Fiscal space is created by the denominator, spent at the roll, and bounded by nothing the statistics publish

Fiscal space is not a published series. It is a relation between four things: the stock a government owes, the flow that adds to that stock each year, the size of the economy the stock is measured against, and the price at which the stock is refinanced as it matures. Three of the four have annual observations in this brief. The fourth has a market yield standing in for it and no series at all for what a treasury actually pays. A structural view of fiscal space has to be built out of the three that are measured, and honest about the join where the measurement stops.

Where this is written from

The vantage fixes everything that follows, so it belongs at the front. I am an analyst writing inside The Cefa Brief, working from published statistics and public documents and from nothing else: annual series redistributed by FRED and retrieved for this site on 19 September 2026, together with each agency's own account of what its series counts. I hold no position in a finance ministry, no access to a debt management office and no sight of an issuance calendar.

That position reads one thing well and is blind to several. It can test whether a claim about national aggregates is consistent with the published record, and it can insist on the unit a figure is stated in, which is where most of the errors in this field begin. It cannot see the maturity profile of the outstanding stock, the average coupon on it, the share falling due in any given year, or the interest a treasury paid. None of those is in these files. Where the argument needs one, the absence is named and the argument stops there rather than carrying on over an estimate.

What creates it happens in the denominator

Start with the stock. Federal debt held by the public and intragovernmental holdings combined runs 28.7, 30.8, 32.7, 35.3 and 37.1 trillion dollars across 2021 to 2025, an increase of 8.4 trillion, or 29.3 percent on the 2021 figure. Over the same five observations federal debt as a share of GDP runs 120.9, 118.4, 117.7, 120.4 and 120.7 percent. The stock added 8.4 trillion dollars and the share ended 0.2 points lower than it began. Across the full six readings, which start at 122.3 percent in 2020, the share never leaves a band 4.6 points wide.

Room of that kind is made in the denominator, and the denominator is measured in current dollars. Gross domestic product in current dollars runs 26.1, 27.8, 29.3 and 30.8 trillion across 2022 to 2025, a rise of 4.7 trillion or 18.0 percent. Output in chained 2017 dollars runs 22.08, 22.72, 23.36 and 23.85 trillion over the same years, a rise of 1.77 trillion or 8.0 percent. Dividing the first growth factor by the second, 1.180077 by 1.080163, leaves 1.092499, a price component of 9.2 percent. Consumer prices, on a 1982-84 base of 100, went from 258.9 in 2020 to 322 in 2025, a rise of 24.4 percent. A price level that climbs shrinks a debt ratio without anyone having decided that it should, and the relief it gives is indistinguishable in the ratio from the relief that real output gives.

What destroys it is the price of the roll

A debt stock is not priced once. It is priced again every time a maturing security is replaced, and the market rate against which that replacement is priced has moved a long way. The ten-year Treasury constant maturity yield runs 0.89, 1.45, 2.95, 3.96, 4.21 and 4.29 percent across 2020 to 2025; the last reading is 4.82 times the first. The effective overnight policy rate runs 0.38, 0.08, 1.68, 5.02, 5.14 and 4.21 percent over the same six years. Debt issued into the front of that window is cheap. Debt issued into the back of it is not. The composition of the stock moves from one to the other on a schedule, and the schedule is the thing that decides when a higher yield becomes a higher bill.

Here the record runs out, and it runs out exactly where the argument would most like to continue. Nothing in this brief records what a treasury pays in interest, what the average rate on the outstanding stock is, or how much of that stock falls due in any given year. Those three numbers govern the speed at which a repricing reaches a budget, and without them the defensible statement is a narrow one: the price of new long borrowing rose from 0.89 to 4.29 percent between 2020 and 2025, and a stock that grew by 8.4 trillion dollars over most of that period is being refinanced somewhere along that path. How much of it has already repriced is not a question these files can answer, and I will not put a figure where the answer should be.

The flow, and an arithmetic that does not close

The annual flow tells a different story from the ratio. The federal surplus or deficit runs minus 3096, minus 2774, minus 1374, minus 1687, minus 1815 and minus 1775 billion dollars across 2020 to 2025. Between 2020 and 2022 the deficit narrowed by 1722 billion. Since 2022 it has sat inside a band 441 billion wide, from 1374 to 1815 billion, and the 2025 reading is still 401 billion above the 2022 one. One check on those numbers is worth doing in public because it fails. Five annual stock observations bracket four annual changes, so the flows to set beside them are the deficits for 2022 to 2025, which sum to 6651 billion dollars. Across those same four intervals the stock rose by 8.4 trillion. The stock grew by about 1.7 trillion dollars more than those four deficits sum to, and this brief holds nothing that reconciles the difference: the deficit series is an annual Treasury figure, the stock series is an annual observation whose formation the file does not describe, and each stock point is rounded to one decimal place in trillions. A smaller version of the same gap sits inside the ratio. The 2025 stock of 37.1 trillion divided by current-dollar GDP of 30.8 trillion is 120.5 percent, against a published share of 120.7. Neither difference is large. Both are named here rather than explained, because explaining them would take documents this vantage does not hold.

What it forecloses, and what it does not

Nothing in the published record marks a ceiling. The share of GDP has six readings between 117.7 and 122.3 percent and there is no line drawn anywhere across them; wherever the constraint sits, it is not in the data, and a writer who puts one there has invented it. Foreclosure works through scale instead. The 2025 deficit of 1775 billion dollars is 1.64 times the 1081 billion of research and development inside GDP that year, and 1.36 times the 1309 billion of nonresidential construction spending. Both of those are economy-wide totals covering every sector rather than federal programmes, and both are stated at a seasonally adjusted annual rate, which makes each point a yearly figure and not a month's; Three agencies, three clocks, and the comparison stands for scale alone. What it establishes is that one year of federal borrowing is larger than the entire national effort in either of the two categories whose returns arrive latest.

The only durable source of fiscal space is the one that takes longest to move. Real output per person runs 61127, 64758, 66026, 67411, 68681 and 69749 chained 2017 dollars across 2020 to 2025, a gain of 8622 dollars or 14.1 percent. That is the line which has to carry a stock that added 8.4 trillion dollars across five observations, and it is the line least responsive to anything settled inside a budget year. My position is that fiscal space should be reported as four figures held apart rather than as one share: the stock, the flow, the denominator with its price component shown, and the rate at which the stock is being refinanced. Compressed into a single ratio it conceals the plainest fact in the window, which is that the share fell in the years the borrowing was heaviest. What this vantage cannot do is say whether the space is running out. That needs the maturity profile, the coupon on the existing stock and the interest actually paid, and none of the three is in these files. The limit is real. It is simply not one I can locate from here.