The debt that finances itself.
The United States is running a structural deficit near 5.8% of GDP — in an expansion, with low unemployment — while net interest has passed defense to become the fastest-growing line in the budget. This is a slow-burn sustainability story, not an imminent-default one — the question is not whether the path is unsustainable, but how, when and through which channel it adjusts.
The striking part: the primary deficit actually narrows this decade. The entire deterioration is interest compounding on a high and rising stock.
Lead with debt held by the public (~101% of GDP) — that is what investors price, and what drives interest rates and the term premium. Gross federal debt (~122%, $39.2tn) adds ~$7.6tn the government owes itself (Social Security and Medicare trust funds) and is the figure the debt ceiling is set against. Both are correct; they measure different things. Conflating them is the most common error in the debate.
Sustainability is about the slope, not the level. A debt ratio of 101% is not, by itself, a crisis number — Japan operates above 200% with low yields, and the US borrows in the currency it prints and runs the deepest, most liquid sovereign market on earth. What makes the path "unsustainable" in the technical sense is the trajectory: debt rising without bound under current policy because the cost of the debt now grows faster than the economy. Read the slope and the r − g gap, not the headline level.
A structural deficit, not a cyclical one
CBO's February 2026 baseline is the anchor, and one contrast inside it is the regime: the headline gap widens through the decade even as the underlying budget tightens.
- Total deficit2026 → 2036, % of GDP · $1.853tn → $3.115tn5.8% → 6.7%
- Debt held by the publicrises every single year — above the WWII peak~101% → ~120%
- Primary deficit (budget before interest)narrows — so the entire widening is interest2.6% → 2.1%
Treasury's own Financial Report calls current policy “unsustainable” in the precise technical sense, not as rhetoric. That single contrast — a tightening primary balance beneath a widening headline gap — is the regime. [HARD]
The word is technical, not editorial. Treasury defines a sustainable policy as one under which debt held by the public is stable or falling as a share of the economy over the long run; "unsustainable" is its negation — a path on which debt/GDP rises without bound under current policy. The claim is about the slope and the r − g gap, never the level. A flat 101% would not be unsustainable; a 101% that climbs every year, faster as it goes, is. This is why the level alone settles nothing — and why the dynamics chapter, not the headline ratio, is the heart of the page. [HARD]
What makes this regime different is that it is structural, not cyclical — and that is the most common misread. Every prior high-debt episode carried its own correction baked in:
- 1946 (~106% of GDP) — reversed by rapid demobilization, which cut spending at a stroke, and a generation of g > r as the post-war boom outgrew the debt.
- 2009 and 2020 spikes — cyclical emergency responses that unwound on their own as the economy recovered and the stimulus lapsed.
The 2026 deficit has no such reversal baked in. It is not a war to be demobilized from or a recession to recover out of; it is a permanent mismatch between an aging population's promised benefits — Social Security and Medicare, on demographic autopilot — and revenue stuck near 17–18% of GDP. Nothing in current law closes that gap; absent a deliberate choice, it widens. A high-debt episode that contains its own correction is a spike. One that does not is a regime.
The path does not stop at 2036. Treasury's long-run stress projection, carrying current policy forward, has debt held by the public continuing toward ~175% of GDP over the following two decades, exceeding ~200% by 2048, and — on unchanged assumptions — reaching an effectively-uninvestable 576% by 2100 [DIR]. These are not forecasts. They are illustrations of what current law implies if nothing changes — warning markers whose purpose is to be falsified by the policy correction they are meant to provoke. Read them as the arithmetic of inaction, not a prediction of it.
The direction is being pushed harder by the policy mix that produced this baseline. CBO attributes roughly $2tn of additional borrowing over 2026–2035 to the 2025 reconciliation act (OBBBA) — which both extended expiring tax cuts, lowering the revenue path, and raised the debt ceiling to $41.1tn — bringing the cumulative ten-year deficit to about $23tn, some $1.4tn above the prior projection [HARD]. This is the observed arithmetic; whether it was sound policy is the question the callout below sets aside.
Treasury Secretary Bessent has stated a "3% deficit goal" and described the deficit as having fallen to ~5.5%. Neither reconciles with CBO's 5.8% — the legislated baseline never falls below 5.6% across the ten-year window. We note the stated aspiration without endorsing the arithmetic, and weight the legislated baseline. Whether to close the gap from the spending side, the revenue side, or both is a values choice; the arithmetic only says the gap exists and is structural.
Where the money goes
The composition is the crux. Roughly 74% of spending is now "mandatory" — Social Security, Medicare, Medicaid and other autopilot programs not subject to annual appropriations. The part Congress actually fights over each year, discretionary spending, is falling as a share of GDP. Revenue, near 17.5%, is roughly at its 50-year average — not unusually low, but far below what the spending commitments imply.
| Category | FY2026 | FY2036 | Direction |
|---|---|---|---|
| Revenues | 17.5% | 17.8% | Roughly flat — near the 50-year average |
| Total outlays | 23.3% | 24.4% | Rising faster than the economy |
| Mandatory (entitlements) | 14.2% | 15.0% | Rising — demographics, on autopilot |
| Discretionary | 5.9% | 4.8% | Falling — the part Congress votes on each year |
| Net interest | 3.3% | 4.6% | Rising fastest — and the most rate-sensitive |
The three fastest-growing lines — Social Security, Medicare/health, and net interest — are precisely the three that are either legally protected or contractually unavoidable. Cutting "waste, fraud and abuse" in the discretionary ~26% cannot arithmetically stabilize a debt driven by the autopilot 74% plus interest. This is the single most important structural fact for any honest consolidation discussion.
Sources: CBO Budget & Economic Outlook, Feb 2026 (By the Numbers) · CRFB on the CBO Feb 2026 baseline. All current, lagged to release cadence.
r minus g — the regime change
Debt sustainability turns on one comparison: the effective interest rate on the debt (r) versus the economy's growth rate (g). When r < g, the ratio shrinks on its own even with a modest deficit — the world of the 2010s. When r > g, the sign flips: the debt compounds faster than the economy, and stabilizing the ratio requires a primary surplus. The US is now in the r > g regime, and that is the regime change.
In plain words, the Blanchard identity says the same thing the household analogy does: a debt grows as a share of your income whenever the interest on it outpaces your pay rise — and the only way to stop the share climbing is to spend less than you earn before counting the interest. For a government that means a primary surplus: revenue above all spending except interest. The US is running a primary deficit, in an expansion, with r above g — all three working the wrong way at once.
The arithmetic, worked through. Net interest over public debt gives the effective rate; set it against growth and the primary balance, and the stabilization gap falls out:
- Effective rate on the debt (r)2026 → 2036, as low-coupon stock rolls higher [HARD]~3.2% → ~3.8%
- Real growth (g)CBO — a mature economy, not a catch-up boom [HARD]1.8%
- Primary surplus needed to hold the ratioat ~120% debt and a 1-point positive r − g~1.2%
- Projected primary balance−2.1%
- Stabilization gap [DIR]≈ the entire defense budget, every year, just to hold flat~3.3%
r − g is not a fixed dial — the debt bends it. Three feedbacks make the gap partly self-inflicted. Crowding-out: a bigger debt stock competing for the same pool of savings nudges r up at the margin. Growth-drag: dollars spent servicing debt are dollars not invested, shaving g over time. The maturity-rollover channel: with a 5.8-year average maturity, the effective rate on the stock lags the market — so the high rates of 2025–26 keep feeding into the average cost for five to six years even if the Fed cuts [DIR]. A benign rate path still locks in a higher debt-service bill than is visible today; the gap is likely to widen mechanically into 2030 regardless of policy.
The cleanest yardstick says we are past the line. Jason Furman's centrist test — depoliticized because it targets a burden, not a debt level — is to keep real net interest below ~2% of GDP. At 3.3% and rising to 4.6%, the US has already breached it [HARD]. By that measure the question is no longer "is the path sustainable" (the consensus answer is no) but "how, when and how painfully does it adjust." Treasury's own Financial Report puts the cost of waiting at roughly 1 point of GDP per decade — a 4.7% long-run gap that becomes 5.6% after a ten-year delay and 6.9% after twenty [HARD]. The arithmetic only gets more expensive the longer the adjustment is deferred.
The fiscal-to-rates bridge, quantified. Federal Reserve research finds a permanent 1-point rise in debt/GDP lifts the 10-year term premium by ~2.5–3.5bp [DIR]. The ~20-point debt rise CBO projects across 2026→2036 implies a 50–70bp permanent term-premium lift — about a fifth to a quarter on top of today's +0.67% ACM reading. That is a steady grind, not an overnight shock; the genuine danger is a non-linear repricing — the market shifting from "absorb the supply at a concession" to "demand a structural risk premium."
The effective rate on the debt climbs from ~3.2% toward ~3.8% while CBO's real growth holds near 1.8%. The moment the rising cost line crosses above the flat growth line is the r > g regime — the point at which the debt begins, mechanically, to finance itself.
Sources: CBO Feb 2026 (effective-rate inputs, primary balance 2.6%→2.1%, 1.8% growth, debt 101%→120%) · Blanchard / PIIE / CEPR (the r − g identity) · Furman, Aspen Economic Strategy Group / World Bank (the real-net-interest threshold) · Treasury Financial Report (4.7% fiscal gap; ~1pp-per-decade cost of delay) · Federal Reserve FEDS 2024-027 (the term-premium elasticity, with material uncertainty). The effective rate is derived; the stabilization-gap arithmetic and the elasticity are [DIR].
The interest-cost spiral — and its circuit-breakers
The loop is a feedback circuit: deficits → more debt issued → higher interest expense → larger deficits → still more debt. The closed-loop accelerant is that net interest is itself a component of the deficit — so once it exceeds the primary balance's capacity to shrink, the total deficit can widen even as the primary deficit improves. That is exactly the 2026–2036 picture, and it is the single most important fact on this page: the primary deficit narrows from 2.6% to 2.1% of GDP while the total deficit widens from 5.8% to 6.7% — the entire wedge is interest. Washington can run a tighter underlying budget and still watch the headline gap grow.
Picture two lines moving apart: the budget before interest is slowly closing, the headline deficit is widening, and the gap that opens between them is the interest bill.
- Primary balance (before interest)a half-point of genuine discipline2.6% → 2.1%
- Total deficit5.8% → 6.7%
- The wedge = net interest3.3% → 4.6%
The fiscal effort is real; the interest tide is simply rising faster. This is what "the loop" means in numbers — not that policy fails, but that policy can succeed at the primary level and still lose ground overall.
The compounding is no longer theoretical. CBO has the US spending $16.2tn on net interest over 2026–2035 — the bill itself compounding at roughly 6.5% a year as the stock reprices, rising from $1.04tn (3.3% of GDP) in FY26 to $2.14tn (4.6%) by FY36. For scale, $16.2tn over the decade is more than the federal government expects to spend on national defense across the same years, and it is money that buys no road, no missile and no benefit cheque — it is rent on past borrowing.
There are two figures, and the difference is structural. Treasury's MTS Table 5 shows gross interest on the debt of $866.8bn FYTD through May 2026 — the cash the Treasury actually pays. The budget-meaningful net figure is $722.7bn. The ~$125.7bn wedge is interest the government pays to itself — chiefly to the Social Security and Medicare trust funds, which hold non-marketable Treasuries and rebate the coupon back into the budget. That subsidy is temporary. As the trust funds deplete through the early 2030s (see the dated clocks below), they stop receiving net interest and begin redeeming principal — converting intragovernmental debt into public debt the market must hold. Net interest then converges toward gross, and a cushion that has flattered the headline for decades quietly disappears.
The structure of the debt — not just its size — is what makes the loop grind. The US carries a 5.8-year weighted-average maturity and leans heavily on short bills, so rates transmit to the interest bill fast on the way up but the legacy low-coupon long bonds roll off slowly on the way down. The asymmetry is the whole story: in 2026 alone the Treasury must refinance an estimated $9–10tn of maturing debt on top of the new ~$1.9tn deficit, and its own quarterly data show roughly $14.6tn of maturities against $15.7tn of gross issuance in a single FYTD window — a violent rollover rhythm in which every quarter a large slice of the stock is repriced to whatever the curve offers that morning.
Two interest measures matter, and the second is the early-warning one. The cost on the stock moves slowly (the 5.8-year ladder above); the cost on the refinancing flow reprices at today's rate immediately — the maturing stock plus ~$2tn a year of net new marketable borrowing the Treasury must place:
- Net new marketable borrowing — FY26TBAC primary-dealer medians$1.95tn
- FY27$2.02tn
- FY28$2.10tn
- Bill share of issuanceabove the TBAC’s own 15–20% guidance band21.7%
That is the “activist issuance” debate in one line: bills ease the auction today but pass Fed-rate changes into the interest bill fastest, while extending into longer coupons would force a term-premium concession. Whichever lever, the flow is where higher rates bite first — the place to watch before the stock-average even moves. [DIR]
Because so much cost sits in the rolling stock rather than the policy rate, even aggressive easing barely moves the bill:
- 100bp Fed cut — year-one saving~$66bn
- 200bp cut — year-one saving~$144bn
- … against a net-interest bill heading through~$1.1tn/yr
With a 5.8-year average maturity it takes five to six years for the stock to fully reprice, so the higher rates of 2025–26 are locked into the average cost regardless of what the Fed does next. The r − g gap is likely to widen mechanically through 2030 even if the Fed eases. The implication is blunt: monetary policy can ease financial conditions, but it cannot ease the fiscal arithmetic.
Think of the debt as a household that financed its life on a ladder of fixed-rate loans taken out when money was cheap. When market rates jump, the family's average payment barely moves at first — only the loans that happen to mature this year refinance at the new, higher rate. But a fifth of the ladder rolls over every year, so the pain arrives in instalments and keeps building long after rates have stopped rising. Now run it in reverse: even if the central bank slashes rates tomorrow, the family is still locked into years of expensive loans signed at the peak. That lag — fast going up, slow coming down — is why a higher-rate regime is, for the Treasury, a one-way ratchet on the next half-decade of interest cost. [DIR]
Mid-2026 is the live demonstration. The Iran-war oil shock pushed CPI to a three-year high of 4.2%, forcing the Fed to hold at 3.50–3.75% rather than cut — and every month of "higher for longer" feeds straight into the rollover cost of the maturing bill and note stock on the entire ~$39tn pile. The war's direct cost is a rounding error; its indirect cost — the higher rate at which the whole debt refinances — is the channel that actually moves the fiscal numbers.
Is it a "doom-loop"? Honestly — not mechanically, not yet, not inevitably. The loop only turns truly explosive if r stays above g and markets begin pricing a risk premium that itself raises r further. The US keeps three circuit-breakers a peripheral sovereign lacks, and each is worth stating plainly:
It borrows in its own currency. Every dollar of US debt is repayable in dollars the US alone prints, so there is no mechanism that can force a nominal default the way a foreign-currency debt can. The constraint is real, but it is the inflation/exchange-rate constraint, not the bankruptcy one — a profoundly different and softer ceiling.
Treasuries remain the global safe asset. In a scare, capital runs toward US debt, not away — so flight-to-quality episodes lower the Treasury's borrowing cost even amid its own deterioration. Mid-2026 proved it: the Iran shock strengthened the dollar and drew sovereign buyers into Treasuries while the fiscal numbers were visibly worsening.
The Fed can become a buyer in extremis. A central bank that can purchase its own government's paper can cap a disorderly auction or a liquidity spiral — the backstop that decides whether stress stays a price move or becomes a funding crisis. It is a last resort with inflationary costs, but its mere existence changes the tail.
The honest label is a grinding loop — a slow transfer of the budget from programs to creditors — not a sudden collapse. The genuine danger is not the trend but a non-linear repricing: the day the market shifts from "absorb the supply at a small concession" to "demand a structural risk premium." That tail is real. It is not the base case.
Sources: CBO Feb 2026 ($16.2tn cumulative; net interest 3.3%→4.6%; primary 2.6%→2.1%, total 5.8%→6.7%) · Treasury MTS Table 5 (gross $866.8bn vs net $722.7bn FYTD; the ~$125.7bn trust-fund wedge) · Treasury MTS Table 9 (net interest > defense) · the maturity / rollover arithmetic ($9–10tn 2026 wall; $14.6tn maturities vs $15.7tn issuance FYTD; the 100bp/200bp savings) is [DIR], from public TBAC and dealer data. The defense bar is approximate (FYTD defense ran below net interest); the household and ladder schematics are [DIR] illustrative. TBAC bill share 21.7% (above its 15–20% guidance band) and the dealer-median deficits ($1.95 / $2.02 / $2.10tn, FY26–28) are [HARD] (Treasury TBAC); the "activist issuance" reading is [DIR].
Who holds the debt — and the marginal-buyer shift
For two decades the marginal Treasury buyer — the bidder who clears the last slice of each auction — was price-insensitive: foreign central banks recycling trade surpluses, and the Fed during QE. Neither bid for yield; both bought for policy. That cohort is now in retreat. As it exits, the Treasury must clear rising supply with price-sensitive buyers — hedge funds, money funds, households — who do bid for yield, and demand a higher term premium. That hand-off, not the absolute stock of debt, is the analytical core linking who holds the paper to the long end of the curve.
Picture an auction where the same ten lots clear every week. For years a central bank took the last three lots at any price, to manage its currency. Now it has stopped. The remaining bidders are private funds who will take those lots — but only at a cheaper price, i.e. a higher yield. The average holder barely changes; the marginal one changes completely. Price is set at the margin, so the whole curve reprices. [DIR]
| Cohort | Share | What it means |
|---|---|---|
| Domestic private | ~55% | Mutual funds, banks, pensions, households — price-sensitive. |
| Foreign (official + private) | ~25–30% | Of the ~$39tn gross stock; ~22% of the marketable stock — retreating in share even as the dollar total hits records (largest holders listed below). |
| Intragovernmental | ~20% | Trust funds owning Treasuries — converts to public debt as they deplete. |
| Federal Reserve | ~13% | SOMA $4.476tn and shrinking via QT — no longer the price-insensitive backstop bid. |
- Largest foreign holders — Japanofficial + private · March 2026 TIC~$1.19tn
- United Kingdom~$897bn
- Chinaits lowest since September 2008~$652bn
- Foreign block, totalnear a record in dollars; ~22% of marketable, down from ~50% in 2008~$9.5tn
Shares are of the ~$39tn gross stock. The foreign block totals ~$9.5tn (Feb 2026, up ~6% year-over-year from ~$8.9tn) — near a record in dollars even as its share falls. [HARD], TIC.
The "$138bn foreign sale" headline — the largest monthly drop since September 2022 — inverts on the full Treasury release (sb0499): officials stepped back, but price-sensitive private money more than filled the gap.
- Headline foreign “sale”−$138bn
- Net result+$150.7bn
- Net private foreign buying+$162.1bn
- Net official outflows−$11.4bn
- Bonds & notes — private boughtofficial institutions sold −$37.9bn+$51.5bn
- Visible official sellers — Japan (yen defense) · China−$47.7 / −$41bn
A composition shift — price-insensitive official holders giving way to price-sensitive private ones — not a buyer's strike. The “$138bn” headline must never stand alone. [HARD]
The data does not support it as an acute threat. China's reductions have been gradual and orderly — from a ~$1.3tn peak to ~$652bn (its lowest since September 2008) over a decade. A genuine fire-sale would crush the value of its own remaining holdings and surge the yuan against its mercantilist interest: a mutually-assured-destruction dynamic that makes a deliberate strike low-probability. The credible concern is passive diminution — foreign holdings growing slower than issuance, so the foreign share grinds down. The TBAC names exactly this: declining international holdings as a share of Treasuries outstanding is "a new factor pressuring term premia higher." A slow grind, not a crisis trigger. [DIR]
The Fed is the other price-insensitive buyer now in retreat. SOMA Treasury holdings stand at $4.476tn (June 11, 2026; ~$6.44tn across all securities), down from the QE peak as the balance sheet runs off. It still buys — modest reinvestment (~$15bn/mo) plus reserve-management purchases — but is a large, no-longer-growing holder, not the automatic backstop it was through 2021. Its custody account for foreign officials, ~$2.9tn, is a further window on the same official step-back. [HARD], Fed H.4.1.
Official foreign institutions (net sellers) and the Fed (QT) — the two price-insensitive cohorts. They bought for policy, not yield, and absorbed supply without demanding a premium.
Private foreign (a record dollar bid, partly Iran flight-to-safety) plus domestic funds and households. Real demand — but it clears only at a yield, lifting the premium auction by auction.
Not fewer buyers — different buyers. The bid is intact; its price-sensitivity is not. A rising risk premium on a still-funded market, the slow channel from cohort mix to the long end.
Sources: Forbes / Treasury ownership shares (domestic ~55% · foreign ~25–30% · intragovernmental ~20% · Fed ~13%) · Treasury TIC, March 2026 (release sb0499 — the private-vs-official split; Japan ~$1.19tn, UK ~$897bn, China ~$652bn, foreign total ~$9.5tn) · Fed H.4.1 (SOMA $4.476tn, foreign custody ~$2.9tn) · TIC historical (foreign share ~22%, down from ~50% in 2008) · Treasury TBAC ("a new factor pressuring term premia higher"). Cohort shares are current, lagged ~6 weeks (TIC). The marginal-buyer mechanism and the China read are [DIR].
What is actually driving it
Four forces push the path, and they do not weigh the same. Read them in order of structural force: demographics is the engine that sets the deficit's floor; tax policy fixes the revenue ceiling the spending then overshoots; defense is, as a share of GDP, a red herring — rising in dollars but shrinking in the budget; and the Iran-war impulse is the live catalyst — trivial in its direct cost, decisive through the rate channel. The first three are the slow regime; the fourth is the thing moving the numbers this quarter.
Demographics and entitlements — the structural engine. Social Security and the major health programs grow from ~11.2% to ~12.5% of GDP over the decade [HARD] — an extra 1.3 points of GDP arriving on no one's vote. With ~74% of the budget now mandatory [HARD] and on autopilot, this is the line a longer-living population writes into law decades in advance. It is the demographic clock no discretionary discipline can offset: the trustees' 2026 report sharpened, not eased, the urgency — the dated trust-fund clocks below are this driver coming due.
Tax policy — the revenue floor. The honest framing is that revenue is not unusually low: it holds near 17.5–17.8% of GDP [HARD], roughly the 50-year average. The problem is a promises-versus-payments gap, not a starved Treasury. The dominant recent shock is the 2025 reconciliation act (OBBBA), which extended expiring tax cuts and adds ~$2tn of net borrowing over the decade [HARD]. Offsetting it, tariffs and customs have become a visibly larger revenue line — but costly revenue: CBO scores tariffs as both adding cash and raising near-term inflation while dampening growth [HARD]. A revenue source that taxes the economy to fund the deficit is not free money.
Defense — a red herring as a share. The nominal figure rises — the FY2027 request is ~$1.5tn [HARD] — which makes "defense spending is exploding" an easy headline. As a share of GDP it does the opposite: discretionary spending, defense included, is falling, because nominal funding grows slower than the economy. The crossover is the fact that lasts: net interest has now passed defense [HARD] to become the larger line — for the first time in modern history, rent on past borrowing outranks the entire military in the budget. As the budget chapter showed, the discretionary lever is far too small to move the total.
The Iran-war impulse — the live catalyst. The direct cost is a rounding error: ~$29bn official / ~$34bn on independent trackers [HARD] against a $7.45tn budget. And it is not yet even appropriated — no supplemental has been submitted to Congress, with internal DoD estimates of a future ask ranging $50–200bn+ [DIR, pending not enacted]. The force that actually moves the fiscal numbers is indirect: the Hormuz oil spike pushed WTI toward ~$100, drove CPI to a three-year-high 4.2%, and held the Fed at 3.50–3.75% rather than cutting [HARD] — and every month of "higher for longer" raises the rollover cost on the entire ~$39tn stock. A war whose direct bill is invisible against the budget reaches the deficit through the rate at which the whole debt refinances.
Demographics is the structural engine (it sets the floor and runs on autopilot); tax policy sets the revenue ceiling (roughly average, deliberately held there by OBBBA); defense is a red herring as a share of GDP (rising in dollars, shrinking in the budget, now passed by interest); and the Iran impulse is the live catalyst — a rounding error in direct cost, but the dominant near-term mover through the oil→inflation→Fed-hold→rollover-cost channel. The first three are why the path is unsustainable; the fourth is why mid-2026 is when it tightens.
Sources: CRFB / CBO Feb 2026 (entitlements 11.2%→12.5% of GDP, mandatory ~74%, OBBBA ~$2tn) · CBO Monthly Budget Review (tariffs/customs and their growth-and-inflation cost; defense outlays and the FY2027 ~$1.5tn request; net interest > defense) · CNN / MilitarySpend.org (Iran direct cost ~$29–34bn; supplemental status — none submitted) · BLS / Fed (CPI 4.2%, funds rate 3.50–3.75%). Iran figures are current on an active conflict; the $50–200bn+ supplemental is pending, not enacted.
The political calendar
The fiscal calendar matters less for any one deadline than for what it reveals: the adjustment is arithmetically simple and procedurally near-impossible.
OBBBA raised the ceiling to $41.1tn on July 4, 2025 — the largest single-dollar increase in history. Gross debt is ~$39.22tn, leaving ~$2.06tn of headroom. The Bipartisan Policy Center and CBO put the limit being reached between late winter and mid-summer 2027, with the X-date later in 2027 after extraordinary measures. The mid-2026 stress is coming from rates and the war, not the ceiling — treating the limit as a 2026 flashpoint is stale framing.
Appropriations have been unusually contentious: a record 43-day shutdown began October 2025, and a missed January 2026 deadline triggered a short shutdown resolved in early February. The structural political trap is what one council member called "reconciliation in reverse" — the only filibuster-bypass vehicle was used to widen the deficit (OBBBA), while the routine appropriations process fights over the shrinking discretionary slice. Consolidation is arithmetically possible but procedurally improbable.
Sources: Bipartisan Policy Center debt-limit watch (ceiling, X-date) · CRFB Appropriations Watch FY2026 (shutdowns). X-date timing is [DIR]; the ceiling and shutdown facts are current.
How fiscal stress reaches markets
Fiscal risk does not announce itself at the front of the curve — that belt is the Fed's, anchored to the policy rate. It surfaces further out, in the term premium — the cleanest market-priced fiscal signal, and the number the Rates desk prices directly. This chapter traces the chain from a 5.8%-of-GDP deficit to the long bond, the dollar and gold, and names the one pattern that would tell you the chain had snapped.
Why the term premium and not the front end? Because the two ends answer to different masters. The Fed owns the short rate; an investor worried about debt cannot express that worry there. What they can do is demand more to hold a thirty-year bond — and that demand is exactly what fiscal supply and the cohort shift (above) push up.
The larger force, and the one Washington controls. A ~5.8%-of-GDP deficit forces the Treasury to issue paper into a market with fewer captive buyers. More supply against a given demand curve clears at a higher yield — on most readings the dominant driver of the move.
The structural force. As official managers diversify, each auction leans harder on price-sensitive private demand. Real but bounded — foreign holdings sit near record highs (foreign share ~22% of the stock). A rising risk premium, not a buyers' strike.
Supply steepens the curve while the most reliable slice of demand flattens out. Either force alone is manageable; together they carry the term premium to a fifteen-year high without a single failed auction — a grind, not a break.
This page reads the 10-year term premium at 0.60–0.70% on the ACM model — its highest sustained level since 2011/2015. The Rates desk's canonical anchor is +0.67% ACM / +0.80% KW, and our band brackets the ACM number exactly: no conflict, two desks agreeing. Morgan Stanley's ~55bp is a single-source point estimate that sits comfortably inside the band; a careful reader treats the band, not any one print, as the signal.
The mid-2026 spike is not one story but two, layered. Driver one — expected short rates: the Iran oil shock pushed inflation to a three-year high and the Fed held rather than cut, so the expected average of future short rates rose — lifting the long yield without touching the premium. Driver two — the premium proper: fiscal supply and the cohort shift raise the compensation investors demand for duration. Both push the long end the same way, which is why the move looks unambiguous — but splitting it between them is genuinely contested. The Fed's own research gives a yardstick for driver two: a permanent 1-point rise in debt/GDP lifts the 10-year term premium by roughly 2.5–3.5bp, so the CBO path (debt up ~20 points by 2036) implies a ~50–70bp structural lift — a steady grind, not a shock.
Intuition says fiscal deterioration weakens the currency. In mid-2026 the opposite held: the Iran shock strengthened the dollar, as flight-to-safety drew sovereign and reserve buyers into dollar assets and Treasuries. Geopolitical stress is, near-term, bullish the dollar/Treasury complex even while the long-run fiscal story is bearish. Two dollar gauges get conflated in this debate — they are different series and must never be swapped:
| Dollar gauge | What it is |
|---|---|
| ICE DXY · ~99.8 (+~8%) | A proprietary, six-currency gauge, heavily euro-weighted — the headline screen number, and the one that moved on the shock. [DIR] |
| Fed broad DTWEXBGS · ~120 | A broad, trade-weighted basket spanning the dollar's actual trading partners — a different series whose level is not comparable to DXY. DTWEXBGS is not "the DXY." |
So the bad scenario is not "no buyers." It is "buyers clear the supply, but demand yield plus a currency risk premium" — a price-clearing event, not a failed auction.
It is tempting to read gold as the mirror of the dollar — up when the dollar is down. The deeper bid is structural: central banks accumulate gold to hedge dollar-reserve exposure, not to abandon it, and that official demand persists even in months when a firm dollar would normally cap the price. Gold is the issuer-free, sanction-proof reserve asset, so its bid tracks the slow erosion of confidence in any single sovereign's paper — a different and more durable force than the day-to-day dollar trade. [DIR]
How would you tell a flight-to-safety event from a genuine fiscal-credibility break? One pattern distinguishes them. In an ordinary stress episode, yields rise and the dollar rises together — havens bid. The tell of a credibility break is the three moving the other way at once: rising yields, a falling dollar and rising gold, simultaneously — capital leaving the bond and the currency and seeking the issuer-free asset. As of mid-2026 that pattern has not appeared: the Iran shock strengthened the dollar and drew buyers into Treasuries — the falsifier of a buyer's strike, observed in real time. The building blocks exist (a fifteen-year-high premium, official net selling, a Fed unable to cut), but the diagnostic pattern is absent — so the buyer's strike stays a low-probability tail to monitor, not a base case to price.
Sources: term premium — NY Fed ACM (canonical +0.67% per R-09 Rates desk; KW +0.80%), Morgan Stanley ~55bp point estimate · transmission elasticity — Federal Reserve FEDS (~2.5–3.5bp per 1pp debt/GDP) · dollar — broad trade-weighted DTWEXBGS (~120) is a distinct series from the ICE DXY (~99.8, proprietary six-currency); the two are never interchangeable. The fiscal-vs-macro decomposition of the premium and the DXY level are [DIR].
Three credible camps
The diagnosis is close to unanimous: CBO, Treasury, CRFB, the Peterson Foundation, the rating agencies — and the doves — agree the current-policy path is unsustainable in the technical sense. The genuine disagreement is not over the facts but over the consequences: how the regime resolves, how probable a disorderly break is, and whether the slow resolution even counts as a "crisis." We present each camp in its strongest form, the mechanism it leans on, and a falsifier — the evidence that would prove it wrong. A position that cannot be falsified is an ideology, not an analysis.
What is settled: an r > g regime, a structural primary deficit, net interest past defense, a shrinking price-insensitive buyer base, and debt held by the public marching from ~101% toward ~120%. What is contested: whether that ends in consolidation, quiet erosion, or a repricing — and on what timetable. The camps are less opposed than they appear: the hawks are right about the trajectory, the doves are right that nominal default is near-impossible, and the muddle-through school is probably right about the form of resolution. They diverge on the probability of the tail and on whether a slow, costly adjustment is a crisis or merely a price.
Fiscal hawksSummers · Furman · CBO · CRFB · PGPF · rating agenciesThe path is unsustainable and the tail is rising — adjust now, because delay compounds.◇ [DIR]
An r > g regime plus a structural primary deficit plus a retreating price-insensitive bid means the term premium climbs, the interest bill compounds, and fiscal space erodes — while the tail risk of a disorderly repricing grows. The remedy is an active primary-balance adjustment of 2–3% of GDP [HARD], the sooner the cheaper; Furman — a sober hawk, not an alarmist — recommends targeting primary balance by 2030.
The cost of waiting is the clinching evidence: Treasury's own Financial Report puts the 75-year fiscal gap at 4.7% of GDP, rising to 5.6% after a decade's delay [HARD] — roughly a point of GDP per decade. Furman's depoliticized yardstick (real net interest below ~2% of GDP) is already breached at 3.3% and rising.
A durable return to g > r — an AI-productivity surge that stabilizes the ratio without policy — or the term premium failing to rise through years of record issuance and a shrinking buyer base (the Japan precedent). Repeated stress episodes that strengthen Treasury demand, as in mid-2026, would defang the buyer's-strike tail.
Doves — the looser-constraint viewKelton · Wray · KrugmanA currency issuer cannot be forced to default; the real limit is inflation, not a debt ratio.◇ [DIR]
A government that borrows in a currency it controls cannot run out of money or be forced into nominal default; the binding constraint is inflation and real resources, not the debt level. The camp is a spectrum: Kelton's MMT reaches it via monetary sovereignty (inflation, not the deficit, is the signal a line has been crossed); Krugman, no MMT adherent, reaches a similar "no solvency crisis" conclusion via own-currency plus safe-asset demand.
The historical record is the evidence: almost every modern debt crisis involved foreign-currency borrowing. The US borrows in dollars it alone prints, so the bankruptcy channel that disciplines a peripheral sovereign does not exist — a softer, inflation-shaped ceiling replaces the hard default one.
A genuine inflationary loss of control — fiscal dominance binding, the Fed unable to stabilize prices without crashing the debt — would prove the dove's own stated constraint had bitten. So would a sustained, disorderly term-premium spike that forces austerity anyway: "cannot be forced to default" is cold comfort if the real burden becomes politically unbearable.
Muddle-through · financial repressionReinhart–Sbrancia · the FTPL school · much of Wall StreetNo clean consolidation, no default — a slow blend of growth, mild inflation and negative real rates erodes the real debt.◇ [DIR]
The binary of "consolidation versus crisis" is a false choice. High-debt advanced economies usually resolve through a blend: some growth, some surprise inflation, and financial repression — negative real rates and captive demand from regulated institutions — that quietly liquidates the real debt over a decade-plus. No dramatic adjustment, no dramatic break; a slow grind that falls on the long-duration saver and is the political path of least resistance.
The post-WWII episode is the template: Reinhart–Sbrancia document the "liquidation effect" by which negative real rates erode a debt stock. The fiscal-theory-of-the-price-level school supplies the why — if fiscal policy will not adjust, the price level must.
The growth leg failing — productivity not arriving — would leave inflation and repression to do all the work, collapsing this view into the dove's inflation problem. So would the repression toolkit proving unavailable in a deep, open capital market (rate caps and captive demand are far harder for the US than for a closed 1950s economy). Its Achilles' heel: the case assumes an orderly process — any episode that goes non-linear falsifies it.
One read of the same tape vindicates each in part. The term premium is rising (hawk-supportive); stress strengthens the dollar and Treasury demand (dove-supportive); the resolution visibly forming is oil-inflation plus a constrained Fed (repression-supportive). The intellectually honest posture is to assign meaningful weight to all three and treat the buyer's-strike tail as low-probability, high-impact — a risk to monitor, not a base case to assume. The single tell that would adjudicate between them — the three-signal pattern described in Transmission — has not appeared.
Sources: hawks — Summers, Furman (Aspen Economic Strategy Group / World Bank), CBO, CRFB, PGPF, the rating agencies. Doves — Kelton and Wray for the MMT form, Krugman for the orthodox-Keynesian form. Muddle-through / repression — Reinhart–Sbrancia (the liquidation effect) and the fiscal-theory-of-the-price-level school, with the Yale Budget Lab AI scenario on the growth leg. Camp characterizations are [DIR] interpretive summaries, not quotations; the cost-of-delay and adjustment-size figures are [HARD], per Treasury and Furman.
Catalysts and scenarios
Six scenarios, council-blended. The likelihoods are deliberately qualitative — relative weighting, not false precision — and every cell is directional. We use ordinal labels (Modal · Moderate · Lower · Upside tail · Fat tail) rather than fabricated percentages, because the underlying judgement is a blend of three independent reads, not a calibrated forecast.
The single framing to carry: the base case is not one scenario but a blend of two — an ongoing muddle-through in which the ratio grinds from ~101% toward ~120% without a break, resolving over years through financial repression: mild inflation tolerance and negative real rates that quietly erode the real value of the debt. That is how the United States resolved its WWII peak, and it is the most historically-tested path — not explicit default, not a clean consolidation. The consequential risk does not live in the base case; it lives in the tails, where a disorderly repricing or a deep recession changes the regime rather than extending it.
It is the resolution mechanism general audiences understand least. It is the post-war liquidation mechanism the camps chapter describes — negative real rates quietly eroding the real debt — but the US version would be subtle: regulatory liquidity rules, reserve-management purchases, inflation tolerance and maturity policy, rather than a 1950s-style control regime. It stabilises the headline ratio while imposing a quiet, real cost on bondholders — which is exactly why it tends to win politically.
Each row gives the cross-asset effect across the three instruments that price fiscal risk — Treasuries, gold and the dollar — and a lead indicator paired with its falsifier: the evidence that would confirm the scenario is forming, and the evidence that would rule it out.
| Scenario | Weight | Cross-asset effect | Lead indicator → falsifier |
|---|---|---|---|
| Muddle-through — BASE | Modal | USTs: term premium grinds higher on supply; auctions clear. Gold: structural bid. Dollar: range-bound. | Lead: deficit holds ~5–7% of GDP; r − g mildly positive. Falsifier: durable r < g returns. |
| Financial repression | Modal blend | Mild inflation + negative real rates erode the real debt; long-duration savers pay. The historically-tested resolution, not a footnote. | Lead: real policy rates held below growth for years. Falsifier: positive real rates sustained. |
| Hard consolidation | Lower | Legislated tax rises / spending cuts (~$827bn/yr, Yale Budget Lab) stabilize the ratio; near-term growth drag. | Lead: a credible multi-year fiscal deal. Falsifier: reconciliation keeps widening the deficit. |
| Recession | Moderate | Auto-stabilizers push the deficit to 9–11%; flight-to-quality lowers Treasury yields even as the fiscal numbers worsen. | Lead: unemployment turns up. Falsifier: soft landing holds. |
| AI / productivity surge | Upside tail | Faster trend growth lifts g above r; the ratio stabilizes without austerity — the benign exit. | Lead: sustained productivity acceleration. Falsifier: growth stays ~1.8%. |
| Buyer's strike | Fat tail | Disorderly term-premium spike; a non-linear repricing from "absorb supply at a concession" to "demand a structural risk premium." | Tell: rising yields + falling dollar + rising gold simultaneously — has NOT appeared. Falsifier: that pattern never forms. |
The same six scenarios collapse into three families once you ask what each does to the dollar/Treasury complex.
Muddle-through + financial repression (Modal blend), with hard consolidation the orderly-but-improbable cousin. The debt grinds higher; the term premium creeps; real rates sit below growth. Treasuries reprice slowly, gold holds a structural bid, the dollar stays range-bound. The regime is prolonged, not broken — the burden shifts quietly to the long-duration saver.
Buyer's strike (Fat tail): a non-linear repricing in which price-sensitive buyers demand a structural risk premium and a currency premium at once — capital fleeing from Treasuries. The three-signal tell described in Transmission still applies; it has not appeared, and mid-2026 stress did the opposite. Low-probability, high-impact.
AI / productivity surge (Upside tail) lifts g above r and stabilises the ratio without austerity — the clean exit. Its mirror is recession (Moderate): the fiscal numbers worsen, yet flight-to-quality pulls capital toward Treasuries and yields fall. Same haven reflex, opposite fiscal sign.
The buyer's-strike tail and the recession tail are, at the level of flows, the same stress pointed at different targets. A US-fiscal scare is a flight from Treasuries — the safe asset is the thing in question, so yields rise and the dollar falls. A non-US risk-off shock — a foreign war, a banking scare abroad — is a flight to Treasuries: the same panic, but the US is the refuge, so yields fall and the dollar firms. This is why the dollar's direction, not the level of yields, is the cleanest read on which kind of stress is unfolding. Rising yields alone are ambiguous; rising yields with a falling dollar are the signature of fiscal credibility being repriced. Mid-2026 was unambiguously the second case — the flight to Treasuries described in Transmission, not the flight from.
Read the matrix as a probability mass, not a point forecast: most weight on the base family (the regime continues and is slowly repressed away); a meaningful but lower slice on recession; thin tails on the benign growth-out and on the disorderly buyer's strike. Hard consolidation (~$827bn/yr of adjustment, Yale Budget Lab [HARD]) is arithmetically the cleanest fix and procedurally the least likely — "reconciliation in reverse" keeps widening the gap rather than closing it.
Sources: scenario framing is council-blended and subjective / illustrative [DIR] — relative weighting, not precision. The base case is a muddle-through × financial-repression blend (the post-WWII resolution, after Reinhart–Sbrancia); the hard-consolidation stabilisation cost of ~$827bn/yr is the Yale Budget Lab estimate [HARD]. The buyer's-strike "tell" — rising yields, falling dollar and rising gold together — is the council's diagnostic trigger, not a forecast; it has not appeared.
The dated clocks — trust-fund depletion
Unlike most of this story, these are hard, dated catalysts. On depletion, benefits are automatically cut to what incoming payroll taxes cover, absent congressional action — and the depletion also converts intragovernmental debt into public debt, the bridge back to the gross-vs-public distinction.
| Trust fund | Depletion | Then payable |
|---|---|---|
| Social Security retirement (OASI) | Q4 2032 | 78% |
| Medicare Hospital Insurance (HI) | Q2 2033 | 89% |
| Combined OASDI | 2034 | 83% |
The CBO baseline assumes scheduled benefits keep being paid in full — even though current law forces the automatic cut to the payable share (78% / 89% / 83%) the moment each fund depletes. The realistic political path is neither extreme: Congress almost certainly authorizes a general-fund transfer to keep cheques whole. That transfer is new borrowing — it converts the ~$7.6tn of intragovernmental holdings into public debt the market must absorb, a cost sitting outside the headline baseline until the choice is made. So the dated clocks are not only an entitlement story; they are a scheduled, off-baseline addition to the public debt this page tracks — the §2 "trust funds owning themselves," coming due. (The dates and payable shares are [HARD]; the general-fund-transfer path is [DIR].)
Sources: SSA 2026 Trustees Report (OASI Q4 2032; OASDI 2034) · CMS 2026 Medicare Trustees (HI Q2 2033). Released June 2026; current.
Winners and losers
Who the regime rewards and who it taxes follows directly from the mechanics above — duration and real rates do the sorting.
Long-duration bondholders (repriced by a rising term premium), future taxpayers (every decade of delay adds ~1pp of GDP to the required adjustment), and savers under financial repression (negative real returns are the mechanism that erodes the real debt).
Short-duration and cash (paid to wait), gold and real assets (the sanction-proof, issuer-free hedge), the near-term dollar (haven bid in stress), and the sovereign issuer itself under inflation (which erodes the real value of its own debt).
Banks (wider margins versus duration losses) and pensions (higher discount rates help funding but mark down asset values) — the net depends on the path of real rates.
Sources: winners/losers are [DIR] interpretive read-throughs of the regime, not investment advice; the repression channel follows Reinhart–Sbrancia.
What we can show live — and what we can't, yet
Feed-state is the status of a tile's number, never of a section. Every section ships its full frame; only the live figure waits on a wired feed. US fiscal data is unusually well-served by free official sources — Treasury's Debt-to-the-Penny is the only daily free fiscal series, and the MTS, H.4.1, ACM and daily-curve endpoints are all free and parseable.
No invented data. Scope boundary: this page covers on-balance-sheet federal debt; contingent liabilities (GSE, FDIC, PBGC, student loans) convert to public debt only when triggered, are not in CBO baselines, and are not inflated into the headline numbers. Figures are as of mid-June 2026 and re-pulled at build; directional figures are labelled [DIR].