Lessons were learned

Posted on: 7 June 2026

A few evenings ago I had dinner with a friend who manages money for one of Britain's large pension funds. Not just any fund: one of the defined benefit schemes, where the employer pays in the contributions and the fund undertakes to give the worker a pension fixed in advance, whatever the markets happen to do in the decades in between. It is a trade that lives on a precise tension and on equally precise interlocking parts; on one side a liability that is certain and sits far off in the future, the pension owed in twenty years, and on the other a pool of money that has to be invested today so as to honour that promise without being swept away by the movement of interest rates.

At some point between courses I asked him whether 2008 had taught the industry anything. He smiled in the way people smile when they have known the answer for too long, and said, almost cheerfully, that the toxic stuff was still there, that the racier end of finance was alive and in good health and had simply changed its name. That smile is where everything else began, because he was right, though in a sense more precise and more unsettling than the one he intended. To see it you have to look inside two mechanisms that almost nobody outside the trade knows in any detail.

He lives in the first of them every day. A defined benefit fund has to protect itself against the movement of rates, because the value of the pension it has promised moves with rates. To do that, British schemes adopted long ago the strategies known as LDI, liability driven investment, which knead together leverage and interest rate derivatives so as to cover a great deal of risk with very little capital. Put without euphemism, a pension fund behaving like a leveraged bank, and it works for as long as rates move slowly. In September 2022 they did not move slowly. The Truss government's mini-budget of 23 September sent gilt yields up at a speed no one had seen, collateral calls were triggered on the derivatives, and the LDI funds found themselves forced to sell gilts at the very moment prices were collapsing, feeding the fall that was strangling them. The Bank of England had to step in on 28 September, buying gilts up to £65 billion to halt the spiral, and an analysis by the Bank itself estimates that forced selling by LDI funds accounted for roughly half the fall in prices over those few days. Keep that picture in mind, because it is the whole point: the part of the pension system that ought to be the most cautious of all had become so loaded with hidden leverage that it could threaten the stability of an entire G7 government bond market.

I came upon the second mechanism by asking how insurers manage to guarantee the lifetime annuities, the ones that pay a cheque for life to someone who has retired. One of the raw materials is the homes of the old. There is a product called equity release, the lifetime mortgage, in which an elderly person puts up their home as security and receives money without making any repayments; the interest rolls up, and the debt is settled when the house is sold on their death. There is even a variant in which the owner sells all or part of the home in return for a sum and the right to go on living there to the end. Insurers buy these loans because their repayment profile, tied to the span of a life, resembles the profile of the annuities they have to pay.

There is something a continental eye still finds strange about pledging the house in order to draw an income, since where I come from the house is the thing you hand on to your children, not a raw material to be fed into an insurer's balance sheet. In Britain it is an ordinary and deep market. So far, in any case, nothing scandalous. The scandal, put more clinically, lies in what is done with these loans next.

A loan of that kind has a defect from the supervisor's point of view: its cash flows are not fixed, because no one knows when the person will die or what the house will be worth, and there is a guarantee that prevents the lender from asking the estate for more than the sale proceeds. An asset like that does not qualify for the capital relief insurers want, the so-called matching adjustment. The solution is twenty years old. You take the loan and you slice it in two. The PRA, the British prudential regulator, set this out in writing as early as February 2015: to make these loans eligible, firms have to split their cash flows into a fixed coupon note and a variable note, sending the fixed one into the favoured portfolio and consigning the variable one, the piece that absorbs the risk, to another part of the business.

Read that sentence carefully, then cast your mind back to 2008. It is identical. You take a risk that is hard to price, you divide it into a senior slice you call safe and a junior slice that keeps all the toxicity, and you arrange for the safe slice to receive the favourable treatment. It is the same alchemy that turned subprime mortgages into bonds carrying a seal of safety, and it is not some niche conference curiosity, because in December 2024 Aviva took the structure into the public market with a securitisation of around £1 billion built on legacy equity release portfolios.

This is why my friend was only half right. They have not changed the name. The name of the mechanism is exactly what it always was, securitisation to obtain capital relief. What has changed is the address. In 2008 the raw material was the houses of American families who could not afford them; today it is the liabilities of pension funds and the homes of British pensioners. They did not change the name. They changed the victim.

And here comes the part that interests me most, because it turns the cliché on its head. We are told that 2008 taught us nothing. That is false. 2008 was learned, and learned superbly, by the regulators, who learned to watch the instruments under the names they carried then, and it is precisely for that reason that the engine moved on. When a rule strikes a label, the financial engineering migrates towards the next label, the one not yet watched. This is a principle known as Goodhart's law: the moment a measure becomes a target, it stops being a good measure. Regulate subprime and the risk reappears dressed as a senior note backed by lifetime mortgages; regulate the senior note and it will reappear somewhere else again. The regulator chases the name while the engine keeps its function and changes its bodywork.

The question, then, is not whether these structures hold in normal times, since in normal times almost everything holds. The question is whether they hold under a fast, correlated shock with no lender of last resort standing behind them. In September 2022 the answer came back, and it was no, the Bank of England had to step in. At the start of 2025 a second and smaller test was absorbed in orderly fashion, with the larger buffers imposed after the fright. The prediction I am willing to make, and one anyone will be able to test over the coming years, is a plain one: the next sharp move in rates will once again produce cascading collateral calls or strains on the property guarantees, and once again someone will have to choose between forced sales and a public rescue. If it does not happen, I will have been wrong, and I will be the first to be glad of it.

My friend, meanwhile, goes on managing money inside that system. He understands exactly how it works, down to the details he set out over dinner with the composure of a man who has handled them for decades. And that is the most uncomfortable thing about the evening: not that someone fails to understand, but that everyone understands, and the machine carries on regardless.