The national debt is usually pictured as a balloon, filling toward the day it bursts. It is the wrong picture. The debt is a balloon with a slow leak. It never pops and it never gets repaid. The air goes out of it a little at a time, into prices, and out of whoever is holding dollars when it does. Nobody is told.
The conditions that have kept that leak painless for 85 years are wearing out. The wearing out is measurable, and it has already begun.
Start by putting down the $40T. It is the least informative number in the entire discussion.
It took this country 192 years, Washington to Reagan, to borrow its first $1T. The most recent $10T took 4.5, which is $6B a day, weekends included. Even that is not the number that matters.
Two Rates
$100T of debt against an $80T economy is a lighter burden than $40T against $20T. Size means nothing without the thing paying for it.
The comparison that decides everything runs between two rates. Economists label them r and g.
r is the average interest the government actually pays across every bond it has outstanding. Not the Fed's policy rate, not the ten-year yield the market charges on a 10-year loan, but the blended cost of the entire pile. By that measure it is 2.75 percent today.
g is the speed of the economy in nominal terms, meaning real growth plus inflation. If output grows 2 percent and prices rise 2.5 percent, g is 4.5 percent, and all of it counts, because the Treasury collects dollars rather than units of output. That is about where it is running.
As long as g sits above r, the economy grows faster than the interest accrues. The debt can climb in dollars every year and still shrink against the thing that services it. No repayment required. You outrun it.
Put numbers on it. The economy grows 4.5 percent a year and the debt pays 2.75. The debt can grow 2.75 percent a year forever, with nothing ever repaid, and still be a smaller share of the economy every year. Nothing is paid off. The thing the debt is measured against grows faster than the debt does, and 2.5 points of that 4.5 is not more output. It is prices. That is the leak: the air leaves through the price of everything, out of whoever was holding dollars.
Let r climb above g, a crossing, and the direction reverses. Interest then compounds faster than the economy expands. The ratio rises under its own power, and it keeps rising even if Congress balances the budget before interest the following morning, because the debt has stopped needing a deficit in order to grow. It grows because it is big and expensive.
Run the leak backwards and prices fall, the debt gets heavier instead of lighter, and every borrower in the country carries it at once. 1929 to 1933 is what that costs.
85 Years
I divided federal interest outlays by the debt outstanding at the start of each fiscal year, and set the result against nominal GDP growth, running 1941 through 2025. 85 years of r-interest against g-growth.
r-interest has been above g-growth in 15 of those 85 years. 18 percent of the time. So the comfortable state, the one where growth outpaces interest, is not some fortunate accident. It is the normal working condition of the American economy across four generations.
The uncomfortable state also refuses to stick around. The longest unbroken run of r-interest above g-growth in the whole record is 3 years, 1985 through 1987. Everything else lasts 1 or 2 years before flipping back.
Ten of the fifteen are recession years.
The crossing rarely happens because interest rates surge. It happens because growth falls out from under them. Nominal GDP shrank 2 percent in 2009 while the interest bill kept accruing at 3.5 percent. Growth hit negative 0.8 percent in 2020.
So the event that tips this country into compounding is not a bond market revolt or a failed auction. It is a garden-variety recession, and those turn up about every 10 years.
A crossing is not a pop. It is the leak widening, because a ratio rising on its own gets settled the only way it ever has, in prices. Every previous crossing closed inside 3 years because revenue covered spending before interest, so nothing pushed against the recovery. The next one lands on a budget where it does not.
The Peg
Between 1941 and 1951 the Federal Reserve pegged Treasury yields. Bills, the short loans, at 0.375 percent, and long bonds at 2.5 percent. The Fed promised to buy however many bonds were required to defend those prices, which meant handing over control of its own balance sheet in exchange for guaranteeing the Treasury cheap money to fight a war.
In 1942 the average rate across the entire federal debt was 1.83 percent while the economy, in dollars, grew 28 percent. Across the full decade r never got above 1.93 percent.
Federal debt fell from 114 percent of GDP in 1945 to 64 percent by 1955. Nearly halved inside 10 years.
Not a dollar of it was repaid. Growth and inflation did the entire job.
So when somebody tells you the country will simply grow its way out of this, they are describing something the United States has already done, carrying a burden about as heavy as today's, inside living memory. The bonds that made it possible were sold to 85 million citizens, and they paid for it.
Two features of that decade do not carry over.
The peg ended, and ending it required a formal treaty between the Treasury and the Fed, the Accord of March 1951. Inflation ran 14 percent in 1947. The debt got inflated away and the people holding the bonds paid the bill in purchasing power. Somebody always does.
And beneath all of it, the government was running roughly balanced primary budgets, meaning revenue covered spending before interest entered the picture. The favorable gap had nothing pushing back against it. Pure tailwind.
Today that gap is still favorable, about 2 points, meaning growth runs about 2 percentage points ahead of the interest rate. The deficit before interest runs about 2.8 percent of GDP. The two very nearly cancel.
Which means the tailwind is real and all of it is used up standing still. Back in 1950 the gap ran 7 points wide with nothing opposing it.
It is narrowing, too. The gap was 9.7 points in 2021 and 2.2 in 2025. 7.5 points handed back inside 4 years.
The Rollover
The government does not reprice when the Fed moves. It reprices when bonds mature.
About $8T of Treasury debt comes due over the next 12 months, $1 of every $5 outstanding, about $22B every day. Each of those dollars is refinanced at whatever the market charges that morning. Everything else keeps paying the coupon, the rate fixed on the day it was sold,, in some cases for another 30 years.
That is why the average rate across $40T sits under 3 percent during a year when the ten-year Treasury yields 4.80 percent. Bonds sold in 2020 and 2021 under 1 percent are still on the books, still paying under 1 percent, and they will keep paying it right up until the day they mature and not an hour before.
A rate rise therefore reaches the Treasury in instalments across years, which is why the damage from a rate cycle is understated while it is happening.
Even if rates never move again, r is rising.
New debt today costs about 4.2 percent. Existing debt averages 2.75. Every maturing bond from the cheap years gets swapped for an expensive one automatically, with nobody at the Fed deciding anything. That drift alone, 2.75 climbing toward 4.2, carries close to $600 billion a year in additional interest and arrives across roughly 5 years.
Nobody votes on it. It happens $22B at a time, every day, without comment.
Who sets the number on any given day is an auction, most business days, with 26 firms required to bid.
The Comfortable Case
The case for not worrying deserves to be stated at full strength.
No threshold exists. The celebrated study putting the danger line at 90 percent of GDP was found to contain a spreadsheet error and does not survive. Japan carries something like 250 percent of GDP and has been past every proposed limit for 25 years without incident. Every previous generation predicted collapse and every previous generation was wrong. And a government borrowing in a currency it issues cannot be forced into default, because it can always create the dollars to pay.
Every one of those statements is true. Not one of them means what it is usually taken to mean.
The breakdown has never once been "we cannot pay." It is always "we paid, and the money stopped meaning much."
What does it mean that the government cannot be forced to default? Set its bond beside a railroad's.
The air leaving the balloon has to go somewhere, and the clearest way to see where is a glass of cranberry juice. Every time somebody takes a sip, it gets topped back up with water. The level never drops. Nobody is ever told there is less juice, and for a long while it still tastes like cranberry. Look at it. The glass is full.
Now run that for 80 years.
The person who poured the first glass knew what cranberry juice tasted like. Their children got it at half strength and were told stories about how it used to taste. Their grandchildren got it at 20 percent and have never once tasted the original. To them, that pale pink water is simply what cranberry juice is. It is not a diminished thing. It is the thing.
This is the part that makes the dilution nearly impossible to stop. You cannot notice a change you have no reference for. A generation that has only ever known cheap money, rising asset prices and a currency that quietly loses a third of its value every 20 years is not being fooled. They are describing their own experience accurately. Their normal is the diluted cup.
The dilution does not need anyone to be deceived. It only needs everyone to be new.
Then one day somebody takes a sip and it is water. Nothing broke on that particular day. What changed is only that somebody finally noticed, and the noticing is what makes it feel sudden. The leak was the same size that day as the day before. Only the reference changed.
That glass has a physical counterpart you can still buy by the roll. In 1965 the silver came out of American coins by act of Congress. Four quarters from the year before are worth about $46 today. Four from the year after are worth one dollar: Four Quarters, One Year Apart.
So the absence of a threshold offers no comfort. It only tells you that whatever you are watching for will not arrive labeled as a number.
Where the Air Goes
So the debt is not a balloon under pressure. Nothing bangs. There is no morning when the country wakes up bankrupt, no default notice, no headline.
It is a balloon with a slow leak. The air goes somewhere.
Debt is a claim against future output.
Output is not an abstraction. It is somebody's Tuesday. It is the hours worked by a person who is eight years old right now, the wages they will earn in their forties, and the share of those wages they will hand over in tax. Every bond sold this week is a small binding assumption about that eight year old's working life, priced and traded before they can spell the word bond.
They never signed anything. Nobody asked. They inherit the arrangement the way they inherit a language, and by the time they are old enough to read the terms, the terms are set and the money is spent.
Claims can be printed without limit. Output cannot, because output is people, and there are only so many of them working only so many hours. When the claims badly outrun what those hours can produce, the correction does not show up as a default. It shows up in the price of things.
Which means the debt never vanishes. It gets paid. The only live question is by whom, and the list has 4 names and no fifth. Taxpayers pay through higher taxes. Beneficiaries pay through reduced spending. Savers and bondholders pay through inflation. Foreigners pay through a weaker dollar.
Taxpayers are the only payer on that list handed a form and told the amount. How that form reached 50 million people.
Printing is not an escape from that list. Printing is the third name on it. Cut the currency's value in half and you cut the real debt in half, along with every pension, wage, savings account and bond denominated in it. A transfer, not a solution.
And this particular transfer is remarkably dependable. Since 1947 the American price level has risen by a factor of 14.4. Across 78 years, prices have fallen in exactly 3 of them: 1949, 1955, and 2009. 4 percent of the time, and each occasion was a crisis.
Prices essentially always rise. That is not a policy failure, it is the mechanism. Nominal growth contains inflation, so a meaningful share of what people mean by growing our way out is inflating our way out. That is how the war debt was settled. It is very likely how this one gets settled.
The cost of that is not collapse. It is 3 quieter losses.
Interest already claims about 20 cents of every dollar of federal revenue and is headed toward 30. The money is not destroyed, but it is spoken for before anybody votes on anything.
Resolution by inflation works beautifully for a country and terribly for a person. The nation will be fine. The retiree holding a 30-year bond issued at 1.8 percent will not be.
Whether the house, the paycheck and the beef rise with it is its own question, and the answer is no, not together. A House Used to Cost 6,900 Hours.
And the third, which I weight highest: in 2008 and again in 2020 this country answered a crisis by spending enormously and instantly, because it had the room to. Heavy debt at high rates spends that room down. The cost of debt is not the debt. It is what you cannot do when the next thing arrives, and something always arrives.
The favorable gap that carried this country through 85 years is down to about 2 points from nearly 10 just 4 years ago, and the deficit before interest consumes all of it. The average rate on the debt drifts upward toward the cost of new money whether anyone acts or not. An ordinary recession flips the comparison immediately, and ordinary recessions arrive about every decade.
Japan held its borrowing costs near zero for 25 years and was fine the entire time, right up until the market repriced it anyway. Nothing broke. It simply became far more expensive, very quickly, after decades of not.
I cannot give you the year. There isn't one, and anyone offering you a date is guessing.
What I can tell you is that the debt does not pop. It leaks, it has started leaking, and the air has to go somewhere.
Somebody breathes it.
Treasury also publishes an average coupon across every outstanding security, which runs nearer 3.4 percent; I use the interest-outlays measure throughout because it can be computed the same way back to 1941. Figures: national debt and interest outlays from the US Treasury and the Federal Reserve Economic Database; r calculated as federal interest outlays divided by gross federal debt at the start of each fiscal year, set against calendar-year nominal GDP growth, 1941 to 2025; deficit and rate-sensitivity projections from the Congressional Budget Office and the Committee for a Responsible Federal Budget; Treasury yields and ten-year ten-year forward rates as of September 2026.
This is general economic writing and not financial advice. Nothing here is a recommendation to buy, sell or hold anything, and none of it accounts for your circumstances.