Sovereign debt: the buyer of last resort

At their September 16, 2026 meeting, Federal Reserve policymakers raised interest rates. Another question surfaced in the discussion, one that has little to do with the direction of the policy rate. Treasury trading was functioning smoothly, a few participants observed, yet the Fed needed to be prepared for an abrupt breakdown. The minutes, released on October 7, recorded suggestions to improve the central bank’s tools and communications for market stress while limiting its footprint in the Treasury market. This was contingency planning, not an announcement of a new intervention. ^[1]

That distinction can seem remote until buyers step away. Across the preceding six installments, this investigation followed a sovereign bond from auction to dealer inventory, leveraged buyers, collateral chains, margin calls and distressed sales that can eventually affect the issuer’s borrowing costs. At the end of the journey sits an institution able to supply money when the market can no longer do so fast enough: the central bank.

Such support can contain contagion. It can also change incentives, shift risk between private and public balance sheets, and influence the prices at which governments borrow. The contracts governing an intervention and the manner in which it is unwound tell us as much as the initial announcement.

March 2020: when the safest asset becomes a source of cash

US Treasury securities serve several purposes at once. They are investments, liquid reserves, and the collateral behind an enormous volume of short-term lending. Those characteristics make them exceptionally useful under normal conditions. In March 2020, as investors raced to raise dollars, they also made Treasuries a readily available asset to sell.

Research at the Federal Reserve Bank of New York describes customer selling that overwhelmed dealers’ capacity to intermediate trades. Liquidity deteriorated, bid-ask spreads widened and the market’s ability to absorb orders declined. Problems in the Treasury market threatened the plumbing of financing transactions across the system. ^[5]

The Fed initially expanded its repurchase agreement, or repo, operations. In a repo, an eligible borrower receives cash against securities under an agreement to repurchase them. Outstanding Fed repos approached $500 billion in mid-March 2020, according to the Fed’s retrospective account. The central bank then dramatically accelerated outright purchases of Treasuries and agency mortgage-backed securities. These instruments work differently. A repo must be repaid and the collateral returned; an outright purchase leaves the central bank holding the asset until it matures or is sold. ^[19]

The Fed’s daily records capture the sudden pace. On March 27, 2020, the data show $71.50 billion in Treasury purchases alongside $41.04 billion in agency MBS purchases, for a combined $112.54 billion on that trading day. The release identifies what the Desk purchased. It does not identify every ultimate investor whose sale was made easier by the resulting liquidity. ^[3]

In a July 15, 2020 retrospective, New York Fed markets executive Lorie Logan reported approximately $1.7 trillion in Treasuries and more than $800 billion in agency MBS bought since mid-March. These figures cover a substantially longer period than the late-March daily spike. A subsequent New York Fed research paper finds that the recovery in market liquidity is consistent with the purchases having been effective. Multiple measures were deployed during the crisis, however, so one cannot mechanically attribute every improvement to a single intervention. ^{[4] [5]}

The Fed’s role was primarily to absorb securities already circulating among investors, rather than simply replacing a private bidder in a Treasury auction. That change can also improve the conditions for subsequent government issuance. Measuring the transmission into new borrowing costs requires separate evidence. The headline purchase total alone cannot provide it.

September 2022: time for the UK pensions system to adjust

The UK crisis provides a tighter case study. A sudden change in the fiscal outlook was followed by a rapid rise in long-dated gilt yields. Pension funds using liability-driven investment, or LDI, strategies faced demands for additional collateral. To raise cash, they sold long-maturity government bonds, driving prices lower and generating further demands for margin. The speed of the move exposed inadequate buffers and operational constraints across parts of the sector. The UK’s system-wide exploratory exercise, published in 2024, subsequently examined stressed interactions between banks and non-bank investors. Its scenario is distinct from the observed 2022 transactions. ^{[21] [17]}

On September 28, 2022, the Bank of England announced targeted purchases in the affected segment of the gilt market. Buying ended, as scheduled, on October 14. The Bank initially offered potential purchasing capacity of up to £65 billion, subsequently increased for the final operations. Its actual temporary portfolio reached £19.3 billion, comprising £12.1 billion of conventional gilts and £7.2 billion of index-linked gilts. The larger number described the possible scale of the backstop; the smaller one describes what was bought. ^{[6] [7]}

The operating rules are central to the story. The Bank sought backstop prices that would support trading without replacing ordinary market demand. It set a firm end date and later resold the bonds as buyers expressed interest. The portfolio was entirely unwound by January 12, 2023. ^[22] Official reports put total sale proceeds at £23.1 billion. In March 2023, senior Bank of England official Andrew Hauser reported that the intervention generated £3.8 billion for the government. That was the outcome of this specific operation, not an expected return on future rescues. ^{[6] [7] [8]}

The intervention allowed LDI funds and their pension fund investors time to reinforce their financing positions. It did not establish a permanent interest-rate guarantee for the British government. A favorable realized return should not obscure the fact that investors experienced losses and the public sector accepted meaningful tail risk while the operation was running. The episode shows what a contained intervention can achieve under particular circumstances. ^{[6] [20]}

Secured lending backstops

Since the crises of 2020 and 2022, central banks have developed facilities intended to curb forced selling without necessarily accumulating vast new outright bond holdings. Their practical importance lies in the terms: who qualifies to borrow, what collateral is accepted, at what cost, and when repayment falls due.

On September 16, 2026, the New York Fed updated the operating details of its standing repo mechanism. It provides two daily windows, morning and afternoon. Eligible counterparties include primary dealers and selected deposit-taking institutions; acceptable securities include Treasuries, agency debt and agency mortgage-backed securities. The published rate rose to 4.00%, effective September 17, 2026, consistent with the September FOMC directive. The FAQ specifies a $40 billion limit per eligible security type per operation, with one proposition allowed for each type. That is a transaction parameter, neither a promise to deliver $40 billion nor an observation of actual facility use. ^{[1] [2]}

A hedge fund holding Treasuries does not thereby receive direct access to the Fed. It may borrow from an eligible bank or dealer, which in turn can use the facility if its conditions are satisfied. The distinction matters: a public liquidity option is being provided to specified intermediaries; the allocation of that financing to the intermediaries’ own clients remains governed by private contracts and prudential constraints.

The Fed also offers a different route for overseas monetary authorities. Through its FIMA repo facility, approved foreign central banks holding Treasuries in custody at the Fed can temporarily raise dollars against those bonds. The arrangement offers an alternative to dumping Treasury holdings on the market to meet an urgent need for cash. Created in March 2020, FIMA repo became a standing facility in July 2021. Eligibility applies to approved official institutions, not private bond funds in general. ^{[15] [16]}

In Britain, a more striking institutional change concerns lending directly to non-banks. Since January 2025, eligible firms have been able to apply to participate in the Contingent Non-Bank Financial Institution Repo Facility, or CNRF. The initial eligible groups include certain pension schemes, LDI funds and insurers. The facility is designed to operate only in exceptionally severe gilt-market disruption that threatens financial stability, when demand for liquidity lies beyond the existing bank facilities and lending can address the dysfunction. Pricing is intended to be unattractive in ordinary markets while providing a usable backstop under stress. Opening the facility to applicants does not itself establish that any loan has been drawn. ^{[18] [9] [10]}

Neither facility erases collateral haircuts or repayment obligations. Their purpose is to substitute conditional liquidity for a sale that a distressed market might struggle to accommodate. A sound institution short of cash presents a different challenge from an insolvent one whose assets no longer cover its debts.

The TPI and euro-area legal constraints

Euro-area governments issue debt in the same currency, but under separate sovereign credits. Their yields may diverge because of genuine differences in public finances, economic expectations and liquidity, as well as market movements detached from fundamentals. The European Central Bank must distinguish those forces while preserving the common monetary stance.

Article 123 of the Treaty on the Functioning of the European Union prohibits the ECB and national central banks from directly purchasing government debt from public authorities or extending them credit. Secondary-market operations fall within a different legal framework with their own constraints. Those rules determine which transactions are possible and for what purpose. ^[14]

On July 21, 2022, the ECB unveiled its Transmission Protection Instrument, or TPI, to counter unwarranted, disorderly market dynamics that seriously threaten the transmission of monetary policy. When its conditions are met, it can purchase public debt on the secondary market, generally with one to ten years of remaining maturity. There is no preset purchase volume, but any activation requires a proportionality assessment and examination of fiscal and macroeconomic eligibility criteria. ^[12]

The ECB’s stated criteria concern compliance with EU fiscal rules, the absence of severe macroeconomic imbalances, debt sustainability and adherence to relevant European policy commitments. An excessive deficit procedure alone does not exclude a country taking the required effective corrective action. Purchases can cease when transmission improves durably or persistent tensions are judged to reflect national fundamentals. The published framework creates a conditional possibility, not an automatic entitlement for a government to obtain ECB financing. ^[12]

The timing of other ECB programmes provides essential context. The Eurosystem stopped reinvesting principal from maturing APP assets in July 2023 and from PEPP assets at the end of 2024. According to the ECB’s 2025 annual report, no monetary-policy asset purchases took place under these two programmes during 2025; the corresponding portfolios continued to decline as holdings matured. That statement concerns the scope and timing of APP and PEPP. It is not a substitute for examining any decision under the distinct TPI framework. ^{[13] [23]}

The hardest decision comes when rising yields reflect several causes at once: persistent inflation, concern about fiscal sustainability, and a temporary shortage of buyers. A central bank can protect the transmission of the common monetary policy. Determining a sovereign’s long-term fiscal capacity on behalf of its government would cross an entirely different boundary.

Who actually benefits from a central-bank backstop?

Distress belongs first to the forced seller. An outright purchase or secured lending facility can let that investor liquidate a position or wait for a less frantic market. Dealers may regain the ability to quote prices; investors may once again face executable markets. Pension schemes and the companies standing behind them may be spared a wider disruption from a short-term scramble for cash. The size of these benefits is an empirical question in every episode.

The gains differ from one participant to another. A seller to the central bank gets a particular transaction price. A repo borrower gets temporary cash and an obligation to repay. An investor who remains exposed to the bonds still bears interest-rate risk. The government’s possible benefit comes through improved market functioning at later debt auctions, not through a direct subtraction of central-bank purchases from its future interest bill.

The 2022 British case also exposes the public risk. Before purchases began, the UK Treasury had provided the Bank of England with a £100 billion indemnity for the operation. The eventual bond purchases amounted to £19.3 billion, and officials later reported a £3.8 billion realized return to the government. These figures measure a contingent risk guarantee, actual transactions and a realized result, respectively. They are not competing estimates of the same quantity. ^{[20] [8]}

Success on one occasion is not a contractual guarantee of success on another. A sustained increase in yields while the bank holds the securities could generate losses; weak collateral or a failing counterparty creates different risks. A backstop that investors can rely on too easily may encourage them to carry smaller liquidity buffers or greater leverage than they otherwise would. That is the economic and political challenge of designing the intervention. ^[20]

Emergency purchases and QE cash transfers

There is another reason to keep the categories separate. The Bank of England’s 2022 financial-stability gilt purchases were narrowly targeted and temporary. The much larger Asset Purchase Facility, or APF used for quantitative easing, or QE, accumulated holdings over many years to influence monetary conditions. The outcome of one exercise is not the balance-sheet result of the other.

On August 4, 2026, the Bank of England published its Asset Purchase Facility report for the second quarter. Net cumulative cash transferred from the APF to HM Treasury had peaked at £123.9 billion at end-September 2022. The direction of quarterly transfers reversed from October 2022. At June 30, 2026, the cumulative net transfer remained positive, at £16.2 billion. Subtracting the two published cumulative values gives £107.7 billion. That difference is a change in cash transfers under the QE-related indemnity mechanism, not a cost of the temporary LDI rescue, and not the comprehensive fiscal cost of QE. ^[11]

There is a mechanical reason those cash flows can reverse. Reserve balances created to finance asset purchases earn interest tied to central-bank policy, while many of the purchased securities pay coupons fixed years earlier. Higher official rates therefore raise the financing cost of the portfolio relative to some of its income. Selling bonds bought when yields were lower can also crystallize valuation losses. The government indemnity covers cash shortfalls under the programme. ^[11]

The Bank’s report is explicit about what the cash transfers leave out. QE may have lowered government borrowing costs and supported output, employment and tax receipts. Its own scenarios leave the combined fiscal effects substantially uncertain. Measuring losses on asset sales while omitting the benefits of cheaper sovereign issuance and macroeconomic stabilization would therefore produce a distorted accounting. ^[11]

A more useful test follows from the emergency facilities: was temporary support priced, collateralized, limited and unwound in a way that avoided a standing subsidy to leveraged private positions? The UK record offers evidence for one design in 2022. It does not establish a universal rule.

An insurance policy needs a price and an expiration date

The three monetary systems now offer different approaches. The Fed used massive outright purchases in 2020 and has since built standing secured lending capacity for eligible institutions. The Bank of England purchased gilts temporarily in 2022 and later offered a contingent lending route for certain non-bank investors. The ECB designed a tool to protect monetary transmission subject to European legal and fiscal constraints. Each framework reflects a different institutional structure. ^{[1] [9] [12] [19]}

A public backstop must be credible enough to halt contagious selling. Its terms must also encourage market participants to finance their own resilience during normal times. The strongest objection to discretionary intervention is that a central bank can end up underwriting positions that were insufficiently robust and encouraging their return. The strongest case for intervening is found in the documented crises: allowing distressed sales to disrupt a systemically important market can impose much broader costs on households and businesses than a tightly designed response. ^{[6] [10] [20]}

Judging the response therefore requires more than the amount of bonds bought. The central questions are who could access the facility, which assets qualified, at what price, for how long, which losses remained private, and how the operation ended. The public record of counterparties and transactions will ultimately determine how much of that allocation can be established independently.

Seven installments later, who holds sovereign debt?

This series began with a government selling bonds. We followed the securities through auctions, dealer balance sheets, investors using borrowed money, collateral exchanges and margin calls. The legal instrument remained the same while its owners, lenders and guarantees changed. Ordinarily these links distribute finance and risk. Under stress they can synchronize the need for a single thing: immediate cash.

At that point a monetary authority may lend against collateral or purchase securities outright. Both can reopen a market when private balance sheets are overwhelmed. Both operate within legal boundaries, monetary-policy objectives and the possibility of transferring risk. March 2020 and September 2022 demonstrate how powerful these tools can be. Their differences explain why the design and exit terms matter as much as the initial commitment. ^{[5] [6] [12]}

Three investigative questions remain. Aggregate holdings data cannot fully identify the final owners of sovereign bonds or reveal every institution’s exposure. Private funding commitments remain only partly visible, despite their importance when a bond has to be sold quickly. And the next major test is still ahead: will the facilities developed since 2022 prevent a new episode of forced selling without another extraordinary expansion of central-bank bond holdings? Existing arrangements are documented; their behavior in an unobserved future crisis is not.

Governments can continue to borrow while investors accept the price offered for their debt. Across seven installments, that price has turned out to depend on intermediaries, financing contracts and collateral requirements largely invisible behind the headline public-debt number. A central bank can prevent a market seizure. It cannot make the economic reasons for a different borrowing price disappear.

Read the series

1. The first buyer
2. Dealers’ balance sheets
3. Leveraged buyers
4. The collateral chain
5. Margin calls
6. The price of getting out

Method and limitations

Sources checked on October 12, 2026. US amounts are in dollars and UK amounts in sterling, with no currency conversion. The first figure reproduces the official columns for 16 trading days from March 13 to April 3, 2020; published cumulative purchases can differ from the sum of rounded daily values. The 29.7% ratio compares UK purchases with the initial £65bn capacity, which was later increased. The £107.7bn difference compares two cumulative APF/QE cash balances, separately from the LDI intervention’s return. Facility diagrams describe published rules, without assuming use or predicting their effect in a future crisis. Central-bank statements and assessments are attributed to their authors.