The Shibboleth Premium

There is a job advertisement doing the rounds — there are, in fact, dozens of them, near-identical, refreshed monthly — for a modelling actuary in a bulk purchase annuity team. The role is what it always is: build and maintain the cash flow models, run the Solvency II balance sheet, feed the IFRS 17 engine, survive the audit. And there, sitting in the essential criteria like a bouncer at the wrong door, is the line: DB pensions experience required.

It is worth pausing on what this requirement actually claims. It claims that the binding constraint on modelling defined benefit annuities — the scarce, hard-won, years-to-acquire skill — is familiarity with the benefits themselves. Not the capital framework. Not the accounting standard. Not the modelling platform. The benefits.

Let us test that claim against the product. A bulk annuity, once it is on an insurer's books, is among the simplest liabilities in life insurance. It is a decrement table, an indexation vector, and a discount curve. There is no policyholder behaviour to model, because the policyholder has no options worth speaking of. There are no unit funds, no bonus philosophy, no lapse dynamics, no premium persistency. A pension in payment does exactly one interesting thing, and it does it once.

The complexity — the real complexity, the kind that generates six-figure errors and career-ending audit findings — lives entirely elsewhere. It lives in the matching adjustment: eligibility, attestation, the consequences of breaching it. It lives in the SCR aggregation, the transitional measures, the risk margin. It lives in IFRS 17's contractual service margin, its coverage units, its transition cohorts, and the miserable reconciliation between the accounting balance sheet and the regulatory one. It lives in the discipline of production reporting itself: run cycles, controls, analysis of change, the accumulated judgement of knowing which number looks wrong before you can prove it.

Every item on that list is the bread and butter of a life insurance modelling actuary. Not one of them is taught in a pensions consultancy.

The fortnight and the three years

So consider the two candidates the industry could hire, and what each must learn.

The life actuary joining a BPA team must learn the benefit taxonomy of UK occupational schemes: GMP mechanics and their equalisation, revaluation in deferment, the zoology of pension increases — LPI at five per cent, LPI at two and a half, fixed, statutory, and the mongrel combinations that vary by tranche of service. This is genuinely fiddly. It is also entirely rule-based. Every one of those rules is written down — in scheme documentation, in legislation, in the benefit specification that accompanies every transaction. There is no judgement in it, no craft, no tacit knowledge. It is reference material. A competent actuary absorbs the working vocabulary in a fortnight and the full grammar within a reporting cycle.

Now run the tape the other way. The pensions actuary joining a life insurer must learn: a production modelling platform (a year to real competence, on a good day); the entire Solvency II balance sheet and its internal politics; IFRS 17 from a standing start; and the craft of regulated financial reporting, which cannot be read in a document because it lives in the scar tissue of people who have closed period-ends under audit. Two to three years to genuine independence, and that is the optimistic case.

The learning-curve asymmetry is not close. It is a fortnight against three years. And yet the industry's hiring filters point in precisely the wrong direction: the scheme-side consultant who has never closed an insurance reporting period walks through the door as a natural fit, while the life actuary who has built the exact machinery the role exists to operate is rejected for lacking "pensions experience." If you set out to design a screen that maximised time-to-productivity while paying a premium for the privilege, you could not improve on it.

Who pays, and who decides

The economics of this are not mysterious; they are merely unexamined. The bulk annuity market has grown explosively — eight or nine insurers scaling their teams simultaneously, all fishing in the same pond, because the "prior BPA experience" requirement guarantees the pond cannot grow. The result is exactly what a first-year economics student would predict from an artificial supply restriction: the anointed circulate between the same handful of employers at escalating salaries, while equivalent skill sits outside the fence, unpriced.

Call it the shibboleth premium: the surcharge an industry pays, indefinitely, to avoid teaching a new hire the word "franking." A firm paying, conservatively, a fifteen or twenty per cent scarcity premium across a modelling team, forever, in order to save a fortnight of familiarisation per hire, has not made a judgement about skills. It has made an accounting error and institutionalised it.

Why does the error persist? Because the person who imposes the requirement does not bear its cost. The hiring manager who insists on the exact-match candidate is buying insurance for himself, not capability for his employer: nobody was ever blamed for hiring the person who had done the identical job across the road. The recruiter, paid on placement, pattern-matches CVs to the incumbent workforce because that is what clears the client's filter. HR converts the whole thing into a keyword search. At no point in this chain does anyone ask the only question that matters — what does this role actually require, and how long does the gap take to close? — because at no point does anyone have an incentive to ask it. It is a principal–agent problem wearing a competency framework.

The sociology underneath

There is a deeper layer still. The bulk annuity industry did not emerge from life insurance; it emerged from pensions consulting. Its founders, its deal teams, its client relationships, its dinner-party circuit — all came from the scheme side. "Pensions person" is the in-group identity, and the hiring template exists to reproduce it. The DB requirement was never a reasoned position about skill transferability. It is the residue of who built these firms, defended after the fact with justifications that dissolve on contact with what the modelling job actually involves.

This is why the requirement cannot be argued away in a screening call. You cannot reason someone out of a position they were never reasoned into. The gatekeeper applying the checklist did not write it and does not understand it; the person who could waive it — the head of modelling who lives daily with the SII and IFRS 17 skills shortage and knows exactly which side of the asymmetry hurts — never sees the CVs the checklist has already killed.

And there is an uglier property of experience requirements generally, which deserves naming. "Must have N years in X" is the most legally comfortable filter ever devised, precisely because of what it proxies for without saying so. It screens for career path, which screens for cohort, which screens for age, background, and route into the profession — all while remaining immaculately defensible on paper. An industry that would never write "people like us" in a job specification has found a formulation that achieves the same result and survives an employment tribunal. Whether any individual firm intends this is beside the point; the filter does not need intent to do its work. Arbitrary experience fences always discriminate. That is what fences are for.

What it costs

The bill is not only paid in salaries. It is paid in the quality of the work. The genuinely hard problems in bulk annuities are drifting steadily away from the benefit side and toward the asset side and the structural side: matching adjustment portfolios stuffed with illiquid assets, funded reinsurance chains, the modelling of things that were never meant to sit behind a pension. These are systems problems — architecture, data pipelines, model design — and the people best equipped for them are precisely the ones the pensions shibboleth screens out. An industry congratulating itself on the depth of its GMP knowledge while its real risks migrate somewhere its hiring filter cannot see is not being prudent. It is checking passports at the loading dock while the factory floor runs unsupervised.

None of this requires bad faith to explain, which is what makes it durable. Every actor in the chain is behaving rationally within their own incentives, and the sum of that rationality is an industry that pays a permanent premium for a two-week vocabulary, filters out the people who have already mastered the hard part, and calls the result a talent shortage.

It is not a talent shortage. It is a door policy. And the club is paying its members handsomely to keep pretending otherwise.