← Back to the Explorer

Fiscal Explorer — how it works

It reproduces CBO’s published baseline exactly, then lets you change the fiscal and economic assumptions and re-solve for what the debt path would look like.

What it reproduces (exactly)

The starting point is not a model of my own. It is CBO’s February 2026 Budget and Economic Outlook baseline, extended with CBO’s Long-Term Budget Outlook. Debt held by the public runs from 99.0% of GDP in 2025 to 120.2% in 2036, 144.5% in 2046, and 175.1% in 2056; by 2056 net interest is 6.9% of GDP and the deficit is 9.1%. Every history and forecast series is built directly from CBO and OMB source workbooks, and matches CBO to the dollar. When you open the tool and touch nothing, you are looking at CBO’s numbers. The one exception is the GDP of years that are over: for those the tool divides by BEA’s actual fiscal-year GDP, revised in September 2026, which is why 2025 reads 99.0% where CBO’s February estimate gave 99.4%. From 2026 on, the GDP is CBO’s own.

The fiscal gap: 2.37% here, 2.33% in Auerbach and Gale

The Explorer reports a fiscal gap of 2.37% of GDP: the permanent cut in the primary deficit, starting in 2027, that would leave debt in 2056 at the share of GDP it was at in 2025. The number usually quoted alongside it is 2.33%. This page used to explain the difference as a calendar-year convention. That was wrong twice over, so here is the correction.

Nothing in the Explorer is measured on a calendar year. The debt, the GDP denominator, the deficits and the interest rates are all fiscal-year series, and the gap is computed fiscal year by fiscal year. The one calendar-year input anywhere near it is CBO’s projected growth of nominal GDP, which its long-term economic supplement reports for calendar years; substituting fiscal-year growth, computed from the tool’s own GDP levels, changes the gap by 0.001 point, and in the direction away from 2.33%.

And 2.33% is not CBO’s figure. CBO’s February 2026 workbooks — the ones this tool is built from — carry no fiscal gap at all, and the measure last appeared in its long-term outlook data a decade ago. 2.33% is Alan Auerbach and William Gale’s, from their March 2026 update on the federal budget outlook: the same idea, a permanent change starting in 2027 that holds debt at its current share of GDP under current law. They compute it on CBO’s extended baseline, so from 2026 on the debt path is the one the Explorer uses — their published table and this tool’s match to within 0.002 points of GDP in every year from 2026 to 2056. Two things differ. Their target year is 2055, not 2056: take the Explorer’s own arithmetic out to 2055 and it reads 2.35%. And the level they hold debt at is CBO’s estimate of its 2025 share of GDP, 99.4%, where the Explorer holds the 99.0% that BEA’s actual GDP gives; holding a lower level takes a slightly larger cut. Hold the Explorer to 99.4% as well and it reads 2.34%.

That leaves about a hundredth of a point, and it is a third convention: how each year’s adjustment is discounted back to the target year. On their target year and their starting level, counting the year an adjustment is made, rather than only the years after it, reproduces 2.33% exactly — but I have not checked that against their formula, so treat it as the likely remainder and not as the answer. Three conventions, then, two pinned down and one not, and no error in either number.

CBO-exact levers vs “Furman rules of thumb”

The levers come in two kinds, and it matters which is which:

A reader can always tell the two apart, because the Furman levers are named as such on screen.

Discretionary spending: inflation to 2040, then GDP

CBO’s baseline grows discretionary spending — defense and nondefense appropriations — with inflation through 2040, so it falls as a share of the economy, from 5.9% of GDP in 2026 to 4.6% in 2040. After 2040 CBO holds it at that constant share of GDP.

The Explorer keeps that rule when you change the economy. If faster productivity or labor-force growth raises GDP, discretionary spending does not respond before 2040; after 2040 it grows with the larger GDP from wherever it stood in 2040. A faster-growth path therefore still ends with a lower discretionary share than CBO’s, because its 2040 starting point was set by inflation, not by the bigger economy.

The Policy setting lets you replace the rule, separately for defense and nondefense and separately before and after 2040. The choices are inflation (CBO’s rule before 2040), inflation plus population, or GDP (CBO’s rule after 2040); choosing GDP from the start holds spending at its 2026 share of GDP. Two combinations are left out on purpose: spending that grows with inflation plus population, or with GDP, through 2040 keeps doing so afterward, rather than stepping down to inflation or up to GDP. Inflation is the GDP price index and population is the Social Security area population, both from CBO’s long-term economic projections. CBO splits defense from nondefense only through 2036; after that each keeps its 2036 share of the total.

Where it’s modeled, not CBO

Two honest caveats, both flagged in the tool as modeled, not CBO:

How the goal-solver works

Most fiscal tools add up policies and tell you the resulting debt. This one runs the other way: you name a goal — hold debt at today’s level, stay below a ceiling, reach a deficit target — and it solves for the fiscal adjustment needed to hit it, on the economy-adjusted path. A ceiling is treated strictly as a limit to stay under: the solver will tighten policy to keep debt below it, but it will never loosen policy just to push debt up toward a ceiling that isn’t binding.

Sources

The tool is refreshed whenever CBO releases a new Outlook, and carries a vintage stamp so you know which release it’s built on.

Found something that looks off, or have a suggestion? I’d genuinely like to hear it — feedback is welcome. — Jason Furman