The chancellor faces a tough, perhaps career-defining or ending, task in the November 26 budget (the budget) to close a fiscal gap estimated to lie in the £20-£40bn range, to meet her own set fiscal operating rules.
These rules combined with government’s self-denying triple tax lock against income, national insurance, and VAT rises, mean that a Hobson’s Choice beckons between imposing especially economically distortive or sapping ‘stealth-type’ tax alternatives and politically unpalatable spending cuts and/or delays to ending the two children benefit and to other poverty lifting measures, while maintaining the confidence of the bond markets.
Various permutations of possible tax increases to make inroads in that gap already have been spun and will certainly continue to spin during the budget run-up, including, as the autumn leaves fall, that she could even jettison that triple tax lock as her Hobson’s Choice becomes ever starker.
Is there an easier or less painful way out?
In August, the Institute of Public Policy and Research (IPPR) published Fixing the Leak (August IPPR), making the case that the fiscal gap could be significantly reduced by two related measures without substantively compromising the Bank of England’s (Bank or BoE) monetary policy (MP) remit.
First, by imposing a levy on the interest income remunerated above two per cent that the commercial banks (CBs, taken to mean in this post by any financial institution with Bank reserve deposits) continue to receive on their QE (Quantitative Easing) related central bank reserves (see dedicated section one central bank reserves sub-section for explanation).
Such a future public recoupment of that continuing Bank windfall subsidy to the CBs, according to August IPPR, should save claims on the current budget, estimated as totalling between £35bn and £40bn this parliament, providing Rachel Reeves (RR) with an additional £5nn to £7bn current budget fiscal headroom.
Second, by halting active sales of government bonds (gilts) purchased between 2009 and 2022 by the Asset Purchase Facility (APF) when it implemented the Bank’s five QE programmes between 2009 and 2022 (see Table 1, and Sections one and two for explanation).
Such limiting of the unwinding QE process – Quantitative Tightening (QT) – August IPPR estimates would possibly save £10bn-plus annually.
Although, as Section three explains, these figures – both in terms of their overall magnitude and their relative apportionment impact on future current budgets and hence the chancellor’s primary fiscal mandate – are sensitive to the future path of Bank rate (BR) and its relationship to long term gilt rates (yield curve), as well as to the actual future pace of QT and its composition split between passive redemption and active sales of gilts – they are undoubtedly substantial and material to the overall public finances.
Variants have been previously proposed or suggested by the New Economics Foundation, and by NIESR authors (see its section nine), and previously by Gerald Holtham (city economist) and Charles Goodhart (former Bank deputy governor and LSE professor), as well as by mainstream financial journalists, notably by the FT’s Chris Giles, as well as by another former Bank of England (BoE) deputy governor, Paul Tucker, in a seminal, detailed, but relatively accessible March 2022 Institute of Fiscal Studies Green Budget chapter (Tucker, reference one).
Nor have they been dismissed out of hand – in contrast to being actively discouraged – by the current BoE governor, Andrew Bailey (the governor, or Bailey) (see Section Two).
But, given the current fiscal and political challenges facing the Starmer government, they are certainly topical and pressing, increasingly picked up in more popular and sometimes informed political discussion this autumn.
In that light, the former Labour Transport Secretary, Louise Haigh, in a September 2025 pre Labour Party conference article argued that the UK should follow the European Central Bank (ECB) and Swiss National Bank and introduce tiered interest instead of full payments on reserves that in the UK cost nearly “nearly £40bn last financial year (2024-25)”.
The crux of her argument is that “subsidising the banking sector at full interest while cutting investment or welfare spending is economically incoherent” while moving to tiered reserve remuneration “would not compromise monetary policy”.
She also went on to argue that if the BoE followed the practice of the US Federal Reserve and European Central Bank to “actively halt “sales of government bonds purchased under QE, the “Exchequer’s losses would be paused”.
Her views can be taken as reasonably representative of a significant segment of the parliamentary Labour Party and perhaps even of the Cabinet in the lead up to the pivotable November budget.
They are certainly not off the wall. Paul Tucker himself had noted back in 2022 (almost pre-echoing Louise Haigh) that if partial remuneration of reserves were implemented for the two years 2023-25, according to the Banks’ own then figures, between £30bn and £45bn could have been saved for the public purse – roughly nine per cent of the annual public spending on health, education, and defence (Tucker reference one, key findings five and six).
The first section of this post provides initial explanatory background on MP and QE and the role of central bank reserves.
The second, focuses on and explains the public finance ramifications of QE as it was implemented through the establishment of the Treasury-indemnified Asset Purchase Facility (APF).
The third, seeks to identify the main uncertainties and possible policy conflicts connected with the proposed and related reforms, focused on their public finance and MP impacts.
The fourth and final section after touching on some of the wider political economy issues connected with QE, draws together the discussion focused on its public finance costs as implemented by the APF, concluding that the later Covid QE rounds came attached with clearly discernible public finance risks, subsequently realised, that should have been addressed in policy consideration and decisions by both the Bank and Treasury at the time.
This, in turn, provokes wider political questions as to why they were not taken on board earlier, either by Rishi Sunak’s administration or by the incoming Starmer administration, notably in RR’s first budget.
Certainly, the right questions were not asked at the right time and/or were not taken seriously enough by the people who ought to have asked them, for whatever combination of reasons.
That horse may not yet have bolted but before it does, the opportunity should be taken this autumn 2025 budget to recycle for public purposes the supra-profits that the reaped from QE-related reserves remuneration CBs – given their source and scale, and their relatively limited distortionary impacts on either MP or the wider economy – without any further wasteful delay.
Although this website is broadly agnostic between proposed variants of imposing a QE-reserves income levy (relatively easy to implement, is customised to cause and remedy, establishing a workable precedent for the future, but proceeds depend on future BR path and implementation could on the margin impinge on Bank MP) and a bank windfall tax (avoids any interaction with the Bank’s MP remit, is more amenable to a political in contrast to a technocratic narrative, but could prove more difficult/problematic to implement to achieve target proceeds allied to possible unintended consequence risk), it is unequivocal that the current public finance position overrides any possible problems, as identified above, in progressing with either with best possible alacrity.
Summary of other recommendations
- The Bank’s mandate should be revised (probably best through the Chancellor’s annual remit letter to the Governor) to ensure that the Bank and Treasury jointly take responsibility for the future public finance implications that arise from the Bank’s MP operations, most pertinently and presently with reference to the operations of the APF concerning active gilt sales (noting that Treasury representative attends MPC and their contribution – limited to public finance implications – should be minuted).
- QT, in words of the governor himself, has limited or tangential impact on the Bank’s MP operation. Active APF gilt sales should be halted or proceed only when future net public finance impacts are minimised on proper and due assessment (probably as best advised by UK Debt Office).
- QE should only be used in future when its expected impacts are clearly identified and justified relative to possible alternatives and should be implemented taking regard to its future impact on the public finances.
- The House of Commons Treasury Committee should step up its scrutiny and review capacity pertaining to the interactions and inter-relationships between fiscal and MP.
- The Bank itself should develop and improve its communications to explain its operation of MP in lay understandable language more honestly and transparently (in this website’s own experience precise mechanism by which the reserves liabilities resulting from Bank QE activity via APF are produced and distributed between CBs, not clearly explained by the Bank).
These recommendations are largely endorsements of proposals made by technical experts but have been assessed as according with and furthering strategic social democratic aims and purposes.
- Background
In 1997 the BoE was given a mandate by the incoming New Labour government to achieve price stability, defined as achieving a forward-looking symmetrical (overshoots and undershoots treated the same) two per cent medium-term inflation target (as it was later revised to), measured by the Consumer Price Index (CPI).
The target was operationalised by its Monetary Policy Committee (MPC) independently adjusting Bank Rate (BR) – the interest rate that the Bank pays on CBs overnight deposits or reserves (see central reserves subsection) – with reference to that primary price stability mandate.
The following ten years were marked by a benign period of non-inflationary sustained growth before it was dramatically burst by the 2008-2009 Global Financial Crisis (GFC) that threatened to turn into a global economic catastrophe worse than the inter-war depression.
In response to that crisis, the BoE slashed BR close to its lower effective or zero bound rate (ELB) – the point where any further reductions would have little or no expansionary impact on real economic activity.
March 2009 marked a policy watershed. Although BR was reduced to 0.5%, QE was also introduced and made centre stage to the Bank’s monetary policy (MP) response to the crisis – a de facto recognition that the ELB had been reached and that ‘unconventional’ or ‘extraordinary’ MP, aka QE, needed to become the primary instrument to encourage economic activity and recovery.
QE, according to the primer published on the Bank’s website, involves the Bank creating money digitally in the form of ‘central bank reserves’ (see sub-section below ) – expanding its balance sheet in the process – to purchase financial assets, mainly government bonds (gilts) of varying duration by including the longest from non-bank financial institutions, such as pension funds and insurance companies, all to lower longer-term borrowing costs for households and companies.
Its aim in 2009 was to inject liquidity into the fragile financial system, lowering long-term interest rates, to encourage economic activity and recovery to avert depression.
As Table1QE records, between March 2009 and December 2012, £375bn of government gilts were purchased across three Global Financial Crisis (GFC) QE programmes.
Although the precise mechanisms by which QE purchases translated into economic activity remain hazy and uncertain (as this website discussed in this pre-Covid 2018 post) the official consensus is that QE1-3 (more especially, QE1) did help to underpin a steady if slow recovery, averting at least depression and deflation.
That said, we don’t know what would have happened without QE and/or if more active fiscal measures instead had been employed.
Although economic catastrophe was avoided, the ensuing decade proved one of fiscal austerity and of stagnant growth and productivity.
The Brexit political drama that David Cameron’s Conservative majority government elected in 2015 ushered in hardly helped, engendering ex ante economic uncertainty before generating generally accepted ex post negative net economic consequences.
Indeed, following the June 2016 Brexit referendum, the MPC decided that a further phase of QE4 was necessary to counter the economic and financial impacts of that uncertainty on the economy and financial system.
The economy then limped on, showing slight signs of intermittent recovery (muted or indiscernible in gdp per capita or productivity terms) until 2020 dawned, when the worst international public health pandemic since the post-first world war Spanish flu outbreak, struck in earnest.
The economically dislocating impact of Covid, associated with a 20% fall in gdp across the lockdown quarter, was met both by a massive monetary and fiscal policy response – this time acting in the same anti-deflationary direction.
Between March 2020 and December 2021, a QE5 programme of £375bn (Covid QE5) rivalling is size the QE1-3 GFC rounds took place.
It brought the cumulative QE purchases of government bonds (gilts) held in the APF (see dedicated sub-section below) to £875bn (plus another £20bn of corporate bond purchases) – an astonishing 40% of gdp.
Some economic and political commentators, noting that the Bank’s QE5 gilt purchases aligned closely with the speed of gilt issuance by HM Treasury made to finance the furlough and other associated Covid fiscal response measures, began to question whether QE had become more a modern digital mechanism for the government to fund its record fiscal deficit than a MP one, cloaking what was in essence, central bank money creation or monetary financing of fiscal deficits (colloquially known as money printing).
Following the near decadal deployment of QE as the primary MP instrument or mechanism, others cast doubt on the economic efficacy of the impact of QE on real output and employment activity over and above shoring up financial market conditions and stability (including any arising from leverage, unsustainable debt or credit growth), which, admittedly, is also now a Bank function and responsibility and part of its mandate.
Such concerns were distilled in a House of Lords Economic Affairs Committee July 2021 report, suggestively entitled Quantitative easing: a dangerous addiction?.
The Treasury Permanent Secretary at HM Treasury between 2005 and 2016, Lord Macpherson, was cited as likening the ability of QE to stimulate spending and investment in stable economic conditions to “pushing on a string”.
According to Macpherson, when QE was first deployed in 2009 — after interest rates had been cut from 4.5% in October 2008 to 0.5% in March 2009 — it “had a real impact” – a point supported by 2011 Bank research that concluded that medium to long-term gilt yields fell by about 100 basis points (one per cent) in response to the first £200 billion QE program, flattening the yield curve (this plots the effective interest yield of gilts, according to their maturity duration).
McPherson went on to observe, however, that when long-term, as well as short-term, interest rates remain sticky at the ELB, the Bank must “buy a great deal of debt to have any impact at all”: in short, diminishing returns bed in.
Later research by the Bank has lent some support to such a conclusion, indicating that subsequent rounds produced more limited muted, if any, impacts on long-term interest rates – one, if not the primary, QE-related objective.
More generally, the BoE’s own Internal Evaluation Office in its 2021 report on the Bank’s approach to QE itself identified several knowledge gaps in the Bank’s understanding of how QE worked in practice, including the relationship between quantitative easing and financial stability, and its consideration of potential monetary–fiscal interlinkages.
In response to such criticisms, the governor advised the committee that QE is most effective where there is “impaired market liquidity”, as in 2009, and then again in March 2020 when there was a “dash to cash” as Covid lockdowns loomed, wreaking financial dislocation, possibly risking insolvency, across bank and non-bank sectors.
Companies then, in the absence of trading revenue, needed and sought finance to cover their expenses.
If not addressed through QE, according to the governor, these presenting risks could have tipped the economy into deflation and depression, notwithstanding the ameliorative impact of the fiscally financed furlough schemes on household budgets and financial security.
He also dismissed the accusation that the Covid QE rounds in effect were designed to accommodate Covid-related fiscal expansion, pointing out that successive decisions were taken by the MPC for reasons then recorded in the relevant MPC meeting minutes, prior to successive Treasury decisions on debt issuance.
The governor’s responses, however, come across as more convincing to the £200bn initial QE intervention beginning in March 2020 than at least the later Covid-related rounds of QE5 by when financial stability had returned, or even to later QE2-3 rounds.
Anyway, by the beginning of 2022, Covid, due to a successful vaccination programme, had been largely defeated both medically and economically, at least in the UK.
Unfortunately, just as the economic outlook brightened, wider post-Covid global supply chain pressures, the invasion of Ukraine by Russia impacting upon energy and food prices, post-Brexit labour market developments amid a domestic fall in the economic participation of older workers, all combined in a perfect storm to produce a cost-push supply side inflation push of a magnitude not experienced in the UK since the early nineties.
This ‘cost of living’ crisis turned the MPC’s attention back to inflation. From its historic low of the 0.1% level maintained between March 2020 and December 202, the MPC began to raise BR in slow and steady 0.25% steps to reach a post-Covid high of 5.25% by August 2023.
The MPC assessed that this bout of inflation was driven more by supply side exogenous (external) factors, such as international energy prices, rather than endogenously (internally) excess demand pressures.
Despite the CPI inflation (identifier D7G7) (inflation) measure used for MP purposes exceeding two per cent on average each quarter from 2021 (Q1) peaking at 10.7% in 2022 (Q4), it did not wish to choke off any recovery by raising rates too fast, in the process forcing up mortgage rates during the cost-of-living crisis.
The inflationary resurgence was interpreted as a transitory short-term event driven by exogenous factors that the MPC expected to abate by rather than persist into the medium term.
It did, however, begin to reverse or unwind QE in October 2022 by allowing previously purchased APF bonds to mature (passive redemption) and by selling them actively to investors in the secondary market (active sales) – the process termed Quantitative Tightening (QT) (see next section for further detail).
By 2024 (Q3) inflation had dropped back close to target, allowing the MPC to reduce BR from its post-Covid high in slow stages to four per cent, where it remains following the September 2025 MPC decision, despite inflation rising steadily to this year to reach 3.8% (12 months to August 2025).
According to the minutes of that meeting, the MPC concluded that on balance that this latest inflationary impulse was unlikely to persist into the medium term in the face of evidence of a weakening jobs market, some abatement of wage inflation, and the continuing impact of MP weighting down on demand and business investment within a subdued overarching growth environment.
Two of its seven members, in fact, voted for a BR reduction of 0.25% noting that “the latest inflation hump was expected to normalise, with the most significant contributions coming from one-off changes in administered prices (water and energy) and global food inflation in a limited set of items, neither related to demand pressure”.
The minutes did, however, emphasise that MP was ‘not on a preset path’, implying that instead of continuing cuts to BR, it could be raised in response to emerging evidence of worsening risk of medium-term inflationary pressures anchored into expectations.
£70bn of QT sales across the October 2025 to September 2026 period, less than the annual £100bn in previous years, was also decided by the MPC, comprising £49bn passive and £21bn active – an increase on the c£13bn of earlier years.
The official stated Bank view (see also below) is that unlike QE – used as a MP mechanism to reduce interest rates and therefore support inflation – the aim of QT is not to affect interest rates or inflation nor to signal MP intentions: it is to provide balance sheet headroom (normalisation) to undertake QE again in future, should that be needed to achieve the Bank’s inflation target.
QT is thus intended and implemented as a gradual and predictable process undertaken during periods of financial calm rather than crisis to build such headroom.
Yet by selling gilts and increasing their supply, given the inverse relationship between the price and effective yield of gilts, QT can act in a different direction to the intended direction of MP, especially when BR is on a downward direction, as it has been over the past two years.
In that light, the Banks Executive Director for Markets in June 2025 advised that monetary policymakers should “consider the interactions of QT and policy rate decisions, especially at a time where these two tools are acting in different directions”(noting that) since potential tightening effects of QT cannot be perfectly offset by cuts in Bank Rate, because they generate a different set of monetary and financial conditions, the combination of tools (that of QE and QT) and their macroeconomic effects must be carefully considered”.
What was missing from the MPC’s discussion – at least as recorded by the September minutes – was the impact of continuing QT sales on the public finances, notwithstanding that the meeting was attended by the Second Permanent Secretary of the Treasury – Sam Beckett.
Central bank reserves
What are central bank reserves and what has made them not only central to the QE/QT process but also to the wider public finances?
Well, let us go to the horse’s mouth. In a May 2024 speech (Bailey, 2024), the current BoE governor, Andrew Bailey (the governor), after confirming that the commercial banks create money simply by extending loans to their customers, identified central bank reserves (reserves) as the most liquid and ultimate form of money that provides the ultimate means of settlement (monetary backing) for all transactions in the economy.
Reserves are commercial/clearing bank (CB) deposits held at the Bank or, as a Bank 2015 Bank Working Paper (WP) put it “current account balances held by commercial banks at the central bank” (see, p10 section ‘commercial bank reserves’ of reference for further detail and explanation). They represent (with banknotes) its main liabilities.
CB transactions are settled “by debiting one commercial bank’s reserve account and crediting another’s”: the settlement function.
The CBs also need to hold sufficient central bank reserves, or ‘liquidity’, as a precaution to meet the potential outflows of money from their customers’ accounts and other prudential requirements.
However, and crucially, in contrast to their distribution, it is only the transactions that the CBs undertake with the central bank (and vice versa, website italics) that can affect the aggregate quantity of reserves.
Instructionally, the Bank did not renumerate reserves until 2006. Rather, it implemented monetary policy by keeping them ‘scarce’ deliberately to maintain a mechanism to transmit BR decisions across the financial system.
Indeed, from the early 1980s the Bank did not set any reserve requirements. Instead, the CBs chose what (non-zero) balance they aimed to hold each day at the Bank: a demand-driven system.
By the early 2000s, they tended to hold only the “bare minimum of reserves required for day-to-day settlement” purposes, as an unremunerated system provided them with strong commercial incentive to seek alternative interest paying destinations for such funds.
The Bank consequently needed to ensure that its aggregate supply of reserves met the daily CB demand for them. That, according to the governor, caused “hyperactivity in the Bank’s monetary operations (open-market operations) …. and persistent volatility in the overnight rate of interest in the money markets”.
Reserve levels became inadequate for financial stability purposes as liquidity management moved squarely into the hands of private institutions and markets during a period when CBs reduced their holdings of public sector liquid assets in favour of wholesale money market funding.
These byproducts of financial liberalisation, the governor in his speech conceded, contributed to the scale of the GFC.
An internal Bank review back in 2005 had earlier concluded that these evolved arrangements were undesirable and inimical to financial system stability.
The Bank’s 2006 related decision to remunerate reserves at the prevailing MPC-set BR was designed to ensure that the CBs (and any financial institution that wished to bank with the BoE,) held reserves sufficient to minimise the prospect of them requiring fluctuating overdrafts with the Bank.
This new system – known as ‘voluntary reserves averaging’ – allowed each CB to set itself an average target level of reserves over the ‘monetary maintenance period’ envisaged as that between MPC meetings; in return, the BoE would pay BR on their reserve balances close to each CB’s target, with available Bank standing deposit and lending facilities also paying and charging rates of interest close to BR.
Then came the GFC crisis and the Bank’s QE response. The subsequent expansion of its balance sheet produced an abundant ‘supply-driven’ reserves system.
CBs consequently no longer needed to pay above BR to borrow reserves for liquidity or other purposes.
Nor, with their reserves remunerated at BR, did they possess any real incentive to lend any excess reserves at below BR.
Meanwhile, interest-rate sensitive market competition to borrow funds for placement as reserves at the Bank ensured near parity on the overnight deposit and the BR reserves remuneration rates.
But as QT progresses, that ‘supply-driven’ reserves system is expected to revert to a ‘demand-driven’ one.
The Bank’s Executive Director for Markets in a July 2024 speech explained that this process is likely to result in a kink in the aggregate CB demand for reserves to occur when reserves ‘abundance’ is replaced by reserves ‘scarcity’.
A prospect, accordingly, requiring new institutional designs of reserves management and policy rate control (see later dedicated Section Two sub-section on Impact on MP, for fuller discussion).
2 The public finance ramifications of QE and the Asset Purchase Facility (APF)
During the five grouped QE programmes (see Table 1), starting in March 2009, the Bank made loans to the Asset Purchase Facility (APF) to finance the QE purchase of government gilts. Such loans to the APF are recorded on the Bank’s balance sheet as assets.
The Bank reserves (see previous dedicated sub-section) ‘created’ or ‘issued’ or even ‘borrowed’ by QE are recorded as a corresponding liability on the other side of the Bank’s balance sheet. These balance sheet identities therefore grow or reduce in tandem: they must balance.
The precise mechanism by which reserves are produced by the Bank remains a bit mysterious to outsiders, at least.
Clearest explanation (but still does not exhaust lay questions, such as is what stopping the seller ‘claiming’ their reserve credits, which the Bank has ‘paid’ for? ) found is from a National Institute Economic and Social Research (NIESR) blog: “The Bank of England paid for the gilts it bought by crediting the reserve balances of the commercial banks where the sellers held their accounts, for the credit of the sellers. It did not absorb the surplus funds on the commercial banks’ reserve balances. Thus reserve balances increased as the amount of QE increased”.
Essentially for the purposes of this post, privately held fixed-rate government debt (purchased gilts with an average maturity of 13 years) was swapped for publicly held floating-rate obligations (renumerated reserves variant with BR changes made by the MPC) making government debt increasingly sensitive to short term interest changes set by BR – a risk transfer that has had and will continue to have potentially very significant public finance consequences, as are set out below: the core concern of this post.
The APF was established in 2009 as a wholly owned subsidiary company of the Bank for the special vehicle purpose to hold the overwhelmingly government gilt assets that the Bank purchased under QE.
It is fully indemnified by HM Treasury (indemnity). Although the Deed of Indemnity remains unpublished, official documents confirm that any financial losses resulting from its activities are borne by HM Treasury. Any gains are owed to HM Treasury.
The indemnity means that the interest rate and associated public finance risk is borne by the Treasury (and by extension the public purse) directly: shortfalls have to met by either taxation or borrowing.
Although the APF does not report a direct profit or loss, each year the indemnity is valued on the difference between the fair value of the APF company’s assets and liabilities and settled accordingly (see 2024 APF annual report).
As such the APF, and by extension the Bank, was and is indemnified against financial risks associated with its activity, although it is required to manage those risks.
BoE loans to the APF to purchase gilts are recorded on the APF balance sheet as liabilities; the resulting APF cash deposits as assets. When purchased, gilts are valued as APF assets at their initial purchase price.
The Treasury pays into the APF the interest due on these purchased APF gilts; on the debit side of the APF accounts, APF outflows to the Bank pay for the Bank’s remuneration of the reserves at current BR.
As a rough indication, the accumulated APF gilts in 2021 earned fixed bi-annual ‘coupon’ payments equivalent to an effective annual interest rate of around two per cent (in 2021, the OBR advised that the average APF gilt effective interest return was then 2.1%, and higher for much of the period between March 2009 and December 2021).
On the outflow side, between March 2009 and December 2021, reserves were remunerated (financed from by APF outflows) at a prevailing BR of 0.5% or below, which between March 2020 and December 2021 fell further to a floor of 0.1%, as the official Bank Rate series catalogues.
That spread or wedge between the gilt interest inflows and renumerated reserve outflows generated APF ‘net profits’, that reached a cumulative total c£124bn by the third quarter of 2022, which, according to the OBR March 2025 Economic and Fiscal Outlook (para 6.12), were then transferred from the APF to the Treasury.
From July 2022 onwards that flow was reversed as the outflow flow of reserve interest payments from the APF to the CBs began to exceed the gilt interest flow into the APF. The previous positive surplus flow was thus transformed into a negative deficit one.
These losses were compounded by gilt valuation losses, which began in October 2022, when QT started.
QT crystallised valuation losses occur either when APF gilts are passively redeemed at their maturity (when purchased at an initial purchase price below their par redemption value); or, when they actively sold on the secondary market below their APF initial purchase price.
Such secondary market sale prices are sensitive to the maturity duration of the sold gilts and, most significantly, to the prevailing BR and to long-term gilt interest rate changes.
The inverse relationship between BR rises and gilt values means, for instance, that a gilt paying one per cent needs to offer an equivalent return to one paying four per cent of the same duration issued at current BR, requiring its price on the secondary market to fall commensurately.
August IPPR advised that APF gilt holdings were worth 26% (mark-to-market measure) less than their initial purchase (buying) price.
In total, according to the OBR, c£86bn was paid by the Treasury into the APF from the beginning of 2023 to 2025 to cover both APF interest and valuation losses.
This transfer flow – depending upon the BR path, gilt yields, and speed (run off) of QT sales (see next section for fuller discussion) – is expected by it to produce continuing APF further outflows of £100bn or more by March 2030 (rising further into next decade).
The OBR forecasts, more specifically, that although the annual transfer from the Treasury to the APF (table 6.4, reproduced in Table 2 below) will fall from an eye watering c£36bn in 2024-25, such losses between April 2025 and March 2030 will remain within a cc£18bn and c£24bn annual range.
Table 2: OBR Forecast of Treasury cash transfers to the APF, 2024-30, £bn
The OBR, when the pre-2022 APF interest net profits remitted to the Treasury are netted off, forecasts a net lifetime APF loss of c£134bn to the public purse.
The final figure, as noted above, will vary and depend upon future movements in BR, long-term gilt rates, and QT run-off, especially future BR.
It is quite possible but unlikely given recent and current conditions that BR falls and remains below in the medium term to below two per cent.
Certainly, it would seem sensible and prudent for public expenditure planning purposes to assume that it will not and that the downside risk of BR fluctuating at around four per cent or that it could go even higher if inflation remains sticky or reignites again should be factored in.
Could have such adverse, even momentous, public finance impacts been averted?
By mid-to-late 2019, pre-Covid, long forward gilt rates were unusually low: the 20-year forward rate was between one and two per cent; the 30-year was between nought and one per cent.
Indeed, the official gilt yield series catalogues that the 30 year gilt yield was less than two per cent between July 2016 and May 2022, falling to 0.6% in May 2020; the Treasury could have then potentially borrowed that month x paying 0.6% interest on x each year until 2052.
The government by issuing long-term bonds to the market could therefore have locked-in such unusually low long forward rates to fund pandemic-related deficit funding, rather than the APF borrowing, in effect, at a floating rate from the BoE to purchase QE5Covid-related purchases.
That meant that the Treasury was taking on the public finance risk that BR over the life of APF held bonds would rise above their long forward fixed rate – a risk that always risked becoming acute with any return to inflation as, in fact, did occur.
By the autumn of 2019, in the informed reading of informed financial insiders, including former deputy governor Tucker, it was, or ought to have been, (see Tucker reference one) clear to both the Treasury and the Bank that the combination of QE and paying interest on the additional £440bn additional reserves created as the result of the CovidQE5 gilt purchases, presented a clear future risk to public finance sustainability.
The governor, in his 2021 evidence to the House of Lords Economic Committee, advised that Covid QE was undertaken – as was GFC QE at the outset – primarily to provide needed liquidity to the financial system.
The magnitude of Covid-related QE (see Table 1), however, seems unnecessarily high – at least when measured against the governor’s objective/yardstick – and in the light of the public finance risks involved with APF gilt purchases – undesirable (although see the governor’s further arguments/points made in his evidence).
Also, by then it was known that the wider MP and the other impacts of QE on the real economy was uncertain and muted, at best.
And, in any case Covid was unlike the GFC was predominately a health rather than a financial crisis at source where the economy was subject to government decreed lockdown, acting on supply capacity directly rather than on aggregate demand, which was supported outside MP by fiscal action.
3. Uncertainties and short-term policy issues
The primary purpose of moving from full to partial renumeration or the reserves income levy is to raise much needed public revenue while minimising interference with the Bank’s MP operations and/or wider adverse economic effects.
This section focuses on, first, explores how the current fiscal framework and its rule ordering affects the interaction and possible trade-off between halting QT active sales or changing the speed of QT run off given the respective measurement and treatment of APF interest and valuation losses under that framework; second, the possible interface between the proposed reserve income levy and MP operation: and, third, the alternative case for a direct windfall tax instead.
Impact on the public finances and the fiscal framework
APF interest and valuation losses produce different impacts on the public finances, at least when as they are measured by the October 2024 fiscal framework iteration.
The Government’s main target measure for its fiscal mandate – the current budget to be in surplus by 2029-30 until 2029-30 becomes the third year of the forecast, when the target becomes a rolling three year one – is driven by APF interest, not valuation losses.
Such losses, according to the OBR, do not affect the measurement of the primary current budget deficit rule, but add (all else the same) to the Public Sector Net Debt (PSND) and to Public Sector Net Financial Liabilities (PSNFL), as changes to them are calculated by “combining changes in borrowing with changes in financial transactions and any valuation effects” (para 6.8, OBR March 2025)
This is significant as PSNFL in October 2024 became the numerator to the supplementary debt fiscal rule.
It requires the PSNFL/gdp ratio to fall by 2029-30, until that year becomes the third year of a rolling three-year forecast, upon which the third year of successive forecasts becomes the rolling target.
To recap, crystallised valuation losses occur when APF gilts are redeemed at a value lower than their initial purchase price, whether passively on their maturity date at their issued par value or when they are actively sold in the secondary gilt market.
In contrast to ‘mark to market’ value losses are measured annually within the APF accounts.
Essentially latter represents a ‘paper’ or accounting loss that may or may not be crystallised – but one that is still measured by the PSNFL – while a realised capital loss represents an actual one (a fall in house and share prices can result in ‘paper’ negative equity but actual gross profit or loss will depend upon realised actual final sale prices; if you bought an ’average’ house in 1990 but sold it in the autumn of 1992 you would have sold at an actual loss but would have realised a substantial real gain if you sold it in 2007, even though it would have a recorded a ‘mark to market’ loss in 1992).
Sales of APF gilts sold at a valuation loss, however, also ends the exposure of the APF to any future interest losses that otherwise would have been incurred on gilts sold between their sell and maturity date.
That effect should dampen – subject, as ever, to current BR and yield curve assumptions – the forecast future APF interest losses by the OBR.
Conversely, a faster pace of gilt QT sales crystallising valuation losses would, according to the OBR in March, improve the primary current budget fiscal rule or headroom position, while having a ‘modest’ negative impact on PSND and PSNFL and, by extension, the supplementary debt rule (para 6.14 and footnote six, p.132).
IPPR August advised that if active QT sales were paused (all remaining APF gilts passively redeemed at par value on their maturity dates), future APF valuation losses would be reduced by about a half (reflecting the difference between current mark to market fair and their redemption par values).
It went on to note crystallised valuation losses covered by the Treasury indemnity by adding to net public debt would impose future debt interest costs onto the current budget (figure 1.2 and table 3.1 and their commentaries).
The OBR in March forecasted (as Table 2 records) annual APF interest losses to fall from £18.5bn in 2024-25 to £3.3bn by 2029-30, based on falling BR expectations and a declining APF gilt stock, with APF annual valuation losses hovering around the £20bn level annually between April 2026 and March 2029.
It also assumed that the £100bn annual programmes of QT sales would continue from September 2025 onwards, with active sales comprising c£48bn of that annual £100bn total (para 6.11 and associated footnotes).
In the event, the September 2025 MPC, as the last section recorded, decided to reduce this coming year’s QT sales (October 2025 to September 2026) to £70bn, increasing active sales from their 2024-25 level of £13bn to £21bn but well below the £48bn assumed by the OBR.
Although, in the light of that decision, the forthcoming November 2025 OBR Budget Outlook is likely to adjust that March forecast – sensitive to any modified assumptions concerning the future speed of the QT run off it may make in any such revised forecast – the trade-off identified above between valuation and interest rate losses and their differential impact on primary current and supplementary debt fiscal rule measurement, where higher/lower levels of active sales should reduce/increase future interest losses, all else the same, will remain.
Ending the full remuneration of reserves through moving to tiered renumeration or – as August IPPR proposes – through a customised reserves income levy, would provide a substantial supplemental source of public revenue to the benefit of the current budget and to RR’s headroom; in effect, in effect if not obviating lessening any trade-off between valuation loss and future interest income liabilities in net public expenditure terms.
On the other hand, to paraphrase, among others, the BoE governor, it is not at all obvious that changing the pace of active QT(run off) — pulling losses forward or pushing them back in time — should matter very much to the chancellor’s fiscal headroom, insofar that the metric that matters is the present value of the crystallised flow of any future valuation losses (future loss has a lower monetary value than one incurred in the present).
The flies in the ointment of that viewpoint include that present values are not used to measure fiscal framework adherence, notably with reference to the primary current budget rule.
In any case, the current unprecedented divergence between BR and long-term gilt rates (steep gilt curve) suggests that the current time is the wrong time to actively sell long-term gilts bought at higher prices than their par values into the secondary market.
Realised prices will reflect current gilt rates (in early September 2025 reaching c5.7.% for 30-year duration gilts, falling slightly in early October c5.5%; compared to c4.7% for 10 year gilts), meaning that the sale price of APR gilts attached with a coupon rate paying less than the prevailing effective interest rate will need to fall to provide a commensurate rate return to other long-term gilts sold in the secondary market.
It is possible and quite likely but uncertain that long-term gilt rates will fall closer to BR (gilt curve flattens), in which case the future valuation losses on APF-held long-term gilts could reduce below what their current assessed present values indicate.
Impact on Monetary Policy
The official Bank view is that completely ending the BoE remuneration of central bank reserves at BR could make MP inoperable.
Essentially, this is because if all reserves were zero remunerated when their supply was above the minimum that satisfied commercial banks’ aggregate demand to settle their everyday transactions and to hold cash as a precaution against potential outflows in times of stress, the CBs would possess an incentive to lend them out, where they could, at best rates higher than the prevailing BR.
Such ‘a minimum level’ is postulated as the ‘marginal’ level of reserves on which MPC decisions on BR would bite for MP purposes.
It broadly corresponds to the Preferred Minimum Range of Reserves (PMRR) measure propagated by the governor and the Bank generally (see below).
Most commentators accordingly argue for the partial remuneration of reserves, not their complete non- remuneration.
Notably, Paul Tucker, a former career Bank insider that rose to become its deputy governor and was once the favourite to succeed Merlyn King as Governor, now a Harvard fellow, recommended ((key finding four, Tucker reference one) that when a central bank’s MP in future significantly employs QE, it then should not remunerate all the reserves held by the private sector but only “whatever fraction of reserves required to establish its policy (BR) rate in the short-term money markets”.
He did, however,regarding its immediate implementation, attach some substantive cavils to that conclusion, covering possible allocative efficiency, MP, financial stability and political economy concerns (key findings eight to 13)
Yet, the current governor has shied away from any move to limit reserves remuneration, even partially.
In Bailey 2024 he advised that it “would complicate the transmission of monetary policy substantially, because that begins with us setting the short-term official rate—the official Bank Rate. That transmits through the interest rate we pay on the reserves that banks hold at the Bank of England … it would complicate and weaken the implementation of monetary policy”.
He went on to explain that the Bank introduced a new weekly Short-Term Repo (STR) facility in 2022, complementing its other reserve supply operations – including its Indexed Long-Term Repo, (ILTR), which supplies reserves for six-months at a price related to demand, against wide a range of collateral.
Lending through the STR in the first instance, and eventually alongside other Bank facilities, according to the Governor, “will start to pick up the shortfall in reserves supply, to meet prudential needs for reserves and maintain monetary control through the setting of Bank Rate, with QT shifting to changing the mix of assets that back central bank reverses, not the level of reserves itself”.
Strategically, he advised, that the Bank is seeking to set up an operating framework that: (a) delivers on our core monetary policy and financial stability mandates, and subject to that; (b) minimises risk to the Bank’s balance sheet; (c) minimises market distortions, and (d) is transparent and accountable, predicated on the Preferred Minimum Range of Reserves (PMRR) concept.
The governor later recognised that quantifying the PMRR would prove tricky, as it “cannot be objectively observed, it is likely to evolve over time, and it will be affected by (Bank)decisions ….. (including) our choices in how we (the Bank) supply reserves, and the price and collateral terms of backing assets”.
In this emerging institutional framework, the setting and path of the PMRR – and by extension the unremunerated level of reserves – consequently would have to be flexible and adjustable.
If the tiered or partial set level of unremunerated reserves was mis-specified and set too high, MP acting through BR could, indeed, be compromised: it follows that the higher the tiered level was set, the closer it is likely to reach or exceed the PMRR.
As the QT process progresses, the gilts held in the APF may eventually fall below the PMRR level, probably sooner rather (that is unless QT sales are paused or pared back).
Other arguments against reserve remuneration include that it would tend to produce economic tightening effects during periods of QE and a loosening during QT periods: opposite to their MP intention.
Forcing unremunerated reserves on the banking system, according to some, also smacks of direct money creation: printing money with its associated bedfellows: financial repression (forcing savers and creditors to accept below market rates of interest) and fiscal dominance (subverting MP to fiscal imperatives that could involve unsustainable spending or an unsustainable mix of spending and taxation) with associated dangers to the Bank’s independent primary inflation remit and associated credibility.
These considerations probably explain why the governor, unsurprisingly, has both espoused and the Bank has developed approaches involving the ‘repo’ mechanisms outlined above that allow BBs to borrow unlimited amounts of reserves against gilt collateral, as these in combination with BR will allow QE to unwind “without the risk of any loss of monetary control on the way to the PMRR”.
What is neglected by the above discussion are the immediate public finance implications of unwinding QT and how to avoid their future reoccurrence in any future QE deployment.
A levy on reserves income and alternative direct fiscal responses
As was set out in the introduction, IPPR August recommended a levy on QE reserves where remunerated at above a two per cent interest rate return.
Although largely mimicking in conception and reach the tiered reserves approach, the levy would be implemented by the government as a fiscal policy measure, avoiding, according to the IPPR, possible conflation and confusion with the Bank’s MP role.
Other possible posited advantages include that it would (be):
- relatively easy to implement using existing Bank financial payment systems;
- difficult if not impossible to avoid, insofar that the Bank can through its monetary operations control the aggregate level of reserves and thus can prevent CBs as a collective group reducing their reserve holdings (see Section One central bank reserves section);
- directly apply and respond to a primary source and driver of rising CB supra profits derived from BR-renumerated reserves renumeration;
- avoid the possible distortionary impacts that a more general tax levied on CB activity and any associated incentives to shift profits abroad or to otherwise avoid incidence, could involve;
- put in place an institutional bespoke mechanism ready to be applied and tweaked when future QE programmes were deemed necessary for MP purposes.
And, not least in practical political terms, it would raise a lot of money – between a cumulative £34bn and £41bn by end of this parliament if implemented this year as designed and assumed, depending on future BR, gilt curve, and QT run off rates and their interaction.
That is a tidy receipt, just what RR needs to help her meet her primary fiscal rule!
But, as the IPPR concedes – as did Tucker in his qualified 2022 espousal of tiered remuneration – such a levy could still potentially impact on the Bank’s MP remit and operations.
Given the MP problems associated with a partial tiered system of reserves, the governor and other commentators have suggested that the public finance angle would be cleaner and better addressed, rather, by a direct fiscal bank levy, or put more politically, by a ‘windfall’ tax (IPPR cites banking company corporation tax surcharge) on the CBs.
They point out that this would fairly and squarely be a fiscal measure levied by the elected government approved by parliament, unrelated to Bank MP mandate or activity.
Like most taxes, a direct bank tax or levy could, however, deflate and/or distort economic activity, namely through its restrictive impact on bank lending and could also provide a possible competitive advantage to non-bank institutions (as could a levy on renumerated reserves), unburdened by reserve and regulatory requirements with possible attendant adverse financial system stability implications.
Its yield could prove less certain and unstable compared to a levy on remunerated reserves, which itself, however, would also be subject to implementation and lobbying pressures, as well as to shifting economic and financial circumstances.
Financial effects of such a tax on CBs would vary with its magnitude and effective incidence.
Effective incidence would depend on their apportionment of the financial hit to be taken by their customers compared to that taken by shareholders and – through reduced bonuses and share awards etc – by powerful internal management stakeholders able to obtain and extract unearned economic rent returns.
Insofar that the retail banking market is oligarchic rather than competitive, its incidence could, or even probably, according to many informed commentators, including Tucker, predominately thus fall on bankers and shareholders rather than on banking activity impacting on real economic outcomes.
Such a windfall tax – as could a levy on renumerated reserves but due to its technical nature, possibly less so – accordingly could be related to a wider Starmerite ‘we are all in this together’ political narrative, where rent-seeking activity and returns are taxed heavier (albeit by different customised mechanisms designed for dedicated sector-specific purposes) than are returns won in competitive markets.
That narrative and principle could cover not only the CBs but also the private utilities, and the operation of the private speculative housing market by the major housebuilders, although such a slogan, as ever, will skim over complexities and risk of unintended consequences.
4 Conclusions and implications
All very complicated and somewhat cloudy then. Uncertainty shrouds; trade-offs should, at least, be recognised.
However, what does shine through in clear light is the intrinsic tendency of QE – as implemented by the APF (as Section One described) – to bequeath considerable latent, increasingly crystallised since 2022 into actual public finance costs (as Table 2QE) sourced from OBR estimates summarised for the remainder of this parliament).
The fiscal public finance implications of QE and then QT – given their overall magnitude and part they will play in the Hobson’s Choice that RR faces in her November budget – have come to a head.
A timely but considered and effective public policy response can no longer be put on one side.
Although not the focus of this post, at an overarching level, the weak substantive evidenced validation of the impact of QE and QT, especially on real economic activity and their wider process linkages to MP, is also striking from the material that it has assembled and used.
Indeed, the governor has himself noted “I am not sure how large a room would need to be to fit in all the people who really understands the Bank’s balance sheet” (Bailey 2024).
Hindsight, of course, is a wonderful thing. And, undoubtedly, the Bank and its MPC face a tough task, where they can be and are dammed if they do and damned if they don’t.
Since 2022 they have had to navigate a near return to stagflation conditions where simply bearing down on short-term inflation could well throw the economic baby out with the bathwater, but ignoring its sticky existence risks embedding inflationary expectations inimical to medium-term sustained non-inflationary growth (for instance, see discussion in the Bank’s August 2025 Monetary Report).
It suggests that the initial success of the post-1997 monetary framework was, at least in part, the product of a fortuitous global political economy backdrop that is no longer with us and is unlikely to return in the foreseeable future.
The core macro-economy orthodoxy for decades has been that there is no long-term trade off relationship between inflation and growth (without price stability, long-term growth will be jeopardised).
It rather begs the question, given the recent dilemmas that recent MP has had to navigate, as to what the length of the medium term should be for MP mandate purposes.
Turning back to the primary but related purpose of this post, the Bank’s post-QE remumeration of its reserves and the complicity of the Treasury in its APF implementation, as well as that (or negligent oversight) of the wider political insider community, provides a prime example of that tendency, appearing as a clear mistake of both commission and omission.
As the Bank insiders and the 2021 House of Lords report referenced in this post have highlighted, the later Covid-related QE rounds came attached with clearly discernible public finance risks, subsequently realised, that should have been addressed in policy consideration and decisions by both the Bank and Treasury at the time, provoking related questions as to why proposals to partially renumerate bank reserves were not taken on board earlier, either by Rishi Sunak’s administration or by the incoming Starmer administration and by Rachel Reeves in her first budget.
Certainly, the right questions were not asked at the right time and/or were not taken seriously enough by the people who ought to have asked them, for whatever combination of reasons.
That horse may not yet have bolted but before it does, the opportunity should be taken this autumn 2025 budget cycle without any further delay to recycle for public purposes the supra-profits that the CBs reaped from QE-related reserves remuneration.
Section Two discussed in some detail the pros and cons of putting a levy on renumerated reserve income above two per cent as recently proposed by AugustIPPR compared to a more conventional windfall tax, with reference to the Bank’s MP operation and the upholding of its primary inflation mandate.
Readers will draw their own conclusions concerning the arguments and considerations made.
The assessment of this website – qualified, where applicable at the margin, by the recognition that the Treasury and Bank shareholders have possession of relevant information not available to this website – that although broadly agnostic between proposed variants of imposing a QE-reserves income levy (relatively easy to implement, is customised to cause and remedy, establishing a workable precedent for the future, but proceeds depend on future BR path and implementation could on the margin impinge on Bank MP) and a bank windfall tax (avoids any interaction with the Bank’s MP remit, is more amenable to a political in contrast to a technocratic narrative, but could prove more difficult/problematic to implement to achieve target proceeds allied to possible unintended consequence risk), it is unequivocal that the current public finance position overrides any possible problems, as identified above, in progressing with either with best possible alacrity.
Summary of other recommendations
- The Bank’s mandate should be revised (probably best through the Chancellor’s annual remit letter to the Governor) to ensure that the Bank and Treasury jointly take responsibility for the future public finance implications that arise from the Bank’s MP operations, most pertinently and presently with reference to the operations of the APF concerning active gilt sales (noting that Treasury representative attends MPC and their contribution – limited to public finance implications – should be minuted).
- QT, in words of the governor himself, has limited or tangential impact on the Bank’s MP operation. Active APF gilt sales should be halted or proceed only when future net public finance impacts are minimised on proper and due assessment (probably as best advised by UK Debt Office).
- QE should only be used in future when its expected impacts are clearly identified and justified relative to possible alternatives and should be implemented taking regard to its future impact on the public finances.
- The House of Commons Treasury Committee should step up its scrutiny and review capacity concerning the interactions and inter-relationships between fiscal and MP.
- The Bank itself should develop and improve its communications to explain its operation of MP in lay understandable language more honestly and transparently (in this website’s own experience precise mechanism by which the reserves liabilities resulting from Bank QE activity via APF are produced and distributed between CBs, not clearly explained by the Bank).