NBER Macrohistory Database
During the first several decades of its existence, the National Bureau of Economic Research (NBER) assembled an extensive data set that covers all aspects of the pre-WWI and interwar economies, including production, construction, employment, money, prices, asset market transactions, foreign trade, and government activity. Many series are highly disaggregated, and many exist at the monthly or quarterly frequency. The data set has some coverage of the United Kingdom, France and Germany, although it predominantly covers the United States. For information see Improving the Accessibility of the NBER's Historical Data , by Daniel Feenberg and Jeff Miron. (NBER Working Paper #5186). Published in the Journal of Business and Economic Statistics, Volume 15 Number 3 (July 1997) pages 293-299.
An analytical index of the NBER database was kept in hard copy form. In some respects it is more informative than the online index, but it is available as a 9-megabyte .PDF file only. It lists a few series that are not available here, and misses a few that are.
The data files are offered in two ASCII formats: Micro-TSP .db format and rectangular .dat format. The data and comment lines in the .db format are together in one file and can be read automatically by time series software packages such as TSP, Micro-TSP, and RATS. The rectangular .dat format makes reading the data using general-purpose statistical software packages easier. The comment lines have been extracted from the data and placed into accompanying .txt files. Jean Roth created the rectangular data format from the .db files in May 2001, so please check the .db file for comparison and report any errors to email@example.com
The .db format begins with comment lines documenting the series. After the text comments the next line is a '-1' for annual data, '-4' for quarterly data and '-12' for monthly data. The next two lines show the start and end date for the series in year.subperiod format. For example 1945.08 stands for August of 1945. For a quarterly series 1945.2 would stand for the second quarter of 1945. The remainder of the file are ASCII numeric values with one observation per line. Missing values are coded as "NA".
[Alan G Isaac has posted documentation for the .db format including the correct representation for missing values, which was incorrect on this site before August 2005. We are gratefull to him for his help.
The rectangular .dat files have a column for the year, a column for the quarter or month, if necessary, and a column of data. The NA missing value codes have been changed to a "." The comment lines are placed in an accompanying .txt file.
Information about seasonal adjustments is available, but in most cases only unadjusted series have been made available here.
Also of possible interest
Last modified April 10, 2008