A Third of the Debt Comes Due Every Year

The Government Accountability Office reported that 33 percent of marketable Treasury debt matures within twelve months, up from 24 percent in 2014. Against total marketable debt of roughly $30 trillion, that is something on the order of ten trillion dollars that has to be refinanced in any given year regardless of what anyone wants, at whatever rate the market happens to be charging at the moment it falls due.

The composition has shifted in the same direction. Treasury bills, meaning debt maturing in a year or less, were 13 percent of marketable debt in 2014 and are about 22 percent now. The Treasury Borrowing Advisory Committee, the body of market participants that advises on exactly this question, has for years suggested a bill share averaging around 20 percent, on the grounds that it offers a reasonable trade off between interest cost and rollover risk. The share now sits modestly above the top of that range and the borrowing need keeps growing. Treasury estimated $739 billion of net marketable borrowing for the July to September quarter and $628 billion for October to December.

The Historical Echo

The sharpest illustration of what short dated debt does under stress is Mexico in 1994, and the mechanism is worth following closely because it is not the one people expect.

Through that year the Mexican government progressively replaced peso denominated Cetes with Tesobonos, short term instruments indexed to the dollar. The switch worked in the sense that it kept foreign money coming in. Tesobonos outstanding rose from $3.1 billion at the end of March 1994 to $29.2 billion by December, and essentially all of it was scheduled to mature during 1995. Meanwhile foreign reserves fell from $29 billion in February to $12.5 billion in early December, then to roughly $6 billion by the end of that month.

Nothing about Mexico's underlying solvency changed in those ten months. What changed was the relationship between how fast the obligations came due and how much cash was available to meet them when they did. The peso's band was widened on the twentieth of December and the currency was floated two days later. A rescue package of about fifty billion dollars followed in January.

The United States is not Mexico in 1994, and the differences are large enough that the comparison would be silly if drawn too tightly. The dollar is the reserve currency, the debt is denominated in a currency the issuer controls, and there is no fixed exchange rate to defend. The transferable lesson is narrower. Short dated debt is not cheaper debt. It is debt whose price is reset continuously, which converts a slow problem into a fast one the moment conditions change.

Where Patient Capital Is Positioning

Two further figures complete the picture. Net interest cost exceeded $970 billion in the 2025 fiscal year, or 3.2 percent of gross domestic product, against $229 billion and 1.3 percent in 2014. In the first nine months of the current fiscal year interest ran to $857 billion, about thirteen percent higher than the same period a year earlier. And in August the Treasury announced it would at least double its long end liquidity support buybacks, from a two billion dollar cap per operation to at least four billion, effective the ninth of September, citing the need for greater liquidity support in the ten to thirty year sectors.

That last item is the one worth noticing. A buyback programme designed to support liquidity at the long end is an acknowledgement that the long end needs support, which is not a statement anyone would have thought necessary about Treasury bonds a decade ago. The market took it well, with the thirty year yield falling nine basis points on the announcement, and there is nothing improper about it. It is simply a new piece of machinery, added because the old arrangement was straining.

For someone thinking in decades rather than quarters, none of this forecasts a crisis and it should not be read as doing so. It describes a financing structure that has become more sensitive to interest rates than it used to be, at a time when the amount being financed is larger than it used to be. Assets whose value does not depend on anyone's ability to refinance, which is what gold, land, energy and the industrial metals have in common, are not a prediction that the refinancing will fail. They are a way of not needing an opinion about it.

The Capital Memo