Chapter 06 · How this is built

Three layers, six chapters, one chain of evidence

Every figure on this site can be traced from headline back to source. This chapter explains how.

Three-layer architecture

Official → Bridge → Modelled

Official

Stats SA P0441, QLFS, QES, provincial GDP, supply-and-use tables. SARS, DHET, NSF, EPWP. Loaded as published, with no transformation beyond unit normalisation.

Bridge

Documented mappings between industry classifications (P0441 ↔ QES ↔ QLFS ↔ SIC-7 ↔ ISIC-4), formality definitions, and the OQSF qualification framework. Each rule reviewable, each edit auditable.

Modelled

Allocation methods that fill structural gaps — quarterly provincial GVA, jobs by province × industry, opportunity-to-supply, retention curves. Always reproducible from the layers below.

"No figure may appear on this site without a provenance badge and a path back to its source."

— Architecture principle

Architecture map (workbook sheets)

Architecture Map
Data flows upward from Official → Bridge → Modelled → Output
	LAYER 1 — OFFICIAL
	fact_gdp_qtr_nat Stats SA P0441 Quarterly GDP/GVA by industry	fact_prov_gdp_annual Stats SA Experimental Provincial GDP (annual)	fact_qes_qtr Stats SA QES Formal non-agri employment + earnings	fact_qlfs_qtr Stats SA QLFS Total labour market + informality + demographics	fact_sut_annual Stats SA SUT Industry × product benchmark matrix
	↓	↓	↓	↓	↓
	LAYER 2 — BRIDGE
	bridge_industry Crosswalk: SIC7 ↔ P0441_10 ↔ QES_8 ↔ QLFS_10 ↔ ISIC4	bridge_formality QLFS total − QES formal = informal_non_qes (by industry)	dim_geography National ↔ Province ↔ Metro ↔ District crosswalk	dim_trade Skills trades → industry groupings (for labour conversion)	dim_series_break Rebasings, redefinitions, QLFS informality change from Q3 2025
	↓	↓	↓	↓	↓
	LAYER 3 — MODELLED
	model_gva_qtr_prov Quarterly provincial GVA = national qtr × provincial annual share	model_jobs_qtr_prov Formal (QES) + informal (QLFS − QES) + total jobs by prov-industry-qtr	model_productivity GVA per worker/hour Comp per hour Jobs per R1m output	model_opportunity_map Demand: pipeline, backlog, implied jobs (SUT-linked)	Skills + Funding + Labour Absorption Pipeline conversion, cost per sustained worker, capital crowd-in
	↓	↓	↓	↓	↓
	OUTPUT
	Dashboard Roll-up panels: Economic Base, Pipeline, Capital Efficiency, Demand	Metrics_Dictionary Every KPI with: definition, unit, source, owner	Treasury narrative Cost per sustained worker, jobs per R1m, demand-to-labour gap	Funder narrative Conversion chain: spend → certified → placed → retained @ 12m	Federation narrative Demand map, trade-level gaps, municipal proxies