There has been a lot of press coverage over the last three years over the losses that the UK government is incurring on its quantitative easing programme, and the related issue of the payment of interest on Bank of England reserve accounts. However it was not clear to me how these issues were related or how this situation came about. As someone who follows the financial press, I expected that I would be able to understand the matter properly with some research, and I think that I now have – and what follows is my attempt to explain it.
There are better explanations out there, I expect – this from the IFS is the best that I have found. But it is a think-tank piece aimed at policymakers already familiar with the core mechanics in play; my explanation here will instead try to explain it to a lay audience.
Background: Fractional Reserve Banking
To begin to understand this topic, you do need to understand fractional reserve banking. This video from Positive Money is the best primer that I have found so far – unfortunately they only seem to have this in video form, and Positive Money is a group with an agenda which you may want to tune out. To briefly recap, most of what you and I consider money – bank deposits – is just entries on a commercial bank’s computer. A bank creates money of this type – which in macroeconomics is called broad money – when it issues a loan: it simply adds the value of the loan to their record for the customer’s account. Only if and when this money is transferred or paid to another bank does the bank actually have to pony up some real money, money that is trusted by other banks – this so called narrow money. In the UK narrow money consists of (Bank of England) notes and coins, and balances on accounts at the Bank of England.
Only banks get to have bank accounts at the Bank of England. The Bank of England does not trust any of the IOUs from the commercial banks; the commercial banks have, or had, to take out a loan from the Bank of England to get a balance there – putting up collateral to do so, usually in the form of government bonds. Using the balances so created, the commercial banks can settle payments by transferring corresponding amounts between their reserve accounts.
The balance for, say, HSBC at the Bank of England is not the total amount gross balances on accounts at HSBC or anything like it, nor do the total balances at the Bank of England equal UK broad money supply. These accounts are in effect the current accounts for the commercial banks themselves, holding working capital used to resolve inter-bank payments; the total narrow money was much, much smaller than the UK’s GDP or the total sterling broad money supply.
One reason that these balances historically were small is that, before 2005, the Bank of England paid no interest on the reserve accounts. A bank with a positive balance was incentivised to make new loans or buy interest-bearing securities to exchange its balance for something that did earn interest. The relevant data series from the Bank of England does not cover the period before 2006 so I cannot quantify how small the reserve account balances were at that time.
2005-6
The Bank of England implemented some reforms in this period, of which the one relevant to our discussion is that it started to pay interest on balances on the reserve accounts (ref).
Even after this change, balances at the bank were still close to zero. Bank of England data shows that typically gross balances across all commercial banks did not exceed £25B in 2006 (ref). That is not nothing, but it is small compared to GDP or to broad money (~£1.3T M4 money in 2025, ref).
The Global Financial Crisis
Then the Global Financial Crisis happens. Interest rates fall to nearly zero, and the Bank’s Monetary Policy Committee finds that it has hit the interest rate lower bound and still needs to do something more.
The Bank of England starts a programme of quantitative easing. Roughly speaking, this is an intended as an injection of money to stimulate the economy and stop a shrinking of the money supply. The Bank, being a bank, creates this money by issuing a loan: it creates a new subsidiary, the Asset Purchase Facility (APF), gives it a reserve account at the Bank, and issues a loan for £200B. The APF then uses this money to buy UK gilts and other investment-grade debt on the secondary market (ref); settling these transactions causes the £200B to move from the APF’s account to the commercial banks’ reserve accounts at the Bank.
Let’s pretend that HSBC and Barclays are the only UK commercial banks for a minute. We started with something like (to be clear, the numbers here are all made up just for illustration):
| Balance | Loan Owed | |
| HSBC | £0.9B | £1B |
| Barclays | £1.1B |
Then the APF is created and funded:
| Balance | Loan Owed | |
| HSBC | £0.9B | £1B |
| Barclays | £1.1B | £1B |
| APF | £200B | £200B |
and then it spends its money – buying bonds from the commercial banks and their customers:
| Balance | Loan Owed | |
| HSBC | £119.9B | – |
| Barclays | £80.1B | – |
| APF | £0 | £200B |
After later rounds of QE, including during Covid, took the total money injected by QE up to £900B at peak, the balances could have looked something like:
| Balance | Loan Owed | |
| HSBC | £519.9B | – |
| Barclays | £380.1B | – |
| APF | £0 | £900B |
This is now a totally different situation from what we previously said were “current accounts for banks”; the commercial banks all have large positive balances on their accounts, far exceeding their working capital needs for reserve currency. And they cannot, collective, rid themselves of these balances even if they want to: HSBC could buy gilts or pay dividends to shed some money, but since the recipients all bank with commercial banks who bank with the Bank of England, it would merely move some positive balance to the other banks. Unless the APF reduces its loan, the others cannot reduce their balances collectively.
This is where the connection between the Quantitative Easing and the interest paid on central bank reserve accounts arises. The Bank now owes interest, paid at or near Bank rate, on these large positive balances. But the bank charges interest on the loan to the APF at around Bank rate. Since the net balances sum to roughly zero, the interest paid to the Bank on the APF loan will pay for the interest paid from the Bank to the commercial banks. The APF in turn owns a lot of interest-bearing securities that pay interest at or above the rates of the gilt yield curve at various maturities.
One more thing to note here: this loan by the Bank of England to the APF is a very different sort of loan from the loans that the Bank made to the commercial banks. Those loans were secured on collateral, but the APF had no assets and owned nothing before the Bank loaned it this money. The APF bought good quality debt, but that debt is not security for the loan, and there is no certainty that the assets in the APF will actually cover the loan; it is something of a coin toss whether the APF assets will yield enough to pay the loan in full, depending on how the interest on the floating-rate loan from the Bank changes over time. The Bank can only offer the APF a floating rate loan at Bank rate, because it needs to pay Bank-rate interest on the positive balances of the commercial banks on the other side of the transaction. So the Bank secures an indemnity from HM Treasury to cover any shortfall in the APF – it simply would not be able prudently to make the loan without it – moving the risk off of the Bank’s balance sheet.
The 2010s
In the 2010s, interest rates stayed low and long-term interest rates – as implied by the yield on long-dated gilts – continued to fall after the crisis. The APF was lending long and borrowing short so, having lent a lot of money of money to the government in 2008-9 on higher long-term rates, it now profited from the lower Bank rate.
The next rule change, then, was made by the Bank and the Treasury in 2012. Rather than have the APF retain profit, all parties agreed that the indemnity from the UK Treasury should be made symmetrical – as well as the Treasury paying into the APF to cover any shortfall, any profit would be paid out periodically (ref). This avoided the APF having a positive balance while the rest of government was in debt, reducing the public debt outstanding. Over the following years, a profit of around £10B/yr was passed from the APF to the Treasury.
Notwithstanding this, QE expanded further during the 2010s as the MPC grappled with the long period of interest rates close to the lower bound, increasing the size of the APF and the size of the loan from the Bank.
During this period, many political commentators argued that the government was in a uniquely strong position. Since interest rates on 10 year, 20 year and even 30 year government debt were below 2%, it was argued that the government had an excellent opportunity to borrow to invest. The government disagreed and continued with austerity, seeking to slow the growth in the national debt – while commentators, mostly from the political left, advocated to spend and invest in infrastructure, healthcare and education.
The actual position, however, was not how it appeared. Although the government was placing much debt at historically low rates of interest, a significant portion of that debt was being bought back by a related entity, the APF. The APF in turn was paying Bank-rate interest on the loan used to finance these purchases. So the APF was transforming the maturity curve of the government debt, turning long-dated low interest debt into an overnight loan at bank rate.
By mid-to-late 2019 – notably, before the 2020 COVID-19 pandemic began – long forward rates were unusually low: the 20-year forward rate was between 1% and 2%, and the 30-year between 0% and 1%. … this implies that, ex ante, it would have been much cheaper to fund the government by issuing long-term bonds to the market, thereby locking in the unusually low long forward rates, than by borrowing at a floating rate from the Bank of England.
Covid & Post-Covid
With interest rates already low, the economic crisis brought about by the Covid pandemic forced the Bank to lower interest rates back to the lower bound once again and then look for more stimulus. Quantitative Easing was called for once again, and this time the bank did another £400B of money injection following the same pattern as the previous rounds. The Bank increased the loan to the APF, and the APF bought more gilts and investment-grade bonds across a wide range of maturities.
The APF was briefly the largest holder of UK gilts. This meant that the UK debt maturity curve was now dramatically different from what the Debt Management Office thought it was: this largest slice of the debt was not maturing in 20 or 30 years regardless of what the bonds said, but was instead financed at the floating Bank rate via the APF. And so, when the post-Covid inflation started, things turned sour quickly.
As the Bank raised interest rates, the APF began to lose money. In 2022, commentators warned that it was losing £10+B/yr and could lose, in the bad case, £40+B/yr. And indeed that did happen: the Treasury indemnity for the APF resulted in a net payment from the Treasury of £44B in 2023-4 , and mark-to-market unsettled losses – that is, the gap between the present value of the APF’s assets and its liability to the Bank – of £178B (ref – see note 8).
Unfortunately we cannot fully reverse this situation. At its core, we thought that we borrowed £900B at favourable low rates of interest fixed for up to 30 years, when it was really a rolling overnight loan. As a result the UK finds itself with a much higher effective rate of interest on the historically high level of borrowing post-Covid than we thought we were getting.
Models
There are a few different mental models that one can use to think about this situation. A model is a way of thinking, and different models can describe the same situation in different ways and may be more or less useful in understanding it or in analysing the consequences of different choices that we might make in future.
Model 1: The APF is losing a lot of money
This model is the model of the APF’s own accounts, and that of many commentators including the Labour Party. It is succinctly expressed in https://www.ft.com/content/dcea7102-f037-4521-9ce9-514c623f492b.
The APF is a kind of interest rate swap. It lent £900B of money at historically low rates of interest, using money borrowed at the overnight floating rate from the Bank of England. This was, by 2019-2021, a rather one-sided bet, since interest rates were low at this point. Whatever the merits of QE, even if more QE happened it would not cause the Bank rate to fall below the lower bound – but Bank rate could go up.
To be clear, there is no way to lose money on the bonds themselves if they are held to maturity; the loss comes from the interest owed on the loan being above the interest returned by the bonds.
The APF can reduce future losses on the interest rate difference by selling bonds before maturity and reducing its outstanding loan at the Bank of England. But because the bonds in its inventory are all trading well below par due to the fall in interest rates, this also realises a large immediate loss which, if markets are efficient, is equal to the estimated loss if the bond was held to maturity instead (again as discussed in https://www.ft.com/content/dcea7102-f037-4521-9ce9-514c623f492b).
In this model, the die is cast and there is nothing directly to do about the losses here – we have to close our position expensively (sell the bonds at a loss to the taxpayer) or let it expire out expensively (pay the interest difference from tax receipts).
Model 2: The Maturity Curve of UK Government Debt is not what we thought
UK government debt that is held by the APF is not really part of the public debt anymore. That debt in effect ceases to exist in its gilt form when it is bought by the APF, since the debt and interest are now owed from the government to itself; it was replaced by a floating-rate debt to the Bank of England. (Pretend for now that the Bank is not itself government-owned; the money does end up paid to private entities regardless.)
The APF was briefly the largest holder of UK gilts in 2021-2. So the UK debt maturity curve was and is now dramatically different from what the Debt Management Office thought it was: this largest slice of the debt was not maturing in 20 or 30 years regardless of what the bonds said, but was instead financed at the floating Bank rate via the APF.
This reduction in the average maturity of the UK’s public debt has been very unfortunate, since we missed the chance to lock in a large amount of public debt at long maturity and low interest. The current losses are not really losses – they are an accounting glitch that shows the opportunity cost of missing this chance back in 2019-21; the cashflow here is just the cost of our very large overnight Bank-rate loan. Because we failed to lock in the debt at low rates of interest and long-term rates are now higher, the die is cast and we cannot do anything about it.
This view is expressed in, for instance, https://www.ft.com/content/45a441c9-26e0-4de3-b985-faafc5968a49, and this points out that one reason that the UK’s losses are so high is that the Bank of England inadvertently made this problem worse by buying longer-duration gilts than similar programmes in other countries.
Model 3: Central bank reserve balances paying interest is the problem
Forget about the APF entirely; it loans money to the UK government (by buying gilts) and the money it borrowed is from the Bank of England which is also part of the UK government. These are all related-party transactions and no profit or loss for the taxpayer can result from this in itself. The point where the lost money leaves the public purse and becomes a private asset is when the Bank of England pays interest on the balances of the commercial banks; that is the leak.
In this model, the banks could be considered to be double-dipping, or to be the beneficiaries of a windfall gain. This is the view of the Greens, Lib Dems, and many commentators (for instance Robert Peston). Banks both earn interest on their Bank of England account balances and, because these balances constitute highly liquid quality assets, they contribute to the banks’ capital ratios and enable them to lend more and earn more interest on their own books. To an extent, that was an intended effect of QE during the crises themselves.
We only started paying interest on these balances in 2005, and not for any reason related to Quantitative Easing; maybe we find a way to stop doing that. Note that, contrary to Reform UK’s manifesto, this is hard in practice – for instance from the Financial Times:
… if all reserves were remunerated at zero and there are more reserves in the system than just about enough, overnight interest rates would collapse to zero…
Subsequent Events
We await the APF’s accounts for 2024-5; last year’s accounts were published in June. Crudely, we would expect the position to be that the Treasury paid in another ~£45B and that between £100B or £150B of future liability remains, depending on how gilt yields have changed against the distribution of assets in the APF. The slight fall in Bank rate may have reduced the implied future losses a little.