Country-attributable OLAF investigation density: a reproducible index from the published anti-fraud record — EU-27, 2014–2024

This paper documents the method, sources, denominator construction, validation record and limitations of a country-attributable investigation-density index built entirely from the public anti-fraud record. For each EU Member State the index divides the number of concluded, Member-State-linked European Anti-Fraud Office - OLAF investigations recorded in OLAF Annual Reports for 2014 to 2024 by the country-attributable EU expenditure compatible with the scope of OLAF's own Member-State reporting: the European Agricultural Guarantee Fund, the European Agricultural Fund for Rural Development, cohesion funds and non-repayable Recovery and Resilience Facility grants. Across the EU-27, 1,017 concluded cases are set against €1,286.1 billion, a weighted average of 0.79 cases per €1 billion. Belgium records 8 concluded cases against €13.95 billion, an index of 0.57 and rank 15 of 27, below the EU-27 average, while over the same period Belgium registered the highest volume of incoming information of investigative interest of any Member State and the highest number of OLAF referrals to the European Public Prosecutor's Office, 81 of 389 across 2021 to 2024. The paper documents a correction: an earlier working version used a Financial Transparency System direct-management denominator that did not match the scope of OLAF's table; it was replaced and disclosed to OLAF in writing before publication, and the structural finding did not change. The method was tested through formal notification of five EU institutions and a right-of-reply exchange with OLAF; no figure has been disputed. The scope limit is stated precisely: OLAF's Member-State table excludes directly managed EU funds by OLAF's own published notice, so the index is a country-attributable expenditure-density indicator, not a full OLAF-scope exposure denominator, and it measures the public reporting architecture, not conduct. Key findings 0.7%. Of the 1,115 signals concerning Belgium in OLAF's own published and disclosed Member-State intake record for 2014–2024, 8 became concluded investigations. The EU-27 average is 21.8%. Both numbers are OLAF's own; neither uses an external funding denominator. Same reporting architecture, opposite outputs. Belgium produced 81 OLAF referrals to the EU's criminal prosecutor across 2021–2024 — the highest of any Member State — against 4 concluded administrative investigations over the same four years. 15th of 27. On country-attributable EU expenditure, Belgium's investigation density is 0.57 concluded cases per €1 billion, below the EU-27 weighted average of 0.79, while registering the highest signal volume in the Union. Two country-level tables gone. OLAF described discontinuing its Member-State intake table after 2018 as an editorial choice, and attributed the absence of the per-country EPPO-referrals table from the 2025 report to the same reasons; the internal decision record is the subject of a pending access request. These are the two series that would let Parliament, the Court of Auditors and the public examine country-level variation. A dated obligation. The Commission has accepted, in writing, an obligation to analyse the variation in fraud reported across the EU by end 2026 (ECA Special Report 26/2025). This paper assembles a public country-level series directly relevant to that obligation. Five institutions notified; none disputed a figure. The method was placed before OLAF, the Commission, the Court of Auditors, the EPPO and the European Parliament. One earlier ratio that used the wrong denominator was corrected, in writing, before publication; the finding did not change.