Paper Firms and Formal Payrolls: Registry Unification, Digital Tax-Risk Enforcement, and the Inactive-Enterprise Wedge Across Uzbekistan's Districts and Sectors

When transition states synchronize corporate registries with automated value-added tax risk scoring, does a steep contraction in active enterprise counts reflect proportional formal-payroll contraction or a two-stage administrative purge of dormant registrations? Combining harmonized district-year panels across 147 boundary-consistent parent jurisdictions covering roughly 84% of national commercial entities and formal payrolls (2016–2025, N = 1,470; payroll window 2017–2024, N = 1,176) with population-matched monthly panels across all 14 regions and nine commercial sectors of Uzbekistan (January 2022–September 2026, N = 7,182), we trace the inactive-enterprise wedge across inter-agency database unification, the January 2023 value-added tax rate cut, and compulsory deregistration. Nationally, commercial inactivity rose from 5.83% at year-end 2022 to 44.86% at year-end 2024 as operating entities fell by 31.38% alongside descriptive national growth in formal workbook employment (+2.95%) and real monthly wages (+14.01%), before receding to 32.06% at year-end 2025 during concentrated register removals. Under region-by-sector and region-by-month fixed effects with exact 14-region wild cluster bootstrap inference, invoice-chain-intensive sectors (industry, construction, and trade) exhibit a 5.56-percentage-point wider inactivity differential during the late-reclassification and deregistration phase (bootstrap p = 0.00012; 95% bootstrap confidence set [4.87, 6.26]), whereas saturating sector-by-month shocks leaves predetermined 2022Q1 regional-sector inactivity and specialization gradients statistically unresolved. Across districts, a 10-percentage-point higher 2019 trade-and-construction share yields conditional 2024 differences of -0.0160 log points in formal employment (bootstrap p = 0.4161; 95% confidence set [-0.0590, 0.0227]; 14-region 80% minimum detectable effect 0.0559) and -0.0281 in real wages (bootstrap p = 0.0908; [-0.0665, 0.0044]), surviving Greater Tashkent transfer restoration and pre-2022 district drift controls.