The essay worth reading
Alex Imas published a piece called "What will be scarce? The economics of structural change and the post-commodity future of work" on his Substack (Ghosts of Electricity) on 14 April 2026. It's the clearest articulation we've read this year of a structural-change thesis for the AI transition — and one we keep bumping into on the Smart Island data without having had the vocabulary to describe it. Worth the twenty minutes.
Imas's core argument, compressed: as AI makes commodities cheap, wealthy societies will reallocate spending and employment toward relational sectors — domains where the human being doing the work is not a means to an end but is the end. Not because machines can't physically produce the outputs, but because people, as they get richer, increasingly want the human element itself. The historical analogy is the shift from agriculture (40% of US employment in 1900, 2% by 2000) to services. We don't eat less food now; we just spend a dramatically smaller share on growing it and a larger share on everything else.
The specific categories Imas names as durable: nurses, therapists, teachers, boutique fitness instructors, personal chefs, bespoke tailors, craft brewers, live performers, spiritual guides, childcare workers, hospitality staff. His best line: "You don't need to be Picasso. You need to be the person whose involvement makes the product feel like it was made for someone, by someone."
The essay is measured, well-grounded in structural-change economics (Baumol's cost disease, income elasticity, nonhomothetic preferences), and honest about its limits — Imas flags that the framework works for developed economies, and that developing nations face messier transitions.
We've read the thesis against a year's worth of Smart Island data on the Isle of Man. It holds directionally. It needs three specific Manx nuances.
Where the data confirms him
1. Our per-SOC AI verdicts map cleanly onto his "relational" category. Every Isle of Man occupation we've classified sits in one of four buckets — resilient, augmented, mixed, at-risk — using the Anthropic Economic Index, Frey-Osborne, and Felten-Raj-Seamans exposure indices weighted by Manx census workforce. The roles that score resilient or augmented are, almost verbatim, Imas's list: nurses, care workers, primary teachers, counsellors, personal trainers, hairdressers, plumbers, chefs, childcare workers, social workers. The roles scored at-risk are, again almost verbatim, Imas's commodity category: paralegal, routine accountancy, marketing operations, junior software, customer service, basic admin. The agreement isn't coincidence — Imas is drawing from the same underlying literature we've built our models on — but it does mean the Island has a quantitative test of his thesis rather than a polemical one. Browse /occupations and filter by AI verdict; the emerald (resilient) and sky (augmented) chips concentrate in exactly the sectors he names.
2. Our shortage list is a shopping list for relational labour. On /education/shortages, the occupations flagged as having live IoM demand today AND long training pipelines are dominated by the relational cluster. Nurses, mental health practitioners, plumbers, electricians, childcare workers, teaching assistants. The 2+ year training pipelines Imas's framework predicts should become scarcer as demand rises — and our data is showing exactly that scarcity. The training-lag reality-check strip on that page tells you 67% of the Island's current shortage demand sits on a 2+ year pipeline; that number is heavy because the resilient/augmented categories Imas describes take time to train into.
3. The UCM-led-skills argument we've made separately lines up with Imas's implicit pedagogy. Our Transition Era piece draws on the Stanford AI Index, WEF Future of Jobs, McKinsey and OECD; it concludes the durable skills are judgement, AI/data governance, personal skills, domain depth, adaptability. That's the same skill profile Imas says wins in a relational economy — because the common thread isn't what you know, it's your ability to be present, to exercise judgement for a specific person in a specific context. The UCM blog post argues this pedagogy is what UCM is structurally built to deliver — Imas's essay is the clearest macroeconomic justification for that argument we've seen yet.
Where the data complicates him
1. His income-elasticity mechanism isn't firing on the Island yet. Imas's thesis leans on a specific behavioural claim: as household incomes rise, people spend disproportionately more on relational goods. That's true in aggregate across developed economies — but the Isle of Man's Personal Income has been essentially flat in real terms since 2019. Treasury data published in April 2026 (which we've discussed under the Transition Era framing) shows real Personal Income at roughly £1.92bn in 2023/24, basically unchanged for half a decade. If incomes haven't been rising in purchasing-power terms, the Imas reallocation mechanism — "wealthier people buy more relational goods" — has nothing to push against. The thesis may be latent on the Island but not yet active. The workforce projections page makes this concrete: the Embracing AI scenario produces the Imas-style reallocation (augmented workers capture productivity uplift, relational sectors grow), but the Status Quo and Hollowing scenarios don't — they just compress the middle and leave the relational roles under-supplied.
2. Imas doesn't address housing, and the Island's housing cost is the binding constraint on relational-sector scale. This is the gap we think is most underweighted in the essay. Nurses, care workers, primary teachers, childcare staff — the people Imas is right to flag as durable — are paid toward the lower-middle of the IoM wage distribution. The Island's housing affordability crisis means that even when employer demand for these roles is high (and it is), the people who'd fill them can't afford to live here. Housing is the rate-limiting step on the relational economy, not AI automation. A framework that predicts relational sectors absorb growing employment needs to reckon with whether those workers can physically live near the people they serve. Imas's essay is silent on this; on the Isle of Man it's the central tension.
3. Financial services — our biggest sector — doesn't fit either of his buckets cleanly. Imas's dichotomy is commodity cognitive work (automatable) vs relational work (not). IoM financial services is a mix of both inside the same firms, sometimes in the same roles. Routine compliance checks, first-pass tax preparation, standard trust filings — that's commodity and AI-automatable. But senior client-facing advisory — the same professional explaining the same Island jurisdictional nuance to the same high-net-worth family for twenty years — is as relational as therapy. And the sector's ITIP contribution (the chunk that funds public services on the Island) rests on the latter not the former. Imas's framework correctly predicts compression at the junior end; it doesn't capture how Manx fin-services firms will restructure to keep the relational senior tier while AI takes the commodity middle. Our sector deep-dives on /data/sector wrestle with this explicitly; the essay doesn't.
Where we extend him
Imas argues reallocation will happen structurally. Our projection model argues it's contingent on policy. The three scenarios on /education/projections are designed to test the same question from the other end:
- Embracing AI (UCM embeds AI fluency across programmes; NAIO drives adoption through trusted industry partners; augmented roles compound productivity) — Real Personal Income rises to ~£2.35bn by 2046. Mid-level holds. Relational sectors grow. This is the Imas reallocation landing on the Isle of Man.
- Status Quo (passive AI adoption, no UCM curriculum pivot, no active NAIO brokerage) — Real income stays flat at £1.95bn. Mid-level thins. Relational demand is there but unfillable. The mechanism is there but can't exert itself because the skills infrastructure hasn't updated.
- AI Hollowing (AI deployment races ahead of reskilling; mid-level collapses; no reskilling pathway) — Real income falls to £1.55bn. Relational sectors shrink because the household wage bill that pays for them is shrinking. Imas's mechanism goes into reverse.
The non-obvious policy implication: on the Isle of Man specifically, which of Imas's futures we land in is determined by UCM and NAIO more than by AI itself. The technology creates the possibility space; the institutional response picks the point inside it. That's not something Imas addresses because it's a jurisdiction-scale question; but on an Island of 85,000 people it's the only question.
The short version
Imas is right that relational sectors are where durable employment and value consolidate as AI eats commodity work. The Smart Island data confirms his categorisation almost role-for-role. What his essay underweights for an Isle-of-Man context: the income-rising mechanism he needs isn't firing on the Island yet; housing affordability is the real rate-limiter on relational-sector scale, not automation; and financial services is a mixed bucket that his dichotomy handles poorly.
What his framework IS very useful for: giving the Island a macroeconomic vocabulary for the case we've been making separately. UCM-led skills pivots aren't a nice-to-have; they're the institutional precondition for the Imas reallocation landing here. NAIO-led adoption-through-industry-partners is the other precondition. /education/transition-era makes the policy case; the Imas essay is the macroeconomic backing.
Read his essay. Then run our workforce projections with Embracing AI selected. The numbers are different. The story is the same.
— Alex Imas, "What will be scarce?", Ghosts of Electricity, 14 April 2026: aleximas.substack.com/p/what-will-be-scarce.