6.1Who was shut, and who was shut out
The two episodes draw their boundaries in opposite ways. In 1907 the boundary decided who would be helped, and it was drawn before the crisis by a private association's rules: members inside, trust companies outside, with the Knickerbocker in the strange position of having met the clearing house's reserve rule without belonging to it. In Sprague's account no one was ordered closed. Closure came to those outside the circle by refusal, by the withdrawal of a clearing arrangement, by a run. In March 1933, the executive order encompassed all deposit-taking entities simultaneously, extending its mandate to national associations, state institutions, trust entities, mutual savings banks, and the Reserve Banks themselves. The sorting into inside and outside was postponed to the reopening, and done by examination.
The contrasting boundaries illuminate the structural friction analyzed by Moen and Tallman (1992), where the exclusion of New York trust companies from direct clearing privileges amplified systemic vulnerability during the initial runs. The 1933 texts look in part like an answer to that gap. Roosevelt's public explanation directly reassured non-member state banking enterprises of liquidity support channeled via member correspondents and the Reconstruction Finance Corporation, outlining parallel state authorization for their resumption. Whether this was meant as an answer to 1907 the texts do not say, and the article does not claim it. What they show is that the question of the non-member, which in 1907 was settled by silence, was in 1933 put in words to the public.
It also speaks to Calomiris and Haber's argument that the fragility of American banking was a matter of political design (Calomiris and Haber 2014). The 1907 boundary was a product of rules that banks had made for their own competition with trust companies, as Sprague tells it. The 1933 boundary was drawn by the state for the whole system at once. The documents cannot show which design was more or less fragile. They show that in both cases the line that decided who was helped was a line of institutional status, drawn before the panic for other reasons, and that in 1933 the state chose to override it at the moment of closing and redraw it at the moment of reopening.
6.2How long
On the question of term, the contrast runs the other way from what a reader might expect. The 1907 closures, as Sprague records them, had no terms set by any authority. The only stated term in the week was a private one, the savings banks' right to sixty days' notice, invoked against their own depositors. The restriction of cash payments in New York began with the issue of loan certificates and had no written end. The Knickerbocker's suspension ended, months later, with a reorganization agreed by its depositors and shareholders.
The 1933 proclamation named its term precisely. By 9 March the Act had replaced it with an emergency period to be set by later proclamation, for the member banks; the address described the change as making it possible to extend the holiday and lift it gradually. The conservatorship provisions of the Act run until the further disposition of a bank's business, without a date, though conservators were permitted to release payments ratably under Section 206. The section on gold has no term at all. The one act in this pair that was given a dated term kept a dated end only until the statute of 9 March made the end a matter for later proclamation, and the instruments that followed named the authority that would decide the end rather than the end itself.
It would be easy to read this as a case of the emergency that stays, the pattern Higgs describes in which the federal government grew in jumps during crises and did not return fully to its earlier size afterwards (Higgs 1987). The texts at hand cannot carry that reading. They show the term opening; they do not show how long the opened term lasted, or whether any of the powers outlived their reasons. Skeptron's record says nothing of what became of the holiday as an act. What the texts are consistent with is narrower: the dated term was the first part of the 1933 programme to give way, and the reason the address gives for abandoning it, the need for time to examine, is the same reason Sprague gives for the failure of 1907, where there was no time at all.
6.3What the documents can and cannot carry for each debate
For the debate over runs and insolvency, the documents carry one thing well. They show that in both episodes the authorities acted before they could distinguish an illiquid bank from an insolvent one, and they show how each handled that ignorance. In 1907 it was handled by not examining, and the institution outside the circle bore the result. In 1933 it was handled by closing everything long enough to examine, and by attaching assurance only to those found sound. The address's own diagnosis contains both sides of the debate: a run on sound banks, and banks badly run. It is not evidence for either. It is evidence that the government of March 1933 did not need to choose between them in order to act.
For the debate over restriction, the documents carry less. Sprague's verdict that the New York restriction of 1907 was discreditable and avoidable, and Friedman and Schwartz's reading that the restriction kept the panic from becoming a wave of failures (Friedman and Schwartz 1963), are both claims about what would have happened otherwise. The excerpt read here is consistent with one narrower claim: the restriction was decided by postponement, foreseen by the committee that delayed it, and given no term. The 1933 proclamation, in authorizing clearing-house certificates under the Secretary of the Treasury, takes the instrument whose delay Sprague blamed and puts it under public authority. Whether that change mattered is beyond the documents.
For the debate over 1933, the documents carry the reasons and the promises: the recitals name both drains Wigmore studies, and the address names the wider lending Silber relies on and gives an assurance confined to the reopened banks. Neither text measures a flow or a deposit; that deposits returned after the reopening belongs to Silber's evidence, not to anything read here (Silber 2009).
6.4The crisis catalogue
The two episodes read in detail take their place within the broader catalogue of eighty-four emergency decisions documented in Chapter 3. Across that corpus, the coder metrics showed varying levels of agreement: while acts and their broad classifications reached 98% and 90% agreement respectively, agreement was moderate at 58% (49 of 84) on the time horizon and the reversal constraint. Furthermore, the catalogue shows that emergency actions rarely stem from an unconstrained single individual; in the decider distribution established in Chapter 3, an individual was implied in only 1 case, whereas institutions, states, and collective bodies predominated whenever deciders were identified.
The catalogue distributions also put the question of term into perspective. Across the eighty-four cases, when an exit was named, it almost invariably took the form of a calendar term of time rather than an objective state condition such as a solvency metric or reserve ratio. The 1933 experience is consistent with this catalogue-level pattern: Proclamation 2039 set an observable calendar term (9 March) under F1, but formulated no state condition for ending the holiday. The transition on 9 March from a calendar term to an administrative licensing regime highlights why F1 is supported across the catalogue: emergency acts frequently initiate intervention under a calendar deadline, but the substantive resolution is deferred to administrative discretion.
Similarly, the modern summaries divide in an instructive manner. The record of 1933 preserves the chronological succession of three dated acts and names the depositors left outside the reopening. It loses the proclamation's initial calendar term and the statutory modification that followed. The record of 1907 preserves the refusal of support and warns against equating lack of time with insolvency, matching Sprague's own reservations, but leaves out the clearing-house reserve rules and the eventual corporate reorganization. Neither record substitutes for contemporary statutory texts. Their significance lies in demonstrating what modern public summaries preserve: specific calendar dates are retained, whereas institutional boundaries are compressed and explicit exit terms are largely omitted. Across the catalogue, this documentary compression is consistent with the finding that most emergency records fail to articulate two distinct dated moments or explicitly identify who bore the costs of intervention.