Historical Chart of U.S. Recessions and Bank Panics

Jeffrey Rogers Hummel emails:

Some of you may be interested in the attached historical chart I use in several of my classes. It lists U.S. recessions, bank panics, and periods of bank failures from the nineteenth century to the present. The list distinguishes between the three categories and varies somewhat from other lists compiled by economists and historians. I've never been fully satisfied with the general approach of either discipline. Economists, as their studies go back in time, have a tendency to rely on highly unreliable data series that exaggerate the number of recessions and panics, something most strikingly but not exclusively demonstrated in the notable work of Christina Romer. Historians, on the other hand, relying on more anecdotal and less quantitative evidence, tend to exaggerate the duration and severity of recessions. What I have tried to do is integrate the best of both approaches, using them to cross check each other. The chart is documented with two pages of extensive notes in which I list references and explain my listing and dating.

One value of the chart is that gives a visual presentation of which recessions were accompanied by bank panics and which were not. Equally important, it distinguishes between bank panics and periods of significant numbers of bank failures. These two categories are the most often confused or conflated, and yet this distinction is critical. Not all bank panics (periods of contagious runs and sometimes bank suspensions) were accompanied by numerous bank failures, nor were all periods of of numerous failures accompanied by panics. In a follow-up message, I will provide an Excel file showing the data I used for identifying periods of numerous failures during the national banking era.

Among other advantages, the chart helps highlight how sui generis the Great Depression was. Not only does the Great Depression have the longest downturn (43 months) but it is one of the few depressions accompanied by both a bank panic and numerous bank failures.

Comments and suggestions are welcome.

*********
Jeffrey Rogers Hummel
Department of Economics
San Jose State University
P.O. Box 4644
Walnut Creek, CA 94596
jeff@jrhummel.com

U.S. RECESSIONS/BANKING CRISES

Date Major Recession (months)1 Bank Panic or Suspension2 Numerous Bank Failures3

1812 none: war–related Aug 1814–Jan 1817

1819 1818–1820 Apr–May 1819

1837 1837–1838 May 1837

1839 none Oct 1839–Mar 1842 1839–1841 (state defaults)

1857 Jun 1857–Dec 1858 (18) Oct 1857

1861 none: war–related Dec 1861

1873 Oct 1873–Dec 1875 (27) Sep 1873

1893 Jan 1893–Jun 1894 (17) Jun–Aug 1893 1890–1896 (agricultural)

1907 May 1907–Jun 1908 (13) Oct 1907

1913 Jan 1913-Dec 1914 (23) Aug–Oct 1914 (no suspension)

1920 Jan 1920–Jul 1921 (18)

1921–1930 (agricultural)

1929 Aug 1929–Mar 1933 (43) Oct 1930–Mar 1933 1931–1933

1937 May 1937–Jun 1938 (13)

1948 Nov 1948–Oct 1949 (11)

1953 Jul 1953–May 1954 (10)

1957 Aug 1957–Apr 1958 (8)

1960 Apr 1960–Feb 1961 (10)

1969 Dec 1969–Nov 1970 (11)

1973 Nov 1973–Mar 1975 (16)

1980 Jan 1980–Jul 1980 (6)

1981 Jul 1981–Nov 1982 (16)

1980–1991 (S & L crisis)

1990 Jul 1990–Mar 1991 (8)

2001 Mar 2001–Nov 2001 (8)

2007 Dec 2007–June 2009 (18) Aug 2007-Dec 2008 2008 (investment banks)

[1] I have almost entirely confined the list of major recessions to those constituting part of a standard business cycle, omitting periods of economic dislocation resulting from U.S. wars, government embargoes, and other obvious supply-side shocks. For the number and dating of recessions from 1948 forward, I have exclusively followed the National Bureau of Economic Research (NBER). But prior to 1929, the NBER notoriously exaggerates the volatility of the U.S. economy, as first revealed by Christina Romer, “Is the Stabilization of the Postwar Economy a Figment of the Data?” American Economic Review 76 (June 1986): 314-34, and further documented by her in “The Prewar Business Cycle Reconsidered: New Estimates of Gross National Product, 1869-1908,” Journal of Political Economy 97  (February 1989): 1-37, and “New Estimates of Prewar Gross National Product and Unemployment,” Journal of Economic History 46 (June 2009): 341-52. Moreover, even after 1929, the NBER reports a post-World War II recession lasting from February to October 1945 that no one was aware of at time, as easily confirmed by looking at the unemployment data. Richard K. Vedder and Lowell E. Gallaway point out in Out of Work: Unemployment and Government in Twentieth-Century America (New York: Holmes & Meier, 1993), pp. 153-157, that this alleged postwar recession is a statistical artifact that varies in severity with the regular comprehensive revisions of GNP-GDP estimates by the Bureau of Economic Analysis (BEA). The BEA’s original estimates showed only a minor downturn, subsequent revisions converted it into a major downturn, and the latest comprehensive (2013) revisions have reduced its magnitude, although not to the level of the BEA’s original estimates.

Prior to 1929, therefore, I have only listed recessions that can be documented with unemployment data or more traditional historical evidence from contemporaries. The unemployment data I have employed are the revisions of J. R. Vernon, “Unemployment Rates in Postbellum America, 1869-1899,” Journal of Macroeconomics 16 (Autumn 1994): 701-14 and Romer, “Spurious Volatility in Historical Unemployment Data,’ Journal of Political Economy 94 (February 1986): 1-37.  I have still accepted the monthly NBER dating, which only goes back to 1857, for those pre-1929 recessions that I consider genuine, with the notable exception of 1873. In that case, the NBER dating (based on the Kuznets-Kendrick series) of a 65-month duration is so inconsistent with other evidence that it was even questioned by Milton Friedman and Anna Jacobson Schwartz in a Monetary History of the United States, 1867-1969 (Princeton, NJ: Princeton University Press, 1963), pp. 40-44, 87-88. This is one of the most striking cases in which some observers at the time and many economists today have confused mild secular deflation with a depression, a confusion exposed by George Selgin, Less than Zero: The Case for a Falling Price Level in a Growing Economy (London: Institute of Economic Affairs, 1997). Even the Kuznets estimates show no decline in real net national product during this recession, and an acceleration of its growth after 1875. I have therefore accepted the NBER dating only for the recession’s beginning (Oct 1873). For dating its end, I have relied upon Jospeh H. Davis, “An Improved Annual Chronology of U.S. Business Cycles since the 1790s,” Journal of Economic History 66 (March 2006): 103-12, who uses Romer's methodology to revise NBER dating, but only on an annual basis. This makes Dec 1875 rather than Mar 1879 the end of the 1873 recession. (To be sure, on a per capita basis real GDP did fall during this recession but was back to the 1873 level by 1876 according to the Romer series. One of the three series provided in Historical Statistics of the United States, Earliest Times to the Present: Millennial Edition, edited by Susan B. Carter, et. al. (New York: Cambridge University Press, 2006) shows real GDP per capita recovering to the 1873 level within a year; the other two show it not recovering until 1876.)

Estimates of U.S. GDP prior to the Civil War are even more problematic, making precise monthly dating of recessions impossible. Writing under NBER auspices, Geoffrey H. Moore and Victor Zarnowitz, “The Development and Role of the National Bureau of Economic Research’s Business Cycle Chronologies,” in The

American Business Cycle: Continuity and Change, ed. by Robert J. Gordon (Chicago: University of Chicago Press, 1986), p. 746, attempts an annual dating of recessions for this period. Their chronology is conveniently summarized in David Glasner, Business Cycles and Depressions: An Encyclopedia (New York: Garland, 1997), p. 731. But Davis’s more realistic revision of the Moore-Zarnowitz pre-Civil War series is even too promiscuous in the number and dating of recessions. So I have relied upon the consensus of standard historical accounts along with the GDP statistics in Historical Statistics: Millennial Edition to determine what qualifies as an actual recession and its annual dating. The one case where I diverge from some (but not all) mainstream historical accounts is the alleged recession during the banking crisis that began in 1839, after the recovery from the 1837 recession. As Friedman and Schwartz, Monetary History, p. 299; Douglass C. North, The Economic Growth of the United States, 1790–1860 (New York: Prentice Hall. 1961), p. 202; and Peter Temin, The Jacksonian Economy (New York: W.W. Norton. 1969), p. 157, have all noted, estimates of real GDP growth over the next four years are quite robust. Thus, 1839-1843 appears to be another case were deflation (in this case, quite severe) is confused with depression. (Admittedly, the estimates of real GDP per capita in Historical Statistics: Millennial Edition do show a decline after 1838.)

[2] The number of bank panics is also often exaggerated. For the post-Civil War period, many authors follow the listing first compiled by O. M. W. Sprague in History of Crises Under the National Banking System (Washington: Government Printing Office. 1910), and some even add in a few more. But Elmus Wicker, in Banking Panics of the Gilded Age (Cambridge: Cambridge University Press, 2000) persuasively demonstrates that the alleged Panics of 1884 and 1890 were really only incipient financial crises nipped in the bud by the actions of bank clearinghouses. For the pre-Civil War period, especially egregious in its listing of panics is Willard Long Thorp’s widely cited Business Annals . . . (New York: National Bureau of Economic Research, 1926), which even mistakenly attributes to the United Sates panics that affected only England (1825 and 1847). For an excellent survey and critique of alternative lists of U.S. bank panics, see Andrew J. Jalil, ”A New History of Banking Panics in the United States, 1825-1929: Construction and Implications,“ American Economic Journal: Macroeconomics, 7 (July 2015): 295-330, along with its online appendix. I have confined my own list to those panics that Jalil defines as ”major,“ with two exceptions. I have omitted the very minor economic contraction of 1833, following Andrew Jackson’s phased withdrawal of government deposits from the Second Bank of the United States, since the impact on banks was almost entirely confined to the Second Bank and its branches. I have, however, included the more pronounced global financial crisis at the outbreak of World War I, in which the U.S. stock market was shut down for four months, although the emergency currency authorized under the Aldrich-Vreeland Act prevented any bank suspension. For the monthly dating of those panics listed, I have relied not on Jalil but on the more conservative dating of Charles W. Calomiris and Gary Gorton, “The Origins of Banking Panics: Models, Facts, and Bank Deregulation,” in Financial Markets and Financial Crises, ed. by in R. Glenn Hubbard (Chicago: University of Chicago Press, 1991), p. 114.

[3] Bank panics, even when accompanied by numerous suspensions (or what Friedman and Schwartz prefer to call “restrictions on cash payments” to distinguish them from government suspensions of redeemability}, do not always result in a major number of bank failures. For instance, Calomiris and Gorton, p. 153. report the failure of only six national banks out of a total of 6412 during the Panic of 1907, or less than 0.1 percent. Of course the Panic of 1907 was concentrated among state banks and trust companies. Unfortunately, as far as I can tell, there are no good time series on the failures of state banks for the period prior to the creation of the Federal Reserve. Yet there were over 12,000 state banks at the outset of the Panic of 1907. One very incomplete estimate of total bank suspensions, state and national, during that panic was 153. Even if all suspensions had resulted in failures, which of course did not happen, we still have a failure rate of 0.7 percent for all commercial banks. Confusion of bank suspensions with bank failures can even infect serious scholarly work, such as Michael D. Bordo and David C. Wheelock, “Price Stability and Financial Stability: The Historical Record,” Federal Reserve Bank of St. Louis Review, 80 (September-October 1998): 41-62, which in its charts showing bank failures is clearly depicting instead statistics on the annual number of bank suspensions.

Similarly, periods of numerous bank failures do not always coincide with bank panics, as the S & L crisis dramatically illustrates. So it is crucial to distinguish between bank panics and periods of numerous bank failures, although specifying the latter requires judgment calls. In addition to traditional historical sources and works cited in previous notes, I have relied upon U.S. Comptroller of the Currency, Annual Report, December 7, 1914 (Washington: Government Printing Office, 1915), v. 2, Table 35, pp. 66-103; Historical Statistics of the United States, Colonial Times to 1970 (Washington: U.S. Bureau of the Census, 1975), v. 2, Series X 741-755; Historical Statistics: Millennial Edition, v. 3, Banks, Series Cj149-157, 158-168, 203-211, 212-224, 251-264. Graham Newell, a former graduate student of mine, compiled data series from these sources for 1871 to 1915 into an Excel file with graphs. Unfortunately, prior to the creation of the Federal Reserve, none of these series are entirely consistent or comprehensive. Particularly “fragmentary and incomplete” is the pre-1921 series on bank suspensions, appearing in the 1975 edition of Historical Statistics, and based on U.S. Comptroller of the Currency, Annual Report, 1931, p. 1040; Board of Governors of the Federal Reserve System, Banking and Monetary Statistics, pp. 283, 292; and U.S. Federal Deposit Insurance Corporation, Annual Report, 1934, pp. 112-113. This is probably why that particular series was omitted from the millennial edition of Historical Statistics.