As of November 12, 2026, the US national debt was $40.61 trillion (Treasury, Debt to the Penny), and over the past year it grew by about $78,871 a second. The 10-year Treasury yield was 5.10% on November 12, 2026.
Treasury borrowing rates since 1981, and where each path takes them. Pick a path above to highlight it.
ExplainersWhat σ means ▸ The 2020 bottom ▸ The shaded band: lenders charge more ▸
Cents of every federal tax dollar that go to interest. 100 means interest takes everything.
The tax rise needed to stop the debt outgrowing the economy, if Congress waits until that year. Same in spending cuts.
ExplainersWhy waiting doesn't get cheaper ▸
The whole menu: arithmetic allows three ways out — adjust now (the tax rise above), adjust later under market pressure (the cost-of-waiting chart), or inflate it away (the first ending). Growth alone can't close the gap while the debt's interest rate exceeds the economy's growth rate — that gap is the spiral. The one thing the arithmetic never offers is waiting for free.
ExplainersFiscal dominance ▸ Financial repression ▸ Could it become hyperinflation? ▸ The UK's 2022 gilt crisis ▸ Collateral damage ▸ Why the Volcker playbook breaks ▸ France, live ▸ No captive buyers ▸ Why the usual brakes fail ▸
These signals don't say whether things are good or bad. They say which way out the system is heading. Financial repression is the quiet version of the first ending: rates held below inflation, so the debt shrinks against the economy while savers lose. These four signals watch for it, with rules fixed in advance. What is financial repression? ▸
Waiting for the first data update.
The crossing date isn't a crash date. It's the point where one of these stops being a choice. Every path below has happened before, somewhere.
These aren't separate futures. A crisis usually forces a choice between the first two, and a market break is often what forces it. And these are the mild versions: every precedent above is a case a country survived. Could it go further, to hyperinflation?
Why there are only three: the Fed has three jobs — stable prices, a stable financial system, and an affordable Treasury market. Past a certain debt level it can defend any two of them, not all three. Each ending is one way of choosing which two. And the damage doesn't wait for an ending: rising yields are already working through banks, stocks, housing and the Fed itself.
ExplainersSocial Security after 2032 ▸
Uses the scenario you picked above. Change the numbers to match your life.
ExplainersWhat σ means ▸ The 2020 bottom ▸ Term premium ▸ Treasury auctions ▸
Each indicator below is scored automatically from the latest data, including when it cuts against the debt-spiral story.
Auction note: a true "tail" compares the auction yield with the market yield just before the sale, which needs paid data. Bids per dollar sold (bid-to-cover) and the share left to primary dealers are the standard free measures.
The 10-year Treasury yield, monthly, since 1981, with a regression channel for each regime. For four decades every refinanced bond was cheaper than the one it replaced; since the 2020 turn, each one costs more. Every projection on this page starts from that turn.
The rule, fixed in advance: every month, each path's forecast from 12 months earlier is checked against what actually happened. The path with the smallest average miss is the one reality is following, whichever it is.
"Said then" is the prediction the clock recorded that day, frozen and never revised. "Says now" is today's prediction for the same path. "Pace check" projects the national debt forward from that day's 12-month borrowing pace and compares it with today's estimate; grading starts once a snapshot is 90 days old.
Federal debt rose from 32% of the economy in 2001 to 98% in 2023. The Committee for a Responsible Federal Budget measured how much each major tax and spending policy since 2001 added, in points of GDP. Together they add about 98 points: without them, CRFB estimates, the debt would have been paid off. Here is the same total split two ways.
The first split totals 97 and the second 98 because of rounding. CRFB also notes that some bipartisan laws extended policies first passed by one party; if those extensions are credited to the party that started them, Republican partisan actions could explain 20 points.
Official estimate when each law passed, over that law's own ten-year window, before interest unless noted. Estimates overlap (some laws extended earlier ones), so they can't be added up. Headline figures: CRFB, "From Riches to Rags," Jan 2024. Each law's source is linked in its name.
Under the hoodDownload the technical methodology (PDF): every equation, assumption and source ▸
Each maturity's monthly yield, with its median line projected forward. Weighted by how much the government owes at each maturity.
There are no knobs to turn. Each assumption is set to the best figure the evidence supports, and the uncertainty that remains is shown openly as the five projection paths.
It starts from the Congressional Budget Office's numbers for revenue, spending and the deficit before interest, and from the latest market data, refreshed daily: Treasury yields from the Federal Reserve Bank of St. Louis, debt held by the public and the average rate on that debt from the Treasury.
Choose "Official" and the model reproduces CBO's long-term outlook closely: debt near 166% of GDP and interest near 36% of revenue in 2056, against CBO's 175% and 37%.
Sources
Every change to the model that can move a date, with its version number. Track-record rows recorded under an earlier version are marked.