Education · Adaptive Capacity

Who's Exposed AND Stuck?

A per-occupation map of AI exposure × adaptive capacity for the Manx workforce. Adapts the Manning & Aguirre (NBER w34705, January 2026) framework — exposure alone is a partial picture; the policy question is which roles face high exposure AND have weak retraining pathways. That's the cell that needs help.

🌗AI Transition Era · context for the analysis below

We're in a weird transition phase — read every signal twice.

Labour market signals and AI disruption are arriving at the same time. UCM is a partner in the response, not the target of analysis. Where we flag a gap, treat it as an opportunity to lead.

Read: Skills that matter in the transition era →Which UCM courses already teach for the AI era

The framework, in plain language

Most AI-impact analysis stops at the question which jobs face high AI exposure?. That's a useful screen, but it overstates the problem. A heavily-exposed worker who has a degree, a mobile skill set, lots of training pathways, and a current of demand pulling them into adjacent occupations is probably going to transition fine. A less-exposed worker with none of those things may be in more trouble than the headline numbers suggest.

Manning & Aguirre (Jan 2026) crossed those two questions on US data — exposure on one axis, an occupation-level adaptive capacity index on the other — and found that of the 37 million Americans in the top exposure quartile, about 26 million also score above-median on adaptive capacity (likely fine), while around 6 million sit in the vulnerable cell: high exposure, low capacity. ~4.2% of their sample, concentrated in clerical/administrative roles. That's the population the policy money should follow.

We've adapted the framework for the Isle of Man using the data we already have — UCM training-pathway counts, IoM census workforce, ASHE/IoM-advertised salaries, training-lag estimates per SOC, sector diversity. It is not a clean replication of their O*NET-derived index; it's a Manx approximation that should be read as direction-of-travel, not a headline number to lift verbatim.

⚠️Vulnerable

32.1%

17 occupations · 3,777 workers

High AI exposure × low adaptive capacity. The cell that needs targeted policy.

🤝Mobile

1.6%

1 occupations · 194 workers

High exposure × high capacity. Will probably transition fine.

🛡️Overqualified

7.6%

4 occupations · 892 workers

Low exposure × high capacity. Could move, doesn't need to.

🏠Sheltered

58.6%

44 occupations · 6,897 workers

Low exposure × low capacity. Stable for now.

📍

32.1% of UCM-mapped Manx workers sit in the vulnerable cell

The US benchmark in Manning & Aguirre is 4.2%. Our number looks higher, but they aren't directly comparable — this measures the IoM SOCs that UCM offers training for (about 11,760 census workers across 66 occupations), not the whole 49,000-strong working population. What the gap signals is real even if the percentages don't line up: the Manx UCM-served workforce skews toward exposed clerical/admin/finance roles, and the retraining pathway counts we use to score capacity are sparse for many of those exact roles. Read this as a hotspots map, not a population census.

Source-paper benchmark context: Manning & Aguirre identify ~6.1m US workers (4.2% of sample) as both high-exposure AND low-capacity, concentrated in clerical/admin roles.

⏳

Why “0-month response” is a fantasy — read the lag column

Adaptive capacity isn't a switch you flip. A vulnerable cell with “low capacity” doesn't become high-capacity the moment a new course launches. The “lag” column on the table below is the fastest qualifying UCM route — the minimum elapsed time between someone starting and them being able to practise the role independently. Plumbing is 36–48 months. Nursing is 36 months. An MBA top-up is 18. Even a short professional certification has its own 6–12 month tail.

Translation: even if every recommendation on this page were funded tomorrow, the supply response is years out. Treat the “vulnerable” cell as policy debt that will mature slowly, not a problem you can rebudget out of.

Courses, not modules. Wherever this page or All Courses reports a “pathway” or “course”, we mean a UCM parent course — a learner-facing programme with a qualification at the end. Module-codes (IS-prefixed sub-units of a parent course), accreditation-of-prior-learning placeholders, and shells with no teaching hours are filtered out at precompute. So when the table says “1 pathway” for a vulnerable SOC, that's 1 actual course someone could enrol in — not 1 module inside something else.

Each course's own length, primary career destination, automation share and lag chip live on /education/courses.

Reading the “Lag” column: < 12 moshort retrains (CPD, short-bursts, top-ups) — capacity expands inside a year.12–24 moFE / part-time HE — a one- or two-year wait for newly trained workers.≥ 24 mofull degrees, apprenticeships, regulated trades — multi-year supply tails. Plumbers don't solve the gap in 0 months.

⏰

If we started retraining today, when does the supply land?

Years to retrain the whole cell at current UCM intake

⚠️Vulnerable

10.9 yr

plus 26% of cell workers with no UCM pathway at all

🤝Mobile

∞

plus 100% of cell workers with no UCM pathway at all

🛡️Overqualified

5.6 yr

every cell worker has a UCM pathway

🏠Sheltered

6.1 yr

plus 44% of cell workers with no UCM pathway at all

How to read this: the number is total cell workers with at least one UCM pathway divided by total annual graduate output across those pathways. Below each bar we report what % of the cell sits in SOCs with zero UCM pathway today — that share isn't in the years-to-absorb at all, because the supply tap doesn't exist. A 30-year vulnerable absorption number plus a 40% no-pathway fraction is a different problem than a 30-year number with 0% no-pathway: the first means the supply tap is missing for nearly half the cell.

Each curve shows the share of that cell's census workforce whose fastest qualifying UCM route finishes by a given month. Read it as: “if every worker in this cell enrolled today on the shortest available pathway, what fraction would be qualified by then?” A flat curve hugging the bottom for 12+ months means the supply response is years out, not months.

Vulnerable cell, month 12

13%

474 of 3,777 workers reachable

Vulnerable cell, month 24

98%

3,684 workers — most FE / part-time HE pathways have landed

Vulnerable cell, month 36

98%

3,684 workers — degree-length and apprenticeship cohorts now graduated

Important: this chart assumes every worker in a cell starts a retraining pathway on day zero — a fantasy ceiling. Real-world reskilling rates are a fraction of that. Treat the curves as the absolute upper bound on supply response: even at perfect uptake, vulnerable-cell capacity expands at the speed UCM courses can graduate cohorts. Funding decisions made today affect 2027–2030 labour-market outcomes; the policy-relevant question is which pathways exist at all, not which ones we can rebudget into next quarter.

🎯

Where the policy money should follow first

The framework above maps every UCM-served SOC into one of four cells. These two lists pull out the highest-priority entries from the vulnerable cell, separated by problem type: the absorption gaps are SOCs where a UCM pathway exists but the supply tap is too small relative to the workforce; the catalogue gaps are SOCs where no qualifying UCM pathway exists at all. Different policy responses, different timelines.

⏳

Absorption gaps — cohort scaling

existing UCM offer, supply tap too small

Policy lever: scale UCM cohort sizes for these courses, or co-fund employer apprenticeship growth. The supply mechanism exists; it just needs more capacity.

📚

Catalogue gaps — courses to add

vulnerable cell, no UCM pathway at all

No vulnerable SOCs are catalogue gaps — every vulnerable SOC has at least one UCM pathway.

Policy lever: catalogue addition first, cohort funding second. There is no supply mechanism today — funding the existing offer doesn't do anything for these workers.

These lists are derived from the rows that follow — sort “Yrs to absorb” descending or filter to vulnerable cell with the “UCM courses” column at zero to see the full picture. The priority cut here is the top 5 of each — the curriculum committee's opening agenda.

Filter cells

Each dot is a Manx occupation UCM teaches toward; size = IoM census workforce in that SOC. Hover for detail. The bottom-right cell — high AI exposure, low adaptive capacity — is the policy hotspot.

Every UCM-served occupation, scored

Cell	Occupation	IoM workers	Exposure ↓	Capacity	Median £	UCM courses	Grads/yr	Yrs to absorb	Lag	Skill xfer
⚠️ Vulnerable	▶3543Project support officers· Marketing	136	90	27	£34k	1	—	no pathway	18mo	117
⚠️ Vulnerable	▶4159Other administrative occupations n.e.c.· Business & Management	10	82	45	£23k	51	—	no pathway	18mo	99
🤝 Mobile	▶2423Taxation experts· Business & Management	194	73	52	£45k	63	—	no pathway	18mo	32
⚠️ Vulnerable	▶5323Painters and decorators· Painting & Decorating	295	70	30	£31k	3	24	12.3 yr	24mo	92
⚠️ Vulnerable	▶5222Tool makers, tool fitters and markers-out· Engineering	18	69	30	£39k	36	—	no pathway	18mo	thin
⚠️ Vulnerable	▶5232Vehicle body builders and repairers· Automotive	3	67	16	£35k	9	—	no pathway	18mo	thin
⚠️ Vulnerable	▶3417Photographers, audio-visual and broadcasting equipment operators· Computing & IT	0	62	38	£30k	33	—	no pathway	18mo	thin
⚠️ Vulnerable	▶2425Actuaries, economists and statisticians· Mathematics	93	57	44	£75k	9	40	2.3 yr	—	72
⚠️ Vulnerable	▶2421Chartered and certified accountants· Accounting	1,019	57	29	£47k	2	32	31.8 yr	24mo	40
⚠️ Vulnerable	▶2422Finance and investment analysts and advisers· Accounting	0	56	15	£48k	2	—	no pathway	24mo	20
⚠️ Vulnerable	▶5231Vehicle technicians, mechanics and electricians· Automotive	305	55	43	£37k	9	40	7.6 yr	18mo	102
⚠️ Vulnerable	▶5313Bricklayers· Bricklaying	87	54	27	£32k	6	40	2.2 yr	24mo	thin
⚠️ Vulnerable	▶5223Metal working production and maintenance fitters· Engineering	260	54	38	£40k	36	—	no pathway	18mo	64
⚠️ Vulnerable	▶5435Cooks· Culinary	17	53	17	£18k	13	—	no pathway	18mo	thin
⚠️ Vulnerable	▶5434Chefs· Culinary	510	52	36	£26k	13	40	12.8 yr	18mo	19
⚠️ Vulnerable	▶1136Human resource managers and directors· Creative & Design	134	52	40	£54k	16	—	no pathway	18mo	74
⚠️ Vulnerable	▶6221Hairdressers and barbers· Hairdressing	474	51	43	£15k	12	40	11.8 yr	12mo	72
⚠️ Vulnerable	▶6121Pest control officers· Childcare	416	51	17	£27k	9	—	no pathway	18mo	13
🏠 Sheltered	▶5221Metal machining setters and setter-operators· Engineering	18	49	33	£35k	36	—	no pathway	18mo	31
🏠 Sheltered	▶2413Solicitors and lawyers· Law	252	48	43	£101k	2	40	6.3 yr	—	45
🏠 Sheltered	▶5316Carpenters and joiners· Carpentry	335	48	37	£34k	18	40	8.4 yr	18mo	thin
🏠 Sheltered	▶3537Financial and accounting technicians· Accounting	4	47	15	£57k	2	—	no pathway	24mo	thin
🛡️ Overqualified	▶3538Business sales executives· Business & Management	70	47	64	£57k	51	40	1.8 yr	18mo	77
🏠 Sheltered	▶3111Laboratory technicians· Science	29	47	35	£27k	15	40	0.7 yr	24mo	thin
🏠 Sheltered	▶5113Gardeners and landscape gardeners· Horticulture	458	46	37	£27k	8	40	11.4 yr	24mo	93
🏠 Sheltered	▶6211Sports and leisure assistants· Sport & Fitness	66	46	31	£14k	15	—	no pathway	12mo	88
🏠 Sheltered	▶5319Construction and building trades n.e.c.· Construction	77	46	44	£34k	11	40	1.9 yr	18mo	101
🏠 Sheltered	▶6146Senior care workers· Nursing	30	46	33	£23k	25	—	no pathway	12mo	47
🏠 Sheltered	▶6145Care workers and home carers· Nursing	989	46	46	£23k	25	40	24.7 yr	12mo	23
🏠 Sheltered	▶6141Nursing auxiliaries and assistants· Nursing	645	46	35	£23k	27	—	no pathway	12mo	69
🏠 Sheltered	▶3534Financial accounts managers· Accounting	437	42	23	£45k	2	—	no pathway	24mo	89
🏠 Sheltered	▶2135Cyber security professionals· Computing & IT	7	42	37	£55k	11	40	0.2 yr	24mo	thin
🏠 Sheltered	▶2134Programmers and software development professionals· Computing & IT	2	42	16	£56k	6	—	no pathway	24mo	thin
🏠 Sheltered	▶2139Information technology professionals n.e.c.· Computing & IT	62	42	38	£50k	6	40	1.6 yr	24mo	68
🏠 Sheltered	▶3112Electrical and electronics technicians· Science	87	42	34	£35k	15	—	no pathway	24mo	102
🏠 Sheltered	▶2126Aerospace engineers· Engineering	32	42	26	£56k	36	—	no pathway	36mo	22
🏠 Sheltered	▶2436Construction project managers· Construction	21	40	37	£77k	11	—	no pathway	—	129
🏠 Sheltered	▶3113Engineering technicians· Engineering	29	40	42	£44k	51	—	no pathway	24mo	thin
🏠 Sheltered	▶3442Sports coaches, instructors and officials· Sport & Fitness	171	39	45	£49k	10	40	4.3 yr	24mo	116
🏠 Sheltered	▶2314Primary education teaching professionals· Childcare	881	39	15	£43k	3	—	no pathway	24mo	20
🛡️ Overqualified	▶3411Artists· Computing & IT	65	39	56	£49k	33	40	1.6 yr	18mo	thin
🏠 Sheltered	▶3223Housing officers· Counselling	0	38	12	£33k	6	—	no pathway	24mo	thin
🏠 Sheltered	▶2315Nursery education teaching professionals· Childcare	122	38	11	£31k	3	—	no pathway	24mo	9
🛡️ Overqualified	▶5249Electrical and electronic trades n.e.c.· Engineering	122	35	59	£48k	43	40	3.0 yr	18mo	66
🏠 Sheltered	▶5312Stonemasons and related trades· Construction	367	34	25	£34k	11	—	no pathway	18mo	69
🛡️ Overqualified	▶1132Marketing, sales, and advertising directors· Creative & Design	635	33	58	£90k	16	40	15.9 yr	18mo	50
🏠 Sheltered	▶6222Beauticians and related occupations· Beauty	1	32	42	£15k	7	40	0.0 yr	12mo	81
🏠 Sheltered	▶5241Electricians and electrical fitters· Electrical Installation	303	31	42	£39k	7	40	7.6 yr	24mo	117
🏠 Sheltered	▶2312Further education teaching professionals· Childcare	28	31	14	£39k	3	—	no pathway	24mo	23
🏠 Sheltered	▶3421Interior designers· Computing & IT	122	29	39	£35k	33	—	no pathway	18mo	thin
🏠 Sheltered	▶3416Arts officers, producers and directors· Computing & IT	32	29	39	£40k	33	—	no pathway	18mo	thin
🏠 Sheltered	▶2119Natural and social science professionals n.e.c.· Science	34	28	23	£42k	15	—	no pathway	24mo	thin
🏠 Sheltered	▶2215Dental practitioners· Nursing	18	27	17	—	3	—	no pathway	—	thin
🏠 Sheltered	▶2219Medical practitioners n.e.c.· Counselling	55	27	44	—	6	40	1.4 yr	—	94
🏠 Sheltered	▶5315Plumbers & heating and ventilating installers and repairers· Plumbing	267	27	35	£37k	5	40	6.7 yr	24mo	70
🏠 Sheltered	▶2443Probation officers· Criminology	25	25	31	£124k	1	12	2.1 yr	—	thin
🏠 Sheltered	▶2111Chemical scientists· Science	11	25	23	£40k	15	—	no pathway	24mo	thin
🏠 Sheltered	▶2113Biochemists and biomedical scientists· Science	10	24	24	£45k	15	—	no pathway	24mo	thin
🏠 Sheltered	▶2473Advertising and public relations professionals· Marketing	0	20	27	—	1	20	0.0 yr	18mo	thin
🏠 Sheltered	▶6143Dental nurses· Nursing	0	20	17	—	3	—	no pathway	24mo	thin
🏠 Sheltered	▶6147Care escorts· Health & Care	0	20	24	—	5	—	no pathway	12mo	thin
🏠 Sheltered	▶6122Childminders and related occupations· Childcare	0	20	38	—	9	40	0.0 yr	18mo	thin
🏠 Sheltered	▶3131IT operations technicians· Computing & IT	0	20	20	£35k	9	—	no pathway	24mo	39
🏠 Sheltered	▶2112Biological scientists· Science	66	20	24	£44k	15	—	no pathway	24mo	thin
🏠 Sheltered	▶2231Midwifery nurses· Nursing	800	16	38	£42k	22	40	20.0 yr	24mo	27
🏠 Sheltered	▶5112Horticultural trades· Horticulture	4	11	11	£25k	8	—	no pathway	24mo	thin

Reading the table: Exposure ≥ 50 = high (red); Capacity ≥ 50 = high (green). Click any column header to sort. UCM courses is the count of parent courses that lead toward this SOC — modules (IS-codes), APEL placeholders and empty shells filtered out. Grads/yr is the estimated annual UCM graduate output reaching this SOC. Yrs to absorb is censusWorkers ÷ grads/yr — the years it would take, at current UCM intake, to retrain the entire IoM workforce in this occupation. The numbers above 20 are the policy hotspots: even a perfect supply response would mature decades from now. Lag is the fastest qualifying UCM route in months. Click any row to see the actual UCM courses leading toward that SOC, with their length and automation chips.

Methodology — every input visible

AI exposure score (0-100)

Frey-Osborne × 0.55 + Anthropic automation × 0.25 + (1 - augmentation) × 0.20

Same composite used on Demand vs AI Risk and Shortages so the threshold (≥ 50) is consistent across the site.

Adaptive capacity score (0-100)

UCM pathway count — 25.0%
UCM annual graduate supply — 15.0%
Median wage (mobility proxy) — 20.0%
Training lag, inverted — 12.5%
Field diversity — 12.5%
Skill transferability — 15.0% (Manning's task-overlap analogue, IoM-grounded)

Each component normalised 0-100 against the IoM observed range, then weighted-averaged.

Cell thresholds: exposure ≥ 50 = high; capacity ≥ 50 = high. The four-cell split is a deliberate simplification — workers near the boundaries shift cells if you change either threshold by 5 points.

About the skill-transferability component: Manx-grounded analogue of Manning & Aguirre's O*NET task-overlap signal. For each SOC we build a vocabulary of skills + knowledge tags observed across IoM job listings (via Job → JobSkill / JobKnowledge), then count OTHER SOCs whose Jaccard similarity is ≥ 0.2. SOCs with fewer than 3 observed skill tags are excluded from scoring (insufficient signal). This is our IoM-grounded analogue of Manning & Aguirre's O*NET task-overlap signal — they ask “how many adjacent occupations share core tasks?”, we ask “how many adjacent occupations show overlapping skills + knowledge in actually-advertised Manx work?”. Limitation: SOCs with very few advertised listings on the Island will under-score, because the skill vocabulary is sparse — we floor SOCs with fewer than 3 observed skill tags to avoid false-positive matches.

Honest limitation: Manning & Aguirre's capacity index uses O*NET worker characteristics (educational attainment, age distribution, prior mobility) that we don't have at SOC level for the Isle of Man. Our proxy uses UCM pathway data + Manx-listing skill overlap instead, which captures a different signal — the ability to retrain INTO a role + observed skill spread across SOCs, rather than the ability of incumbents to transition OUT of one. The "vulnerable" cell here is therefore better read as "exposed roles where the retraining infrastructure is thin AND the skill set sits inside a narrow corridor" than as "exposed workers who personally can't adapt". Source: NBER Working Paper w34705.

🌍

Want the wider picture? This page covers the 66 UCM-served 4-digit SOCs (11,760 census workers). The whole IoM workforce sits on a parallel page — /education/adaptive-capacity-census — at 3-digit minor-group granularity, including the no-pathway clusters where UCM doesn't teach toward the role at all. Same framework, wider denominator. Read both.

Where to go next: Census-wide quadrant · Transition Era · 20-yr projections · Shortages · All UCM courses · All careers