A TALE OF TWO DEBTS
Dickens opens A Tale of Two Cities in 1775, the year the American war began. That war was financed twice over, once in London and once at Versailles, and eventually settled at Paris in 1783 at ruinous expense to Great Britain and France.
Six years later one of those states summoned an assembly it had not convened since 1614 to resolve a fiscal problem and did not survive the attempt. The other had accumulated a far larger burden relative to its economy and paid every cent of it. Dickens does not claim the English were better. His England is complacent, disorderly and violent. The distinction between the two was not virtue, and it was not the size of the debt either, which is where intuition fails first.
On the eve of the Revolution, French debt stood at roughly 65% of national income and British debt at roughly 135%. France collapsed carrying half of what Britain carried safely, and Britain went on to carry far more: by 1820 it stood near 175%.
The Debt-to-GDP ratio, the most common debt sustainability measure, can be reduced in only two ways. Britain worked mostly on the denominator, growing the economy faster than the debt, while repaying about a fifth of the nominal amount, until the ratio fell below 30%. France focused on the numerator and got rid of the debt itself: it printed paper money until the paper was worthless, sold off the land it had seized from the Church, and in 1797 simply cancelled two thirds of what it owed. That is how France reached 9% by 1802.
The deeper difference was maturity. By 1815 two-thirds of British debt consisted of 3% coupon perpetual bonds and therefore with no rollover obligation. Consequently, the state owed a coupon forever and never a principal repayment. Conversely, France funded itself increasingly with floating debt, short paper refinanced continually, until it could no longer be refinanced and was converted by decree into long-term obligations. This occurred in 1759, 1770 and again in 1788. The crisis of 1789 was not a crisis of how much France owed. It was a crisis of when.
The Price, Not the Quantity
Let’s fast forward to today and have a look at the winner of the Independence War: the US Congressional Budget Office projects Federal debt held by the public increase from 101% of GDP this year to 120% by 2036, passing the wartime peak of 106% set in 1946. Over the same period the primary deficit falls, from 2.6 to 2.1% of GDP, while net interest rises from 3.3 to 4.6 per cent. Washington is not proposing to borrow more recklessly. However, it will be forced to pay considerably more for what it has already borrowed.
That price has been set publicly over the past fortnight. On 12 August the Treasury sold 10-year notes at the highest cost since 2007; on 13 August it sold USD 25 billion of 30-year bonds at 5.22%, the highest yield at any thirtyyear auction since 2001. Five weeks earlier the thirty-year had cleared at 5.06% with strong demand. The long bond has since reached 5.31%, the highest since 2007. No coupon was missed. The market is certainly not pricing default.
If not default, then what? A long yield decomposes into the average short rate expected over the life of the bond, plus the compensation demanded for locking the money up. That second component, the term premium, is model-dependent. The FED's measure stood in August 2020 at -0.66%, at a time when investors paid for the privilege of holding duration. With the July level being at +0.87%, this is no longer the case.
There is little consensus on the causes: some point to a less predictable Federal Reserve communication framework under Chair Kevin Warsh; others to rising national expenditure on themes such as healthcare and defence; and a growing number of analysts to a secular deanchoring from the dollar and from US debt. The debate will not be settled soon. What matters is the Treasury's reaction.
The Shortening
he government must raise whatever it needs. It sells notes and bonds on a published schedule, and anything those sales do not cover has to be raised in Treasury bills, which mature within a year. Bills reached 21.7% of the debt outstanding at the end of April, close to where they have sat for two years. The Treasury Borrowing Advisory Committee, whose members are drawn from the banks, dealers and asset managers on both sides of the market, has named as its longerterm target between 15 and 20%. The Committee first judged an elevated share appropriate in 2020, when borrowing needs were extraordinary, and said then that it should come down over time to leave Treasury room in a future crisis. Six years later it has not come down. Reducing it would mean selling more longdated debt and pushing long yields higher still, and there is a congressional election in November.
Bills mature within a year and must then be replaced by new bills at whatever rate prevails. With a fifth of the debt in that form, a rise in short-term rates reaches the interest bill within twelve months, and the budget with it. The same weakness had been breaking Versailles for thirty years, through three forced conversions, and the American war finished it.
On 19 August the Treasury doubled the ceiling on buybacks of long-dated debt from USD 2 billion to at least USD 4 billion, with the acknowledged purpose of holding long yields down. The advisory committee had said buybacks were appropriate for liquidity but not for altering the maturity profile. The thirty-year yield fell 9.2 basis points on the announcement, then returned to 5.27% within two days. Reaching for the maturity profile rather than the deficit is not new. In 1951, released from a decade of pegged yields, the Treasury persuaded holders of medium-dated bonds to swap them for non-marketable paper running nearly thirty years, so that the supply would not have to be tested in the market. Then it lengthened, since 2020 it has shortened. Either way, what the Treasury adjusts is the maturity of the debt, never the size of it. The move at the long end is a price for duration, not a bet on policy rates, and a divergence shows it. Soft inflation, weaker employment and falling retail sales should have pulled the whole curve lower. The front end obeyed, with the two-year at 4.18 %. The long end did the opposite. A second, separate signal points at the currency rather than at the maturity. The dollar index fell 0.87% last week while gold rose more than 5% to a 3-month high near USD 4 636. Investors are not only demanding more yield; they are also questioning the strength of the dollar in which that yield will be paid. Both readings could be wrong, because rising long yields are not always a warning. They also rise when growth accelerates. If artificial intelligence delivers measurable productivity gains sooner than expected and capital expenditure stays elevated, higher real yields may simply price a faster economy. That is the British path in modern form: in the Debt-to-GDP equation the denominator grows and the ratio consequently improves. The case is not dismissed. It requires a productivity outcome not yet visible in the data, and the fiscal projections do not depend on it.
Who Pays?
Suppose it does not occur. A state with a debt it cannot grow out of and will not tax its way out of has one remaining course, and the United States has taken it before, within living memory.
The Federal Reserve went down this road from April 1942 until March 1951, willingly at first. To fund the war cheaply, the Treasury issued bills at 0.375% and bonds at 2.5%, and the Fed made those rates possible by standing ready to buy the paper in the market at prices consistent with them. The war ended and the long-bond ceiling did not. The Fed raised the bill peg in July 1947, but the 2.5% cap on long-term bonds held and enforcing it obliged the FED to absorb bonds instead. So, the buying continued while inflation ran at 17.6% in the year to June 1947 and 9.5% in the year to June 1948; a holder of the 2.5 per cent long bond lost more than a fifth of his purchasing power across the two years.
The result was a debt neither repaid nor defaulted upon. It stood at 106% of GDP in 1946 and near 23% of GDP by 1974. Primary surpluses did part of the work; the rest came from inflation and from interest rates held below it. Every coupon was honoured. The bondholder was paid.
Diluting a debt on that scale required two things: a debt too large to refinance comfortably at market rates, and a government facing a bill it was unwilling to fund by taxing. In 1942 the bill was a war. Both conditions are in place. The third element, the bill, arrived in February.
The Strait of Hormuz crisis prompted the White House to ask Congress for USD 87.6 billion on top of a record USD 1.5 trillion defence request. The amount is not the point, and it never was in 1942 either. What matters is that a government which will not raise taxes and cannot cut has to find the money somewhere, and there are only two parties who can provide it: the taxpayer and the bondholder. Taking it from the first requires a vote. Taking it from the second does not. Choosing between them is a political act, and this is an election year. France did not fall because i ts debt was unpayable. It fell because it could not agree who would pay.
This letter began with two ways to reduce debt. It ends with two ways to hold one. As investors, we hold short maturities, and we avoid the long end. Yields at the front end are high enough to absorb a good deal of further rise before income turns into loss, and no such cushion exists at thirty years. Where the next decade takes the American debt is not something we can know, and 1789 is a reminder that those closest to the question rarely do. Being paid to wait, and being free to change our minds each time the paper matures, is the position that does not require us to.
Download the article in PDF format
