Sources & methodology.

Model moi-anchored-2024-v1: 2,106 applications at MOI, reported by its press department in Pravda on 15 April 2024. The separate 960 applications at district offices are excluded. We assume end March 2024 as the reference because the extraction date was not stated.

Estimated open workload = reference + reported applications − grants − other outcomes. Complete-year other outcomes are allocated across months in proportion to grants; a zero-grant year uses equal monthly allocations. In incomplete years, grants are scaled by the preceding two complete years’ other-outcome/grant ratio. Monthly timing is estimated retrospectively from the current capture. No negative balance is clamped.

Descent workload is a modeled allocation using its share of grants over the last 12 complete months at the selected reference. That same share is held fixed for “At that time”, “Latest” and “Change since then”. It is not observed Descent intake or pending stock. All-pathway pressure remains the common flow context.

Months of output = modeled workload ÷ average monthly output over 12 complete months. All pathways uses estimated results; Descent uses reported grants only. This is workload context, not cases ahead, a personal wait or a decision date. Submission, MOI receipt, decision, notification and oath/collection are separate milestones; the grant feed does not establish those event dates.

Modeled decision timing is a separate conditional oldest-first receipt-cohort scenario. Receipt is assumed on the first day of the following month; a known month override uses the same month-start convention. Opening work is the anchored prior-month-end workload, with receipt-month and subsequent output deducted once; later arrivals are excluded. Reported outcomes are decision proxies, not verified decision dates. The rate is total output divided by total working days across the four complete months before the first of the current Bratislava month. Remaining work is projected from that month-start at this pooled rate. Working days exclude Slovak nonworking holidays, including year-specific exceptions; monthly output is not daily observation. Results display one modeled month, not a probability range. Missing output or pace months make the calculation unavailable; current-month data is excluded from both deductions and pace. All pathways uses modeled results. Descent freezes its trailing twelve-month grant share at the prior month-end to allocate opening workload, then uses only Descent grants for deductions and pace. Separate Descent other outcomes are unknown and not deducted; this is not an observed queue. Calendar projections are retrospective from the current capture, not archived forecasts; a reached cohort does not establish an individual decision.

Working-day calendar sk-241-20251101-v1 uses the 2024 rules and rules effective November 2025, cross-checked with the NBS calendar. No substitute Mondays are added. The supported 2024–2199 calendar assumes currently enacted recurring rules remain unchanged in future years; it is not a forecast of holidays or staffing.

The model assumes compatible application and grant units and one grant record per closure. The source stock may include suspended or approved applications awaiting collection. No family adjustment is applied: resident age statistics have a different population and no parent links. A transfer of their estimated under-14 shares (13.6% in 2024; 15.2% in 2025, assuming uniform ages within bands) would be sensitivity analysis, not a correction.

The 2024 annual report, pages 37–38, counts 1,331 grants versus 1,333 in the feed, but 33 rejections / 124 stopped cases versus 13 / 95. The 49-case difference is about 1.6% of a 3,100 workload estimate, not an estimate of total model error. Undefined scope differences prevent substituting those outcomes. July 2026 may mark a process change; no speed gain is assumed.