County-Scale Registered Population Density in China (2005–2020): Spatial Patterns, Cluster Transitions and Socioeconomic Associations

County-level population concentration and change have different implications for territorial planning. This study examines registered population density and local classifications in 2005, 2010, 2015 and 2020 for 2121 linked counties, alongside a separate 2020 socioeconomic analysis of 2074 counties using original non-missing inputs. Positive spatial association persists in all four snapshots. Under inverse-distance weights, 163 counties are High–High (HH) at both endpoints, 93 enter HH and 10 leave it; exact memberships are sensitive to alternative neighbourhood rules. Fixed-footprint modelled estimates closely reproduce density rankings but give less consistent endpoint changes. Hongkou District remains HH in every tested year and neighbourhood, while its official registered count decreases by 11.3% between 2005 and 2020; a separate 2010–2020 census-resident comparison also records decline. The case distinguishes persistent concentration from continued growth. Geographically weighted regression describes varying GDP and tertiary-sector associations, predominantly positive bed-count associations and negative area associations. Size baselines and remaining spatial dependence limit interpretation of strong fit and local coefficients. Planning applications should verify population definitions, territorial continuity and actual service use before translating a cluster label into expansion or consolidation.