7.1Bagehot on both sides of the debate
This chapter places the study's reading in the debate over whether a common rule steadies a market or shakes it, and asks what the two records add to it.
On this reading Bagehot stands on both sides of the debate. With the tradition that Kindleberger's history of crises and of the lender of last resort describes (Kindleberger 1978), he asks for a rule that the public can know in advance. The public has a right to know, he says in Chapter VII, whether the holder of the final reserve acknowledges its duty and is ready to perform it. With Daníelsson and MacKenzie, he shares a concern at one remove. Chapter VII describes a board that acts on the accepted rules of its time and place, and a reserve ratio, stated by the Bank's own spokesmen, that had no agreed meaning. It also says the public could not tell how the Bank would act in a panic. The study does not say that the ratio caused that uncertainty. What divides the two kinds of rule is direction and position: a rule for one holder, pointing against the crowd, against a measure many apply to themselves.
This refines Daníelsson's point (Daníelsson 2002). The danger of a shared model is not only that it is wrong in a crisis: applied by many, it makes them act as one. This study's reading of Bagehot adds the step that a system acting as one can be steadied only if some holder is placed to act otherwise.
7.2A shared measure without a holder who stands apart
Chapter VII describes a panic as the moment when the many already act alike, and the Bank of England becomes the only lender. The rule exists because one holder stands apart from that common move.
The first record describes a common move with no such holder. In it, a supervisory authority recommends one risk measure. Volatility rises and every bank's limit is breached together, and WSTiger's example judgment estimates the scale of the sales that would follow. The remedies the record offers are forms of variety: differentiated responses, methods that differ. That is a coherent answer, and a different one from Bagehot's. Variety spreads the holders out before the strain arrives; Bagehot's rule accepts that they will move together and places one holder against them. This study does not rank the two.
This study reads the record in the light of MacKenzie's claim that a financial model can change the market it describes (MacKenzie 2006). Its measure does not only observe the volatility. Through the limits, it produces the sales whose scale the judgment then estimates.
7.3Who bears the simultaneous sale
In Bagehot's account, those who bear a panic are those who cannot get money in time. Chapter VII traces a chain: one dealer is refused the securities he needs, so he cannot pay a second, who cannot pay a third in turn, numbered by Bagehot as No. 1, No. 2 and onward. The cost falls on whoever stands further down that chain.
The first record's Example judgment, WSTiger's dated reading rather than a finding, estimates the scale of the market sales. It names no one who bears them.
The second record names the account that needs the margin now, and describes delays in withdrawal, transfer or deposit. It does not say who bears the delay or who should carry the cost instead, and this chapter does not either.
7.4What the ledger shows, and what it cannot
In plain words: neither record says who decides, and neither settles who pays.
In ledger terms, no record in the set names its decider in its own words (0 of 53). For the first record the two coders, two models of different families, split on the decider: one read the supervisory authority, the other the banks. This study reads that split as consistent with the record's silence, and no more. On the bearer, both records are unsettled. For the first, one coder read the market at large and the other could not tell. For the second, one read the holder of the funds and the other the operator who needs the margin now.
The ledger's entries for the options (in the appendix) are readings of single cells, not counts. This study draws one point from them. The coders read the two records' loss-cutting options in opposite ways on resilience: for the first record they read the option to differentiate responses as gaining it, and for the second they read the total-asset basis as giving it up. Across all 53 records the pre-registered hypothesis H8, that the options cutting the immediate loss most often give up evolvability and resilience, is read as supported (24 for, 13 against, 16 not applicable; 65%).
Every record carries a revisit line (53 of 53, 100%), and the ledger finds 5 of 53 (9%) that code reads as stating a number, threshold or time. For the first record that reading is a quirk of the test. The code matched a Korean word in the line that also appears in the term for a limit breach, and the line has no figure, threshold or time in it. This study therefore does not treat the record as stating a numeric threshold: it is a match on a word, a property of the test and the record form, and not a finding about decisions.
7.5An alternative reading
The records could instead illustrate Bookstaber's account of complex, tightly coupled markets (Bookstaber 2007). They fit it in part. Yet Bookstaber's account, as this study uses it, places the cause in innovation and complexity. The records place it elsewhere: in a common measure in the first, in time and route in the second. Bagehot's account, where the cause is the system's one store and the remedy is one holder, fits the absence the records leave more closely.
The alternative holds better in one place. The first record's Insight, that "decisions that are rational for each party move in the same direction at the same moment", sits nearer to Bookstaber and Daníelsson than to Bagehot. It describes the common move, not the holder who might stand against it.