Wednesday, October 31, 2012

Historical US Unemployment Data

Step right up, and don't be shy
Because you will not believe your eyes
--The Tubes

Have been wanting to stitch together historical US unemployment rates for some time. Stitch is the operative word because this series has been subject to myriad methodological changes over the years making pure apples-to-apples comparisons impossible.

Researchers have taken various shots at reconstructing historical data in a manner that makes them comparable to the BLS U3 unemployment series extracted from the Current Population Survey. The BLS series has served as the 'headline' number since 1940, although this series itself has undergone significant methodological revisions that cast doubt about its integrity (some concerns outlined here and here).

Lebergott's (1957, 1964) studies of early US labor markets produced perhaps the most widely cited data on pre BLS unemployment estimates. Using census and other economic data along with interpolation techniques, he estimated unemployment rates during several periods in the 1800s (interesting chart here), and annual estimates from 1890 to 1940.

However, many researchers took issue with Lebergott's work, particularly because Lebergott employed far different methodologies than the BLS when generating his estimates. This led to several studies seeking to modify Lebergott's series to better coincide with the 'modern' BLS series. Romer's (1986) work is often regarded as the decisive revision of Lebergott's annual data from 1890 - 1930. Coen's (1973) estimates from 1922 - 1940 also have garnered validity.

The following graph shows the results of combining the various work.

Although the presentation suggests a uniform series, nothing could be further from the truth. This is actually a composite of 4 series:

1890 - 1930: Romer's (1986) modification of Lebergott's (1957, 1964) annual estimates
1930 - 1939: Coen's (1973) modification of Lebergott's (1957) annual estimates
1940 - 1946: BLS 'war time' estimates which count persons aged 14 and over in the civilian labor force
1947 - present: BLS U3 series which counts persons aged 16 and over in the civilian labor force

Given the academic brainpower involved in trying to 'normalize' the historical data, the combined series warrants some degree of face validity. However, there will always be limitations concerning apples-to-apples comparability. Heck, there are even legitimate questions about the extent to which the current BLS U3 accurately captures the unemployment construct.

Nonetheless, there is some utility here, I think. We'll examine in some future posts.


Coen, R.M. 1973. Labor force and unemployment in the 1920's and 1930's: A re-examination based on postwar experience. Review of Economics and Statistics, 55(1): 46-55.

Lebergott, S. 1957. Annual estimates of unemployment in the United States, 1900-1954. In The measurement and behavior of unemployment. Princeton, NJ: Princeton University Press.

Lebergott, S. 1964. Manpower in economic growth: The American record since 1800. New York: McGraw-Hill.

Romer, C. 1986. Spurious volatility in historical unemployment data. Journal of Political Economy, 94(1): 1-37.

1 comment:

dgeorge12358 said...

Only politics prevents the BLS from releasing the correct number. The lack of transparency here in the data preparation allows for direct political manipulation.
~John Williams, Shadow Statistics