The consensus response to the administration’s growth argument is that it is arithmetically impossible. That response is too quick. The arithmetic is not impossible. It has been done before, within living memory, in this country, and the mechanism that did it was mostly not legislation. What makes the current version unlikely is not the math. It is that the math requires three separate things to go right in sequence, and the last time they did, two of them were accidents.
The gap is smaller than the headline number suggests
Interest expense over the trailing twelve months is running near $1.4 trillion against $40 trillion of gross debt — an effective rate of roughly 3.5%. If AI-driven productivity gains are real and large, nominal GDP running 5% to 6% is not a fantasy number. That is roughly 3% real plus a 2.5% price level, which is close to what the late 1990s actually delivered.
Against a 3.5% effective cost, nominal growth in that range clears the hurdle by 150 to 250 basis points. On a $40 trillion stock, that is $600 billion to $1 trillion a year of ratio relief before a single spending decision is made. The debt does not shrink. The denominator outruns it. That is the entire trick, and it is a real trick, not a rhetorical one.
The passive mechanism is the important part
The interesting feature of the 1990s fiscal turnaround is how little of it was decided. The 1993 tax increases contributed roughly 0.7% of GDP in revenue — about one tenth of the total improvement. The rest came from an economy that grew faster than anyone had modelled and poured receipts into the Treasury that nobody had projected.
The numbers are worth stating plainly. Between 1992 and fiscal 2000, receipts grew at 8.0% a year while outlays grew at 3.3% a year, less than half the 7.3% pace of the preceding twelve years. Federal spending as a share of GDP fell for eight consecutive years and reached its lowest level since 1966. The unified balance swung from a deficit of 4.5% of GDP in 1992 to a surplus of 2.3% in 2000 — a 6.8 point move. Debt held by the public went from 49.4% of GDP in fiscal 1993 to 33% in fiscal 2001.
And the line that matters most for the current argument: net interest was $198.7 billion in 1993 and roughly $210 billion by 2001. Essentially flat in dollars while the economy grew by half. That is what “receipts growing into a fixed spending base” looks like when it works. Nobody cut the interest line. It simply stopped mattering, because everything around it got bigger.
The transmission channel was capital gains. A booming equity market generated realisations that repeatedly exceeded CBO’s projections. The current cycle has an obvious analogue in the concentration of index gains in a handful of names whose holders will eventually sell.
Condition one: the productivity has to be real
Nonfarm business productivity in the late 1990s ran 2.1% on a four-year average and 2.8% in the strongest year, against roughly 1.25% from the 1970s through the early 1990s. That acceleration is the whole thing. It is what let receipts grow at 8% while outlays grew at 3.3% without a fight.
The claim on the table is that AI delivers a comparable step-change. The claim is plausible. Enterprise deployment is broad, the capital is committed, and the displacement is already visible in specific labour markets. What has not happened is the acceleration showing up in the aggregate productivity statistics. Until it does, the fiscal case rests on an assumption rather than a series. That is not a reason to dismiss it — the 1990s acceleration was also invisible in the data until it was in the data — but it is the condition, and it is unmet as of today.
Condition two: it has to arrive before the effective rate converges on the marginal rate
This is the condition the 1990s did not have to satisfy, and it is the one that bites hardest now.
The 3.5% effective rate is a legacy artefact. It reflects a decade of sub-2% coupons still sitting in the book, and it is rising toward the marginal rate mechanically as that book rolls. Weighted average maturity on marketable debt is near 70 months. The thirty-year has been printing above 5.3%. Every month that passes converts a little more of the cheap stock into expensive stock, with no policy decision required and no way to stop it.
So there are two convergences running against each other. Productivity converging upward toward the level that widens the tax base, and the effective rate converging upward toward the marginal rate. The fiscal case wins only if the first one arrives first. Clinton inherited debt at roughly half of GDP and a falling rate environment. The current setup is the inverse on both counts.
Condition three: Congress cannot spend the receipts
The part of the 1990s story that gets least attention is that the restraint was institutional. The Budget Enforcement Act caps and PAYGO were in force, and the GAO’s own retrospective credits them alongside growth for the surpluses. When those controls expired, the deficits returned immediately — accompanied by tax cuts, new spending, and capital gains receipts that came in below projection.
Nothing equivalent is in force now. There is no cap regime, no PAYGO with teeth, and no successor under discussion. A surge in unexpected receipts arriving into that environment does not automatically go to the debt. It goes wherever the legislature sends it. The 1990s surpluses required a mechanism that made spending the windfall procedurally difficult, and that mechanism was removed twenty years ago.
What would confirm it
Three conditions, each individually plausible. The productivity gains are a reasonable bet. Some of the debt was issued long enough ago that the rollover clock has years left. And Congress has occasionally restrained itself. Jointly, they are unlikely — not because any one is far-fetched, but because they have to hold simultaneously and they are not correlated in the helpful direction.
The series that settles it is not GDP and not the debt total. It is the wedge between receipts growth and outlays growth in the monthly Treasury statement. The 1990s ran that wedge at 8.0% against 3.3% for eight straight years. Anything approaching a sustained four-point gap would mean the mechanism is live and the story is real. Anything narrower means the growth is arriving and being spent, which is the outcome the arithmetic cannot survive.
