Up 70% in dollars. Down 33% in gold.

• The S&P 500 rose about 70% in dollars over five years and fell about a third measured in gold. Both are true, and only one appears on a brokerage statement.
• Gold passed US Treasuries as a share of central bank reserves for the first time since 1996, at 27% against 22%. The automatic bid for US debt is thinning, and the 30-year yield is at its highest since 2007.
• Washington’s answer is to move borrowing from the long end to the short end, where the Fed sets the rate, with stablecoin reserves as the new captive buyer. Holders of long-dated paper pay for it.

A clip of JD Vance has been circulating this month in which he calls the dollar’s reserve-currency status a resource curse, and compares it to coal in Appalachia. Coal made a handful of out-of-state companies rich, left the counties it came from with no tax base, and moved on. His argument is that being the world’s money did something similar to the American industrial economy.

Worth being precise about the clip, because a lot of the commentary is not. He said it in 2023, as a senator, at an event hosted by American Moment. It resurfaced in the middle of August 2026 and turned into an argument about his 2028 positioning. It is his view, but it is not a new announcement of policy.

The idea underneath it is older than either of them. Robert Triffin set it out in 1960: the country that supplies the world’s reserve asset has to run persistent deficits to supply it, and that steadily works against its own tradeable sector. Economists have argued about the Triffin dilemma for sixty-five years and still disagree. Plenty of them think Vance has the analogy wrong, and that cheap borrowing is worth more than the manufacturing it costs.

We are not going to settle that. What interests us is narrower and more useful, and it starts with a number almost nobody looks at.

Take the five years to the end of August. The S&P 500 went from 4,523 to 7,677. That is a 70% gain, one of the better five-year runs an investor could have had, and anyone holding it felt richer at the end than at the start.

Now price the same index in gold instead of dollars. In August 2021, buying one unit of the S&P 500 cost about 2.49 ounces. Today it costs about 1.66. Measured that way, the same holding is down roughly a third.

Both statements are true. The index did rise 70%, and its holders did lose about a third of their purchasing power against the one asset central banks have been buying. The difference is entirely in the yardstick, and your brokerage statement only shows you one of them.

Two honesty notes, because this comparison gets abused. The first is that these are price-index figures, so dividends are excluded. Reinvest them and the gold-relative loss narrows, though it does not disappear. The second is that gold is not a stable ruler either. It set a record near $5,600 earlier this year and has since given back a meaningful part of that. Choosing gold as the denominator is a choice, and measured against consumer prices the picture is far less dramatic.

What makes the gold comparison worth running anyway is who is doing the buying. This is not retail sentiment. It is central banks.

For eighty years the United States had a customer for its debt that never had to be persuaded. Countries that wanted to hold reserves held dollars, and holding dollars mostly meant holding Treasuries. That is the structural bid no other issuer gets, and it is the real content of the phrase exorbitant privilege.

It is thinning. On ECB figures, gold reached 27% of global central bank reserve assets at the end of 2025, up from 20% a year earlier, while the Treasury share fell from 25% to 22%. That is the first time since 1996 that gold has outweighed Treasuries in official reserves. Among 76 central banks surveyed, 89% expect their gold holdings to rise over the next year and 74% expect lower dollar holdings over five years.

Keep the scale straight, though. Dollar-denominated assets as a whole are still the largest block at 42% of reserves, and nothing here says the dollar is finished. What it says is that the automatic part of the bid is going away, and that the buyer of last resort now has to be paid.

You can see the price of that in the long bond. The 30-year Treasury yield touched 5.31% in mid-August, its highest since 2007, with the debt approaching $40 trillion and interest costs running $857 billion in the first nine months of the fiscal year, up 13% on the year before.

This is where it stops being commentary and becomes something you can watch happen.

The Treasury cannot set long-term interest rates. Those are set at auction by pension funds, insurers, foreign central banks and hedge funds, and right now those buyers are asking for more. What the Treasury can do is stop borrowing at the long end and borrow at the short end instead, where the Federal Reserve sets the price.

That is what is underway. On 19 August the Treasury doubled its buyback programme for older long-dated securities from $2 billion to at least $4 billion, with the enlarged operations scheduled from 9 September to 4 November. Scott Bessent has been explicit about the mechanism and even gave it a name, the Treasury Twist, buying long-dated debt and paying for it with short-term issuance. He has also built the Treasury’s cash balance at the Fed to around $950 billion, against a target of $550 to $600 billion under the previous administration, and has signalled it could help fund the operation.

Retiring long debt and replacing it with bills is not free. It swaps a rate locked for thirty years for one that resets every few months. That is a deliberate trade: pay more now for the ability to have the rate lowered later by an institution that can be leaned on, rather than by an auction that cannot.

A strategy of funding a government on short-term paper needs somebody to hold trillions of it without haggling. That buyer is being legislated into existence, and most people file it under crypto.

The GENIUS Act requires payment stablecoins to be backed one for one by high-quality liquid assets, and the permitted list is short: dollars, demand deposits, short-term Treasury bills, and Treasury repo. Bills held must have 93 days or less remaining. Roughly $230 billion of dollar-pegged stablecoins were in circulation by the first quarter of 2026, and private forecasts of the additional bill demand this creates by 2030 range from $400 billion to $2.3 trillion.

Think about who that buyer is. Someone in Buenos Aires or Lagos holding a dollar stablecoin is not comparing yields. They want a dollar that works on a phone, and they will hold it at zero interest because zero in dollars beats what their own currency is doing. No government can instruct them to sell. It is the closest thing to a replacement for the automatic bid that anyone has come up with.

The last step needs no announcement. If the debt sits at the short end, the short rate is administered, and inflation runs above it, the balance shrinks in real terms every year while every payment is made on time. Economists call it a negative real rate. It happened after the Second World War, when a comparable debt burden was roughly halved inside a decade, and the people who paid were holders of long bonds: pension funds, insurers, and anyone who moved to safety at the wrong moment.

What this is not. It is not a forecast that the dollar collapses, and it is not a price target for gold. The dollar remains 42% of global reserves, no rival is close, and gold has already shown this year that it can fall hard while the argument for it stays intact. Anyone selling you a number off this story is selling you something.

What it is, is a reminder that a return has two parts and most people only track one. If your assets rise 70% in a unit of account that is being deliberately managed downward, you have to ask what happened to the unit before you decide how you did. The people running the policy are not hiding this. Bessent named the operation himself.

For us the practical response is the one we keep arriving at, and it is duller than the story. Hold things that are not all denominated in the same unit, and do not confuse a nominal gain with a real one. That means real assets that produce income rather than only price, currency exposure that is not entirely dollar-linked, and a hard-money allocation held for the job it does rather than for its quarterly performance. Gold does that job precisely because it has no deficit to fund and no election to win.

Watch three things over the next year. Whether the buyback programme is extended past 4 November. Whether the bill share of the debt keeps climbing. And whether gold’s share of official reserves rises again in the 2026 data. If all three go one way, this stops being a theory about the yardstick and becomes the arithmetic of everyone’s portfolio.

This analysis is for informational purposes only and is not personal investment advice. Valid as of publication date; conditions evolve. Past returns are not indicative of future results. Pressure-test the framing against your own thesis before acting on it.