The Panic of 1907 and the Run on New York's Trust Companies

Key Takeaways

• A curb-market mining stock brought down the payments system. United Copper fell from $36 to $10 a share on 16 October 1907, and the banks controlled by the men behind the corner held $71 million of assets against more than $2 billion at all New York banks and trusts. What propagated was the discovery of who was connected to whom, not the loss.
• The regulatory gap did the rest. From 1906 New York required trusts to hold reserves of 15 percent of deposits, only 5 percent of it vault currency, while central reserve city national banks held 25 percent in legal tender or specie and alone belonged to the clearing house.
• Two thirds of the equity decline preceded the panic. The Cowles all-common-stock index peaked at 83.8 in September 1906 and was 62.2 by September 1907 before reaching 52.3 in November, so 68.6 percent of the 37.6 percent peak-to-trough fall was complete before Heinze's corner collapsed. The money market, by contrast, seized: monthly call money on mixed collateral went from 4.00 percent in September to 21.00 percent in October, and the daily rate jumped from 9.5 to 70 percent the day Knickerbocker closed and touched 100 percent two days later.
• The rescue was a syndicate, not a balance sheet. The Treasury placed $37.6 million with New York national banks between 21 and 31 October and supplied $36 million in small bills, and by mid-November its own working capital was down to roughly $5 million. The clearing house waited five days before issuing loan certificates, which Sprague called the worst error of the crisis.
• Solvency was never the binding constraint for the New York banks. They lost $30 million of reserve in the week to 2 November, then only $9 million more by 23 November, and held $251 million when the currency premium ended. They restricted cash payments for two months anyway.
• The system healed by importing gold. The monetary gold stock rose from $1.490 billion at the end of October 1907 to $1.643 billion at the end of March 1908, a gain of $153 million or 10.3 percent, while the Bank of England's day-weighted minimum discount rate went from 4.53 percent in October to 7.00 percent in December.

What Happened During the Eleven Days of October 1907?

Between Wednesday 16 October and Saturday 26 October 1907, New York moved from a failed speculation in one mining stock to a general suspension of cash payments by the banks at the centre of the national payments system.

F. Augustus Heinze had made his money in a Butte, Montana mining dispute, settled out of court in February 1906 when he sold his copper interest to Amalgamated Copper for a reported $25 million. He took the proceeds to New York, bought his way onto the boards of eight banks and two trust companies, and became president of Mercantile National Bank in February 1907. The method was what contemporaries called chain banking: buy stock in one bank, pledge it to borrow, buy stock in the next, so that every institution in the chain was collateral for the one after it. When the corner failed, depositors did not reason about copper. They reasoned about which banks Heinze sat on the board of, and then about his associate Charles W. Morse, a director of seven New York banks who controlled three outright.

Chronology of the acute phase

Dated events with the day of the week, reconstructed from the three institutional accounts listed in the References below.

Why Did a Failed Copper Corner Bring Down Banks?

Because the corner was financed on margin by men who also controlled deposit-taking institutions, and because it failed in public.

Heinze believed brokerage houses were lending out his United Copper shares to short sellers. He had the brokerage Gross and Kleeberg buy 6,000 shares at ascending prices, so that he would repurchase his own stock from the shorts at higher levels and draw more shorts in, then called in his shares to force settlement. The mistake was factual: more stock was available than he thought. A transfer agent then refused to move a block of 17,830 shares, which read as evidence that somebody else controlled the float. Meanwhile the raw copper market was being handled by an Amalgamated subsidiary, the United Metals Selling Company, which the Pujo Committee later established had sold only 5 million pounds between April and August 1907 against a normal range of 150 million to 250 million pounds, then 93 million in October alone.

On 16 October the brokers who had bought United Copper on margin had to sell to meet their own obligations. The stock went from $36 to $10, Gross and Kleeberg suspended, and Otto Heinze and Company closed. That would have been an ordinary broker failure. What made it a banking event was that the reorganisation of Mercantile National, forced on it by the clearing house as the price of support, exposed the interlocking directorates linking banks, brokerages and trusts across the city. Depositors did not need to understand copper, only a list of names.

The clearing house handled the bank half competently, examining the Heinze-linked member banks over the weekend of 19 and 20 October, announcing them solvent, promising a $10 million support fund and requiring Heinze, Morse and E. R. Thomas to leave banking. Mercantile reopened on the Monday and the run stopped, and Sprague's judgment is that as late as 21 October nothing resembling a systemic panic had touched the banks.

Why Were the Trust Companies the Weak Point Rather Than the Banks?

Trust companies were state-chartered competitors for deposits that had grown far faster than the banks and were regulated far more lightly. New York trust assets rose 244 percent in the ten years to 1907, from $396.7 million to $1,394.0 million, while the city's national bank assets grew 97 percent, from $915.2 million to $1,800.0 million. A category that had been an estate-management business in the 1890s was by 1907 doing almost everything a bank did except issue notes.

What it did not have was cash. New York first required trusts to hold reserves of 15 percent of deposits in 1906, only 5 percent of it vault currency; before that they held whatever they thought prudent. Central reserve city national banks had to keep 25 percent in legal tender or specie. The Federal Reserve's own history puts the practical contrast at roughly 5 percent against 25 percent, and adds the detail that makes the arithmetic dangerous: trust deposits were payable in cash on demand, exactly like bank deposits.

The trusts were also load-bearing for the stock market in a way invisible from their balance sheets. National banks were prohibited from making uncollateralised loans or guaranteeing broker checks drawn on insufficient funds. Trusts were not. A broker would borrow from a trust without collateral, buy securities, pledge those securities for an overnight call loan from a national bank, and repay the trust out of the call loan, all inside one business day. The exposure lasted hours and looked riskless. It was also the first link in the chain, and when trusts stopped extending that daylight credit the whole sequence stopped with them. They were the marginal supplier of liquidity to the largest equity market in the country, funded on demand, holding almost no cash, with no access to a lender of last resort. All four facts were public. None said when.

Why Did the Clearing House Refuse to Lend to the Knickerbocker Trust?

Because the New York Clearing House was a private association of member banks and Knickerbocker was not a member. Neither were most of the city's trusts. The refusal was not a judgment about Knickerbocker's assets but a statement about the boundary of a club.

Knickerbocker was the third largest trust company in New York. The run appears to have begun on Friday 18 October, when Barney's connection to Morse was reported, and the National Bank of Commerce had been lending privately to cover the withdrawals. On Monday 21 October it asked the clearing house for a loan on Knickerbocker's behalf, was told the association's resources were reserved for members, and then announced it would stop acting as Knickerbocker's clearing agent. That announcement was read exactly as it was meant to be. The request went to Morgan, who sent Benjamin Strong of Bankers Trust, later the first head of the Federal Reserve Bank of New York, to value the books. Strong could not reach a conclusion in the hours available; Chandler's account, cited in the Federal Reserve's own history of the episode, records that Knickerbocker had run out of cash and closed its doors before even a cursory examination could be completed. Morgan declined, and that deterred everyone else. On Tuesday the trust paid out $8 million in roughly three hours and suspended just after noon.

Suspension is not failure. Knickerbocker reopened in March 1908 after a capital infusion of $2.4 million, the strongest available evidence that it was closer to illiquid than insolvent. But suspension did to New York what a failure would have done, because depositors elsewhere could not tell the difference and had no time to find out.

Hindsight check. The temptation is to say Morgan should have lent. He was being asked to commit money he did not fully control, within hours, on the word of an examiner who said explicitly that he could not form a view, to an institution that had already run out of cash. Nearly the same question was put to three officials over a weekend in March 2023 and answered the other way, and it took a statutory systemic risk exception plus a par-value lending facility to do it. In 1907 there was no statute to invoke and no facility to open. Cognitive biases in trading covers that error.

What Happened at the Trust Company of America After Knickerbocker Closed?

The run moved to the second largest trust in the city within a day, and this time the money was found.

On Wednesday 23 October the New York Times ran its report of Knickerbocker's suspension beside a headline describing the Trust Company of America as the current sore point of the panic. The phrase came from George W. Perkins, one of Morgan's partners, and Barney also sat on that company's board, so the link was real. Whether naming a sound firm near Morgan's offices was an attempt to concentrate the panic somewhere it could be visibly defended, or simply a serious mistake, was disputed then and since.

The withdrawals are documented day by day: approximately $1.5 million on Tuesday 22 October, another $13 million on the Wednesday from a base then near $60 million, and a further $8 million to $9 million on the Thursday. Across the two weeks the run lasted, the Trust Company of America reportedly paid out $47.5 million, roughly four fifths of the base recorded on that Wednesday.

This time Strong finished the job. He arrived after two in the morning, worked through the assets, and reported that afternoon that the company was basically sound and deserved help. Morgan moved about $3 million to it just before closing, enough to let it open the next day, and John D. Rockefeller deposited $10 million with Union Trust and said publicly that he was behind Morgan. The difference between the two trusts was not solvency and not size. It was whether an examiner could form a view in the time available, and that is why the private clearing house model failed here: the association held detailed current knowledge of its members' assets, and nobody held that for the trusts.

Why Did the Call Money Rate Reach 100 Percent on 24 October?

Because the trusts had been the first supplier of credit to the stock market and were now calling loans rather than making them, while country banks simultaneously demanded liquidation of their New York call loans and shipment of the cash home.

Call money on mixed collateral averaged 3.06 percent in August 1907 and 4.00 percent in September. In October it averaged 21.00 percent, more than five times the September figure, and stayed in double digits through December. Averages hide the intraday reality: the Federal Reserve's history records the daily maximum jumping from 9.5 percent to 70 percent on 22 October, the day Knickerbocker closed, and reaching 100 percent two days later, with periods when there were no offers at that rate at all.

Thursday 24 October is the day the exchange nearly stopped. The opening call rate was 6 percent. Ransom H. Thomas, the exchange president, noticed that no money was being offered; a bid of 60 percent went out and drew nothing. The last recorded transaction of the day was back at the opening 6 percent, which tells you the tape was no longer describing a functioning market. Thomas called Stillman, Stillman sent him to Morgan, and while he was travelling the rate touched 100 percent. Morgan told him to announce that $25 million would be available on the floor, and his partner Charles Steele arrived with a list of national banks that had promised it, including $4 million from First National and $8 million from National City. The market took $18.95 million.

Friday's rescue had different terms. About $10 million came from the Morgan group, $2 million from First National and $500,000 from Kuhn, Loeb, and this time Morgan let the market set the rate, which stayed near 50 percent for most of the day, attaching conditions designed to stop the money financing speculation: no margin purchases and cash sales for investment only. Through all of it the exchange stayed open. There was no circuit breaker, no halt and no closure, and what kept it trading was a list of subscriptions read out on the floor, because the machinery described under market structure and execution did not yet exist.

Whose Money Was J. P. Morgan Actually Directing?

Mostly other people's, and a large part of it belonged to the United States government. This is the most distorted fact about 1907.

Treasury Secretary George Cortelyou met the New York financiers on the evening of Wednesday 23 October and deposited $25 million of Treasury funds in national banks the next morning. Between 21 and 31 October the Treasury placed $37.6 million with New York national banks and supplied $36 million in small denomination bills. Sprague, working from the Secretary's own response to a Senate resolution of December 1907, put it more starkly: about half of all the money the New York banks paid out during the two weeks ending 2 November came from the government. Then it ran out. By mid-November the Treasury's working capital had fallen to roughly $5 million, which is precisely the constraint a central bank exists to remove: a Treasury spends a fixed stock of cash and a central bank creates reserves.

The legal position of those deposits mattered less than it looked. Treasury money could go only to national banks, and direct use of it in the stock market was prohibited. But Cortelyou told the Pujo Committee in 1912 that the intended use was not specified when the funds were credited, and that the financiers decided where the money should go. Pressed by counsel Samuel Untermyer for the split of the $25 million between individual banks, he could not recall the transactions and did not know whether the Treasury held records of them.

This does not diminish Morgan. Coordination was the scarce commodity in October 1907 and he supplied it. But the record shows a Treasury without the power to create money handing its entire liquid position to a private committee with no legal accountability for it and no record of how it was split, and that, rather than the rescue's success, is what Congress took away.

Why Did New York City Itself Come Close to Default?

Because a municipality funding itself by rolling short-term obligations faces the same run as a trust company, and in late October 1907 there was no bond market to roll into. The city needed $30 million. It had tried to sell bonds that summer and found no market even then, and had been paying its bills with short-term borrowing since, so those obligations were maturing with no cash behind them. Mayor George McClellan went to Morgan on Monday 28 October, and Morgan's reasoning was simple: a default by New York City in the middle of a banking panic would put the crisis beyond anyone's ability to manage.

The solution on 29 October was a mechanism rather than a cheque. Morgan, Stillman and Baker agreed to underwrite a $30 million, 6 percent city bond issue in which the major banks would take pro rata shares and deposit the bonds with the clearing house, which would issue loan certificates against them in an equal amount and credit those to the city's accounts at First National and National City. The city got spendable balances, the banks got an asset they had not paid cash for, and almost no currency moved.

The clearing house, which a week earlier had refused a trust company on the ground that membership defined its obligations, was now monetising municipal bonds for a borrower that was not a bank at all. The boundary that killed Knickerbocker turned out to be movable once the alternative was a sovereign-adjacent default.

What Was the Tennessee Coal and Iron Deal, and Why Did Roosevelt Allow It?

It was a rescue of a brokerage conducted entirely in securities, because no cash was available to conduct it in cash, and it cost the Roosevelt administration a piece of its antitrust policy.

Moore and Schley had borrowed $25 million from New York banks against a large block of Tennessee Coal, Iron and Railroad Company stock, and those loans were about to mature. The position was worse than a single pledge, because the same stock was also collateral for loans the firm had made to its senior partner Grant B. Schley, Baker's brother-in-law. If the firm liquidated to repay the banks, the price would collapse, the collateral behind a set of already strained call loans would fall with it, and the money market would tighten again at the worst moment.

The plan came from Morgan's attorney Lewis Cass Ledyard. United States Steel would buy the Tennessee Coal and Iron shares and pay with its own highly rated 5 percent gold bonds. The elegance was that no currency changed hands in a market that had none: a distressed, illiquid equity position was exchanged for a marketable bond of a first-class issuer. Moore and Schley was saved without a forced sale, and United States Steel absorbed a competitor. The deal closed on Monday 4 November, after President Roosevelt agreed not to oppose it on antitrust grounds. The same day, trust presidents organised by Morgan funded a $25 million loan for the Trust Company of America and Lincoln Trust, and the panic began to ease.

The antitrust concession was the price, paid by a President who had built his reputation on trust-busting. Whatever a rulebook says in normal conditions, the binding constraint in the acute phase is who has capacity and what they want in exchange.

What Were Clearing House Loan Certificates, and Why Was Issuing Them Late the Central Error?

A clearing house loan certificate was an interest-bearing claim that member banks could use instead of currency to settle balances with each other, freeing the currency they would otherwise have needed for that purpose to be paid across the counter to depositors. Under the 1907 resolution the loan committee accepted bills receivable and other approved securities from members, issued certificates at up to 75 percent of the collateral's market value, and charged 6 percent a year.

The Loan Committee's own report, reproduced in Sprague's study for the National Monetary Commission, gives the arithmetic precisely. The first issue was on 26 October 1907 and the final cancellation on 28 March 1908. The gross issue was $101,060,000, or 9.38 percent of the clearing house banks' total net deposits of $1,066,865,900, and the maximum outstanding at any one time was $88,420,000 on 16 December 1907, or 8.28 percent. Collateral passing through the committee's hands aggregated $453,000,000, of which $330,000,000, or 72.92 percent, was commercial paper.

The device was not new. The clearing house had used it in 1873, 1884, 1890 and 1893, which is why Sprague's verdict is so blunt: the failure to issue certificates at least as early as Tuesday 22 October was the most serious error made during the crisis. Had they come on the Tuesday, he argued, the banks could have lent more freely and the money pools would have been unnecessary. That is a lender-of-last-resort argument made in 1910, three years before there was one.

The clearing house president's later explanation, quoted by Sprague, was that the committee postponed the issue day after day hoping the panic would subside, honouring drafts against a falling reserve until that reserve showed a deficit of $53,000,000, at which point hesitating further would have been folly. Sprague appends a one-sentence footnote that does more damage than a chapter of argument: it was not until two weeks after the issue of loan certificates that the bank statement showed a deficit of $53,000,000. The number the committee gave as its trigger had not yet occurred when it acted.

Why Did Solvent Banks Stop Paying Out Cash?

On the same Saturday it authorised loan certificates, the clearing house restricted the convertibility of deposits into cash. Banks kept taking deposits and clearing checks; what they stopped doing was handing over currency on demand in full. The restriction spread nationally through correspondent banking, because a New York bank that will not ship currency against a country bank's balance forces that country bank to do the same to its own depositors.

A dollar of currency and a dollar of deposits stopped being the same thing. From 31 October to 31 December 1907 inclusive, currency was regularly bought and sold at a premium in New York against certified checks, exactly two months and roughly twice the duration of the equivalent episodes in 1873 and 1893. A. Piatt Andrew's contemporary survey concluded that nearly $500 million of cash substitutes circulated across the country as a principal means of payment, including small-denomination certificates, certified checks and payroll scrip. A national payments system had, for two months, no single national means of payment.

Sprague's charge is that most of this was unnecessary, and the reserve data is the reason. In the week ending 2 November the New York banks lost $30 million and still held $224 million, against a legal reserve deficit of $38 million. By 23 November the reserve had fallen by only $9 million more, then rose every week to $251 million when the currency premium ended. His comparison is with London, where the much smaller banking reserve of the Bank of England fell from £24,000,000 to £17,000,000 in the same fortnight without anyone contemplating suspension.

Behind that is a reserve requirement that behaved as a floor rather than a buffer. The NBER series for the ratio of reserves to net deposits at central reserve city national banks shows 26.2 percent in August 1907, 22.2 percent in December, then 29.7 percent by May 1908. The banks preferred to break a promise to depositors rather than a percentage, then rebuilt the percentage well above the 25 percent minimum while output was still contracting.

How Did Gold Imports and the Bank of England Enter the Story?

The gold standard is what eventually ended the panic, and it ran in both directions: it made the autumn of 1907 unusually tight beforehand, then supplied the cash that repaired the system. The preceding chapter is a Treasury policy that worked too well. In the spring of 1906 Secretary Leslie Shaw subsidised gold imports through finance bills to relieve what he judged a domestic shortage, pulling in about $50 million in a little over a month between April and May. Britain lost the gold, so the Bank of England raised its discount rate in late 1906 and threatened a further increase unless American finance bills were paid at maturity without renewal. Finance bills were therefore suspended during 1907, removing the arbitrage that normally reduced actual shipments of metal, and that summer the United States exported $30 million of gold to London despite relatively high domestic interest rates. New York entered its autumn crop-moving squeeze with an unusually thin gold position.

Once the banks restricted cash payments the flow reversed hard. The currency premium raised sterling exchange to levels that would normally have sent gold out of the country, and gold kept arriving anyway. Sprague's explanation is trade rather than arbitrage: the excess of merchandise exports over imports in the last two months of the year went from $119,000,000 in 1906 to $208,000,000 in 1907, with essentially no finance bills outstanding to absorb it. The monetary gold stock records the result: $1.490 billion at the end of October 1907, $1.562 billion at the end of November, a single-month gain of $72 million or 4.8 percent, and $1.643 billion by the end of March 1908, adding $153 million or 10.3 percent in five months.

Britain paid for it. The Bank of England's minimum discount rate, as a day-weighted monthly figure, was 4.53 percent in October 1907, 6.75 percent in November and 7.00 percent in December, then fell to 4.00 percent by February 1908 and 2.50 percent by June. A country with no central bank imported the liquidity it needed from a country that had one, and the country that had one raised rates into its own slowdown to stop the drain. That channel runs again through the 1929 crash under a reconstructed version of the same standard.

How Far Did Share Prices and the Real Economy Actually Fall?

Further in the real economy than in the stock market, and the market had done most of its falling before anyone had heard of United Copper.

The Cowles all-common-stock index, later maintained by Standard and Poor's, peaked at 83.8 in September 1906, was 62.2 by September 1907 and bottomed at 52.3 in November. The full peak-to-trough decline is 37.6 percent, and 68.6 percent of it was complete in the month before the corner failed. October 1907 cost 10.9 percent. An investor who first noticed something wrong when Knickerbocker closed had already missed two thirds of the loss.

The National Bureau of Economic Research dates the business cycle peak to May 1907 and the trough to June 1908, a contraction of thirteen months following a thirty-three month expansion, so the peak precedes the panic by five months. The NBER index of industrial production and trade is expressed as a percentage of trend rather than as a level; it stood at 112.9 percent of trend in May 1907, was still 105.4 percent in October, then fell to 85.7 percent by December and 81.0 percent in March 1908. Two months either side of the panic cost 19.7 percentage points of trend. The Federal Reserve's history records industrial output falling 17 percent in 1908 and real GNP falling 12 percent, a contraction exceeded in United States history only by the Great Depression.

Monthly readings through the panic and the recovery

All five series come from the NBER Macrohistory Database. The stock index has 1935 to 1939 equal to 100; industrial production and trade is a seasonally adjusted percentage of trend; gold is the end-of-month monetary gold stock.

Read the table across and the sequence appears in it: call money spikes in October while the index is still falling, gold turns in November, and commercial paper, the price of credit to businesses rather than brokers, peaks in December. By June 1908 call money averaged 1.52 percent, half its level in August 1907, while the production index was still 15.9 percentage points below trend.

Business failures record the corporate damage on their own schedule. R. G. Dun's counts total 11,721 for 1907, then 15,690 for 1908, a rise of 33.9 percent, and the worst month is January 1908 at 1,949, three months after the panic and one month after the currency premium ended. Recovery took years: the stock index did not print a monthly reading above its September 1906 peak until August 1909, twenty-one months after the trough and 62.7 percent above the low, and that is also when the industrial production and trade index first returned to 100 percent of trend.

Which Warning Signs Were Visible Before October, and Which Only Afterwards?

The 1907 case is unusually clean here. The structural warnings were not merely visible, they had been published, and they still gave no usable timing.

Visible in advance

• The reserve gap was a matter of state law and the clearing house perimeter was public. Anyone could read that New York trusts held 5 percent of deposits as vault currency against 25 percent for central reserve city national banks, that trust deposits were payable on demand exactly like bank deposits, and that most trusts were outside the association whose resources were reserved for members.
• Share prices had been falling for thirteen months and money was already tight. The index had lost about a quarter from its September 1906 peak by September 1907, including a 9.8 percent fall in March that contemporaries called a rich man's panic, while New York commercial paper rose from 5.47 percent in May to 6.81 percent in September and the country entered the crop-moving season with a gold position depleted by the summer's shipments to London. Heinze's reputation was in the newspapers too: the house of J. S. Bache had cut business relations with him in February 1907.

Only visible afterwards

• Which institution would break first, and how. Nothing on 15 October pointed at Knickerbocker specifically; what made it the first casualty was a personal association with Morse, a clearing agent's decision to resign, and an examiner who ran out of hours. Nor was it knowable in the queue outside its doors that the outcome would be suspension rather than failure, and that it would reopen in March 1908 after a $2.4 million injection.
• That the clearing house would delay five days. The certificate mechanism had worked in four previous crises, so assuming it would be used promptly was the reasonable expectation, and it was wrong. Equally unknowable was that the eventual fix would be a central bank: the Federal Reserve Act was six years away, was not the plan the National Monetary Commission produced, and was contested at every stage.

An investor who understood every item in the first list would have known the system was fragile, and would still have had no way to know that the detonator would be a copper stock or that it would go off on 16 October. That distinction between reading a structure and timing an event is why risk management is a question about survivable position size, not forecast accuracy.

How Did the Panic of 1907 Produce the Federal Reserve?

Not directly and not quickly. Congress passed the Aldrich-Vreeland Act on 30 May 1908, seven months after Knickerbocker suspended. It provided for emergency currency and created an eighteen-member National Monetary Commission, chaired by Senator Nelson Aldrich, to work out what the country's banking and currency laws required. That commission is why this article can be written from primary material at all: it spent three years on hearings and European visits and published a shelf of monographs, one of which is Sprague's history of crises under the national banking system.

Aldrich unveiled his own plan in January 1911, and after a year of revision it went to Congress in 1912 proposing a National Reserve Association. Its origins included a deliberately secret meeting on Jekyll Island, Georgia, in November 1910, attended by the investment banker Paul Warburg and the Treasury official Abram Piatt Andrew among others, whose participants used first names only and said they were duck hunting, on the theory that a plan visibly written by bankers would not survive public scrutiny. They were right. The plan proposed a forty-six member board with only six government appointees, was attacked in both chambers for handing power to the largest institutions, and died with the 1912 election.

What passed instead came from a House subcommittee under Carter Glass, who wanted autonomous regional banks and no central coordinating board, modified by Woodrow Wilson, who insisted on a government-appointed supervisory board, and by Senator Robert L. Owen, who capped the number of Reserve Banks at twelve. The Federal Reserve Act was signed on 23 December 1913.

Moen and Tallman's explanation of why reform happened this time is the sharpest available. The New York Clearing House bankers had opposed a central bank for decades because they were, in effect, running one for themselves. What changed in 1907 was that the panic took hold among institutions outside their membership. It was not the failure of the private lender of last resort that changed minds, but the discovery that its perimeter no longer matched the financial system.

Where Does the Comparison With 2008 and 2023 Break Down?

The Federal Reserve's own history makes the 2008 comparison explicitly and it is a good one at the level of mechanism. Both crises began at institutions outside the effective lender of last resort: the trusts of 1907 borrowed short and supplied the daylight credit that let brokers buy securities, and the investment banks of 2007 and 2008 borrowed short in repo from money market funds and hedge funds to finance asset-backed securities. Both were run by their own funders, and in both cases trouble away from the payments core then wrecked it. The 2008 financial crisis case study covers the modern side in detail. Three things break the analogy, and they are the interesting parts.

Knickerbocker suspended and Lehman failed. Suspension froze depositors out and preserved the institution; Knickerbocker reopened five months later with $2.4 million of new capital, while Lehman's remains were sold off and its customers waited nearly six years for payment. The 1907 toolkit contained an option, temporary closure, that modern resolution law does not really offer.

The 1907 authorities could not create money. The Treasury ran out at roughly $5 million of working capital by mid-November. In the 2023 regional banking stress the constraint was legal and political rather than arithmetic: the Federal Reserve opened a facility lending against securities valued at par, and the Treasury Secretary, the Federal Reserve Chair and the FDIC Chairman jointly invoked a statutory systemic risk exception to guarantee uninsured deposits. Neither instrument had a 1907 equivalent, which is why 1907's answer was a syndicate and a two-month suspension of cash payments.

Nobody in 1907 could be excluded from the consequences. There was no deposit insurance, so a depositor's entire balance was an unsecured claim on an institution they could not examine, and New York savings banks fell back on their right to demand sixty days' notice of withdrawal, a contractual gate rather than a guarantee. FDIC deposit insurance, which insures deposits to at least $250,000 at each insured bank, did not exist until 1933, and that change is why United States bank runs since then have been concentrated among uninsured balances rather than general.

Common Myths About the Panic of 1907

"J. P. Morgan bailed out the United States with his own fortune." He supplied coordination and judgment, both genuinely scarce, and committed his own firm's money. But the Treasury put $37.6 million into New York national banks in eleven days, Sprague estimated that about half of everything those banks paid out in the two weeks to 2 November came from the government, and the pools Morgan assembled were subscriptions from other institutions. Morgan decided where money went. He did not, for the most part, own it.

"The panic was caused by the copper corner." The corner was the detonator, not the charge. United Copper traded on the Curb, the banks connected to Heinze held $71 million of assets against more than $2 billion citywide, and the equity market had already fallen about a quarter. What the corner did was force an examination that revealed who was connected to whom.

"The New York banks had run out of money." They held $224 million of reserve on 2 November after their worst week, lost only $9 million more over the next three weeks, and finished the year at $251 million, and they restricted cash payments for two full months anyway. The binding constraint was a 25 percent statutory reserve ratio the banks treated as untouchable, not the absence of cash.

"Depositors panicked irrationally." A depositor at a New York trust in October 1907 held an on-demand claim against an institution holding roughly 5 percent of deposits as vault cash, with no insurance, no access to the clearing house, and no way to tell a solvent trust from an unsound one in the hours available. Queueing was the correct individual response, and the reforms that followed addressed the structure rather than the sentiment.

What a Reader Can Actually Carry Forward

A gold standard, a private clearing house and a banker working from a personal list of who could be telephoned for money are not coming back. Four things underneath them still carry.

What generalises

• Ask which lender of last resort covers the thing you are exposed to. The single fact that determined who survived October 1907 was clearing house membership. Not asset quality, not capital, not size. The modern form is whether an entity can reach a central bank facility directly, through a parent, or not at all.
• An institution that cannot be examined quickly cannot be rescued quickly. Knickerbocker and the Trust Company of America were both illiquid and both had a Morgan examiner in the building. One got a verdict in time and one did not.
• A reserve that may never be spent is not a reserve. New York's banks suspended cash payments for two months while holding a reserve that fell about 4 percent from its early November level and then rose. A buffer treated as a floor produces the behaviour it exists to prevent. Separating a reported cushion from a spendable one is the whole point of stress testing and scenario analysis.
• Recovery has more than one clock. Share prices bottomed in November 1907, commercial failures peaked in January 1908, output bottomed in March, the NBER trough is June, and the index did not regain its old peak until August 1909. Any single statement about how long 1907 took to recover picks one of those and hides the rest.

What does not generalise

• The private rescue. It worked once, under a specific concentration of authority, and both the participants and Congress concluded it was not repeatable. Treating it as a template rather than as evidence of an institutional gap gets the episode backwards.
• The suspension of cash payments, and the gold cure. Currency and deposits were separable under a gold standard with no central bank, and there is no modern route to a general restriction of convertibility by private agreement. Nor does any modern economy repair a liquidity crisis by importing metal on the back of a trade surplus over five months.

The one question worth asking now

The 1907 material has one answerable form, and it is a question about counterparties rather than about history repeating. For each institution holding something you own or owe, who lends to it on the worst possible Monday morning, and is that a legal entitlement, a facility it must qualify for, or somebody's discretion? In October 1907 the answer for the New York national banks was a private association they belonged to, and for the trust companies it was nobody. Every subsequent banking crisis in this library, from the savings and loan crisis onward, has turned partly on that question, and it requires forecasting nothing. Terms used here are defined in the investment and trading glossary.

References

Every figure on this page was verified against the following sources, each retrieved on 26 August 2026:

• Federal Reserve History: The Panic of 1907: the corner failure and the events of 18 to 26 October, the reserve contrast, the daylight-loan mechanism, the daily call rate at 9.5, 70 and 100 percent, Knickerbocker's reopening on a $2.4 million infusion, its note recording Lester Chandler's finding that the trust had run out of cash before a cursory examination could be finished, and the 1908 falls in output and real GNP.
• National Monetary Commission: History of Crises Under the National Banking System, by O. M. W. Sprague, 1910: the Clearing House Loan Committee report behind every certificate figure quoted here; the Treasury deposits and money pools of late October; the November and December reserve figures; the clearing house president's $53,000,000 explanation and Sprague's footnote contradicting it; the currency premium dates; the Bank of England reserve; and the merchandise trade totals.
• Federal Reserve Bank of Atlanta: Lessons from the Panic of 1907, by Ellis W. Tallman and Jon R. Moen, Economic Review, May and June 1990: the Heinze settlement, directorships and chain banking; the United Copper prices and copper sales volumes; every Trust Company of America withdrawal figure; Strong's examinations, Morgan's $3 million and Rockefeller's $10 million; the Treasury totals; the 24 and 25 October exchange sequence; the New York City bond mechanism and the Moore and Schley transaction; the trust asset growth figures and the 1906 New York reserve rule; the Shaw finance bill policy; Andrew's estimate of currency substitutes; and the Pujo Committee testimony.
• National Bureau of Economic Research: US Business Cycle Expansions and Contractions: the May 1907 peak and June 1908 trough, and the expansion and contraction lengths derived from the surrounding turning points.
• Federal Reserve History: The Federal Reserve Act Signed by President Wilson: the Aldrich-Vreeland Act, the National Monetary Commission, the 1911 plan and 1912 National Reserve Association proposal, the Jekyll Island meeting, the Glass and Owen bills, and the signing on 23 December 1913.
• Federal Reserve Bank of St. Louis, NBER Macrohistory Database, for every monthly series quoted here: Call Money Rates, Mixed Collateral for United States, Series M13001USM156NNBR, Commercial Paper Rates for New York, NY, Series M13002US35620M156NNBR, Index of All Common Stock Prices, Cowles Commission and Standard and Poor's Corporation for United States, Series M1125AUSM343NNBR, Index of Industrial Production and Trade for United States, Series M1204AUSM516SNBR, Monetary Gold Stock for United States, Series M1476AUSM027NNBR, Number of Business Failures, All Commercial for United States, Series M0929AUSM155NNBR, Minimum Rate of Discount, Bank of England for Great Britain, Series M13013GBM156NNBR and Ratio of Reserves to Net Deposits, National Banks, Central Reserve Cities for United States, Series M14060USM156NNBR. The St. Louis Fed hosts rate-limit repeated automated requests, and the FRASER pages answered a plain user agent when they had refused a browser one, so a timeout on any of these is not evidence of link rot.
• FDIC: Deposit Insurance: the statement that deposits are insured to at least $250,000 at each insured bank. This is a statutory amount that Congress has raised before and can raise again, so it is stated as the figure in force at the review date rather than as a permanent feature.

Figures deliberately not stated. No Dow Jones Industrial Average level is given for 1906 to 1909, because no source verified here supplied a daily index; the Cowles monthly index is used instead and labelled as such. No count of bank suspensions, no national deposit total, no unemployment rate and no figure for gold imported specifically from Britain appears, for the same reason. The intraday call money path on 24 October is given in the terms the two institutional accounts use, which differ on whether the 100 percent print was the day's maximum or a level reached at a particular hour.

Frequently Asked Questions