Europe's rearmament: what the ECB knows and what it doesn't

On 17 August 2026, a member of the European Central Bank's Executive Board stood in front of a room of academic economists in Dublin and worked through two dozen slides on Europe's defence build-up. Philip Lane was not there to move markets. He was at the joint annual meetings of the European Economic Association and the Econometric Society, and he was there to show his working.

The slides matter for what they decline to claim. Europe has committed to the largest peacetime military expansion in its post-war history. At The Hague in June 2025, NATO members pledged to reach 5% of GDP on defence and security-related spending by 2035 — 3.5% on core defence, 1.5% on infrastructure and resilience — replacing a 2% guideline that most of them had spent a decade failing to meet. The political argument for this has been made loudly. The economic argument has mostly been asserted: rearmament will create jobs, revive industry, and pull a stagnant continent forward.

Lane's presentation puts numbers against the assertion. Some support it. Several complicate it. And one slide, titled without euphemism Substantial uncertainty surrounding fiscal multipliers of defence spending, gathers published estimates of what a euro of military spending does to GDP and shows them spanning a range from roughly minus 2.5 to plus 3.5.

Who is spending, and who cannot

The countries raising defence spending fastest are, with few exceptions, the ones best able to afford it.

Lane presents two scatter plots covering the euro area's NATO members. The first sets the change in defence expenditure between 2022 and 2026 against each country's sovereign debt-to-GDP ratio at the end of 2025. The second sets the same spending change against the average government budget balance across 2025 and 2026, with countries grouped by debt level — below 60% of GDP, between 60% and 100%, and above 100%.

The pattern in both is the same. The largest increases sit with Estonia, Luxembourg, Bulgaria, Lithuania, the Netherlands and Latvia. The smallest sit with Greece, Italy, France and Belgium. Debt levels run in almost the opposite direction. The build-up is being led by countries with fiscal room and by countries close enough to Russia to treat the question as immediate — categories that, in the Baltics, happen to coincide.

Greece is the instructive exception. Measured as a share of GDP, Greek defence spending has long been among the highest in the euro area, a legacy of its relationship with Turkey rather than of anything decided in 2022. Measured as a change since 2022, it is among the smallest. High spending and high debt can coexist. High spending growth and high debt mostly do not.

This has a straightforward consequence that the presentation does not spell out. The euro area's collective defence capability is being built disproportionately by its smaller and less indebted members, while its largest southern economies — France and Italy between them accounting for a substantial share of euro area output — are moving slowly. A capability gap that follows a fiscal fault line is a durable one.

The escape clause that made the borrowing possible

Most of this new spending is being financed with borrowing that the European Union has explicitly agreed not to hold against member states.

The mechanism is the national escape clause, a provision of the EU fiscal framework that grants additional flexibility equivalent to 1.5% of GDP from 2025 to 2028 where a country can show an increase in defence spending against a selected base year. Lane's slide on public investment splits the euro area into countries that have activated the clause and countries that have not. Fourteen of the twenty-one euro area members have activated it.

The base year is where the accounting gets interesting. For most countries the reference point is 2021 — before the invasion of Ukraine, which makes almost any subsequent increase count. For Bulgaria, Spain and Greece the base year is 2024, a far more recent bar that credits only the most recent acceleration.

Bulgaria's presence on that list is itself new information. The euro area now contains twenty-one countries; the presentation notes that Bulgaria was not required to submit a draft budgetary plan for 2026 because it had not yet joined the currency union in 2025. A country that spent the 1990s dismantling a command economy is now inside the euro and inside the same fiscal flexibility arrangement as Germany.

The clause solves a legal problem, not an economic one. Permission to borrow and capacity to borrow are different things, and the second is decided by bond markets rather than by the European Commission. Italy and France did not decline to raise defence spending because Brussels forbade it.

German order books are outrunning German factories

The clearest evidence of a physical constraint in the entire presentation comes from German manufacturing statistics.

Lane shows two charts built from Federal Statistical Office data, both measuring change against 2021 levels for three categories: explosives, weapons and ammunition, and other transport equipment — a category that includes military fighting vehicles alongside ships, aircraft and rail equipment. One chart tracks domestic orders. The other tracks production. The 2026 figures cover January through June.

Orders have risen by multiples that reach toward 300% above 2021 in the strongest categories. Production over the same period has risen by considerably less — the production chart's scale tops out at less than half the level the orders chart requires. The difference between the two lines is a backlog.

Demand growing faster than capacity adjusts through prices and delivery times rather than through volume. This is not a speculative concern; it is embedded in the ECB's own research method. When ECB staff estimate the macroeconomic effect of defence spending across EU countries, they construct an "adjusted" measure of capital spending that shifts the timing of the fiscal shock using a country-specific delivery-delay proxy built from the Stockholm International Peace Research Institute's Arms Transfer Database. The budget is committed in one year. The vehicle arrives in another. The economy feels the two events at different moments, and an analysis that ignores the gap will misread both.

The banks are barely in the room

European banks are financing almost none of this.

Drawing on AnaCredit, the ECB's loan-level credit register, Lane shows lending to the defence sector against total bank lending to non-financial corporations. Lending to defence is measured in the tens of billions of euros. Total corporate lending is measured in the thousands of billions. The defence share of euro area bank lending to companies sits well below one percent. The country breakdown concentrates what little there is in France, Italy, Germany and Spain, with France the largest on both sides of the ledger.

Over the same period, turnover among defence-sector firms that borrow from euro area banks rose substantially. The sector has been growing. Bank credit to it has not been growing in proportion.

The money is arriving through capital markets instead. A targeted defence equity index, indexed to 100 at the start of 2025, has pulled far away from both the broad euro area market and the technology sector, with the gap widening sharply after the German fiscal announcement in March 2025. On the debt side, the ECB identifies firms "exposed" to defence by whether they mention related terms in earnings calls — about 7% of listed euro area firms as of the second quarter of 2025 — and finds their net issuance of debt securities running well above historical averages.

One detail in that comparison deserves attention. On the ECB's measure, the debt gap between exposed and non-exposed firms widened after 2022, while the gap in capital expenditure between the two groups stayed negative across the period shown. Firms with defence exposure have been raising capital faster than they have been converting it into productive capacity — which is consistent with the German order backlog rather than a separate puzzle.

A build-up financed through equity valuations and bond issuance behaves differently from one financed through bank credit. Bank lending is slow and sticky. Market funding is fast and reprices on sentiment. If the political consensus behind European rearmament weakens, the funding channel currently carrying most of the weight is the one that would notice first.

March 2025: the day the bond market repriced Europe

The largest financial event of Europe's rearmament happened before most of the money was spent.

On 5 March 2025, Friedrich Merz announced a €500 billion infrastructure fund and a plan to exempt defence spending above 1% of GDP from Germany's constitutional debt brake. The ten-year Bund yield rose roughly 30 basis points in a single session — the largest one-day increase since German reunification was agreed in 1990. Lane's slide places that move against the distribution of daily changes since January 2020, and it sits at the extreme of the range.

The more revealing chart is the decomposition. The ECB splits the day's move in euro area overnight index swap forward rates — market-implied expectations of future short-term interest rates — into a real interest rate component and an inflation compensation component, across horizons from one year ahead to nine years forward. Separating the two matters, because they mean different things. A move driven by inflation expectations is a monetary policy problem. A move driven by real rates is a permanent change in the cost of capital across the entire economy.

The modelling that accompanies it is explicit about which direction the composition pushes. Using a version of the ECB's EAGLE model extended to include household preferences over safe assets, staff simulate a 1% of GDP fiscal expansion lasting twenty-five years. Government consumption raises the real interest rate. Government investment eventually lowers it, because investment adds productive capacity while doing less to stimulate private demand in the short run — so the real rate has to fall to bring private demand back up. The ECB notes that without the safe-asset feature in the model, the long-run effect on real rates would be a little over half as large.

For a household this is the channel that matters most. Real interest rates are what mortgages, business loans and government debt service are ultimately priced against. Whether Europe's build-up raises them permanently or lowers them eventually depends on what the money buys.

What economists actually know about defence multipliers

The fiscal multiplier is the number the entire economic case for rearmament rests on, and it is the number economists agree on least.

Lane devotes a full slide to collecting published estimates. The American literature runs from Barro and Redlick's work on temporary versus permanent military spending, through Ramey's military news shocks, Nakamura and Steinsson's use of prime military contracts, Ben Zeev and Pappa, and Antolin-Diaz and Surico. The cross-country literature includes Gechert and Rannenberg's meta-study, Olejnik on central and eastern Europe, and recent EU-wide estimates from Garcia and co-authors and from Furceri and co-authors. The estimates, plotted on one axis, span from around minus 2.5 to plus 3.5.

The spread is not sloppiness. It reflects genuine state dependence, and the studies that split their samples show where it comes from. Gechert and Rannenberg separate expansions from recessions. Garcia and co-authors split by whether a country has ample or limited fiscal space, and by whether it has high or low import reliance. Olejnik separates personnel spending from equipment and infrastructure. Each split moves the answer.

Against that backdrop, the ECB's own model simulations are more contained. Five models — ECB-REBASE, ECB-BASE, ECB-MC, DREAM and HANK — simulate defence spending rising from 2% of GDP in 2025 to 3% by the beginning of 2028 and holding there. The resulting GDP multipliers cluster below 1.4, and the effects on harmonised consumer price inflation stay under half a percentage point on an annual average basis.

The near-term baseline is smaller still. Taking the additional defence spending actually committed across the euro area since the February 2025 Munich Security Conference — national defence plus support to Ukraine — the Eurosystem projections put the annual fiscal impulse at a fraction of a percent of GDP through 2028, with corresponding effects on real GDP growth and inflation measured in fractions of a percentage point. The impulse is modest and drawn out. Anyone expecting European rearmament to function as a growth programme is expecting more than the ECB's own projections deliver.

Where the money goes decides whether it works

The sensitivity analysis is where the presentation becomes genuinely prescriptive, without ever using prescriptive language.

Lane includes a table showing how the GDP multiplier, household consumption and business investment respond to different assumptions. A higher share of public investment raises all three strongly. Spending that complements private goods raises all three strongly. Spending targeted at low-income sectors raises consumption but reduces business investment. Higher import content lowers everything. Anticipation of future monetary tightening lowers everything. Anticipation of tax rises to close the resulting deficit lowers everything, most severely of all.

The import result comes with a number attached. The ECB's benchmark calibration assumes 16% import content for defence spending. Raising that assumption to 50% — closer to what buying American aircraft and munitions actually implies — cuts the multiplier materially. Every euro spent on imported equipment buys capability without buying much domestic activity. That is a defensible trade-off. It is not the trade-off European governments have been describing to their voters.

Composition has been moving slowly. Across 1995 to 2024, euro area defence spending averaged 53% personnel, a declining share; 24% intermediate consumption — fuel, maintenance, the operational inputs of running armed forces — and rising; 19% investment; 3% research and development; and 3% other, including military aid to Ukraine, which has risen since 2021. Longer-run projections drawn from Janes databases show procurement's share rising toward 2035. Whether it rises fast enough to shift the multiplier is an open question.

Lane also presents work on the industrial base itself: where the suppliers and subsidiaries of Europe's largest defence firms are located, how the productivity of those firms compares with regional manufacturing averages, and how net job creation in the automotive industry between 2018 and 2023 relates to defence's share of regional employment. The last comparison poses the question directly. Europe has a shrinking car industry releasing skilled manufacturing labour and a growing defence industry that needs it. Whether the second absorbs what the first releases is a regional question with a different answer in Bavaria than in Piedmont.

The one channel with a clear return

The strongest positive result in the presentation attaches to the smallest line in the budget.

Research and development has averaged about 3% of euro area defence spending over three decades. Eurosystem estimates model what would happen if Europe matched U.S. levels of public sector defence R&D. Public spending crowds private R&D in rather than out, so total research spending rises by more than the public contribution alone. The GDP gains come substantially through total factor productivity — the residual that captures how efficiently an economy converts inputs into output — rather than through simply adding labour and capital.

The cumulative return starts negative and turns positive over the horizon modelled, rising toward roughly two euros of GDP per euro of public R&D invested. That shape is the whole problem. Research costs money before it produces anything, and the payback arrives outside the electoral cycle of the government that funded it. It is the easiest line to cut and the hardest to defend, which is a reasonable explanation for why it has sat at 3% for thirty years while personnel absorbed more than half.

Defence is the smallest of three funding gaps

Europe's defence debate is happening inside a larger fiscal question that gets discussed far less.

Lane presents estimated annual spending needs across three strategic transitions — green, digital and defence — each broken into private funding, existing EU funding, and the remaining public funding gap. Defence is the smallest of the three. The green transition is by a distance the largest. Taken together, the needs run to several percent of GDP annually.

These transitions compete for the same public euros. A defence euro financed by reprioritising a national budget is a euro not spent on grid capacity or broadband. The ECB research this draws on proposes a sequence rather than a single answer: use existing EU and national financing mechanisms first, then reprioritise national budgets, then build genuinely joint strategic financing, alongside reforms to make the EU economy more integrated and productive. That is a polite way of saying that Europe cannot fund all three transitions from national budgets alone at current levels of integration.

What this means for ordinary people

Strip out the modelling and three things in this presentation touch a household directly.

Interest rates are the main transmission channel. Whether Europe's build-up permanently raises the cost of borrowing across the continent depends on composition. Weighted toward salaries, fuel and maintenance, the ECB's model says real rates rise and stay higher — which shows up in mortgage renewals and business loan pricing across all twenty-one euro area countries, including those spending nothing extra on defence. Weighted toward equipment and research, the effect eventually reverses. That decision is being taken now, in national procurement offices, mostly without public debate.

The inflation effect is small in aggregate and concentrated in specific sectors. The Eurosystem's own baseline puts the price effect of committed additional defence spending at fractions of a percentage point. Where prices genuinely move is where orders are running ahead of production — ammunition, explosives, military vehicles — and almost nobody buys those directly. The broader inflation risk from the current build-up is, on the ECB's numbers, modest.

How governments talk about paying for it matters as much as what they spend. The sensitivity table is blunt on this: if households and firms expect deficits to be closed with future tax increases, they save more now, and the multiplier falls. A government that funds defence while credibly explaining that it will not be paid for by an income tax rise gets more economic activity per euro than one that leaves the question open. Fiscal credibility is not an abstraction here; it is a line item.

For anyone tracking this over the next two years, four indicators do most of the work. The ratio of orders to production in German manufacturing data will show whether capacity is catching up with demand or falling further behind. The defence share of euro area bank lending to companies, currently well under one percent, will show whether the banking system is joining the build-up or leaving it to equity markets. The R&D share of defence budgets will show whether governments have absorbed the one clearly positive finding in the ECB's work. And the fate of the national escape clause after 2028 will show whether the current arrangement was a bridge or a permanent change to how Europe accounts for military spending.

Lane's audience in Dublin was academic economists, which is probably why the presentation could afford to be so plain about how much remains unknown. A slide showing that credible estimates of your central parameter range from strongly negative to strongly positive is not the sort of thing that survives translation into a press conference. The build-up will proceed regardless. Whether it leaves Europe with a stronger economy or simply a larger debt stock is being determined right now, in procurement decisions about domestic or imported, equipment or research — none of which will make a headline, and all of which are already visible in the ECB's charts.