The 2022 Gilt Crisis: When Pension Hedges Broke the Bond Market

At eleven o'clock on the morning of Wednesday 28 September 2022, the Bank of England published a short statement that reversed the direction of its own policy. Six days earlier the Monetary Policy Committee had raised Bank Rate to 2.25 per cent and confirmed that the Bank would begin selling gilts out of its portfolio in October. Now the Bank said it would buy them instead — up to £5 billion a day, for thirteen working days, in maturities of twenty years and longer. Its reason had nothing to do with inflation or the exchange rate. "Were dysfunction in this market to continue or worsen," the statement read, "there would be a material risk to UK financial stability."

What had become dysfunctional was the thirty-year gilt, the least exciting instrument in British finance. Its yield had risen from about 3.5 per cent on the Thursday before to just above 5.1 per cent that morning, a level last seen in 2002. Gilts move perhaps ten basis points on a busy day. They had moved more than a hundred and fifty in four sessions, and the selling was coming from the institutions that are supposed to be the natural buyers of long-dated government debt: British corporate pension schemes, which between them owned a substantial share of the market and were being forced to sell into it.

Twenty-Six Minutes

Kwasi Kwarteng had been Chancellor of the Exchequer for seventeen days when he stood up in the Commons on Friday 23 September 2022. The document he presented was formally titled The Growth Plan 2022 and was not a Budget, a distinction with a procedural consequence: it carried no accompanying forecast from the Office for Budget Responsibility, the fiscal watchdog created in 2010 precisely so that chancellors could not mark their own homework.

He spoke for about twenty-six minutes. The additional 45 per cent rate of income tax would be abolished. The planned rise in corporation tax from 19 to 25 per cent was cancelled. The 1.25-percentage-point rise in National Insurance, introduced six months earlier, was reversed. The basic rate of income tax would fall to 19 per cent a year early. Stamp duty thresholds rose, the cap on bankers' bonuses went, and a package of investment zones was announced. "We need a new approach for a new era, focused on growth," Kwarteng told the House.

The measures came to roughly £45 billion a year in foregone revenue by 2026-27, on top of the Energy Price Guarantee announced on 8 September, whose cost over six months was put at around £60 billion. Paul Johnson of the Institute for Fiscal Studies described it that afternoon as the biggest tax-cutting event since Anthony Barber's budget of 1972, a comparison no British chancellor wants, since the Barber boom ended in the secondary banking crisis and a three-day week.

Gilts sold off that afternoon and sterling fell to $1.0857. The real move came when Asia opened on Monday: the pound printed $1.0350 in thin trading, its lowest level against the dollar since decimalisation in 1971. On Tuesday 27 September the International Monetary Fund issued a statement about a G7 economy of a kind normally reserved for emerging markets, saying it did "not recommend large and untargeted fiscal packages at this juncture". Britain's last encounter with the Fund on these terms is recounted in the 1976 sterling crisis.

Journalists reached for the currency because currencies are legible. The damage was in the gilt curve, and it was being done by a mechanism almost nobody outside the pensions industry could describe.

What a Pension Scheme Was Actually Doing

A defined-benefit pension promise is a long-dated, inflation-linked liability. A scheme that owes a retired engineer £20,000 a year, uprated with prices, for the next thirty years holds something that behaves exactly like a very long index-linked bond — and under the accounting rules that took hold in Britain after FRS 17 in 2000, that liability had to be discounted at market rates and carried on the sponsoring company's balance sheet. A fall in long yields raised the present value of the promise. Deficits appeared in corporate accounts without anything happening in the business.

Boots drew the logical conclusion first. In 2001 its scheme, advised by John Ralfe, sold essentially its entire equity portfolio and bought long-dated and index-linked gilts, matching the liability rather than trying to outrun it. The industry that grew up around that idea called itself liability-driven investment, and it had an arithmetic problem from the start. Matching a thirty-year liability pound for pound with thirty-year gilts means holding almost nothing else, and a scheme in deficit cannot afford to abandon the growth assets it is relying on to close the gap.

Leverage solved it. Instead of buying £100 of gilts, a scheme could post £20 of collateral, enter a gilt repo or an interest rate swap, and obtain the same £100 of interest-rate exposure — leaving £80 to invest in equities, credit and property. Sarah Breeden, then the Bank's executive director for financial stability strategy, later put the typical leverage in the pooled corner of the market at three to seven times (Breeden, 2022). Two decades of falling yields, accelerated by quantitative easing after 2009, made the hedge look like the single best decision any trustee board had taken.

By 2021 the Investment Association put the notional value of liability-driven mandates at close to £1.5 trillion, against roughly £400 billion a decade earlier. Around 60 per cent of the schemes using the strategy did so through pooled funds — vehicles shared by many small employers, run by a handful of managers, and slow to react, because a capital call on a pooled fund has to travel out to dozens of trustee boards and back.

Leverage has a price, and the price is collateral. When gilt yields rise, the value of the hedge falls, and the counterparty on the other side of the repo or the swap wants more margin. A scheme that cannot post it must reduce the position. Reducing the position means selling gilts.

The Loop

That is the whole mechanism, and it is a loop. Higher yields trigger collateral calls; collateral calls force gilt sales; gilt sales push yields higher; higher yields trigger further collateral calls. Gabor Pinter's post-mortem for the Bank found that pension and liability-driven investors were consistent net sellers of long-dated gilts through the episode, and that dealers — constrained by their own balance sheets — absorbed only part of the flow before stepping back (Pinter, 2023).

Jon Cunliffe, the Deputy Governor for Financial Stability, set out the sequence in a letter to the Treasury Committee on 5 October. Thirty-year yields had moved around 130 basis points in the four days following the fiscal statement, against a normal daily move of roughly 10 (Cunliffe, 2022). Index-linked gilts, where pension demand is most concentrated and the free float smallest, were worse: real yields on long linkers rose by more in a week than in the preceding decade, and some auctions found almost no bid at all.

Monday 26 September was the day the market's plumbing began to fail. Repo counterparties widened haircuts. Some LDI managers were unable to complete sales at quoted prices. Outside the wholesale market, lenders withdrew more than 1,600 residential mortgage products in five days, according to Moneyfacts, because nobody could price a five-year fix against a swap curve that was moving in figures rather than basis points.

By the evening of Tuesday 27 September, the Bank was being told by managers that a further rise of around 100 basis points would leave a number of pooled LDI funds with negative net asset value. Those funds would have been unable to meet collateral obligations the following morning and would have started winding up — selling gilts into a market that had no capacity to take them. The Financial Policy Committee's record of 12 October described the dynamic in flat institutional prose as a self-reinforcing spiral.

Eleven O'Clock

Yields opened higher again on the Wednesday morning. The Bank acted at eleven, and the design of the intervention mattered as much as the fact of it. Purchases were temporary, time-limited to 14 October, funded from central bank reserves but explicitly separated from the asset purchase facility used for monetary policy, and indemnified by the Treasury. They were aimed only at the long end, where the problem was. The Bank also postponed the start of active gilt sales from 3 October to 31 October, which it was careful to describe as a deferral rather than a reversal.

Thirty-year yields fell roughly a hundred basis points on the day, the largest single-session decline in the instrument's recorded history, closing near 3.9 per cent. The Bank had bought almost nothing. What it had sold was a floor — the knowledge that a bid existed at some price relieved managers of the need to hit every bid immediately, which was precisely the behaviour driving the spiral.

Central banks had done something structurally similar before: an intervention justified by market functioning rather than the stance of policy, addressed to a leveraged non-bank that had become systemically large by accident. The leveraged-hedge-fund version of the same problem is set out in the collapse of Long-Term Capital Management, and the Bank's last domestic emergency of comparable speed is described in the run on Northern Rock.

Three Days Left

Relief did not hold. Schemes used the breathing space to raise cash rather than to rebuild collateral buffers, and by the second week of October long yields were climbing back towards their pre-intervention levels. On 10 October the Bank doubled the maximum daily auction size to £10 billion. On 11 October it extended the operations to index-linked gilts, after a session in which long linkers sold off by a margin that had no precedent in the series.

Andrew Bailey was in Washington for the annual meetings that evening, and he chose to say in public what the Bank had been saying privately. Asked whether the facility would be extended past Friday, the Governor was blunt: "My message to the funds involved and all the firms involved managing those funds: you've got three days left now. You've got to get this done." Sterling fell on the remark and the criticism was immediate, but the deadline held, and the funds finished their rebalancing inside it (Bailey, 2022).

The operations closed on 14 October as announced. Against a headline capacity of £65 billion, the Bank had bought £19.3 billion of gilts, of which under £2 billion were index-linked. It sold the holdings over the following months at a profit of roughly £3.8 billion.

Forty-Nine Days

Politics finished what the bond market started. Kwarteng was summoned back from Washington and sacked on 14 October, thirty-eight days into the job — the second-shortest chancellorship in modern British history. Jeremy Hunt replaced him and reinstated the corporation tax rise the same afternoon.

On 17 October Hunt scrapped almost everything that remained, some £32 billion of it, in an emergency statement delivered before the markets had opened. Truss resigned three days later outside Downing Street, conceding that "I cannot deliver the mandate on which I was elected by the Conservative Party." She had been Prime Minister for forty-nine days, the shortest tenure in the office's history, and had been outlasted by a lettuce in a newspaper's livestream — a detail that obscured how narrowly the episode had been contained.

What Was Left

The schemes themselves came out ahead, which is the counterintuitive coda. Higher yields cut the present value of pension liabilities faster than they cut the value of assets, and the aggregate funding position of Britain's defined-benefit universe improved sharply over 2022, with the Pension Protection Fund's index of some 5,000 schemes finishing the year above 130 per cent funded. Nobody's pension was cut. The danger had never been solvency; it was the speed at which a solvent institution could be required to find cash.

Regulators drew the line there. The Financial Policy Committee recommended in March 2023 that liability-driven investment funds hold collateral buffers sufficient to withstand a rise in gilt yields of at least 250 basis points — more than double the shock most funds had been carrying in September 2022 — and the Pensions Regulator wrote the standard into its guidance that April (FPC, 2023). The Bank followed with its first system-wide exploratory scenario, an exercise designed to model how banks and non-banks would behave towards each other in a stress, on the grounds that the 2022 episode had been invisible to any supervisor looking at one sector at a time. The same blind spot between banking and market regulation is visible in the 2023 failures of SVB and Credit Suisse.

Nothing in the crisis involved a default, a fraud, or an institution that had lied about what it owned. Every hedge performed exactly as designed. Thirty years of falling yields had simply taught a whole industry to calibrate its margin against a world in which long gilts do not move a hundred and thirty basis points in four days, and the Growth Plan supplied the four days.