Atlas · Cohort Insights — Country Programmes

Programme overview

Aggregate signals from Atlas-verified youth cohorts. No individual candidate data is exposed in this view; production access is gated by NGO / government credentials.

Cohort population

Atlas live

2,341

verified Atlas-Card holders, 30-day window

[1] Atlas measurement layer · 2026-04-25

Youth NEET rate Fresh · 2025

GSS Q3 2025

21.5%

young people aged 15–24 not in employment, education or training (1.34M persons)

[2] Ghana Statistical Service Q3 2025 · released Dec 2025

Informal employment

ILOSTAT 2024

88%

share of total employment in the informal economy

[3] ILO ILOSTAT · informal employment rate

Why this cohort matters

Programme rationale

Across the country, 88% of working-age youth earn income outside formal credentialing systems. Atlas builds an interoperable skills passport that maps informal experience to ISCO-08 occupation codes and ESCO skill clusters, surfacing each young person's first formal-economy fingerprint.

Cohort signals on this page are intended for programme design, not outreach. Individual identifiers and contact data are not surfaced. To engage candidates, NGO partners can request a Squad introduction through the consent-based intake flow described in the Methodology tab.

Top 5 ISCO occupations in cohort

Atlas live

ISCO-08	Occupation	Cohort %	Median wage^[4]	Auto-risk^[5]

Why this cohort matters at scale ILO 2026

Employment & Social Trends 2026 · Jan 2026 release

Global context for programme officers. Source: ILO Employment and Social Trends 2026.

Hackathon-grade rubrics. Cohort STEP-equivalent scores are calibrated against Ghana STEP 2013 anchor responses. They are not psychometrically validated. Production deployment requires IRT calibration with ≥1,000 players over a 6-month panel. See Methodology for the limitations register.

Skills supply versus sectoral demand

Verified Atlas cohort, mapped against ILOSTAT 2024 sector employment and World Bank Enterprise Surveys (WBES) skill-gap reporting.

Atlas informal-skill clusters

Atlas taxonomy · ISCO-08

10 clusters mapping informal-sector occupations to ISCO sub-majors. Each card shows cohort share, female share, wage band, automation risk, and demand. Click a card to drill into ESCO skills.

🔎

Showing 5 of 10 clusters

Cohort distribution by skill cluster vs. employer demand

Atlas × ILOSTAT × WBES

Blue = verified cohort supply. Gold = WBES employer demand (% of firms reporting unfilled positions in cluster). Cyan = ILOSTAT YoY sectoral growth.

Most-detected ESCO skills

Atlas measurement

Skills extracted from cohort narratives via structured ESCO classifier. Ranked by detection frequency.

ESCO code	Skill	Tier	Detected (cohort %)

Programme implication

Where supply gaps exceed 30 percentage points relative to WBES demand, targeted training is recommended. This cohort shows the largest gap in software / ICT, where employer demand outpaces verified supply by +57 pp.

Wages and economic opportunity

Median monthly wages by ISCO occupation, with formal / informal differentials. Source: ILOSTAT 2024.

Wage benchmarks by occupation

ILOSTAT 2024

Local currency, monthly. The formal-informal gap is the principal lever for cohort upgrading.

ISCO	Occupation	Informal median	Formal median	Gap

AI wage premium scenario

Atlas modelled

If 50% of the cohort moves up one AI Tier, aggregate monthly wage uplift.

AI fluency stratification

Distribution of AI Tier (0–4) across the verified cohort. AI fluency is currently the highest-leverage individual skill in the global labour market and is unmeasured in any other public dataset for LIC + LMIC youth.

AI Tier distribution

Atlas — original measurement

AI Tier definitions

Atlas methodology

Tier	Profile	Wage premium
0 — Unfamiliar	No AI tool exposure	0%
1 — Curious	1–2 trial uses, no iteration	+5%
2 — Active	Weekly+ use, iterates, has go-to use cases	+20%
3 — Power user	Multi-tool, multimodal, verifies output	+50%
4 — Builder	Builds workflows, automations or code on AI	+100–300%

Policy implication. Raising the cohort's modal AI Tier from 1 to 2 (a single-tier shift for the median worker) plausibly delivers more aggregate wage uplift than any single-skill training programme tested in the past decade in this country. Atlas measures and tracks this lever; existing labour data does not.

Automation exposure

Frey-Osborne automation probability, LMIC-adapted (ILO 2018, reappraisal 2024). Identifies durable tasks that resist automation.

Automation risk by ISCO occupation

Frey-Osborne LMIC 2024

Probability that the occupation is automated within a 5-year horizon. Durable task examples shown for each.

Durable-task strategy

For high-risk occupations (auto-risk > 70%), cohort training should prioritise the durable adjacent tasks — interpersonal service, judgement under context, multilingual customer relationships, and AI orchestration — rather than the exposed routine tasks themselves.

Demography & gender

Wittgenstein Centre 2030 cohort projections, Women Business & the Law 2024 legal-gender score, and ILO/WDI female labour-force benchmarks. All figures hardcoded from public sources at submission.

Youth population (15–24)

UN DESA · WPP 2024

6.7M

22% of total population — Ghana 2024 mid-year estimate.

verified UN World Population Prospects 2024

Female cohort share

UN DESA

49.7%

Of youth aged 15–24 in Ghana, 3.34M are women.

verified UN WPP 2024 (mid-year estimates)

Youth NEET rate

WB WDI 2024

28.1%

Aged 15–24, not in education, employment or training. Of those, 64% are women.

verified World Bank WDI 2024 · ILOSTAT

Wittgenstein 2030 projection

Wittgenstein Centre

46%

of women aged 20–24 will hold secondary+ education by 2030 in Ghana (+12pp from 2020).

verified Wittgenstein Centre · SSP2 medium scenario

Implication. Credential differentiation is becoming harder. AI fluency, ISCO-mapped attestation and skill specificity replace "I have a diploma" as the labour-market signal of the next five years.

Women, Business & the Law 2024

WBL · World Bank

76.9/100

Gender legal score across 8 dimensions: mobility · workplace · pay · marriage · parenthood · entrepreneurship · assets · pension.

verified WBL 2024 · World Bank

No legal restrictions on female employment in measured occupations.

Atlas privacy default for female users: Tier 2 (pseudonymous).

Female labour-force participation (FLFP)

WDI · ILO 2024

Female labour-force participation in Ghana stands at 63% of working-age women, against an estimated male rate of ~68%. The 5pp differential is one of the smallest in West Africa, but masks a 4× wage gap in informal markets that Atlas surfaces at the cluster level.

Women

63%

Men (estimate)

~68%

Why this matters. The Bangladesh gender labour gap (47pp) is one of the largest in South Asia. Atlas auto-shifts to Tier 1 privacy (no real names, hashed phones) for female users in BD, PK, IR, AF, MM — driven by the WBL 2024 score < 60.

Secondary education completion · women aged 20–24

UNESCO UIS · 2023

Year	Women	Men	Gap (pp)
2010	22%	31%	-9
2020	34%	39%	-5
2030 (proj.)	46%	48%	-2

Skill-gap demand · gender breakdown

WBES 2024

Ghanaian SMEs report 30% of unfilled roles are in female-friendly sectors (textiles, hospitality, retail, basic ICT). Atlas surfaces this in the Dealmaker (5223) and Storyteller (5141) clusters.

Same-occupation pay gap

-14%

Mobile money: women

55%

Methodology & limitations

How Atlas measures, what is calibrated against what, and what we do not yet know.

Measurement layers

Layer	What it produces	Method	Calibrated against
Skills extraction	ESCO codes + ISCO-08 classification	LLM tool-calling on player narrative	EU ESCO API · ILO ISCO-08
STEP-equivalent rubrics	Numeracy, problem-solving, Big-Five proxies, grit	Multi-dimensional LLM-judge (0–3 per dim)	Ghana STEP 2013 anchor responses
AI Tier composite	Tier 0–4	5-input weighted score (inventory · prompt · duel)	Behavioural consistency vs self-report
Calibration check	Confidence calibration signal	Round-3 partial-truth choice	Behavioural — unfakeable
Wage benchmarks	Local currency monthly median	ISCO match → ILOSTAT lookup	ILO ILOSTAT 2024
Automation exposure	5-year automation probability	ISCO match → Frey-Osborne LMIC adapted	FO 2013 · ILO 2018 · FO 2024 reappraisal

Limitations register

Limitation	Current state	Required for production
Psychometric validation	Hackathon-grade rubrics; not validated	≥1,000 players × 6 months IRT calibration
Cross-cultural validity	Calibrated for English / Twi register	Native-speaker calibration in 20+ languages
Anti-gaming detection	None — players can over-claim	Behavioural consistency layer; longitudinal tracking
Adversarial robustness	LLM judges are prompt-injectable	Adversarial fine-tuning + rule-based guards
Score stability	One-shot measurement	Test-retest reliability via daily streak

Data refresh policy

Annual cadence. Source data is refreshed once per year, aligned to the publication calendars of the primary providers (World Bank WDI April-May, ILOSTAT rolling, ILO flagship January, BBS / GSS / GSO Q4 country releases, WBL annual edition).

Last refresh: 26 April 2026
Next scheduled refresh: April 2027
Maintainer: Atlas data team — annual provider sweep, not a live API

Atlas-original cohort metrics (Atlas-Card population, AI Tier distribution, ESCO detection frequency) are produced from quest completions and refreshed at the same annual cadence to maintain version-stability for cited research.

Republic of Ghana