Sarah Breeden, who is currently the Bank’s Executive Director for Financial Stability Strategy and Risk [FSSR], has been appointed as Sir John Cunliffe’s successor as the Deputy Governor with responsibility for Financial Stability, starting on 1 November 2023.

Authors: Stephen G. Cecchetti and Jens Hilscher*

Published: 20.03.2024

The Bank of England’s bond sales are having an enormous impact on the current UK government’s budget. This is because the agreement between HM Treasury and the Bank of England states that the losses are compensated by transfers as they accrue. From 2013 to 2022, the Bank of England made positive transfers, increasing government revenue by a total of 6.23% of UK GDP.  But starting in 2023, profits turned to losses. The current yield curve implies an expected value of these losses of about 7.5% of GDP.

*Image of the Bank of England, London, UK sourced via Canva Pro

There are three types of losses: (i) the Bank pays a higher rate on funding than it collects from coupon payments on bonds it holds (negative carry), (ii) for bonds purchased at a premium, there are realized losses at maturity, , (iii) the Bank chooses to sell bonds and realize losses that would otherwise be classified as (i) and (ii) in the future. The sum total of these losses will be roughly the same regardless of whether the Bank’s Monetary Policy Committee chooses to actively sell bonds or allow them to simply mature. Importantly, the decision to sell means bringing the losses forward in time in a way that can have significant fiscal consequences. Under the current policy of reducing the size of its bond holding by roughly £100 billion per year, the losses through 2031 will be £191 billion (all three types combined), compared to £78 billion under a policy of passive runoff (only the first two types of losses). That is, the excess fiscal cost to HM Treasury is an average of £14 billion per year – or 0.45% of UK GDP!

Better, in our view, would be to smooth out the realized losses by allowing the Asset Purchase Facility to run down slowly over the next several decades. In this post we discuss the sources of the losses, how the UK is choosing to account for them, and why the decisions of the Bank of England are having an unnecessarily large short-term fiscal impact.

Beginning during the Global Financial Crisis of 2008-09 and continuing through the COVID pandemic, central banks around the world engaged in large-scale asset purchase programs that significantly increased the size of their balance sheet. In the case of the Bank of England, bond holdings peaked at nearly £900 billion at the end of 2021, roughly doubling the overall size of the balance sheet from its end-2019 level. When they initially purchase the bonds, the policymaker’s objective is to reduce longer-term interest rates, compressing some combination of sovereign term spreads and risk premia on private sector bonds. If successful, these policies stimulate aggregate demand, stabilizing inflation, growth and employment, and the financial system. Ideally, this reduces the length and severity of recessions, thereby increasing aggregate production and tax receipts.

Once the threat of recession is over, and the interest rates rise back to their normal level, the central bank will start to suffer losses as the interest income from the bonds they hold will be less than the interest expense of the financing. As a result, remittances from the central bank to the fiscal authority shrink. Depending on the indemnity agreement with the finance ministry, as well as the accounting rules that are in place, central bank losses (negative payments) trigger an explicit fiscal expenditure either immediately or at some time in the future. That is, central bank balance sheet policies have direct fiscal consequences.

Turning to the specific case of the Bank of England, in a letter dated 29 January 2009, Chancellor of the Exchequer, Alistair Darling, set out the terms of the Asset Purchase Facility,  a subsidiary of the Bank of England, and the legal entity that makes the actual asset purchases. The letter describes the size of the facility, originally £50 billion; the eligible securities, including corporate bonds, commercial paper, syndicated loans, and certain asset backed securities; that the facility is financed by borrowing from the Bank of England; and, importantly, that the Government indemnifies the Bank of England, so the government bears the risk.

The mechanics of this are as follows. The Bank of England lends funds to the facility, which then purchases UK gilts. The facility receives interest income in the form of coupon payments on the gilts and pays interest at the cash rate (the overnight policy rate) to the Bank of England. The UK Treasury receives any surplus and must make up any shortfall. Importantly, these positive or negative transfers occur every three months.

The consequences of this arrangement are straightforward. During the period of very low policy rates, the facility generated a positive cash flow for the fiscal authority (approximately 6.23% of GDP). But, as policy rates rose over the last few years, the flow turned negative. To give a sense of the size of this swing, we can do a simple calculation for 2024. Start with the fact that the face value of the holdings at the beginning of 2024 are £639 billion with £46 billion maturing during the year. These holdings yield coupon interest income of £15.2 billion. Next, note that the price at which the bonds were purchased, their historical cost, totals £744 billion of which £49.6 billion will mature. Using the OIS forward curve which reflects the expected future policy rate – 5.06% at 6 months and 4.23% at 12 months at the beginning of 2024 – this implies an interest cost of £33.6 billion. So, the net interest expense (the negative carry) in 2024 is expected to be £18.4 billion. In addition, during the course of the year, £46 billion will mature and generate a realized loss (the purchase price minus the par value) of £3.5 billion. Adding these together, we see that the total loss for the current year is £21.9 billion, or 0.80% of UK GDP.

Figure 1 shows the actual and projected transfers between the Bank of England and HM Treasury from 2013 to 2071 – when the last bond matures – as a fraction of UK nominal GDP. For 2023, we report the actual losses associated with current policy of sales, while starting in 2024, we compute losses assuming a completely passive policy in which all bonds are held to maturity. As we mentioned at the outset, from 2013 to 2022, the Bank of England made positive transfers to HM Treasury. In fact, from 2014 to 2022 (ignoring the large positive value in 2013) average transfers were 0.44% of UK GDP per year. Looking at the losses, we divide these into the net interest expense (in orange) and the capital loss realized at maturity (in red). For 2023, the total loss is 1.43% of GDP. Our estimates based on the average purchase premium and realized losses from sales suggest that roughly half of this is from realized losses associated with selling long-term securities purchased at a price in excess of the face value (a premium) that were sold at price less than the face value (a discount).

Starting in 2024, estimates are based on the current yield curve. That is, we compute the financing cost at the expected future interest rate and then subtract the coupon income. To this we add the realized losses associated with the fact maturing bonds were purchased at a premium above the face value but only return the face value. These (undiscounted) losses through 2071 sum to –7.46% of GDP. This should be balanced against the combination of profits over the past decade (+6.23 percent of GDP) and the stabilization benefits from accommodative monetary policy.

Figure 1: Bank of England Asset Purchase Facility Net Profits and Losses, 2013 to 2071 (% of GDP)

Note: Data through 2023 are the net transfers from the Bank of England to HM Treasury as reported by the UK Office of National Statistics. For 2023, the division between the carry loss and the capital loss is the authors’ estimate. Data for 2024 to 2053 are authors’ estimates based on the information on the holdings of the Bank of England’s Asset Purchase Facility, combined with a forward interest curve as published by the Bank of England, assuming that all bonds in the APF are held to maturity. Data for GDP through 2028 are from the IMF. From 2028 on, nominal GDP is assumed to grow at a 3.6% constant rate consistent with assumptions in Office of Budget Responsibility (2023).
Source: UK Office of National Statistics, Bank of England, IMF World Economic Outlook Database, and authors’ calculations.

The financing arrangements for the APF mean that HM Treasury must make transfers to the Bank of England in compensation for the losses in the Asset Purchase Facility. These payments create a fiscal expenditure – both the negative cash flow and the realized capital losses. At one level, this is purely bookkeeping, as the amounts are exactly what the UK debt manager would have to pay if they had issued debt with the maturity structure of the consolidated balance sheet of the government. But from a political perspective, there is obviously a fiscal problem as the APF’s losses show up as an explicit budget expenditure for the government. And, as we discuss below, there may be an economic problem as well.

It is worth emphasizing that the estimates in Figure 1 represent the slowest possible path for the runoff of the APF. Because the current policy of the Bank of England is to sell bonds before they mature, these losses are being realized more quickly. The end-of-2023 Asset Purchase Facility Quarterly Report suggests a base case in which the combined maturities plus sales will be roughly £100 billion face value per year until nothing is left. These sales result in the realization of fairly large and immediate losses. To see how large, we note that the bonds currently in the APF were purchased at an average premium of roughly 15%. Sales in 2022 and 2023 were at a discount of 25%. That is, on average, bonds purchased at £115 are being sold for £75.

Figure 2: Bank of England Asset Purchase Facility Net Losses with and without sales, 2024 to 2031 (% of GDP)

Note: We assume that the APF sells £100bn each year. The total is a combination of bonds that mature, paying the face value, and sales. We assume that the APF sells bonds across the board and measure the total size of sales using the average purchase price. Losses then have three components: net interest expense (negative carry), capital losses from maturing bonds, and capital losses from sales. We assume that the bank sells only bonds with maturities of 2032 and beyond since the other bonds will naturally mature within the time span it takes to unwind the APF. Projected sale prices are calculated by discounting bond cash flows using the forward gilt curve published by the Bank of England. Cost of carry is based on the forward OIS rates. Nominal GDP projections are based on a 3.6% annual growth rate.
Source: Bank of England and authors’ calculations.

To understand the implications of this policy, we plot projected measures of the losses with and without sales in Figure 2. We focus on the period 2024 to 2031, which is when the APF’s assets reach zero if the sales plus maturing securities sum to £100 billion. Over this 8-year period, the undiscounted losses without sales average £9.7 billion per year. By comparison, with sales, the average is £23.8 billion per year. As a percent of GDP, the path with sales averages 0.78% of GDP per year, while the path without sales implies average losses of 0.35% of GDP. (Since UK government spending is roughly 45% of GDP, these numbers represent rough 0.8% and 1.7% of total fiscal expenditure.)

Importantly, the discounted sum of these losses does not depend on whether there are sales or not. But we see three reasons to believe that the timing does matter. First, so long nominal growth exceeds the nominal interest rate on average, the slower path pushes the losses out far in the future where (we hope) GDP will be higher making them less painful. Second, since there are constraints on the government’s ability to borrow, the need to fund the APF’s losses from current tax revenue is reducing expenditure on everything else. And third, related to this last point, the UK economy may be in a recession, so the marginal value of government expenditure is relatively high making this a particularly bad time to be forced to use revenue to fund APF losses. Put slightly differently, a smoother path for loses puts less short-term pressure on fiscal finances.

To summarize, at the beginning of 2024, the UK’s APF has large holdings with significant unrealized losses. The institutional arrangement between the Bank of England and HM Treasury means that APF losses are covered by HM Treasury as they occur. The current schedule of bond sales, together with this institutional setup, implies substantial deficits for the fiscal authority.

Indeed, the UK government is well aware of these implications:

“When the Bank of England voted to implement quantitative tightening in September 2022, it put the UK economy in uncharted territory — particularly in terms of the decision to actively sell gilts back to the market.

No major central bank has pursued QT in this way. Both the Federal Reserve and the European Central Bank have opted only for the passive method of allowing their bonds to mature without replacement.”

Harriett Baldwin, chair of the House of Commons Treasury select committee, 7 February 2024.

One potential response by HM Treasury could be to simply roll over the shortfall by issuing new debt. However, this path will increase public sector borrowing.

This leads us to the following set of conclusions. First, since the UK government owns the Bank of England, there is no way to escape the losses implied by the large-scale asset purchase program. Furthermore, for the central bank to be effective in meeting its price stability objective, fiscal backing is essential. But the decisions of the Monetary Policy Committee to actively sell bonds, rather than passively allow them to mature, means larger short-term losses and bigger transfers from the UK Treasury. As MP Harriet Baldwin says so clearly, this is something other central banks are not doing. Is it really necessary? Our conclusion is no. Even if the goal of the Bank of England is to reduce the level of commercial bank reserves in the UK financial system, they have other tools – like repurchase agreements – that they can use to avoid outright sales that are so destabilizing to the UK government’s budget.

This post is an excerpt from a longer work entitled “Fiscal Consequences of Central Bank Losses,” by the same authors.

*Stephen G. Cecchetti is Senior Research Consultant at King’s College London Business School Professor and Rosen Family Chair in International Finance at the Brandeis International Business School, and Chair of the Advisory Scientific Committee of the European Systemic Risk Board; Hilscher is Professor of Agricultural and Resource Economics at the University of California, Davis.