Seven observations
and predictions,
fifteen years on.
In 2009–10, as part of an MBA at Warwick Business School, I built a model of how the Manx economy actually works: four kinds of industry competing for the same limited people, homes and money. In January 2011 a discussion paper applying that model went to the Isle of Man Treasury. Both documents made dated, checkable predictions, on paper, supervised, timestamped. Fifteen years of official data now exist that neither document could have seen. Here is the score. Misses included.
How to read this page: every claim below is quoted verbatim from the 2010 dissertation or the 2011 Treasury paper, old-fashioned punctuation and all. The verdicts are mine; the data isn’t. Sources: Isle of Man National Income accounts, the censuses, Treasury’s own Pink and Blue Books, ONS, Moody’s, and the HMRC–Isle of Man customs accounts. Where a figure can’t be checked against a source independent of the Isle of Man Government, I say so rather than leaving you to assume it can.
1 · The Island would run out of workers, and its lead over the UK couldn’t last
“It finds at structurally-full employment, reached on the IOM somewhere between 1994 and 2002, both hypotheses result in allocative inefficiencies and are damaging. In either case, the Island cannot sustain per-capita growth ahead of the UK in the long-run.”
The model names the limit explicitly: “the limit to growth is recognised as the cost of labour.”
Full employment, 1994–2002 — confirmed, almost to the year. Manx unemployment peaked at 4.9% in 1993, fell to 3.5% by 1996 and 0.6% by 2000, and has never come off the floor since. It was 0.7% when Moody’s reaffirmed the Island’s credit rating in December 2025 — a report that names labour shortages as our single biggest challenge. The Island’s reserve of spare workers was used up inside the window the model named, and we have been at full stretch for a quarter of a century.
The limit being the price of labour — confirmed. Manx and UK median full-time earnings are now at parity, and have been for as long as the published record runs: within a percentage point in 2016, in 2021 and in 2025, with the 2025 survey noting its own gap sits inside the margin of error. The Island’s old advantage — cheaper labour than its neighbours — is gone. Recruiters report salary rises of around 25% in a few years in trust and corporate services.
The convergence call — right about the mechanism, late on the timing. Measured from 1998, Manx income per head pulled roughly 70% ahead of the UK’s by 2019/20. That is the opposite of what I expected, and it held for two decades — sustained, in large part, by the same attraction-led policy toward e-gaming and other footloose sectors that runs through the rest of this page. Then it turned: by 2023 nearly two-thirds of that lead had unwound, and it stood around a third ahead in 2024. The correction the 2011 paper said “must ultimately result” did arrive. It simply took until 2020 to start.
So the limit was real and correctly dated, and the correction was real and badly timed. I said the Island couldn’t sustain a widening lead; it sustained one for twenty years, and I’d have lost money betting against it. What I wouldn’t have lost is the direction.
What “futile” meant, and didn’t. The futility claim was about industrial targeting, and it meant ultimately futile: government picking sectors and incentivising them into an economy that had already hit its limits of people, homes and capital produces gains that don’t last. The short-term successes along the way — e-gaming’s rise chief among them — aren’t evidence against that. They were bubbles, on the model’s own terms, and bubbles burst. It was never a claim that government itself is pointless. There was, and still is, a full agenda of active work that isn’t targeting: opening market access, cutting the penalty of distance through connectivity, making sure capacity actually exists, building houses, and keeping the economy balanced toward the UK.
2 · Favour one industry and it prices out the rest
“If one industry is comparatively incentivised, it will price out other industries.”
e-Gaming and ICT took an extra 11.8 percentage points of the whole economy between 2007 and 2023 — from the Isle of Man’s own sector accounts. On an island already at full employment, an incentivised industry doesn’t grow alongside the others. It bids the same workers, the same homes and the same capital away from them, and the pay data shows where they went: legal and accountancy services, medical services and e-gaming sit at the top of the Manx earnings table.
The counter-signal is worth stating too, because it complicates the story. In 2025, 27.4% of Manx employees earned below the Living Wage, up from 20.2% a year earlier. This isn’t a labour market where everyone’s pay got bid up. It’s a barbell — a well-paid top, a fast-growing low-paid tail, and a squeeze in between.
3 · And the rest decline in a predictable order
“Industries collapse in sequence — beginning with those most exposed to substitution — e.g. retail and tourism.”
The census employment shares tell it in the order the dissertation named: manufacturing from 8.2% of the workforce to 4.7% between 2001 and 2021, retail from 9.3% to 8.3%, tourist accommodation from 1.7% to 1.3% — all while the workforce itself grew 11%. Relative decline, correctly ordered by exposure to substitution. Between the 2011 and 2021 censuses alone, transport lost 1,190 jobs, banking 1,076, insurance 605, wholesale distribution 489, manufacturing 236, and agriculture and fishing 186.
I’d note one revision, and one open question, rather than a clean climbdown. The revision: on the census employment data alone, “collapse” is too strong a word. These are relative declines within a workforce that grew. The open question: the Chamber of Commerce puts job losses at over 1,000, and none of that shows up in the official unemployment rate — because that rate counts Jobseeker’s Allowance claimants, and most of the professionals in the sectors losing jobs don’t qualify for JSA. On that reading, “collapse” isn’t drama, it’s a long slide the headline figure is structurally unable to see. I can’t currently tell you which reading is right, because we don’t measure the labour market properly — the same gap I flag on Our Place in the World: real unemployment and underemployment statistics, not just a JSA count.
4 · The VAT arrangement was a loop that fed itself, until the UK curbed it
“The VAT arrangement is a reinforcing loop — spending raises the Island’s share, which funds spending — curbed with the UK-enforced 2009 renegotiation.”
Customs and Excise receipts climbed steeply from the late 1990s to a mid-2000s peak, reaching roughly 72% of everything Treasury took in, then were cut back hard after the UK rebased the formula in 2009 and again in 2011 — a loss independently reported at around £349m a year in real terms, a 46% drop, between a quarter and 40% of revenue. FERSA, from 2016, now throttles the loop: a five-yearly survey with a retrospective true-up. The replacement arrangement was executed on 11 April 2025 with a £3.116m backdated adjustment, confirmed in the HMRC account audited by the UK National Audit Office.
The mechanism is visible in the record, and the UK closed it much as the paper expected. I’ve left the precise annual receipt figures off this page: the versions I hold don’t reconcile cleanly with the independent series, and I’d rather show you a mechanism I can stand behind than four decimal places I can’t.
5 · The Island would export its young and import experience
“Industrial targeting… leads to a forced emigration of young Manx people in favour of immigration of experienced individuals.”
Deaths have exceeded births every year since 2016 — in 2024, 351 more deaths than births. Every bit of population growth now comes from net inward migration, running at roughly 650 a year. The direction is as predicted.
The composition — precisely who leaves and who arrives — is harder to evidence from published statistics than I implied in 2010. I’ll claim the outcome. I won’t claim the mechanism is proven until the data can carry it.
6 · The world would eventually close in on tax-driven growth
“The international response (multi-lateral tax code simplification) will take some time, although it must ultimately succeed.”
The OECD’s Pillar Two global minimum tax applies here from 1 January 2025 — a 15% floor, with the Isle of Man adopting both the income inclusion rule and a domestic top-up tax, forecast to raise £31m in 2026-27. Slower than I expected, and it arrived by a route I didn’t anticipate. But it arrived.
7 · The crash I predicted hasn’t happened
“Localised economic crash is certain if current policy continues.”
It didn’t happen. What came instead was a long rotation — e-gaming replacing what went before, and the FERSA-era rebuild of public revenue — followed by contraction from 2020 rather than a sudden break. “Certain” was too strong a word, and the trigger I named didn’t fire the way I described it. I record this as a miss and I’m not going to argue it into a hit.
The honest reading is that this and Prediction 1 are the same mechanism seen twice. The 2011 paper said the price gap between the Island and the UK “must ultimately result in market correction”, and that when a crash came the size of that gap would govern how long it lasted. What arrived was the correction without the crash: a four-year grind from 2020 rather than a break. Better for the Island than what I forecast. Still not nothing.
The one recommendation I’d most like back
“Economic figures produced to international standards; leverage the Worldbank or UK’s ONS.”
— governance recommendations, discussion paper to the Isle of Man Treasury, January 2011
It hasn’t happened. No international body checks how the Isle of Man’s national accounts are put together. We are assessed on plenty — the IMF looked at our financial regulation in 2003 and 2009, MONEYVAL evaluates us on money laundering and is back this October, the OECD Global Forum monitors tax transparency — but never on how we measure our own economy. Manx figures do appear in World Bank tables. They are there because we supplied them.
We cannot join the IMF: membership turns on controlling your own external relations, which is why Hong Kong, Aruba and Greenland are not members either. We do not need to. Article IV surveillance, the IMF’s annual economic check-up, can be extended to a territory if the member state asks. The Netherlands gets it for Aruba, Curaçao and Sint Maarten, and those reviews look directly at national accounts. Britain asked for Anguilla and Montserrat in 2011. It has never asked for us, and we have never asked it to.
It would cost us: officials’ time, and real work building balance-of-payments statistics we do not currently produce. It would also mean a published grade on our data that we could not control. Worth it. Being marked by someone who does not work for us is the point, not the drawback.
The model itself
In 2026 I rebuilt the 2010 model as working code and ran it against the historical record. Fed nothing but UK growth, the structure reproduces the shape of the Island’s path — the rise and amputation of the VAT receipts, and the population plateau. The rebuilt model and its interactive simulator are live on the Observatory, clearly marked draft and not to be relied on — that label stays up until the re-estimation against the Isle of Man’s own constant-price accounts is finished and published in full.
The argument does not rest on the arithmetic anyway: on an island with fixed people and homes, promoting a sector moves workers and houses around rather than adding any. Homes first, then people, then growth. A working paper is in preparation. The dissertation, the 2011 Treasury paper, the model code and the sourcing are available to anyone who asks — and I would sooner someone checked them than believed me.
This page is not an “I told you so”. Being wrong would have been the cheaper lesson. The point is narrower: the Island’s situation was diagnosable in 2010, in writing, on the record — and the cost of not acting on that diagnosis is what every other page of this site is about.
What a voter is entitled to ask is not whether I get things right. It is what I do when I don’t. Everything I have had to correct is dated and public on the Corrections Log.