The expensive exit from quantitative easing

THE EXPENSIVE EXIT FROM QUANTITATIVE EASING

Record weekly short-term lending shows how hard it is to shrink the Bank of England's balance-sheet—and who ends up paying.

THIS WEEK the Bank of England lent £126bn of freshly created money to Britain's banks for seven days. In exchange it took custody of government bonds that the banks will reclaim when the loan matures. Repo, as such sale-and-repurchase swaps are known, rarely moves markets. But the size deserves attention. The weekly operation has just set a record, capping a year of escalating bids. What looks like plumbing is a verdict on the long, expensive process of unwinding a decade of quantitative easing.

For most of the 2010s the Bank financed the state by creating reserves—the electronic money that banks hold at the central bank—and using them to buy gilts. Quantitative easing turned the government's borrowing into a deposit base, easing the Treasury's path through a low-rate world. The reflux is proving harder to arrange. Under quantitative tightening the Bank sells gilts back into the market and lets others mature, draining reserves from the banking system and aiming to cut its bond holdings by £70bn in the year to September, to a target of £488bn. In the twelve months to February reserves fell from £706.7bn to £643.5bn. Then the banks did something awkward: they bid the money straight back.

The facility through which they do this—the Bank's short-term repo—is full-allotment. Any bank with top-quality gilts to pledge receives everything it asks for, priced at Bank Rate, the interest rate the Bank charges for central-bank money. Generous terms mean big bids are not, by themselves, a sign of panic. Officials insist the surge reflects design, not distress. They are moving to what they call a "repo-led, demand-driven" framework, in which banks borrow liquidity at a visible price rather than wallowing in a surplus the Bank created years ago. In principle, that is a cleaner way to run a monetary system.

Yet the scale tells a different story. In August 2025 banks bid £75.7bn; in September a record £87.1bn; in January £100.9bn; in June £122.9bn; this week £126.073bn. The acceleration is the point: the reservoir of extra reserves is shrinking faster than expected, and banks' need for them looks structural, not cyclical. They want the money for settlement, for regulatory buffers and for the precaution that a decade of abundance made habitual. The result is a system in which the Bank spends its weeks selling Britain's bonds and then lends them back. In the three months to June it unloaded £6.1bn of gilts through five modest auctions; in a single week this month it lent roughly twenty times that amount. That is not a return to the pre-crisis normal, in which a central bank hovers in the background of money markets. It is a new equilibrium of permanent dependence, rationed weekly at a price.

The dependence has a bill, and it falls on the taxpayer. The Bank's purchases sit in a facility indemnified by the Treasury, which means any losses on them are ultimately financed by the public. For years the arrangement paid: cumulative remittances to the Treasury peaked at £123.9bn in 2022. As interest rates rose the flow reversed; by June the net transfer had shrunk to £16.2bn, and the Bank expects transfers to keep running against it until the unwind ends. Under its own scenarios the lifetime loss is £60bn-£120bn, depending on how quickly rates fall.

On top of this, the Bank pays interest on the reserves banks hold. The New Economics Foundation, a leftish think-tank, put the annual cost to the public finances at about £26bn a year. With Britain's debt stock near £3trn and the interest bill at risk of eclipsing Italy's, as the Telegraph fretted in June, these costs arrive at the worst possible moment.

To be sure, the case for simply finishing the job is real. Faster unwinding reduces the interest-rate risk the public sector bears, and the Bank's own arithmetic suggests the full fiscal effect could be anything from a loss of £70bn to a gain of £65bn—roughly a wash. But the politics of that trade are asymmetric. The losses arrive now, in cash, on the government's borrowing requirement, precisely when gilt supply and debt service are straining; the savings, if any, are hypothetical, deferred and diffuse. And the record repo shows the drain has a floor: the Bank cannot hand control of the money supply back to the market faster than the market is willing to take it. That is why "finish the job quickly" is not obviously a money-saver.

The deeper problem is institutional, and it is the one the Bank's governors least like to advertise. A central bank that lends the banking system's entire marginal liquidity back each week has concentrated power, not dispersed it. In 2022 the Bank discovered how fragile that perch is, when a slump in long-dated gilts forced it to intervene in the very market it was trying to shrink. Since then its Financial Policy Committee has fretted publicly about the resilience of gilt repo—the market on which this whole operation depends—considering mandatory central clearing and minimum haircuts on trades that sidestep it. The questions only harden. What happens to the weekly ration if a big shock arrives between auctions? At what point does the promise of full allotment become a subsidy, as it did during the crisis years?

The repo-led regime still deserves support. It prices central-bank money in the open rather than burying the cost in a bond portfolio, and it lets the Bank shed its holdings without pretending the plumbing does not matter. But clarity has a price. The new regime makes the trade visible, and what is visible cannot be hidden from voters. The Treasury should plan for years in which the Bank's operations impose costs on the public finances—and budget for them—instead of treating those costs as an accident of interest-rate arithmetic. The Bank, for its part, should stop implying that balance-sheet "normalisation" is nearly complete while its weekly rations stand at records.

QE was sold as a crisis measure that would be reversed once the emergency passed. The reversal has happened. The emergency, it turns out, never quite left. The Bank can shrink its bond pile; giving up its job as the market's weekly lender is proving slower and more expensive than anyone hoped. Quantitatively tight, the Bank has never had so many dependants. The exit from QE does not end. It is paid for one week at a time.