4.1Bagehot's rule and what a record would need to show it
Bagehot's starting point in Lombard Street: A Description of the Money Market, published in 1873, is a fact about the structure of the London money market. He argues that no bank in London, and none outside it, holds any considerable sum in hard cash above what its daily business needs, except the Banking Department of the Bank of England. Every other banker keeps its reserve there. The result, in his account, is that the reserve of the Bank of England is the banking reserve of all London, and beyond London of all England, Ireland and Scotland. The duty and the risk both follow from that structure: if the holders of the ultimate reserve refuse to lend while an alarm is running, the alarm becomes a panic.
To lessen the disadvantages of credit as much as possible, Bagehot writes, "we must keep a great store of ready money always available, and advance out of it very freely in periods of panic, and in times of incipient alarm." For the rate he writes: "Very large loans at very high rates are the best remedy for the worst malady of the money market when a foreign drain is added to a domestic drain."
In Chapter VII of the same book he states two rules for the advance. The first is that the loans should be made only at a very high rate of interest, which operates, as he puts it, as a heavy fine on unreasonable timidity: it keeps out those who borrow out of caution rather than need, and protects the reserve. The second is that at that rate the advances should be made on all good banking securities, and as largely as the public ask for them. His reason is specific. The object of the advance is to stay alarm, and nothing should be done to cause alarm; the way to cause alarm is to refuse someone who has good security to offer, and in half an hour at a moment of terror that refusal is carried everywhere. Bad bills and bad securities may be refused without making the panic worse, because in his reading the unsound minority is afraid even to look frightened for fear its unsoundness be detected; the majority to be protected are the sound people with good security to offer.
The announcement is a claim about knowledge as much as about lending. In Chapter VII Bagehot argues that the holders of the ultimate reserve should lend to all that bring good securities quickly, freely and readily, and that the public has a right to know whether the Bank of England acknowledges that duty and is ready to perform it; on that point, he writes, it is very uncertain. He is equally plain in Chapter II that advances made hesitatingly incur the evil of the advance without obtaining its advantage, and that what is needed to stop a panic is the impression that though money may be dear, money is to be had. In later usage the whole doctrine — a central bank lending when no one else will — is called the lender of last resort.
Skeptron's records are its readings of public sources, made in Korean and read here in an English translation prepared once for this study; Skeptron's answers are its own and are dated. The readings of Bagehot and of the other works in this chapter are this study's interpretation of them. Five of the ledger's eighty-four cases are read here. Of the five records, one has its own source text, Alan Greenspan's testimony of 1 October 1998 before the House Committee on Banking and Financial Services on the private-sector refinancing of Long-Term Capital Management, which survives in full. For the other four, this study holds public texts of the same episodes rather than the record's own source. For the record of 1907 it holds O. M. W. Sprague's History of Crises under the National Banking System of 1910, a volume prepared for the National Monetary Commission, in Chapter V, "The crisis of 1907". For the Lehman record it holds the Federal Reserve Board's press release of 14 September 2008 on its liquidity facilities. For the record of the bank failure of 10 March 2023 it holds the FDIC's press release of 10 March 2023 on the Deposit Insurance National Bank of Santa Clara and the Joint Statement of the Department of the Treasury, the Federal Reserve and the FDIC of 12 March 2023. For the Swiss record it holds two Federal Council enactments of March 2023. The Swiss National Bank's own statement of 19 March 2023 does not accompany this study. The other four cases are read from their records, from the reading notes and from the public texts just named.
What would a record have to hold for a later reader to tell whether Bagehot's rule was kept? At least five things. First, who asked for the advance: which bank, which trust company, which subsidiary, and whether the asker stood inside the circle of institutions the lender treated as its own. Second, on what security the advance was made, and whether that security was good in the ordinary sense — pledged and convertible — or something else. Third, who was refused, and whether the person refused had good security to offer, because Bagehot's argument turns on exactly that fact. Fourth, at what rate the advance was made, and how that rate stood against the rate in the market before the panic, since the high rate is what distinguishes a rescue from a free gift. Fifth, until when: what term the advance carried, what condition would end or withdraw it, and who bore the cost while it ran. To that list this study adds the elements its own axes ask for: whether the record presents the act as temporary, whether it names the condition or date of its end, who it names as bearing the cost meanwhile, and whether it sets the figure announced apart from what was actually drawn, paid or settled. None of those elements asks whether an act was wise. Each asks what a record owes a later reader who wants to judge it.
None of the five records read in this chapter holds all of them. Two hold almost none. One holds the shape of a decision and refuses the inference its shape invites. One holds two lines of support with different borrowers and different legal bases. One holds a boundary between legal entities and a set of named facilities, and refuses to let that boundary be read as a blanket answer.
4.2The Knickerbocker Trust in 1907
A trust company in New York at the turn of the twentieth century was an institution that took deposits and did banking business without being a national bank. A clearing house was an association through which banks settled what they owed one another each day; membership carried both privileges and obligations. The record's date is 22 October 1907, and the record's own scenario reads:
1907-10-22 · Knickerbocker's refusal of support and the stoppage of business on October 22. The clearing house, which had held back resources in support of member institutions, refused the request of the non-member Knickerbocker. After being unable to confirm the payment capacity of the books within a short period, financial support also fell through, and business was suspended on the 22nd.
The record's second answer gives the same connection in slightly different translated words: the clearing house is described there as having set aside resources to support member institutions, and as being unable to determine the payment ability of the book in a short time, with the funding support then failing and business stopped on the 22nd. This study keeps both renderings as they stand and does not merge them into one wording. The record's third answer refuses a specific inference:
No. Separate the verification status of liquidity and payment capacity, and the eligibility for support. Do not read the annualized rate of a short-term funding rate as a daily realized yield.
Liquidity — the ability to pay on the day — is kept apart from payment capacity, which is nearer to solvency, and both are kept apart from eligibility for support, which is a question of membership and rule rather than of assets. The record's answer to its third question is that the fact of insufficient time for review may not be turned into a confirmed finding of asset impairment.
Sprague's account is read here in this study's own words, in Chapter V of the 1910 volume. The Knickerbocker Trust Company was the third largest trust company in New York, with deposits of $62,000,000. The connection of its president with some of the Morse enterprises engendered distrust, which showed itself in a succession of unfavourable clearing balances. On Monday 21 October the National Bank of Commerce announced that it would discontinue clearing for the Knickerbocker on the following day. An unofficial committee representing a few trust companies and banks was not given an opportunity to examine the company's affairs until the last moment, so that it would have been difficult if not impossible to take definite action. Nothing was done beyond a reassuring statement by the directors announcing the resignation of the company's president. On Tuesday, after a run of three hours during which $8,000,000 was paid out, the company was forced to suspend.
Sprague writes that whether the company could have been assisted is not clear, but that assistance would have been of advantage both to the banks and to the other trust companies. He notes that the size of the company alone made assistance an undertaking of no little difficulty, and that the condition of its assets cannot have been hopelessly unsatisfactory, because the company was able to resume business in the following March under a plan of reorganization agreed upon by its depositors and shareholders — a plan that, he adds, showed the assets were then far from liquid. He states that had the Knickerbocker been a bank and a member of the clearing house, it is highly probable that it would have been assisted, following the precedent of 1884, and that apparently no proposal to assist the company was considered. His explanation is institutional rather than personal. Relations between banks and trust companies had been strained for years, the banks complaining of unfair competition with institutions not required to hold large cash reserves; in 1903 the clearing house had adopted a rule requiring trust companies that cleared through its members to accumulate a reserve, and rather than submit to that requirement nearly all of the trust companies gave up clearing-house privileges, the most important exception being the Knickerbocker itself, which on Sprague's account might therefore have deserved peculiar consideration. On the question of blame he judges that the failure of the clearing-house authorities to act was doubtless the most natural course, though unfortunate in its consequences, and can hardly be regarded as blameworthy.
Sprague returns to the question in general terms at the end of the episode. Unless it is possible at the outset to take measures almost certain to restore confidence in threatened banking institutions, he concludes, it is far better to leave them to their own devices, by liquidation or reorganization; the money later furnished to two other trust companies saved those companies from failure, in his view, but served no purpose of general importance. He also records the clearing house president's own explanation of the delay: the committee postponed the issue of clearing-house loan certificates from day to day, honouring drafts against a rapidly falling reserve until it showed a deficit of $53,000,000, and then concluded that it would be folly to hesitate longer. Sprague's own judgment on that decision is that the failure to issue certificates at least as early as Tuesday was the most serious error made during the crisis, because earlier issue had in past panics had a calming effect.
Two later readings bear on the same episode. Moen and Tallman's 1992 article studies the panic of 1907 with attention to trust companies, which held lower reserves than national banks and had no access to the New York Clearing House; the authors argue that the panic began with runs on trust companies, concentrated among them, and that this institutional gap in access to clearing-house support helps explain why the panic was severe. Bruner and Carr's 2007 narrative describes the failed attempt to corner the stock of United Copper, the run on the Knickerbocker Trust Company and its spread to other trust companies and to the stock market; it describes how J. P. Morgan and other bankers organised support, pooling resources to lend to institutions and to the stock exchange, and how the Treasury added funds. Each note answers a different question: Moen and Tallman ask why the panic fell hardest where it did; Bruner and Carr ask what the private support that followed consisted of.
Within hours on 22 October 1907 a refusal and a stoppage followed one another, and the record gives no terms for either. Both readings find an act, but they differ on the kind of act; they agree that the act is not presented as temporary, that no condition for its end is named and that nothing is said of what became of it, and they differ on who bore the cost. This study reads the refusal as a decision with a deadline measured in hours: the clearing arrangements for the company were announced on Monday to end on Tuesday, and by Tuesday afternoon the books had not been examined and the company had suspended. Bagehot's rule, as he states it, is about a central bank lending at a high rate on good securities to all who bring them; the 1907 record shows the same question asked of a private association that had set aside resources for its members. What the rule would require of such a refusal — a rate, a security offered, a reason, a term — the record does not hold, and the record's own third answer declines to supply what its silence might invite.
4.3LTCM in 1998
Long-Term Capital Management, commonly written LTCM, was a hedge fund whose partners included leading academics. The record's date is 23 September 1998, the date of the agreement it records; the source it checked is a page of congressional testimony by Alan Greenspan, dated 1 October 1998, which the record keeps separate from the event date. The record's second answer reads:
The October 1 testimony explains that the private parties agreed to commit approximately $3.5 billion of capital, that the existing equity interests were substantially diluted, and that control passed to a committee of outside investors.
Its third answer draws the line on which the whole case rests:
The testimony distinguishes that the Federal Reserve's funds were never provided or proposed. The coordination role, the actual source of funds, the moral hazard debate, and liquidation risk must each be reviewed separately.
The record's scenario adds a note on how the figure is held:
The approximately $3.5 billion rounding of the cited testimony is retained. The $3.6 billion from another source and values with a different precision are not merged as if they were the same source data without error.
LTCM is the one case among the five this chapter reads closely in which the actor's own text survives in full, and that text describes an act of another kind from a rescue: no public money was lent. The chairman of the Federal Reserve described the Federal Reserve Bank of New York's role as facilitation, stated that no Federal Reserve funds were put at risk, and granted the question facilitation raises, moral hazard. This adds to the debate over the lender of last resort a case in which the doctrine's terms — a rate, a security, a borrower — do not arise at all, because the emergency answer took the form of a meeting rather than a loan, and in which the actor defended that form before Congress eight days later.
The record distinguishes three things: the coordination role, in which the Federal Reserve Bank of New York brought the fund's creditors together; the actual source of funds, which the record says was private; and the two matters it lists separately, the moral hazard debate and liquidation risk. The distinction between coordination and provision is the one the record insists on most strongly, and it is a distinction about who paid, not about who met in a room.
The testimony opens with the purpose of the Federal Reserve's involvement: "The Federal Reserve Bank of New York's efforts were designed solely to enhance the probability of an orderly private-sector adjustment, not to dictate the path that adjustment would take," and adds in the same paragraph that "no Federal Reserve funds were put at risk, no promises were made by the Federal Reserve, and no individual firms were pressured to participate." It describes the fund — its leverage through securities repurchase contracts and derivatives, its mathematical models of behaviour, its return of capital to investors at the end of 1997, and the losses that followed the Russian default — and explains the judgment behind the intervention: officers of the Federal Reserve Bank of New York believed that "an orderly resolution" was to the advantage of creditors and of other market participants rather than a forced liquidation in which cross-default clauses set off cascading defaults. It gives the agreement: "On September 23, the private sector parties arrived at an agreement providing a capital infusion of about $3-1/2 billion in return for substantially diluting existing shareholders' stake in LTCM," with control passing to a committee chosen from outside by the new investors. Its next sentence is the one on which the record's third answer turns: "This agreement was not a government bailout, in that Federal Reserve funds were neither provided nor ever even suggested." Only then does the testimony take up the question the record lists separately: "Of course, any time that there is public involvement that softens the blow of private-sector losses--even as obliquely as in this episode--the issue of moral hazard arises."
The record and the testimony agree in sense and differ in wording. The record says the funds "were never provided or proposed"; the testimony says they "were neither provided nor ever even suggested." The record says "approximately $3.5 billion"; the testimony says "about $3-1/2 billion." The record says control passed "to a committee of outside investors"; the testimony says "a committee determined from the outside by the new investors." The testimony argues about moral hazard rather than merely granting it, holding that the optimum degree of risk aversion should be governed by rational judgments about the market place, not by the fear flowing from fire sales; and its conclusion limits the episode as a whole: "The efforts last week were limited to facilitating a private-sector agreement and had no implications for Federal Reserve resources or policies."
Two later readings bear on the episode. Roger Lowenstein's 2000 account describes the rise and collapse of the fund, the heavy losses it took after the Russian default of 1998, the way its highly leveraged trading spread losses through the markets, and the fact that in September 1998 the Federal Reserve Bank of New York brought together the fund's creditors to arrange a private rescue and recapitalisation. Lowenstein's is a journalistic account; its note says it is often used as a case both of leverage and of the role of authorities in organising a private-sector solution, and gives no rate or collateral, describing the episode without the terms of the recapitalisation. Donald MacKenzie's 2006 study argues that financial economics does not merely describe markets but helps to shape them, and treats the collapse of LTCM as one of the cases where models and their wide use affected outcomes; again the note gives a reading of why the episode matters, not the terms of the agreement.
The record names the moral hazard debate as one of four things that must each be reviewed separately, and gives no more: it does not state what the debate's terms were, who held which position, or what evidence would settle it. This study reads the naming as a boundary rather than as an answer — a statement that the question exists and that the answer is not in this record. The actor's own text does take a position on it; the record does not report that position, and this section has given it only as what the testimony says.
The act itself, the kind of act, whether it was presented as temporary, the condition of its end, what became of it and who bore the cost are each left undecided between two independent readings of the record, which disagree on each of them. Two things are settled by both: two dated moments are named — the agreement of 23 September and the testimony of 1 October — and a single figure is given. Across the whole ledger one case in forty-two supplies all of what a later reader needs to judge an emergency act by its own end. This study reads that shape as the record's own: dates and a sum, held apart from the questions of terms and bearers that the two readings could not settle. The record also warns that the coordination role, the actual source of funds, the moral hazard debate and liquidation risk must each be reviewed separately, and that values with different precision are not to be merged as though they were one source's data.
4.4Lehman in 2008
A date in the history of rescues can carry two different acts, and the archive can keep them apart. This section adds to the debate over the lender of last resort a case where the actor's own text of one day survives beside a record of the next: the Federal Reserve Board's release of Sunday 14 September 2008, which widened the collateral its facilities accepted and added one new exception carrying a stated date, and Skeptron's record of Monday 15 September 2008, which holds a bankruptcy on one side of a legal boundary and the use of facilities already in place on the other. The comparison is between what an authority wrote down about terms on the eve of the event and what the record of the event itself holds of them.
Skeptron's record of 15 September 2008 keeps two entities apart, and this chapter and this study keep them apart throughout. The holding company of Lehman Brothers filed for bankruptcy on 15 September 2008. Its dealer subsidiary, a separate legal entity, used the Federal Reserve's existing credit facilities. Lehman as a whole neither received nor was refused a loan; the two things belong to two legal entities and two kinds of act. The record's second answer states the distinction:
The headquarters filed for bankruptcy on September 15, but the dealer subsidiary used existing credit facilities such as the TSLF and PDCF. The legal entities and the means of support must be seen separately.
Its third answer refuses a generalisation and gives its reason:
That generalization cannot be made. Since the borrowing legal entity, facility, balance, and date differ, rescue decisions and market liquidity provision must be distinguished.
A dealer subsidiary in this sense is the arm of a securities firm that buys and sells securities for customers and for its own account; it settles what it owes through the market's own machinery. "Existing credit facilities" means facilities already in place and open to institutions that qualified, not a new rescue created for this case. The record names two of them by abbreviation, the TSLF and the PDCF, expands neither, and gives no rate, no collateral, no balance and no date for either.
What the record does not hold is as plain as what it holds. Both readings mark the act as not presented as temporary, name no condition for its end, say nothing of what became of it and name no second dated moment: the single day of 15 September carries the record, with no decision date and no settlement date set apart from it. Of the forty-two records in this study that show an act by an authority, one lets a later reader judge the act by its own end, and in most of the cases with an act the record names no observable condition for its end. In most of those cases the record also names no one as bearing the loss while the act was in force.
The actor's own text for the day before survives, and it must be read first. The Federal Reserve Board's release of Sunday 14 September 2008 announces, in its own order, what was changed on its facilities and on one statutory limit. It states that the Board had been working closely with the Treasury and the Securities and Exchange Commission, and holding ongoing discussions with market participants, including through the weekend, to identify potential market vulnerabilities in the wake of an unwinding of a major financial institution and to consider appropriate official sector and private sector responses (Federal Reserve Board, press release of 14 September 2008). The release names no institution.
It then describes three changes to collateral. The collateral eligible to be pledged at the Primary Dealer Credit Facility, the PDCF, "has been broadened to closely match the types of collateral that can be pledged in the tri-party repo systems of the two major clearing banks"; previously, the release states, "PDCF collateral had been limited to investment-grade debt securities". The collateral for the Term Securities Lending Facility, the TSLF, "also has been expanded; eligible collateral for Schedule 2 auctions will now include all investment-grade debt securities"; previously only Treasury securities, agency securities, and AAA-rated mortgage-backed and asset-backed securities could be pledged. Auctions under Schedule 2 were to be held weekly rather than every two weeks, and the amounts offered were raised so that the total offered under the TSLF rose to $200 billion from $175 billion; Schedule 1 auctions remained at a total of $50 billion.
The fourth change is of a different kind, and the release says so. The Board "adopted an interim final rule that provides a temporary exception to the limitations in section 23A of the Federal Reserve Act", allowing insured depository institutions to provide liquidity to their affiliates for assets typically funded in the tri-party repo market. On that exception the release gives its own term: "This exception expires on January 30, 2009, unless extended by the Board, and is subject to various conditions to promote safety and soundness."
Read in its order, the release separates three things: a widened collateral base on two facilities already in place, whose widening carries no term; a ceiling raised by $25 billion, stated as what was offered rather than what was drawn; and one new exception with a date attached. It gives no rate for any advance and records no refusal of any institution. The facilities the record of the next day names — the TSLF and the PDCF — are the two the release describes, which makes the release the nearest surviving text of what "existing credit facilities" meant on 15 September. What the release holds that the record does not is a condition of ending, written by the actor, on one of its four measures.
The Financial Crisis Inquiry Commission's report of 2011, an official account written after the event for the United States Congress, is the third document here. Its reading note describes a commission that concludes the crisis was avoidable and was caused by failures of financial regulation and supervision, by failures of corporate governance and risk management at financial institutions, by excessive borrowing and risky investments, and by the lack of transparency in the market for over-the-counter derivatives and in shadow banking; it describes the failure of Lehman Brothers and the rescues of others, and it was divided, with dissenting statements by some commissioners. This study reads that report as an account of causes rather than of a single act's terms. A causal report must place the failure of one institution beside the support given to others in one list; the record of 15 September holds a boundary between two legal entities on one day. The two documents are therefore different kinds of evidence about the same period, and neither supplies what the other holds: the report names no facility, no rate and no condition of ending for the Federal Reserve's decisions of that Monday, and the record names no cause, no reason and no bearer.
This record shows the boundary of the circle rather than the terms of the advance: who stood inside it — a dealer subsidiary qualified to borrow — and who stood outside it — a holding company in bankruptcy. On Bagehot's rule, which asks at what rate, on what security, and with what announcement, the record holds nothing, and the release of 14 September supplies terms for collateral and one exception's end date while supplying no rate and no refusal. That is the strongest thing the pair can be tested on, and it is not enough to test the rule itself.
4.5Silicon Valley Bank and Credit Suisse in March 2023
Two records fall in the same March, ten days apart, and describe different kinds of act. Skeptron's records name dates, entities and boundaries; the FDIC's release, the joint statement of three United States authorities and the two Swiss enactments name terms, deadlines, sums and, in the joint statement, who was protected and who was not. This study reads the contrast as the chapter's strongest evidence on what a record owes a later reader. A receiver is a body appointed to take charge of an institution's affairs while its position is settled; in the first record it is the Federal Deposit Insurance Corporation, the FDIC. The record's date is 10 March 2023, and its second answer reads:
Records the fact that on March 10, 2023 the closing authority closed the bank and designated the FDIC as receiver. It is not the same day's event as the subsequent asset acquisition contract.
Its third answer holds a specific boundary:
Deposit type, insurance coverage, follow-up actions, and asset recovery are separate. This document is used to verify the date of the closure and the change of the managing entity.
The asset acquisition contract that follows the closure is a later event on a later date, and the record says so. It refuses the inference that the loss of deposits, or the scope of compensation, may be drawn from the single fact of closure; the document verifies two things only — the date of the closure and the change of the managing entity.
Synthetic-scene notice. This scene is this study's construction from the record and adds no fact beyond it. On Friday 10 March 2023 a state chartering authority in California closes a bank in Santa Clara, and a federal deposit insurer is named receiver and transfers the insured balances to a bank it creates for the purpose. Two days later three national authorities — a treasury secretary, a central bank chair and a deposit-insurance chairman — issue one statement together, naming who will be made whole and who will not. Six days later still, at eight o'clock in the evening in Bern, the Federal Council brings an ordinance into force on emergency liquidity lending and federal default guarantees, and brings an amendment into force on 19 March, also at eight o'clock in the evening. What each of these actors waited for was the next business day; what each wrote down was what that day would look like for depositors, for shareholders, and for the acquiring bank.
The FDIC's release of 10 March states that "Silicon Valley Bank, Santa Clara, California, was closed today by the California Department of Financial Protection and Innovation, which appointed the Federal Deposit Insurance Corporation (FDIC) as receiver," and that the FDIC "created the Deposit Insurance National Bank of Santa Clara (DINB)" into which it "immediately transferred to the DINB all insured deposits of Silicon Valley Bank." For insured depositors it gives a deadline: "All insured depositors will have full access to their insured deposits no later than Monday morning, March 13, 2023." For the rest it gives a different treatment: "The FDIC will pay uninsured depositors an advance dividend within the next week. Uninsured depositors will receive a receivership certificate for the remaining amount of their uninsured funds. As the FDIC sells the assets of Silicon Valley Bank, future dividend payments may be made to uninsured depositors." On the sums it gives two dated figures and one admission: the bank had "approximately $209.0 billion in total assets and about $175.4 billion in total deposits" as of 31 December 2022, and "At the time of closing, the amount of deposits in excess of the insurance limits was undetermined."
The joint statement of 12 March 2023, issued by Secretary of the Treasury Janet L. Yellen, Federal Reserve Board Chair Jerome H. Powell and FDIC Chairman Martin J. Gruenberg, says who bears what. It records that "Secretary Yellen approved actions enabling the FDIC to complete its resolution of Silicon Valley Bank, Santa Clara, California, in a manner that fully protects all depositors," that "Depositors will have access to all of their money starting Monday, March 13," and that "No losses associated with the resolution of Silicon Valley Bank will be borne by the taxpayer." It then names those left out: "Shareholders and certain unsecured debtholders will not be protected. Senior management has also been removed." And it names where the loss goes: "Any losses to the Deposit Insurance Fund to support uninsured depositors will be recovered by a special assessment on banks, as required by law." It adds that the Federal Reserve Board "will make available additional funding to eligible depository institutions."
Andrew Metrick's 2024 article describes how a bank with a concentrated base of uninsured depositors and large unrealised losses on securities faced a very fast run, and how the authorities responded with a systemic risk exception to protect all depositors and with a new lending facility for banks; the note says the article discusses the regulatory and supervisory failures and the questions the episode raises for deposit insurance and run risk. This study reads Metrick's account, which covers causes and responses across the episode, as different in kind from a record of a single day's act; neither supplies the other's terms. The joint statement of 12 March is one of the few texts in this study that names a bearer in its own words.
The second record is dated 19 March 2023 and concerns the Swiss National Bank's statement on support in the acquisition of Credit Suisse. Its second answer reads:
The SNB stated that UBS and Credit Suisse could receive support in the form of a total limit of CHF 100 billion, holding priority claims in the event of bankruptcy. Separately, it stated that a loan of up to CHF 100 billion with a federal guarantee could be provided to Credit Suisse.
Its third answer opens with a single word and then an instruction:
None. Distinguish the borrowing entity, legal basis, collateral and guarantee, limit, drawdown, and acquisition price. At the time of the announcement, the support is not used as a guarantee of future profitability.
The record does not expand the word "None.", and does not say what it qualifies. The two lines differ in every respect the record names: the first is a total limit of CHF 100 billion available to UBS and Credit Suisse together, with priority claims in the event of bankruptcy; the second is a loan of up to CHF 100 billion with a federal guarantee, to Credit Suisse alone. The record gives no rate, no term, no condition of withdrawal and no drawdown for either; it states that the limits cannot be added together as an executed amount on the same day, or treated as the acquisition price; and it states that at the time of the announcement the support is not to be used as a guarantee of future profitability. Both readings find the figure announced and the figure available kept apart, while the act is not presented as temporary, no condition for its end is named, and who bore the cost is left undecided between the two readings.
The Swiss enactments beside the record supply what the record does not. The Federal Council's emergency ordinance of 16 March 2023 is in force at 20.00 on 16 March, and its Article 15(2) states, in German, "Sie gilt für die Dauer von sechs Monaten ab Inkrafttreten"; this study's English for it is that the ordinance applies for six months from its entry into force. Its Article 4 lists conditions for a federal guarantee: that a substantial injury to the Swiss economy and financial system cannot otherwise be averted, that the loans are suitable and necessary for the borrower's continued business, and that the financial supervisor confirm the borrower's solvency or the existence of a restructuring plan. It provides that no legal claim to a federal guarantee arises.
The amendment of 19 March 2023, in force the same day at 20.00, adds a third instrument. Its Article 14a permits the Confederation to give the acquiring bank, in a merger between systemically important banks supervised by the financial regulator, a guarantee against losses on assets of the acquired bank to be wound down. Its Article 14a(2) states, in German, "Die Garantie zur Verlustabsicherung beträgt höchstens 9 Milliarden Schweizer Franken."; this study's English for it is that the loss-protection guarantee amounts to at most 9 billion Swiss francs. The same Article sets out the condition of payment: that "alle abzuwickelnden Aktiven wurden definitiv verwertet" — all the assets to be wound down have been definitively realised — and that "die übernehmende Bank nach Absatz 1 hat auf den abzuwickelnden Aktiven definitive Verluste von 5 Milliarden Schweizer Franken getragen" — the acquiring bank has itself borne definitive losses of 5 billion Swiss francs on them — with the financial regulator confirming both. Its Article 10a lets such a merger proceed without the usual shareholder resolutions and makes the regulator's decisions binding on the commercial registers. These German texts are official enactments; the English beside them is this study's own.
The Swiss record and the Swiss enactments name different sums and must not be merged: the record holds two lines of up to CHF 100 billion each, while the ordinance names a maximum amount the National Bank may pay and the amendment names a guarantee capped at 9 billion francs behind a first loss of 5 billion borne by the acquiring bank. One record in this study meets all of what a later reader needs to judge an emergency act by its own end, and neither March record meets them here.
The two March records answer a panic question in two different ways: one closes an institution and appoints a receiver; the other lends to an institution behind a guarantee and a claim rank. Neither names a condition of its end. What the actors' own texts supply the two records lack differs in kind. In the American case the texts name the deadline of Monday 13 March for insured depositors, an advance dividend and a receivership certificate for the uninsured, shareholders and certain unsecured debtholders left unprotected, and a special assessment on banks to bear any losses to the deposit-insurance fund. In the Swiss case the enactments carry the missing terms, not the central bank's statement: a term of six months, a cap of 9 billion francs, and a first loss of 5 billion borne by the acquiring bank before the guarantee pays. This study reads the pair as showing what a record can separate — dates, entities, limits — and what it leaves unnamed, which is who paid while each act was in force.
4.6The debate over Bagehot's rule
The dispute this section serves is about terms. The classical doctrine of the lender of last resort states what an institution holding the system's reserve owes, and specifies the terms: a penalty rate, good collateral, and an announcement made in advance. The competing readings ask what function the institution performs and who authorised it, and relocate the duty's justification rather than deny it. None of the five records of this chapter holds the terms the doctrine specifies, so they cannot settle it; what they show is how often an emergency act's record is built around a boundary — who stood inside the circle, who outside — rather than around a rate.
The classical reading of the doctrine yields three testable claims about any rescue: whether the borrower was admitted or refused by reference to solvency and security, whether the rate stood above the market rate as a penalty, and whether the terms were announced before the borrowing began. Bagehot's own words, set out in Chapter 4.1, are the source of all three; the classical reading's distinctive addition is Bordo's argument that the announcement, not the loan alone, does the work.
The other side is not one position. On the function-first reading the lender's terms are consequences of what the institution is for, not terms of a contract between lender and borrower. Perry Mehrling's 2011 book changes the object again, arguing that modern finance rests on market-based credit and funding rather than on bank deposits alone, so that the central bank's role has become that of dealer of last resort for markets as well as lender of last resort to banks; on this reading the crisis of 2007 to 2009 was a breakdown of market liquidity that the Federal Reserve met by acting as a dealer.
The five records bear on these positions in shape rather than in terms. A private clearing house holding back resources for its members and refusing a non-member on 22 October 1907. A central bank officer organising a private recapitalisation in September 1998, whose testimony to Congress states that no Federal Reserve funds were provided or ever even suggested, and grants that public involvement softening private losses raises the question of moral hazard. A holding company's bankruptcy on 15 September 2008 beside a dealer subsidiary's use of facilities already in place. A bank closed on 10 March 2023 with the Federal Deposit Insurance Corporation named receiver. Two liquidity lines of up to CHF 100 billion each announced on 19 March 2023, one carrying priority claims in bankruptcy and one a federal guarantee. Each record answers a version of Tucker's question — who authorised this, and on what written terms — and in three of the five cases the actor's own text answers it directly, with a stated term where the record has none.
Where the records cannot bear is narrower than it looks and more damaging to the classical test. None names a rate. None names a security offered or accepted, or records a collateral assessment. The 2008 release of Sunday 14 September comes closest, supplying a widened collateral base on two facilities and one exception dated 30 January 2009; it gives no rate and records no refusal. The Swiss ordinance of 16 March and its amendment of 19 March name a term of six months, a cap of 9 billion francs and a first loss of 5 billion borne by the acquiring bank before the guarantee pays. Bagehot's rule is a rule about terms, and terms are what these records do not hold. This study reads the five records as evidence that the doctrine's terms can be written down — in a release, in an ordinance — and that they usually are not, a fact about the records rather than the doctrine.
4.7The debate over Lehman and over runs
This section speaks to two disagreements about the Federal Reserve's Monday of 15 September 2008: whether Lehman Brothers could have been rescued, and beneath it a question about what a run is and what stops one. The record holds what an actor wrote down and what a later archive made of it; it holds nothing of the law, the collateral or the judgment on which the accounts disagree.
Ben S. Bernanke's 2015 memoir, by the chairman at the time, describes the decisions of those years: the support of Bear Stearns, the choice not to rescue Lehman Brothers, the rescue of American International Group, the emergency lending facilities created and the asset purchases that followed. As a participant's account it reflects its author's own view.
Laurence M. Ball's 2018 book disputes that account. The disagreement is about law, collateral and judgment under time pressure on 15 September 2008 — about what could be known and decided in a weekend, not about what followed.
Charles W. Calomiris and Stephen H. Haber's 2014 book argues that banking stability is shaped by politics as well as economics, the rules reflecting bargains among rulers, bankers, borrowers and voters; they compare the United States, with frequent crises and a fragmented banking structure, with Canada, which had fewer. None of them reads a single day's terms.
Skeptron's record settles neither question. It holds a date, two legal entities, two named facilities and an instruction not to generalise; it holds no collateral assessment, no rate, no reason and no bearer. Its own third answer refuses the blanket claim that no loans whatsoever were made, for a reason the record gives:
That generalization cannot be made. Since the borrowing legal entity, facility, balance, and date differ, rescue decisions and market liquidity provision must be distinguished.
That record is Skeptron's reading of public sources, made in 2026, not a document of 2008; it cannot show what the Federal Reserve knew, could have known or was bound by that Monday. The nearest actor's text to the Monday is the Board's release of Sunday 14 September 2008, which widened the collateral its two facilities accepted and added one exception that "expires on January 30, 2009, unless extended by the Board, and is subject to various conditions to promote safety and soundness"; Chapter 4.4 reads that release in full and shows that it names no institution and no rate. Of the forty-two records in this study that show an act by an authority, one lets a later reader judge the act by its own end; in most of the cases with an act the record names no observable condition for its end, and in most it names no one as bearing the loss while the act was in force.
The runs literature turns on why a loss of confidence spreads and what removes it; the Ball–Bernanke debate turns on what a lender was bound to do when it spreads. Both concern terms and conditions this record does not carry. Relief of debtors is a much older kind of emergency act, and the next chapter takes it up.