Tracker · The debt and the dollar

Interest on the national debt

The federal government is paying $1.25tn a year to carry its debt, which is 20.9% of everything it takes in and 3.84% of the economy. That is more than the $1.20tn it spends on national defense, and has been since Q1 2024. It is also less of federal revenue than the 28.6% it took in Q2 1985.

318 quarters since Q1 1947

This is the number that decides where the interest bill goes next, and it is the one most commentary skips. Market yields moved a long way in a short time, and the effective rate has barely moved by comparison, because a Treasury issued at 2% in 2020 keeps paying 2% until it matures. The rise already in train is therefore mostly mechanical rather than a forecast: it arrives as old debt is refinanced at newer rates, over years. The same mechanism works in reverse, which is how the 1990s brought the share down without repaying anything.

Does interest on the debt really cost more than defense?

Yes, on this basis. In Q2 2026 interest ran at $1247bn a year against $1198bn for national defense, and interest has been the larger of the two since Q1 2024, 10 quarters. It briefly crossed once before, in Q1 1998, and fell back. Both figures come from the same national accounts table, so they are genuinely comparable, which is more than can be said for most versions of this comparison.

How much of federal revenue goes on interest?

20.9% of it, or about 21 cents of every dollar the federal government takes in. That is high and it is not a record. The peak was 28.6% in Q2 1985, and today sits at the 76th percentile of the record rather than at the top of it.

Why is the interest bill rising so fast?

Two things at once. The stock of debt grew, and the average rate the government pays on it rose off a floor. The effective rate, meaning interest actually paid divided by debt outstanding, is 3.16% today against a record high of 13.98% in Q2 1982. It moves slowly because most of the stock was issued years ago at older rates, so a rise in market yields reaches the interest bill over a decade rather than in a year.

Is the interest bill at a record?

In dollars, yes, and dollars are the least informative way to look at it because the economy and the budget are both larger than they have ever been. As a share of federal receipts it is 20.9% against a 1985 peak of 28.6%. As a share of GDP it is 3.84%. Both of those are the comparisons worth making, and neither is a record.

What would bring it down?

Lower market rates feeding through as old debt matures and is refinanced, faster nominal growth in receipts, or a smaller deficit. The 1990s did all three at once, which is why the share fell from 23.9% across that decade to 15.6% in the next. None of that required the debt to be repaid.

Does a rising interest bill mean gold goes up?

The record does not show that link. The interest share peaked in the mid 1980s and gold fell for the fifteen years that followed. A fiscal constraint and a gold thesis are separate arguments, and a page that runs them together is selling rather than reporting.