European security, reconsidered

European security is discussed almost entirely in the language of politics: which government is committed to what, which summit produced which language, which election threatens which arrangement. That register is not wrong, but it has a short half-life. The structural picture underneath moves slowly and explains more, and it is worth setting out separately from the news.

The arrangement as it actually stands

Strip the institutional detail and the continent's security rests on a single arrangement: European states have, for three generations, under-provided their own defence because an external power provided it for them, at a cost that power was willing to bear for its own reasons.

That is not a criticism. It was rational for everyone involved. For the European states it freed enormous fiscal capacity for domestic priorities and removed the security dilemma among themselves, which had been the continent's defining problem for centuries and which the arrangement genuinely solved. For the external guarantor it purchased influence over the foreign policy of a large bloc of wealthy states and denied the region to a rival.

What matters analytically is that this was a bargain with conditions, and the conditions were about interest rather than affection. Which means the durability question is not about sentiment or about any particular administration. It is about whether the interests that made the bargain worth funding still hold.

The defensive-realist reading

Offensive and defensive realism diverge on how much states need, and the difference produces genuinely different European forecasts.

The offensive reading expects a great power to seek regional hegemony where it can, which implies continuous pressure on the continent's eastern approaches and a security competition that does not stabilise. The defensive reading expects states to seek an appropriate amount of power, enough for security, because seeking more triggers balancing that leaves you worse off. On that reading, expansion beyond what security requires is self-defeating, and states that attempt it usually discover this expensively.

I find the defensive version the better description of Europe specifically, for a structural reason: the continent is unusually well suited to defence. It is geographically fragmented, densely populated, institutionally dense, and its states are wealthy enough to arm seriously when they decide to. Conquest and holding are both very expensive there. That does not make aggression impossible, since states miscalculate and the theory explicitly allows for it, but it does mean the returns to it are poor and the balancing response is fast.

The corollary is the uncomfortable half. If defence is structurally favoured, then the deterrent does not require the current arrangement to be preserved in its current form. It requires someone to provide it. The question is only who, at what cost, and how quickly the transition can be made.

The security dilemma on the continent was solved by an outside guarantor, not by the states resolving it among themselves. That distinction is dormant, not closed.

Burden-sharing is a symptom

The recurring dispute over defence spending is usually framed as a matter of fairness or of political will. Structurally it is neither. It is the predictable output of an arrangement in which the guarantee is a public good and the incentive to free-ride is built in.

Under-provision is what the arrangement was always going to produce, and the periodic disputes about it are not a malfunction. They are the mechanism by which the guarantor extracts payment for a service, and they have been recurring for decades under every combination of governments.

The forecastable part is that the pressure is one-directional and slow. The guarantor's strategic attention is finite and increasingly claimed elsewhere; the relative weight of the region in that calculation declines over time regardless of who is in office. European defence spending therefore rises structurally, unevenly, and more slowly than the rhetoric suggests, because it competes with domestic fiscal commitments that are politically far harder to move, and because the free-rider incentive persists right up until the guarantee is actually gone.

The variable that decides it

If I had to name one thing to watch, it would not be spending levels, alliance communiqués, or any election.

It is whether the continent's states converge or diverge in their threat assessment. Alliances follow threat, and the states on the eastern flank, the states on the southern flank, and the states in the interior do not face the same one. Their exposure differs by geography, energy dependence, trade orientation and history. Convergence produces a coalition that can substitute for an external guarantee. Divergence produces a set of states each seeking its own arrangement, and that is the configuration in which the older continental dynamic reappears, because the security dilemma among European states was suppressed by the guarantee rather than dissolved by it.

Everything else is downstream. Spending follows threat perception. Institutional integration follows threat perception. The credibility of any commitment follows from whether the committing state's public shares the assessment.

What the frame does and does not deliver

Three things, none of them dramatic.

The direction of travel is toward more European self-provision, slowly, unevenly, and with fiscal consequences that land on sovereign accounts already carrying commitments that are politically difficult to reduce. That is a slow-moving credit story, not an event.

The tail risk is not invasion. It is fragmentation: the scenario in which threat perceptions diverge far enough that states pursue individual arrangements, and the continent's internal security dilemma reactivates. Low probability, very high impact, and it belongs in a reverse stress test rather than a base case.

And the timing is genuinely indeterminate, which is the standing limitation of the whole frame. Structure tells you which way the pressure runs; it does not tell you when the thing gives way. In practice that means this read is useful for deciding what to hold and useless for deciding when to trade, and the two should not be run off the same analysis.