The mini-budget, and the buyers who had to sell.

On 23 September 2022 the Chancellor, Kwasi Kwarteng, set out a Growth Plan. It cancelled a planned rise in corporation tax, brought forward a cut in the basic rate of income tax, abolished the additional 45p rate, and cut stamp duty. Alongside it sat the energy price guarantee, a large subsidy already trailed. The measures were permanent tax cuts, not a one-year rebate, and they were published without a forecast from the Office for Budget Responsibility.

Gilts fell in price the same day, which is the same thing as the yield rising. Sterling fell too. The long end of the curve moved more than the ten-year. That is the clue to who was selling.

A defined-benefit pension promises incomes decades ahead. The present value of those promises — the liability — rises when gilt yields fall, and falls when yields rise. A scheme that wants its assets to move the same way holds long gilts, or contracts that behave like long gilts. That approach is called liability-driven investment, LDI.

Many schemes did it with leverage. They posted gilts as collateral, borrowed cash in the gilt repo market, and used the cash to buy more gilts. Or they entered interest-rate swaps: the scheme receives a fixed rate and pays a floating rate, so a fall in yields produces a gain that offsets the rise in the value of the pensions. A modest pot of assets could then hedge a much larger liability. The extra exposure is funded by borrowing, and the borrowing is secured on the gilts.

Leverage works in both directions. When yields rise, gilt prices fall and the swap moves against the scheme. The repo lender, or the bank on the other side of the swap, then demands more collateral. That demand is a margin call, and in this market it is in cash, often the same day.

The price of a gilt is set by whoever is just willing to buy the next bond and whoever is just willing to sell it. Those are the marginal buyer and the marginal seller. On a normal day a pension fund selling a few billion of long gilts finds a dealer, an insurer, or an overseas investor on the other side, at a yield only a little higher.

After 23 September the sellers were not choosing a new strategy. LDI funds had to raise cash for margin. The asset they could sell in size, quickly, was the long gilt itself. Selling pushed the price down, which raised the yield, which triggered another margin call. The people who would normally have been the marginal buyers stepped back. Dealers did not want to hold the flow on their own balance sheets. With no one ready to take the other side except at a much higher yield, the long end gaped.

This is why a speech about tax could move a thirty-year yield by more than a ten-year yield. The leveraged hedges sat at the long end, because that is where the pension payments are. The ten-year rose as well — it is on the chart — but the thirty-year is the line that shows the squeeze.

This was not a restart of quantitative easing. QE was a monetary-policy programme, wound down on a published timetable. These purchases were a financial-stability operation: a temporary buyer, at the long end, so pension schemes could post collateral and reduce leverage without a fire sale setting the price. The Bank said the operation would end on 14 October. It did. The bonds were held in a separate portfolio and were sold in the following months, which is why they are kept out of the monetary-policy totals on the QE and QT page.

Within weeks the tax measures were mostly withdrawn, the Chancellor was replaced, and the Prime Minister resigned. Yields fell back from the peak. They did not return to where they had been at the start of September, and the chart from there to today is the rest of that story: a higher level, through the years of quantitative tightening, not a permanent spike at the mini-budget print.