2.1Run or exposure
The first debate is about what a banking panic is. On one side stands the model of Diamond and Dybvig. A bank that offers deposits withdrawable on demand while holding assets that cannot be sold quickly at full value provides a real service, the insurance of liquidity, but it also admits two outcomes: one in which only depositors who need cash withdraw, and one in which every depositor runs because each expects the others to run. In their model a run can occur without any fundamental weakness in the bank, and deposit insurance or a lender of last resort can remove the bad outcome, though such arrangements raise problems of incentives (Diamond and Dybvig 1983). Gorton carries the idea from the nineteenth century to 2007. For him a financial crisis is a run on short-term debt, whether nineteenth-century bank deposits or the repurchase agreements of 2007 and 2008, which happens when holders lose confidence that the debt can be redeemed at face value; crises are a recurrent feature of systems that issue such debt, not rare accidents, and regulation and government guarantees have been responses to such runs (Gorton 2012).
On the other side are accounts that look for the panic's cause in particular banks, places and institutions. Wicker, examining the banking panics of 1930 to 1933 with data on failures across regions, finds distinct waves rather than one continuous collapse, and finds that the failures of the first wave, in 1930, were concentrated in particular regions and tied to the failure of specific banks and their correspondents; he questions the view that a general contagion produced all the closures (Wicker 1996). Calomiris and Haber move the cause further back. In their account the stability of a banking system is shaped by politics: the rules of banking reflect bargains among rulers, bankers, borrowers and voters, and the frequent crises and fragmented banking structure of the United States, compared with Canada's, follow from political coalitions. Crises are products of institutions that political actors designed or tolerated (Calomiris and Haber 2014).
Moen and Tallman's study of 1907 sits between the two sides and is the one closest to the first case read here. They study the panic with attention to the trust companies of New York, which held lower reserves than national banks and had no access to the New York Clearing House. They argue that the panic began with runs on trust companies and was concentrated among them, and that this institutional gap in access to the clearing house and its support helps explain why the panic was severe (Moen and Tallman 1992). Their argument is about a run, but a run whose force was decided by who stood inside and who stood outside an institution.
A closure document, written before the debate's evidence exists, cannot settle it. It can show how its author described the panic at the time, and what was done about the difference between an illiquid bank and an insolvent one before anyone could tell them apart.
2.2The restriction as remedy or as error
The second debate concerns the restriction of cash payments: the practice, in the American panics Sprague describes, by which banks refused or rationed the payment of currency to their depositors. Sprague's History of Crises under the National Banking System, prepared for the National Monetary Commission, describes the course of the crises of 1873, 1884, 1890, 1893 and 1907, the response of the New York Clearing House, including the issue of clearing-house loan certificates, and the restriction of cash payments by banks; it argues that the system lacked an elastic currency and a central source of reserves, so that panics recurred (Sprague 1910). In the chapter on 1907 read in Section 4 below, Sprague is severe about the restriction itself and about the delay that preceded it.
Friedman and Schwartz read the same restrictions differently. In their monetary history, the restriction of cash payments in 1907 was a response that kept the panic from turning into a wave of bank failures, while showing the weakness of the system; and the panic helped lead to the Federal Reserve Act. For 1930 to 1933 they argue that a series of banking panics and a falling money stock turned a recession into the Great Depression, that the Federal Reserve failed to act as lender of last resort to the banks, and that the bank holiday of 1933 closes the story of the panics (Friedman and Schwartz 1963).
A restriction is a closure of one kind: the bank stays open for some business and shut for the business its depositors most want. Whether it was a remedy or a mistake depends partly on what would have happened without it, which no document can show, and partly on what was decided and when, which the documents can.
2.3What began the holiday and what ended the panic
The third debate is about 1933. Wigmore asks whether the holiday reflected a run on the dollar, a flight from the dollar into gold and foreign exchange, as well as domestic runs on banks. Examining the evidence on gold and currency flows, he argues that foreign and domestic drains on gold and on the banks were both significant in the weeks before the closure, and his article is part of the debate on how far the crisis of early 1933 stemmed from fear that the dollar would be devalued (Wigmore 1987). Silber asks the question from the other end: why the reopening worked. He argues that the reopening restored confidence not only because banks were declared sound, but because the government and the Federal Reserve signalled an implicit guarantee of deposits in the reopened banks, backed by the Federal Reserve's willingness to lend against a wider range of assets under the Emergency Banking Act; deposits flowed back after the reopening (Silber 2009).
These are not opposed answers to one question. They are answers to two questions, one about the cause of the closure and one about the cause of the recovery, and the documents of March 1933 speak to both: the proclamation states its reasons, and the radio address of 12 March states what the reopening was meant to signal.
2.4The debates side by side
| Question | One reading | Another reading | What a closure document can show |
|---|---|---|---|
| What is a panic? | A run that can strike sound banks (Diamond and Dybvig 1983; Gorton 2012) | Failures tied to particular banks, regions and institutional design (Wicker 1996; Calomiris and Haber 2014); a run whose force followed access to the clearing house (Moen and Tallman 1992) | How the authority described the panic when it acted, and whether it sorted the illiquid from the insolvent before, during or after the closure |
| Was the restriction of payments a remedy? | It kept the panic of 1907 from becoming a wave of failures (Friedman and Schwartz 1963) | It was an avoidable step, worsened by delay (Sprague 1910) | Who decided to restrict, when, and whether any term was given |
| What began and what ended the 1933 panic? | Foreign and domestic drains together (Wigmore 1987) | An implicit guarantee on reopened banks, with wider lending (Silber 2009) | The stated reasons for closing, and the stated terms of reopening |
A last strand of writing supplies a frame for comparing the two episodes. Bagehot's rule, as it is usually summarised, says that in a panic a central bank should lend freely, at a high rate, on good collateral, and announce beforehand that it will do so (Bagehot 1873). Bordo argues from historical experience that a credible lender of last resort can prevent panics from spreading and that banking systems with such a lender tended to have fewer panics (Bordo 1990). On this article's reading, the nearest thing to such a lender in New York in 1907 was the clearing house, a private association of its members. In 1933 there was a central bank, and the bank holiday closed it too.
2.5Hypotheses on emergency acts: F1 to F5
The ledger evaluates emergency decisions across eighty-four cases under five pre-registered hypotheses, F1 to F5. A hypothesis is supported when 60% or more of applicable cases support it, opposed when 40% or less do, and unclear between those thresholds or when unclear cases outnumber decided ones. Strict evaluation requires explicit textual statement in both independent codings, whereas lenient evaluation also includes inferences implied by the historical setting.
Hypothesis F1 states that in most cases showing an act, the case text itself, apart from recurring editorial frames, does not name an observable condition, date or event on which the act would end or be revised. Across the catalogue, F1 supports under strict evaluation (supports 28, against 4, unclear 29, not applicable 23; share 88%) and under lenient evaluation (supports 30, against 6, unclear 8, not applicable 40; share 83%). Emergency decrees rarely establish explicit benchmarks for their termination. When evaluating an individual case, documentary findings are consistent with this general tendency, as in 1907 where no exit condition was formulated.
Hypothesis F2 states that in most cases presenting an act as temporary and recording its outcome, the act lasted longer than first stated or became permanent. Very few cases in the catalogue record both conditions: under strict evaluation F2 is unclear (supports 0, against 0, unclear 5, not applicable 79), and under lenient evaluation it remains unclear (supports 0, against 1, unclear 3, not applicable 80; share 0%).
Hypothesis F3 posits that the case text names two different dated moments among an act's announcement, its entry into effect, and its final settlement. The catalogue evidence goes against F3: under strict evaluation it fails to pass (supports 12, against 23, unclear 26, not applicable 23; share 34%), and under lenient evaluation it likewise goes against (supports 13, against 28, unclear 3, not applicable 40; share 32%).
Hypothesis F5 asserts that the case text does not name who bore the loss or cost while the emergency measure was in force. The catalogue overwhelmingly supports F5: strict evaluation yields supports 31, against 1, unclear 29, not applicable 23 (share 97%), while lenient evaluation yields supports 35, against 3, unclear 6, not applicable 40 (share 92%).
2.6The F4 test on announced and settled figures
Hypothesis F4 tests whether emergency measures specify financial magnitudes with precision: in most cases that give a figure, does the case text set the figure announced apart from what was available, drawn, paid or settled later?
The catalogue evidence goes against F4. Under strict evaluation, only 7 cases support F4 while 26 go against, 24 are unclear, and 27 are not applicable (share 21%). Under lenient evaluation, 7 support, 31 go against, 16 are unclear, and 30 are not applicable (share 18%). In most historical records, an authority announcing an emergency intervention names a single lump sum or headline commitment without distinguishing immediate disbursements from contingent caps, secondary tranches, or final recovery values. Where an individual crisis record provides financial sums, its specifics are consistent with this general catalogue pattern.