Does the national debt ever have to be paid back?
Starting from the rollover machinery that refinances maturing bonds, this piece answers the question head-on: why net repayment is historically rare, why that is no licence to borrow without limit, and what the famous 60% and 90% thresholds really are.
GLOBAL · 5 min · Updated 2026-07-17
A government's debt does not work like a household's
When a person takes out a loan, the contract comes with a repayment schedule built around a human lifetime and a finite income. The bank knows the borrower will one day retire and one day die, so it designs the loan to bring the principal home before then. A state starts from a different premise altogether. It has no retirement date and no lifespan, and its power to tax passes from one generation to the next. Sovereign bond markets were therefore never designed around a day on which the principal ceases to exist.
When a government bond matures, the money to redeem it comes, for the most part, from issuing a new bond. This is the rollover. From the creditor's side, sovereign debt is less an obligation to be extinguished than a safe asset that pension funds, banks, insurers and central banks actively want to keep holding. So long as markets trust the state's finances, new buyers queue up as old bonds fall due. That is why national debt is managed not as a bill that must one day be settled in full, but as a balance that keeps revolving.
'Paying it back' means two different things
The first meaning is the redemption of individual bonds. A maturing bond is paid, principal and interest, without exception — the moment that fails is the moment of default. But since much of the money comes from freshly issued debt, redeeming individual bonds does not shrink the total. The second meaning is a net reduction of the stock itself: net repayment. This is what most people picture when they say a country 'paid off its debt,' yet what actually happens, almost always, is the first.
Countries that have sustainedly repaid their debt in net terms are rare in the historical record. The United States famously extinguished virtually all federal debt in 1835 under the Jackson administration — and even there the condition did not last. The realistic path most advanced economies have taken looks quite different: as in the decades after the Second World War, the nominal stock was held steady or even grew, while economic growth and inflation slowly worked the debt-to-GDP ratio down. Not paying the debt off, but diluting its weight by enlarging the denominator.
So can it grow without limit? No.
That debt can be rolled over does not make it free. The first constraint is interest. As the stock grows, so does the share of tax revenue that leaves through interest payments, and with it shrinks the room for schools, welfare and investment. Whenever the interest rate sits above the growth rate of the economy, a snowball effect sets in: the ratio climbs on its own even without new borrowing. Debt that need not be repaid and debt that costs nothing are two very different things.
The second constraint is confidence and prices. The rollover machine rests on the assumption that markets will keep buying. Shake that confidence and yields jump; higher yields make the public finances look worse, which shakes confidence further — a self-fulfilling spiral. A country that borrows in its own currency can lean on its central bank as buyer of last resort and thereby avoid formal default, but the bill is presented to its citizens instead, as inflation and a weaker currency. A country with heavy foreign-currency debt lacks even that safety valve, and can be pushed into crisis far more abruptly.
How much is dangerous? The truth about 60% and 90%
The most famous threshold is 60% of GDP, adopted by the 1992 Maastricht Treaty as a reference criterion for joining the European monetary union. The prevailing view, however, is that this figure was less a product of economic derivation than a political compromise reflecting where member states' finances happened to stand at the time. The other celebrated number is 90%: in a 2010 paper, Reinhart and Rogoff reported that growth slows markedly once public debt passes 90% of GDP, and the finding became a mainstay of the austerity debate.
In 2013, Herndon, Ash and Pollin identified a spreadsheet error along with contested data choices and weighting in that study, and scepticism about the robustness of the 90% threshold has since become widespread in the discipline. The prevailing academic view today is that there is no universal threshold valid for every country. Japan has rolled over a general-government debt exceeding twice its GDP (on IMF definitions) at low interest rates for decades, while some emerging economies have fallen into external-debt crises at far lower ratios. The threshold is set not by a single number, but by a country's structure.
A checklist for the reader: look at structure, not just level
First, watch the trend of the ratio rather than its level. A country with high but stable or falling debt-to-GDP and a country climbing steeply from a low base carry different kinds of risk. Second, set interest costs against tax revenue. The larger the slice of revenue consumed by interest, the more mechanically the government's room for manoeuvre narrows — and this ratio often signals trouble earlier than the headline stock does.
Third, look at who holds the bonds. Debt held mainly by domestic investors and the central bank behaves differently under stress from debt that depends on short-term foreign money, even at identical ratios. Fourth, check the currency of denomination: local-currency and foreign-currency debt follow different grammars of crisis. WorldRealDebt attaches a source, a reference date and an interpolation method to every figure it publishes precisely so that readers can run these four checks for themselves.
Sources and verification
Sources: the fiscal reference criteria of the Maastricht Treaty (1992), the Reinhart & Rogoff (2010) and Herndon, Ash & Pollin (2013) debate, the IMF Global Debt Database, and the WorldRealDebt /methodology/ documentation. Ratios cited in the text are approximate; exact values and reference dates should be checked against each institution's original data.