The Dollar's Two Prices
Debt reading 18 Aug 2026
Paper: ECB WP No. 3174
Is the dollar strengthening or devaluing? Both at once — because it is being measured against two different things. And the mechanism everyone suspects is real; it just isn't the printing press.
DXY around 99.1–99.5, up 1.7% over twelve months. Real 10-year yields near 2.4%, the highest at auction since 2008. Fed funds futures put 56% odds on a hike in September, though Kalshi and Polymarket sit near 48–49% — a coin flip, not a settled expectation. Either way, not what monetisation looks like.
Gold +29% and silver +67% over twelve months. Central banks bought a record 288.9t in Q2 2026 while prices were falling. Gold has passed US Treasuries in official reserves — 27% against 22% at end-2025, per the ECB — though the ECB is clear this mostly reflects gold's price, not reallocation.
One of the three claims is right — and it is the one that matters
The popular framing goes: print dollars to cover $40 trillion of long-term debt, and convert that debt to short-term. Those are three separate assertions with three different verdicts.
“They're printing more dollars”
Literally false, substantively right — the first version of this
section scored it false on a technicality, which was a mistake worth correcting in place.
The literal part: no printing press is running. QT ended 1 December 2025
after $2.4tn of runoff, and reserve-management purchases have since
been cut to zero for 14 August – 14 September 2026, from $10bn the
previous month. Total assets are $6.73tn (H.4.1, 26 Aug 2026), with
reserves at $2.925tn, down 9.3%
over the year. Chair Kevin Warsh built his public case on the argument that the post-2008
Fed created fiscal dominance by becoming the largest buyer of Treasuries.
The substantive part: T-bills are money-like. A four-week bill in a money
fund is functionally a demand deposit. QE swaps bonds for reserves; bill-heavy issuance
swaps bonds for bills. From the consolidated government balance sheet these are the same
operation — both shorten the duration of public liabilities and strip duration out of
private hands. The Fed is not doing it. The Treasury is. Calling that “not
printing” is accurate about the instrument and evasive about the effect.
“They're making the debt short-term”
This is the real story. Bills are now about 22% of marketable debt, above the Treasury Borrowing Advisory Committee's preferred 15–20% band. Goldman put 2026 bill supply at roughly $827bn against $360bn in 2025; four-week bills average $94bn per auction, the Treasury's single largest offering. Then on 19 August 2026, after the 30-year hit a 19-year high, Bessent more than doubled debt buybacks to ≥$4bn per operation from 9 September, targeting the 10–20y and 20–30y sectors — retiring long bonds and funding it short. Reporting suggests he could draw on the roughly $1tn Treasury General Account to do it.
“Deliberate devaluation is the plan”
The framework exists and its author has a seat. Stephen Miran's User's Guide to Restructuring the Global Trading System — the so-called Mar-a-Lago Accord — pairs tariffs with dollar devaluation, terming foreign reserves into ultra-long bonds, and a “user fee” on foreign Treasury holdings. Miran chairs the Council of Economic Advisers. None of it has been implemented. But the fact that it is discussed at all is itself part of what the long end is charging for.
The mechanism isn't inflation. It's debt-management engineering: shorten the maturity, buy back the long end, and manufacture new structural demand for bills. That last step — where the new bill buyers come from — is precisely what the ECB paper below is about.
The paper isn't about money printing. It's about who buys the bills.
Ferrari Minesso & Siena, Private money and public debt: U.S. stablecoins and the global safe asset channel, January 2026.
Under the GENIUS Act, a dollar stablecoin must be fully backed by Treasuries of 93 days or less. So every coin issued to someone in Lagos or Buenos Aires is a forced T-bill purchase. This is not an accident of the design — it is the design. Bessent has publicly pointed to a path toward a $4tn stablecoin market; Brookings and the Aspen Economic Strategy Group estimate $400bn–$2.3tn of incremental bill demand by 2030; Senator Hagerty says he drafted the bill partly to create exactly that structural bid.
The paper builds a three-country DSGE model to ask what that costs. Its answer, in one line: yes, it works — and here is the bill.
| The paper's finding | Mechanism | Where we actually are |
|---|---|---|
| Yields fall in steady state Dollar-positive | Permanent new demand for short Treasuries compresses the risk-free rate. Calibrated at −30bp on the US premium once stablecoin cap reaches $2tn. | Cap ~$308bnGrowth stalled just above $300bn since April 2025 |
| The dollar's footprint widens Dollar-positive | Dollar access for people with no US bank account, settled in real time. Deepens dollar intermediation geographically. | 57.1%USD share of allocated reserves, Q1 2026 — up from 56.4%; the IMF attributes about half to FX valuation |
| Monetary policy loses real traction Cost | On a Fed hike, foreign holders redeem to bank FX gains; issuers dump bills. Yields rise more, the dollar rises less. US output falls about 30% less than it otherwise would. | Not yet bindingBelow the paper's own $500bn “moderate” case |
| Stability condition tightens non-linearly Cost | The Taylor coefficient needed for a stable equilibrium rises with adoption. Fine at 1–2.5 up to $2tn. At 10% of the debt market it needs to be about 6 — an unreachable rule. | 0.8% of debt$2tn would be 5.3%; 10% is twice the top projection |
| Two-way spillovers amplify Cost | Foreign payment and risk shocks transmit into US yields in real time, not through slow portfolio rebalancing. | Confirmed, weaklyLocal projections 2018–2025 show US pass-through amplified; the euro-area leg has the right sign but isn't significant |
A stablecoin redemption wave makes issuers sell bills and convert dollars into foreign currency at the same instant. Yields up and the dollar down, together, from a single shock. That correlation is the signature of a funding crisis rather than a normal cycle — and the paper shows a channel that manufactures it structurally. It is currently dormant: at $308bn we sit well below the regime where it bites. The warning is about the path, not the present.
Breakevens settle the argument
If the market believed the debt would be inflated away, it would say so in one specific place. It isn't saying it.
| Instrument | Level | What it tells you |
|---|---|---|
| 2-year | 4.35% | Rose 8bp on Warsh's Jackson Hole remarks. Policy-sensitive, and it is pricing hikes. |
| 10-year | 4.72% | Up 48bp over twelve months. Not a rally. |
| 30-year | 5.21% | Touched 5.33% in the week of 17 August — a 19-year high. All the pain is here. |
| 10y term premium ACM, 13 Aug | 0.80% | Firmly positive after a decade near or below zero. This is the fiscal risk charge. |
| 10y real yield TIPS auction, 23 July | 2.44% | Highest at auction since October 2008. |
| 10y breakeven | ~2.30% | Anchored. Debasement would put this at 3.5–4%. |
The curve is steep because the risk is at the far end
US Treasury constant-maturity yields, 28 August 2026
Read the breakeven line again, because it is the single most important number on this page. Real yields sit at 17-year highs while inflation expectations sit at 2.3%. The bond market is not pricing an inflationary escape from the debt. It is charging a real premium — for supply, for duration, for fiscal credibility. Those are different diagnoses with different hedges.
This is also why the buyback didn't hold. On 19 August the 30-year fell 9bp to 5.196% on the announcement. Within twenty-four hours it had erased the move entirely and traded back above 5.25%. Buybacks change who holds the duration; they don't change how much duration exists. There has been a buyers' strike in the 10–30 year sector since late June, and $4bn an operation does not clear it.
For positioning: a 30-year at a 19-year high yield means long Treasury prices are near multi-decade lows. The market is paying roughly 2.4% real to take US fiscal risk, which is historically generous. The same fiscal risk is what could take it to 3%.
Right thesis, dangerous instrument
Both metals fell hard on Friday, on hawkish Fed news. That tells you something important about what is actually in the price.
Twelve-month change
To 28 Aug 2026
Drawdown from the 29 January record
Peak to 28 Aug 2026
The structural case is genuine and it is not a retail narrative. Central banks bought 288.9 tonnes in Q2 2026, up 62% year on year — and they bought into falling prices, which is what a policy bid looks like rather than a momentum bid. Poland took 51t, China 33t. But the bid is lumpier than that headline implies, and adversarial checking sharpened this: Q1 2026 was only about 57 tonnes, full-year 2025 fell to 850t after three straight years above 1,000t, Turkey sold or loaned 130t in early 2026, and 2025's single largest gold buyer was Tether — a stablecoin issuer, not a central bank. Silver is heading into a sixth consecutive annual deficit, a gap around 67 million ounces against supply near 1.05 billion, with solar, electronics and datacentre build-out consuming metal structurally rather than cyclically.
Now look at January and February, because it is the most useful thing that has happened to this trade. Both metals went vertical into 29 January. Gold then fell 21% from its record; silver fell about 30% in a single day and 41% from its high. The BIS wrote up the mechanics: leveraged retail ETFs and margin liquidation, with the CME lifting gold margin from 6% to 8% and silver from 11% to 15% at the top, forcing the unwind.
The thesis was right and the instrument still nearly killed you. Everything in the debasement story stayed true through February; the position did not survive it. And Friday's move — metals down 3–4.5% on hawkish Fed commentary — says a meaningful share of today's price is a rate-cut bet, not a pure debasement bet. Debasement is the floor. The rate path is the volatility.
Strengthen against currencies. Devalue against assets. Both, at once.
Next six to twelve months, the dollar more likely firms than falls in FX terms. Futures price a 56% chance of a September hike (prediction markets say 48–49%). Real yields are at 2.4%. Headline CPI is 3.4% with core at 2.5%, and Warsh has said plainly that inflation “has not meaningfully slowed.” Sell-side forecasts cluster in a 94–101 DXY range. A hawkish Fed defending the currency is the structural opposite of monetisation.
Structurally, the dollar keeps losing to real assets — and the driver is fiscal, not monetary. Debt of $40.05tn. A deficit of $1.9tn, 5.8% of GDP. Net interest of roughly $1.0tn in 2026, which is 19 cents of every federal tax dollar, on a path to $2.1tn by 2036. That arithmetic does not need a printing press to erode purchasing power. It only needs time. Gold is pricing it. Breakevens are not, yet.
These are not contradictory findings. The dollar is strong against the euro and the yen because every major sovereign has a version of the same fiscal problem, and weak against gold because gold has no sovereign. Debasement is a purchasing-power phenomenon, not an FX phenomenon. Most of the commentary you will read conflates the two, which is why it reads as contradictory.
Today, yields up, dollar up and gold up are three separate stories — policy, carry, and fiscal hedging. The configuration that means something has broken is yields up while the dollar falls, together, sustained. That is what a genuine loss of confidence looks like, and it is exactly the correlation the ECB paper's stablecoin channel manufactures once adoption gets large enough.
Six numbers, ranked by how much they would change the answer
These are ordered by information value, not by how likely they are. The first one is worth more than the other five combined.
10-year breakeven breaks above 2.75%
The bond market conceding that the debt gets inflated away rather than paid or grown out of. This is the number that flips the whole thesis from “real risk premium” to “debasement,” and it would validate the debasement framing outright.
30-year above 5.5% while DXY falls
The bad correlation going live. Rising long yields alongside a falling currency is the emerging-market signature, and it is the one configuration that would make both the FX call and the bond call wrong at the same time.
Bills push past 25% of marketable debt
Or buybacks scaling beyond roughly $10bn per operation, funded from the TGA. Either is an escalation of the maturity-shortening bet — and it converts the debt stock into a floating-rate liability repriced against the Fed every few weeks.
Stablecoin cap breaks decisively above $500bn
The paper's own threshold for where the global safe asset channel starts to matter. Growth has been stalled just above $300bn since April 2025, so this is currently a dormant risk, not a live one.
Movement on Warsh's Treasury–Fed Accord revision
He has proposed updating the 1951 Accord to give Treasury a say over major changes to the Fed's balance sheet. That is the institutional door to monetisation. If it opens, claim one on this page stops being false.
Central bank gold buying drops below 200t a quarter
The structural bid under the metals fading. This is the one that would undercut the gold leg of the argument rather than confirm it — worth watching precisely because it cuts against the thesis.
Sources
ECB Working Paper 3174 — Ferrari Minesso & Siena, Private money and public debt, Jan 2026
CBS News & Al Jazeera — $40tn debt, 18 Aug 2026
Peter G. Peterson Foundation — net interest costs and CBO projections
CNBC — Treasury doubles buybacks, 19 Aug 2026
Euronews — yields erase the buyback rally
CNBC — TGA as a buyback funding source
CNBC — Warsh, Bessent and the fiscal-dominance question
CNBC — Warsh at Jackson Hole, 28 Aug 2026
FOMC minutes, 28–29 July 2026 — funds rate at 3.50–3.75%
Wolf Street & SVB — end of QT, reserve management purchases
Zawya — bill share of marketable debt, 2026 supply
Trading Economics — 2Y / 10Y / 30Y yields, 28 Aug 2026
FRED & MacroMicro — breakevens and ACM term premium
Trading Economics — gold $4,454.08 close, 28 Aug 2026
Trading Economics — silver $66.15 close, 28 Aug 2026
USAGOLD — pre-speech spot levels, 28 Aug 2026
BIS Quarterly Review — the leveraged-retail mechanics of the Jan–Feb 2026 crash
BullionVault — gold −21%, silver −41% from the 29 Jan records
eToro — central bank buying, 288.9t in Q2 2026
Silver Institute via Scottsdale Mint — sixth consecutive deficit
IMF COFER — USD 57.13% of allocated reserves, Q1 2026
Barchart & Vantage — DXY levels, late Aug 2026
Politico Pro & Yahoo Finance — Bessent and Hagerty on stablecoin Treasury demand
Belfer Center & Flossbach von Storch — the Miran framework, assessed
BLS — CPI 3.4% headline, 2.5% core, July 2026
A note on the numbers. Gold and silver sources disagreed for 28 August because the market moved intraday: USAGOLD reported $4,608 and $69.35 before Warsh spoke at Jackson Hole, Trading Economics $4,454.08 and $66.15 after. Both are used above, labelled. The 12-month change figures are Trading Economics'; one secondary source claimed gold was up ~100% over twelve months, which is inconsistent with a $4,454 price and a 29% gain and has been discarded. August CPI is not out until 11 September, so the inflation reading here is July's. All market levels are as of the 28 August close and will move. Companion letters: The Rupee at Seventy to One Forty and the Fiscal Stress Board. This is analysis, not investment advice.