Problem
The synthetic population is built on the ONS VAT/PAYE-registered frame (~2.7m businesses). Below the VAT threshold this excludes unregistered sole traders — roughly half the below-£85k business population and the group that produces most of the administrative cross-threshold cliff (Tax Policy Associates 2018-19 decomposition: the sole-trader series falls 8.7k→2.9k firms/£1k at the threshold, while companies+partnerships ≈ our frame's levels). Consequences, all currently disclosed but not modelled:
- The distribution's cross-threshold level ratio is much shallower than all-business charts (raised in review by 2026-07-02).
- Threshold cuts (sweep rows below £90k) omit newly covered unregistered businesses, understating revenue gains.
- Registration/location behaviour of the most threshold-sensitive population is outside the data.
Proposed next build
Extend the frame to the full business population (~5.5–5.6m, DBT Business Population Estimates):
- Add a BPE-based unregistered stratum below the threshold (BPE by employment status × broad turnover bands; sole-trader-heavy).
- Near-threshold £1k-band structure for the full universe is directly available: the OBR Chart C / TPA-style HMRC tabulations count all businesses in tax records — use them as level targets (no shape-only rescaling needed once the frame matches the chart universe).
- VAT-liability targets unchanged (registered firms only, ≥ threshold — same universe as now).
- Registration status: mandatory above threshold; below, voluntary registration only within VAT-relevant subpopulations.
This makes threshold cuts costable on the right base, restores the administrative cliff as a frame-correct target rather than an out-of-frame comparison, and is the natural design for the populace port (see populace#258).
Refs: #21 (frame disclosures in Fig. 1 and Section 3), #23 (OBR fine targets), review thread 2026-07-02.
🤖 Generated with Claude Code
Problem
The synthetic population is built on the ONS VAT/PAYE-registered frame (~2.7m businesses). Below the VAT threshold this excludes unregistered sole traders — roughly half the below-£85k business population and the group that produces most of the administrative cross-threshold cliff (Tax Policy Associates 2018-19 decomposition: the sole-trader series falls 8.7k→2.9k firms/£1k at the threshold, while companies+partnerships ≈ our frame's levels). Consequences, all currently disclosed but not modelled:
Proposed next build
Extend the frame to the full business population (~5.5–5.6m, DBT Business Population Estimates):
This makes threshold cuts costable on the right base, restores the administrative cliff as a frame-correct target rather than an out-of-frame comparison, and is the natural design for the populace port (see populace#258).
Refs: #21 (frame disclosures in Fig. 1 and Section 3), #23 (OBR fine targets), review thread 2026-07-02.
🤖 Generated with Claude Code