The Elastic Perimeter: What Counts as Defence When the Target Binds

NATO’s spending guideline is defined on a reported statistic: defence expenditure “according to an agreed definition” must reach 2% of GDP. Compliance can therefore be produced on two margins, by spending more or by widening what is counted as defence. This paper documents the second margin. Using five successive vintages of the NATO expenditure compendium (2023–2026) and the national accounts counterpart of defence spending (COFOG division 02) for 24 European allies, it establishes three facts. First, reported burdens accumulate at the guideline as it becomes politically binding: in the August 2025 compendium, 14 of 30 reporting allies place their 2025 estimate within [2,00; 2,05] per cent of GDP, six report exactly 2,00, and none reports below the line. Second, the gap between NATO-definition and national-accounts defence spending widens from a median of 0,08 percentage points of GDP in 2014 to 0,45 points in 2024, with 62% of allies exceeding a 0,25-point wedge. Third, the mass at the line partly melts on revision: first estimates in [2,00; 2,05] are revised down by 0,112 points on average and 35% of them end below 2,00, against 0,041 points and 17% just above the band. A fixed-effects gradient links wedge growth to the lagged distance from the guideline. The 2026 shift to a 3,5% core target plus an undefined 1,5% “defence- and security-related” category institutionalises the elastic perimeter rather than closing it.