The Century Bond

The problem

The United States owes something on the order of $30 trillion to the public, and the weighted average maturity of that debt is roughly six years.

Read those two numbers together, because together they say something the headline debt figure does not. The country is not carrying a fixed obligation. It is carrying a rolling one. Every year, roughly a quarter of the federal debt matures and must be sold again to somebody, at whatever price the market demands on the day of the auction. Roughly $9 trillion of Treasury securities come due in any twelve-month stretch.

That is not a debt. That is a subscription, renewed continuously, at a rate nobody controls.

The consequence is that federal interest cost is not a policy choice made once. It is re-made every year, in public, in front of an audience. Net interest now runs around $1 trillion a year and has passed national defense as a line item. A one-point move in rates does not affect the debt gradually; it walks through the whole stock in about six years.

Meanwhile the things the borrowing pays for last far longer than six years. A dam, a grid interconnection, a port, a reactor, a rail alignment. We finance century assets with six-year money and then act surprised when the financing terms drive the politics.

Every other proposal in this catalog that involves building something runs into this. № 023 established the two locks that govern the Fund. This one addresses the other side of the balance sheet: not how much we owe, but how long we owe it for.

The proposal

Issue ultra-long Treasury debt, out to 50 and 100 years, and use it to term out the rollover. Not to borrow more. To borrow for longer.

Let this be unambiguous, because the objection is obvious and correct: this proposal does not raise a single additional dollar. It changes the maturity of dollars already borrowed.

How it would work

1. A regular, small, predictable auction. Treasury's own advisory committee has warned for years that a sporadic novelty issue finds no buyers. The answer is a scheduled 50-year and 100-year auction in modest size, quarterly, announced years ahead, with the explicit intention of building a liquid curve rather than opportunistically catching a low. Predictability is the product.

1. Refunding only. Proceeds retire maturing shorter paper. The statute authorizing the issue caps total outstanding debt unchanged by the operation, so this can never become a device for spending more. A century bond that funds new deficits is a different and much worse proposal, and it is not this one.

1. Match it to the asset. Priority use is refinancing debt associated with genuinely long-lived federal capital: transmission, water, nuclear, ports, the federal building stock. A country that will use an asset for eighty years has a defensible reason to pay for it over eighty.

1. A demand side that already exists. Pension funds and life insurers hold liabilities running fifty and sixty years and cannot currently match them with Treasuries, because none exist past thirty. They buy duration abroad or synthesize it with derivatives. This gives them the safest long asset on earth in their own currency.

1. Nothing enters the Fund. Per № 023, lock one: no borrowed dollar ever becomes principal of the American Permanent Fund, and that includes these. This proposal is about the debt's shape, and the Fund is not permitted to touch it. If you read this as a way to leverage the endowment, read № 023 again.

The numbers

Term out $2 trillion of debt from an average six-year maturity into fifty- and hundred-year paper, roughly 7 percent of the public debt, phased over a decade.

The direct interest effect is probably a cost, not a saving. Long bonds normally yield more than short ones. At a term premium of 50 basis points, $2 trillion of extension costs about $10 billion a year in additional interest.

Ten billion a year, to buy what?

To buy the removal of $2 trillion from the rollover queue for a century. The value is not in the coupon, it is in the variance. Consider a rate shock of 200 basis points sustained for a decade. On six-year paper, that shock reaches the full $2 trillion within six years and costs $40 billion a year by the end. On hundred-year paper it reaches none of it, ever. The extension pays for itself, several times, in exactly the scenario that most threatens the fiscal position, and costs $10 billion a year in the scenario where nothing goes wrong.

That is insurance, priced the way insurance is always priced: a certain small premium against an uncertain large loss. We are not going to dress it up as a free lunch, because it is not one.

The honest objections

"Treasury has studied this repeatedly and concluded there is no demand. You are ignoring the people who actually sell the bonds." This is the strongest objection and it is grounded in fact: the Treasury Borrowing Advisory Committee has looked at ultra-longs more than once and advised against, on the grounds that demand is thin, the issue would price poorly, and the liquidity premium would exceed the benefit. Our answer is that every one of those studies assumed an occasional opportunistic issue, which is precisely the format that guarantees illiquidity. Austria, Belgium, Ireland and Mexico have all sold hundred-year paper. The UK ran perpetual consols from 1751 until 2015. The instrument is not exotic; the American version of it has simply never been offered on a schedule. If a committed, calendared program still finds no bid after three years, the program should be ended and this proposal is wrong.

"You would be locking in today's rates for a century. If rates fall, that is a catastrophic error." True, and symmetrical: if rates rise, it is a historic bargain. We do not know which, and neither does anyone selling you a view on hundred-year rates. This is the argument for the modest, phased, calendar-driven size in item 1 rather than a single opportunistic $2 trillion strike. Dollar-cost averaging across a century-long program is the only honest posture when the forecast horizon is a hundred years and nobody has ever had one.

"Argentina sold a hundred-year bond in 2017 and defaulted in three years." It did, and the anecdote deserves its place here rather than in a footnote. It is also the argument for the instrument rather than against it: investors bought century paper from a serial defaulter, which tells you what the appetite looks like for century paper from the issuer of the reserve currency. The relevant caution is different and worth stating: Austria's 2117 bond lost roughly three-quarters of its market value when rates rose. Ultra-long bonds are savagely volatile for the holder. That is the buyer's risk, it is why the buyer demands a premium, and it is priced into the $10 billion above.

"This is financial engineering that changes nothing real. The debt is the debt." Partly fair. The stock is unchanged, and no maturity schedule fixes a primary deficit. But "changes nothing real" is wrong: it converts a variable-rate obligation into a fixed-rate one, and any household that has moved from an adjustable mortgage to a thirty-year fixed understands exactly what was purchased. Nobody claims refinancing eliminates the mortgage.

"Why would a nonpartisan policy lab spend a proposal on debt plumbing?" Because the plumbing determines what is buildable. Every ambitious thing in this catalog is downstream of the country's cost of capital and of the annual political spectacle of refinancing. Boring instruments are how patient countries stay patient.

Sources

• Debt held by the public and weighted average maturity of marketable Treasury debt (treasurydirect.gov, Treasury Bulletin)
• Net interest outlays exceeding national defense outlays; Congressional Budget Office Budget and Economic Outlook (cbo.gov)
• Treasury Borrowing Advisory Committee, recurring assessments of ultra-long issuance and expected demand (treasury.gov)
• Sovereign century bonds: Austria (2017, 2020), Belgium, Ireland, Mexico, Argentina (2017); UK consols, 1751–2015
• Proposals № 023 (The Debt Covenant, whose first lock forbids borrowed principal); № 044 (The Atomic Compact); № 045 (The Permitting Clock)