percent sits only a little above consumer-price inflation, which was 3.4 percent in August. The September minutes record that several officials regard the current rate as “not restrictive or only mildly restrictive.” Measured that way, the Fed is still closer to trailing inflation than to leading it.
The long-term real yield tells a different story. The ten-year real yield began the year at 1.94 percent. By the end of September it was 2.93 percent, with half a point of that rise coming in September alone. On October 5 it reached 2.95 percent, the highest since November 2008. The ordinary ten-year yield reached 5.31 percent the same day, its highest since 2002. On the measure that has mattered to gold, 2026 so far looks like 2022 and not like 2004. Gold’s price agrees. It was down 4.4 percent for the year at the end of September and roughly a quarter below January’s record.
Two things are different from 2022, and they pull in opposite directions.
The first is the reason long-term yields are rising. In 2022 they rose because the Fed was catching up with inflation at great speed. This time the Fed’s own minutes, describing the weeks before the September meeting, attribute part of the rise to other causes, including “uncertainty related to the U.S. Treasury’s announcement and implementation of the buyback program, and competition for capital from heavy private debt issuance to finance the development of artificial intelligence (AI) infrastructure.”
Early October supplied a test of that reading. A weak jobs report on October 2 cut the market-implied odds of a second hike this month from about three in four in late September to under one in five. The two-year yield, which tracks expectations for the Fed, slipped from its late-September peak. Long-term yields kept rising, and the thirty-year reached 5.67 percent on October 7, its highest in more than two decades.
A real yield that is high because the central bank has beaten inflation is a clean headwind for gold. A real yield that is high because lenders want more compensation for holding long-dated government debt is harder to read, because the same worry sends some of those lenders to gold. That was the story of August, which we set out in Gold and the Shadow of Fiscal Dominance.
The second difference is who is buying. In 2022 gold funds were losing metal. In the third quarter of 2026 they took in a record $31 billion, including $10 billion in September while the price was falling, and China’s central bank made its largest purchase in about three years. We look at that split between sellers of futures and buyers of metal in Who Sold Gold in September?
Three questions to ask instead
None of this says what happens next, and the Fed has made a point of not telling anyone. Kevin Warsh again declined to submit a rate projection of his own in September, and we have written about his decision to drop forward guidance. September’s consumer-price figures are due on October 14 and the next decision on October 28. Whatever follows, history suggests three better questions than “did they hike?”
Is the Fed getting ahead of inflation or following it? A policy rate that rises more slowly than prices is the pattern of the 1970s and of 2004. One that rises well above them is 1981.
What are long-term real yields doing? The Treasury publishes the ten-year real yield every business day. It has explained more of gold’s behavior in hiking cycles than the Fed’s own rate has.
Who is on the other side? In 2022 and 2023, official buying was large enough to offset a two-point rise in real yields. Whether fund investors and central banks keep absorbing what rate-sensitive traders sell is the open question of this autumn.
For an individual owner, the record supports a modest set of conclusions. A rate hike is not, by itself, a reason to sell gold or to buy it. A hiking cycle can still hurt badly along the way. The 15 percent drawdown of 2022 happened inside a cycle that gold finished in profit, and anyone holding a position too large to sit through it sold near the low.
The one clear case in which tightening broke gold for a generation required a central bank willing to hold real interest rates far above inflation for years. The Fed says it will deliver price stability. If it does so in that manner, the 1980s are the precedent. If it gets there slowly, or if bond investors decide the larger problem is the government’s debt, the 2000s are. You can see what gold did in each of those decades, in real terms, in our gold returns calculator, and how rates feed through to the price in our guide to interest rates and gold.
