The European Common Agricultural Policy (CAP) commits, in its Treaty foundation, to a fair standard of living for the agricultural community and, in its post-2014 architecture, to enhanced territorial cohesion. Yet repeated reform cycles have left the regional concentration of payments in many Member States visibly untouched. This paper asks why. We document the persistence of the territorial concentration of CAP transfers across the 13 Greek NUTS-2 regions over the 2010–2025 period (€47.65 bn cumulative), identify the CAP design mechanisms that mechanically reproduce it, and quantify how much of the observed aggregate stationarity is the artefact of compositional shifts versus genuinely offsetting forces. Using the universe of payment disbursements aggregated to 13 NUTS-2 regions and 51 NUTS-3 prefectures, we (i) test for σ- and β-convergence and Lorenz dominance, (ii) decompose Theil-T between and within regions and across Pillar I/Pillar II, and (iii) run four counterfactual simulations: Pillar II share held at its 2010 level, Article: 17-style capping at a 12–15% NUTS-2 ceiling, an Article: 29-style lower-tail floor, and a concentration-elasticity perturbation of the top region. The territorial distribution of support proves strikingly stable: standard inequality measures stay within a narrow band for sixteen consecutive years, and the ranking of regions barely changes, so formal convergence tests detect no narrowing over time. Three messages follow. First, this persistence is not accidental but built into the architecture of the CAP—through historical-reference entitlement values, the per-hectare logic of the Basic Payment Scheme, the geographic concentration of coupled support in cotton and livestock, and the cadastral fragmentation of the island prefectures. Second, the apparent stability conceals two large and opposing forces: the post-2014 expansion of Pillar II has reduced regional disparities, while a widening of the Pillar I distribution has increased them by almost the same amount, so aggregate stationarity reflects policy effort cancelling out, not the absence of it. Third, the instruments already in the CAP toolbox have real redistributive power: capping the largest region’s envelope and redistributing the surplus to lagging regions, or introducing a lower-tail floor, would roughly halve measured inequality. Therefore, the spatial concentration of CAP transfers in Greece is a designed equilibrium rather than an unsolved residual, and reducing it requires instruments that act asymmetrically on the top of the distribution.
I Kaimakamis (Thu,) studied this question.