Economic Effects of Runs on Early 'Shadow Banks': Trust Companies and the Impact of the Panic of 1907（NBER Working Paper 18264）

Claims Supported by This Source

1. The Panic of 1907 was fueled by a run on New York City trust companies, the "shadow banks" of their time, but the shock that triggered the run was unrelated to the trust companies' non-financial corporate customers (Abstract).
2. In 1890, the loans of New York trust companies were half those of the city's national banks, but by 1906 they were of the same magnitude. Trust companies were chartered under laxer state laws and argued that they needed lower cash reserves because they had smaller check-clearing volumes and did not hold deposits from out-of-town banks, and thus did not join the New York Clearing House, which could act as a lender of last resort. The city's national banks were required to hold reserves of 25% of deposits, while trust companies had no reserve requirements until 1906, and thereafter 15% (of which one-third in cash) (Section 2, note 11).
3. On October 16, a run began on Mercantile National, and the Clearing House provided a loan. On October 20, the Clearing House announced the solvency of its associated member banks, conditional on the resignation of all of Mercantile's directors. On October 21, Knickerbocker announced the dismissal of its president, Barney (due to his "relationship with Mr. Morse"), and the termination of its clearing agency by the National Bank of Commerce. That night, the Clearing House refused a loan, and it closed on October 22. On October 23, the panic spread to other trust companies (Section 2).
4. Table 1: Deposits at 38 New York City trust companies fell by an average of 32.1% between August 22 and December 19, 1907. The 11 firms with connections to Morse et al. (shared directors/officers, deposits) saw a 53.4% decline, while the 27 without connections saw a 23.4% decline. The connection indicator explains over half of the variation, and this does not change when financial indicators are added (Section 2, Table 1).
5. Emergency loans led by Morgan and partially financed by the Treasury prevented further closures of trust companies, but the runs did not stop immediately, and the Trust Company of America withstood "the heaviest run that any bank in the United States had ever stood up to that time." The runs ended with a final syndicate on November 6. On October 26, the Clearing House issued loan certificates and commercial banks suspended cash payments; full resumption nationwide occurred in January 1908 (Section 2).
6. Note 1: The NBER business cycle peak was in May 1907. The following year, real GNP fell by 11%, industrial production by 16%, and the unemployment rate nearly doubled (Balke–Gordon 1986, Davis 2004, Romer 1983). The business cycle trough was in June 1908 (Section 7).
7. The stock prices of small firms connected to the most affected trust companies fell by 10.4 points, ROE by 13.1%, and dividend rates by 22%; average interest rates rose by 8.3%, and investment rates decreased by nearly 50%. The $128,500,000 reduction in investment by these firms accounts for 18.4% of the $700,000,000 reduction in investment by non-financial corporations in 1908. The effects lasted for more than five years (Abstract, Section 5).
8. Conclusion: Strict regulation encourages the growth of high-risk institutions outside the regulatory perimeter. A members-only lender of last resort (the clearing house) worsened the crisis for non-members, and the refusal to support Knickerbocker led to contagion. The rescue stopped the bank runs but could not prevent a real contraction. The runs ended in November 1907, payments resumed in January 1908, and the business cycle troughed in June (Section 7).

How to read the verification levels. "Read Through" means the source was read in its entirety. "Bibliography & Abstract Read" means the bibliographic information and abstract were confirmed. "Via Course Material" means the source was referenced through the textbook or course text. The "Text Hash" is the SHA-256 of the main text from the publisher's page, excluding tags and whitespace, used to detect if the source has been modified.