5.1The claim
Chapter II and Chapter XI of Lombard Street describe a system in which reserves were not held where they were counted. Country bankers kept their reserves in London; London bankers kept theirs at the Bank of England or with the bill-brokers; the bill-brokers kept almost none, because they paid interest on all they held. One store of cash stood behind many claims, and the claims were pressed together. Bagehot shows that in such a system a reserve that is one party's asset is another party's liability, and that the act of strengthening one reserve in a panic drains another. The record Risk Propagation of Cross-Margin describes the same shape inside a single trading account, where several positions share one pool of collateral. The case adds to the debate a demonstration that the coupling the second side describes is a matter of where reserves are kept and on whose books they appear, and that Bagehot traced it in the arrangements of his own market.
5.2All London, and all England
Bagehot's general view moves outward from the Bank of England in steps. The London banks keep their principal reserve on deposit at the Bank's Banking Department, which is convenient and has a cost:
But those who keep immense sums with a banker gain a convenience at the expense of a danger. They are liable to lose them if the bank fail. As all other bankers keep their banking reserve at the Bank of England, they are liable to fail if it fails. They are dependent on the management of the Bank of England in a day of difficulty and at a crisis for the spare money they keep to meet that difficulty and crisis.
The danger is not only that the depositing banks might lose their money. Bagehot says the main effect is something else:
But the danger to the depositing banks is not the sole or the principal consequence of this mode of keeping the London reserve. The main effect is to cause the reserve to be much smaller in proportion to the liabilities than it would otherwise be.
The reason is that the Bank lends a part of what is deposited with it. A reserve placed with another lender becomes part of that lender's lending. What is counted as reserve by the depositor is, on the books of the holder, a liability partly matched by loans. Bagehot carries the chain past London to the country banks, and past them to Scotland and Ireland, until the Banking Department's reserve is "the banking reserve not only of the Bank of England, but of all London--and not only of all London, but of all England, Ireland, and Scotland too." The figure that the depositors each count as their own is, at the bottom of the chain, a single store.
5.3The second-class dealers
Bagehot then describes how a panic spreads through such a chain. His account of the auxiliary dealers in credit shows collateral at work:
In ordinary times, the intercourse between the two is probably close enough. The little dealer is probably in the habit of pledging his 'securities' to the larger dealer at a rate less than he has himself charged, and of running into the market to lend again. His time and brains are his principal capital, and he wants to be always using them. But in times of incipient panic, the minor money dealer always becomes alarmed. His credit is never very established or very wide; he always fears that he may be the person on whom current suspicion will fasten, and often he is so.
The small dealer is a leveraged lender in the plainest sense. Securities are pledged to borrow, and the borrowing is lent again. In ordinary times the pledge is a formality. In a panic the dealer's first move is to ask the larger dealers for advances, and the larger dealers must ask whether refusal would be safer than lending, since, in Bagehot's words, "A panic grows by what it feeds on". His answer is the rule for which the book is known, stated first in its general form:
A panic, in a word, is a species of neuralgia, and according to the rules of science you must not starve it. The holders of the cash reserve must be ready not only to keep it for their own liabilities, but to advance it most freely for the liabilities of others. They must lend to merchants, to minor bankers, to 'this man and that man,' whenever the security is good. In wild periods of alarm, one failure makes many, and the best way to prevent the derivative failures is to arrest the primary failure which causes them.
The condition "whenever the security is good" is Bagehot's own, in this chapter. It concerns the quality of the security, not its price on the worst day; Case one has shown that, in a general panic, good securities cannot be sold without the Bank's lending. That is where Bagehot's rule and a margin call part.
5.4The bill-brokers and the collateral security
Chapter XI takes the bill-brokers as the clearest case of dependent dealers: specialists who lend more than their own capital by borrowing from bankers and pledging the bills they hold. Bagehot traces a change in how the bankers who lent to the brokers regarded the bills pledged with them. In the older practice the lender judged each bill. In the modern practice, the lender relied on the broker:
He does not exercise much judgment of his own on the bills deposited with him: he often does not watch them very closely. Probably not one-thousandth part of the creditors on security of Overend, Gurney and Co., had ever expected to have to rely on that security, or had ever given much real attention to it.
A few paragraphs on, Bagehot describes the lender who came to take the broker's guarantee "and only to look at the bills as a collateral security". Read together, these sentences describe collateral that was counted and not watched. The lenders held bills as security and expected never to rely on them. The value of the security mattered only on the day the broker failed, and on that day, as the account of 1866 elsewhere in the chapter makes plain, the standing of many firms changed at once:
On the other hand, in years like 1866 a contagious ruin destroys the trustworthiness of very many firms and persons, and often, especially, of many who stood highest immediately before.
A bill is a promise by a merchant. If the trustworthiness of "many who stood highest" falls at the same time, the bills pledged as security fall with the broker who pledged them. That is the co-deterioration of Case one, at one remove.
5.5A reserve that is someone else's liability
The centre of Chapter XI is Bagehot's account of why the brokers were peculiarly exposed under the one-reserve system. The brokers kept little cash, because they paid interest on what they held. Their depositors were, in large part, bankers:
But under a one-reserve system of banking, the position of the bill-brokers is much more singular and much more precarious. In fact, in Lombard Street, the principal depositors of the bill-brokers are the bankers, whether of London, or of provincial England, or of Scotland, or Ireland. Such deposits are, in fact, a portion of the reserve of these bankers; they make an essential part of the sums which they have provided and laid by against a panic. Accordingly, in every panic these sums are sure to be called in from the bill-brokers; they were wanted to be used by their owners in time of panic, and in time of panic they ask for them.
The same sum is a reserve on the banker's books and a call liability on the broker's. The banker has laid it by "against a panic", and so will ask for it in a panic, which is when the broker can least pay: in Bagehot's words, "A great additional strain is thrown upon them just at the moment when they are least able to bear it".
Under a sounder arrangement, Bagehot says, money at the brokers would be one of the banker's investments of cash, "it would not be a part of such cash". He then joins profit and danger: "The deposits of bill-brokers and the profits of bill-broking are increased by our present system, just in proportion as the dangers of bill-brokers during a panic are increased by it." The same arrangement that fills the brokers' books in ordinary times empties them in a panic. Bagehot treats this as a fact of structure, not a moral failing, and ends the chapter by saying that the anomaly can be lessened but not cured, since "In every market a dealer must conduct his business according to the custom of the market, or he will not be able to conduct it at all."
The chapter's summary of the double effect is the sentence closest to the second side of the debate anywhere in Lombard Street:
So that our one-reserve system of banking combines two evils: first, it makes the demand of the brokers upon the final reserve greater, because under it so many bankers remove so much money from the brokers; and under it also the final reserve is reduced to its minimum point, and the entire system of credit is made more delicate, and more sensitive.
The phrase "more delicate, and more sensitive" names, in Bagehot's terms, something close to what Perrow's vocabulary calls tight coupling (Perrow 1984). Perrow wrote about high-risk technologies, not money markets, and the comparison is this article's. What Bagehot adds to it is the detail of how the coupling arises: from reserves placed with another party who lends them.
5.6The record: one loss, common collateral
WSTiger's record Risk Propagation of Cross-Margin sets the same structure inside one account:
Assumption: multiple positions use common collateral, and losses on one instrument have reduced the account's margin headroom. / Required evidence: margin sharing scope · position-level P&L and maintenance margin · account-level stress results / Tags: prop trading systems · cross-margining · common collateral
Cross-margining, in plain terms, means that the margin requirement is set for the account as a whole and every position draws on the same pool of collateral. Its advantage is the advantage of keeping a reserve at the Bank of England: less idle collateral is needed, because gains on one position offset losses on another. Its cost is the one Bagehot named for London's banks, that each is "liable to fail if it fails". The record's question:
Does a loss on one position affect the remaining survival margin of other positions?
The options are Manage individual stop-loss risk only; Add account-level stress testing; and Continuously monitor collateral headroom. The example judgment reads:
Manage the stop-loss risk of individual positions together with the account risk of common collateral. Do not consider the impact of collateral deficiency blocked merely because other trades are profitable.
The second sentence answers a temptation the record plainly expects: to look at the profitable trades and conclude that the account is safe. In Bagehot's chapter, the brokers' profits rose with their dangers; a good return on the arrangement was not evidence that it was safe. The record's judgment refuses the same inference inside an account.
5.7What the record adds and where it stops
The record adds one thing Bagehot does not: a second line of defence that may never be reached. Individual stop-losses, the record implies, are meant to close a losing position before it does great harm. The Invalidation field, read in Case three, asks for the path on which the shared collateral runs short before the stop-losses act. Bagehot's dependent dealers had nothing like a stop-loss; their only defence was to ask a larger dealer for an advance. The record's setting is more elaborate, and its worry is the order in which defences fail.
Where it stops is the same edge as before. The account's common collateral is the whole of the reserve in view. Nothing in the record says what lies beyond it, or whether the lender that set the margin is itself a party to whom the account's collateral is, in Bagehot's phrase, a portion of a reserve laid by against a panic. On Bagehot's reading, the account's lender is in the position of the banker who has placed money with the bill-brokers, and the account is in the position of the broker. The record writes the account's side of that relation and not the lender's.
This case adds to the debate a reading of Bagehot as an analyst of coupling through reserves. The second side's claim that finance makes its own fragility finds a concrete form in his bill-brokers: a reserve counted by one party is a call on another, so that the effort to strengthen one weakens the other at the same moment. The record carries that form into a single account and leaves the lender outside the frame.