Thursday, May 27, 2010

About That 'Long Depression' of the 1870s

Over at Angry Bear, Spencer asks a question:
[W]hat about the recent argument by Bryan Caplan... that the decade of 1870's was the peak for Libertarian freedom and economics[?] Maybe, but I wonder if he is even aware that economic historians label the 1870's as the "Long Depression". I find it really amusing that he so proud of what others call a depression.
I cannot speak to the historical peak of the libertarian movement, but I do want to address the claim that "economic historians label the 1870's as the Long Depression." That may have been the case in the past, but recent scholarship completely refutes this notion. According to a number of studies there simply is no evidence for a prolonged recession in the 1870s. I have covered this issue before and have the reposted that discussion below. Before turning to this discussion let me point out that Joseph H. Davis in his Journal of Economic History paper notes that the cycle dates set by the NBER in 19th century, which do show a prolonged recession in the 1870s, are flawed because they relied heavily on (1) qualitative information which tended to notice downturns more than upturns and on (2) nominal measures rather than real ones. Therefore, the NBER cycle dates for the 19th century are not reliable. Now here is my previous discussion:
About that Great Recession of 1873...it did not last until 1879 and it is not the longest U.S. economic contraction on record. One would not know this, though, by looking at the NBER's business cycle dates. These dates show this economic downturn lasted a record 65 months from October 1873 through May 1879. Therefore, it is understandable why observers like Paul Krugman, Matthew Yglesias, and Robert Shiller continue to invoke this period in their discussions of the current economic crisis. These dates, however, are wrong according to a series of papers published by Joseph H. Davis ( 2004, 2006 ). Using a new and more robust measure of industrial production for the Postbellum period, Davis shows most of the NBER recessions during this time are overstated. In the case of the 1873 downturn it only lasted 2 years. The popular Balke and Gordon (1989) real GNP series for this period similarly shows only a 2-year recession following the 1873 economic downturn while the famous Romer (1989) real GNP series shows no recession at all during this time. Unfortunately, the NBER has not revised these dates and, as a result, it continues to add confusion.
The figure below shows the log version of these three series for the Postbellum period. Nowhere in this figure is there 5 year + economic downturn in the 1870s. (Click on figure to enlarge.)


Update: Commentator ECB points us to this interesting NY Times piece on the 1870s.

15 comments:

  1. This brings back happy(?) memories of my undergrad degree. One of the questions on my Econ History final exam was "Was there a Great Depression in Britain in the late 19th century?" I wish I could remember my answer (it was 30 years ago!) But this prompted me to look up one of the books we had to read, by SB Saul. He argued in 1969 that there wasn't one. That was for the UK.

    ReplyDelete
  2. Technically it may not be a recession, but it was still one of the worse periods of economic performance in US economic history.
    Moreover, there is no definition of a depression, as for example, the Great Depression of the 1930s also had the strongest peacetime period of economic growth on record from 1933 to 1937.

    That is my point, that libertarians are bragging about how well their approach really worked in what is still a period of miserable economic performance.

    ReplyDelete
  3. Spencer, thanks for visiting. On what basis can this time be called one of the worse periods of economic expansion? The only data I know of that relates to this period are the reconstructed GNP and industrial production measures I referenced above. None of them indicate this period, other than the 1873-1874, to be an especially bad economic period. Is there something else that suggest so, other than the NBER cycle dates?

    ReplyDelete
  4. You are right, I studied economic history in the 1960s.

    But I do have a question for someone who has seen the more recent data.

    The most recent stock market data, Shiller's data for example, shows that the stock market was much more volatile prior to WW II and there is a very strong tie --not exact of course -- between the stock market and the real economy.

    How do the people who have developed the new estimates of more stable economic data reconcile this with the continued record showing much more stock market volatility prior to WW II -- or do they even attempt to?

    As a semi-retired business economist I do not have access to the JPEg sources.

    ReplyDelete
  5. The data on real per capita gdp growth that I published at Angry Bear shows the 1870s to be a period of below par economic growth.

    That data is from http://www.measuringworth.org/datasets/usgdp/result.php

    We seem to be typing past each other and replying to the comment before last rather than last.

    ReplyDelete
  6. The image on your original post is no longer there so I do not know what it shows.

    ReplyDelete
  7. Spencer,

    That is a fair point: economic growth may have been positive but below trend during the 1870s. I will take a look at that.

    On your stock market question, here are my tentative (possibly wrong) answers. First, the entire financial system itself was unstable in the Postbellum period due to the poor design of the National Banking system. The system led to periodic shortages of currency and financial panics. I assume some of the stock market volatility reflects this fact. Second, there has been more active macroeconomic management since WWII. Maybe this has kept volatility down. (Though the recent crisis raises questions to me as to how well this has really worked.)

    Also, I would note that though the data suggests the 1870s period may be better than originally thought there is mixed evidence for the entire pre-WWII period. The Romer series shows overall improvement for the period while the Balke and Gordon series shows little change.

    ReplyDelete
  8. When I joined the investment business in the 1970s it was just the standard assumption that there was a roughly a four year economic -stock market cycle, and this was a big improvement over the pre WW II era.

    This held up very well from WW II to the start of the great moderation in the early 1980s.

    I actually published a paper in Business Economics -- the only one I ever did -- where I showed that the longer frequency of economic -stock market expansions after 1983 was worth a few extra points on the stock market PE.
    This made the market closer to reasonably valued in the late 1990s, rather than significantly overvalued using the traditional relationship between interest rates and PEs.

    ReplyDelete
  9. Charles Morris (is this the same one that wrote Trillion dollar meltdown) had a very pertinent article in the NY Times back in
    2006 on this era:
    http://www.nytimes.com/2006/06/02/opinion/02morris.html

    ReplyDelete
  10. David,
    Very relevant post as Scott Sumner and I are currently quibbling over the Long Depression.

    Davis’ revisions of the manufacturing data strongly suggest that the NBER needs to revise its dates for business cycles from that period.

    Going by Balke and Gordon’s data on GNP (probably the current accepted standard) I estimate that real GNP was about 13% below trend in 1878 which is probably consistent with about 10% unemployment. Thus in terms of peak labor market distress it was probably comparable to the recession of 1982-83 or our current one. Thus it ranks among the top five recessions in US history (the others being the Great Depression and the Panic of 1893 of course).

    Mark A. Sadowski

    ReplyDelete
  11. Hey Mark, take a look at link provided by ECB. I now have it referenced above in the post as an update.

    So where are you and Mark discussing this issue?

    ReplyDelete
  12. David,
    Thanks, I've read that article.

    Scott and I have gone back and forth over the Long Depression here:

    http://www.themoneyillusion.com/?p=5248

    He pointed out this post there. I don't think we're that far apart (any of us).

    Mark

    ReplyDelete
  13. Yes, the NBER does record a long contraction in that period (after 1873). The NBER peak is listed as October 1873 and the trough is March of 1879. However, other sources disagree. The other sources are reasonably consistent and appear to reflect considerable scholarship. A few other sources are:

    1. Measuring Worth provides a real GDP series from 1790 to present (http://www.measuringworth.org/usgdp/). The data sources are described in detail at http://www.measuringworth.com/uscompare/sourcegdp.php. Notably, the data from 1869 to 1909 is from Gallman. See "Gallman, Robert E. "Gross National Product in the United States, 1834-1909." In Output, Employment, and Productivity in the United States After 1800, Dorothy S. Brady, editor, 3-76. New York: Columbia University Press (for NBER), 1966.". This series only shows one year of actual contraction (1875) and the decline is very small.

    2. Angus Maddison (now deceased) provides real GDP data for many countries going back centuries. See http://www.ggdc.net/maddison/ for the data in Excel spreadsheet format. Maddison only shows one year of actual contraction (1874).

    3. Historical Statistics of the United States has many GNP series. See http://hsus.cambridge.org/HSUSWeb/toc/showTable.do?id=Ca169-240 for a list. The standard series (Ca208-212) all show a significant decline from 1873 to 1874 and GNP (not GDP) growth thereafter. This set of series is attached as a PDF.

    4. Another source, "Banking Panics of the Gilded Age" discusses the economy of the 1870s in some detail. GNP statistics for each year are provided based on Romer and Balke-Gordon. See table 2.7 on page 30. A copy of that table is attached. Balke-Gordon shows a decline in GNP from 1873 to 1874, Romer does not. Note that this book discusses the Panic of 1873 in some detail and the author, Elmus Wicker, disputes the conventional interpretation of a major economic decline.

    ReplyDelete
  14. Yes, the NBER does record a long contraction in that period (after 1873). The NBER peak is listed as October 1873 and the trough is March of 1879. However, other sources disagree. The other sources are reasonably consistent and appear to reflect considerable scholarship. A few other sources are:

    1. Measuring Worth provides a real GDP series from 1790 to present (http://www.measuringworth.org/usgdp/). The data sources are described in detail at http://www.measuringworth.com/uscompare/sourcegdp.php. Notably, the data from 1869 to 1909 is from Gallman. See "Gallman, Robert E. "Gross National Product in the United States, 1834-1909." In Output, Employment, and Productivity in the United States After 1800, Dorothy S. Brady, editor, 3-76. New York: Columbia University Press (for NBER), 1966.". This series only shows one year of actual contraction (1875) and the decline is very small.

    2. Angus Maddison (now deceased) provides real GDP data for many countries going back centuries. See http://www.ggdc.net/maddison/ for the data in Excel spreadsheet format. Maddison only shows one year of actual contraction (1874).

    3. Historical Statistics of the United States has many GNP series. See http://hsus.cambridge.org/HSUSWeb/toc/showTable.do?id=Ca169-240 for a list. The standard series (Ca208-212) all show a significant decline from 1873 to 1874 and GNP (not GDP) growth thereafter. This set of series is attached as a PDF.

    4. Another source, "Banking Panics of the Gilded Age" discusses the economy of the 1870s in some detail. GNP statistics for each year are provided based on Romer and Balke-Gordon. See table 2.7 on page 30. Balke-Gordon shows a decline in GNP from 1873 to 1874, Romer does not. Note that this book discusses the Panic of 1873 in some detail and the author, Elmus Wicker, disputes the conventional interpretation of a major economic decline.

    The first three data sources listed above also provide GDP/GNP data for the 1890s. The downturn in the 1890s (1892-1894) appears to have been much deeper and followed quickly by another contraction.

    The Measuring Worth series shows a two year economic contraction in the 1890s (1892-1894) and it is much deeper (GDP falls from $339.3 to $304.5 in 2005 dollars). Another decline occurs from $339.2 billion in 1895 to $333.6 billion in 1896.

    Angus Maddison also shows an economic contraction from 1892 to 1894 and it is much deeper (GDP falls from $245.76 billion in 1892 to $227.13 billion in 1894 measured in 1990 international Geary Khamis dollars). Another decline occurs from $254.55 billion in 1895 to $249.38 billion in 1896.

    The series in the Historical Statistics of the United States all show a substantial contraction from 1892 to 1894. Some of the series (Ca211 and Ca212) show another decline after 1895. One series (Ca208) does not. However, even Ca208 shows almost flat output between 1895 to 1896.

    Yet another source provides some perspective on the economy in the 1870s and 1890s. See http://www.nber.org/data/industrial-production-index/ip-total.html for an industrial production index from 1790 to 1915. Industrial production does fall from 1873 to 1875. Thereafter it grows. By contrast, the fall in industrial production after 1892 is greater and a second dip after 1895 is also apparent.

    Let me offer one more indirect set of data points. Both the crisis of 1873 and 1892 triggered shifts in U.S. politics. However, the political impact of the crash of 1892 was much greater. The Republicans gained control of House in 1894 and essentially controlled the Federal government until the Depression. No comparable shift occurred after 1873.

    ReplyDelete
  15. The first three data sources listed above also provide GDP/GNP data for the 1890s. The downturn in the 1890s (1892-1894) appears to have been much deeper and followed quickly by another contraction.

    The Measuring Worth series shows a two year economic contraction in the 1890s (1892-1894) and it is much deeper (GDP falls from $339.3 to $304.5 in 2005 dollars). Another decline occurs from $339.2 billion in 1895 to $333.6 billion in 1896.

    Angus Maddison also shows an economic contraction from 1892 to 1894 and it is much deeper (GDP falls from $245.76 billion in 1892 to $227.13 billion in 1894 measured in 1990 international Geary Khamis dollars). Another decline occurs from $254.55 billion in 1895 to $249.38 billion in 1896.

    The series in the Historical Statistics of the United States all show a substantial contraction from 1892 to 1894. Some of the series (Ca211 and Ca212) show another decline after 1895. One series (Ca208) does not. However, even Ca208 shows almost flat output between 1895 to 1896.

    Yet another source provides some perspective on the economy in the 1870s and 1890s. See http://www.nber.org/data/industrial-production-index/ip-total.html for an industrial production index from 1790 to 1915. Industrial production does fall from 1873 to 1875. Thereafter it grows. By contrast, the fall in industrial production after 1892 is greater and a second dip after 1895 is also apparent.

    Let me offer one more indirect set of data points. Both the crisis of 1873 and 1892 triggered shifts in U.S. politics. However, the political impact of the crash of 1892 was much greater. The Republicans gained control of House in 1894 and essentially controlled the Federal government until the Depression. No comparable shift occurred after 1873.

    ReplyDelete