Inflow is not Entry: Competition without a Referee in the European Defence Fund, 2021–2025

The European Defence Fund (EDF), the Union's first large budget for defence research and development, brings in many newcomers but little entry. We use all 4,184 participations in its five calls (2021–2025), consolidated to corporate groups with time-varying ownership and linked to the precursor programmes. From 2023, groups already present lead 97 per cent of the funding in projects of €20 million or more, although they make up 85–87 per cent of the large-company groups taking part in those projects; newcomers lead at about a quarter of the rate their presence implies. Persistence sits with groups that predate the fund. Almost half of the large groups in its first call had precursor funding, and nine in ten of them return within two calls, against 17–19 per cent of the large groups new to EU defence funding that entered in 2022 or 2023. Much of the apparent entry is the arrival of new subsidiaries of established groups, and the share of participants new to the fund falls from 62 to 38 per cent between 2022 and 2025. These findings bear on the consolidation of Europe's defence industry. The Draghi report asks for scale where needed, and notes that the United States, after consolidating its industry, now resists further consolidation for the sake of competition and innovation. In the EDF's large projects, the concentration of design leadership that consolidation would bring is already observed, without any merger, although a capability advantage of incumbents is not excluded. The Council's text for the defence window of the European Competitiveness Fund (ECF) fixes a single award criterion, the quality of implementation, which favours groups that have implemented before. We read the evidence as competition without a referee: member states may set the Treaties aside for the production of and trade in arms, decide through co-financing what is developed, and hold a veto over every award list. An entry model that extends an earlier model of US acquisition to a market of national segments separates geographic spreading from leadership reserved for incumbents. In its calibrated scenarios, spreading does not substitute for open leadership, and consolidated production remains compatible with contestable design; because the separation of design from production is not observed, the model does not place Europe in a regime. We register three predictions for the defence window of the ECF before its first results are published. They are decided by interval and separate reserved leadership from the capability explanation. The evidence extends to material published on 1 November 2026.