Lost and Found: Ancient BEA County Personal Income Data (1929, 1940, 1950, 1959, 1962, 1965, 1967-1984)

A key data set used in my dissertation on highways and local economic growth about 25 years ago was a personal income and earnings by industry file of U.S. counties that covered the early beginnings of BEA data collection efforts.  Since that time, the data has vanished from the public domain without a trace. It’s been replaced with the current 1969-2000 and 2001-2017 local personal income series.  The latter series use different regional accounting methodologies than the earlier series.  So, the two series 1929-1984 and 1969-2000/2001-2017 can’t simply be spliced together.

I am posting the data here it the hopes that somebody finds it useful (e.g., historians, regional scientists, members of the general public).   I discovered an old SAS version of the file on a 3-1/2 inch floppy disk, which was fortunately still convertible into a more accessible CSV (comma-separated values) format.  Unfortunately, the original suppression codes were not retained, either because my data conversion did not preserve the missing values, the missing value indicators were never available in the first place, or missing values were never properly written out into the SAS file.  Therefore, data users will need to realize that a field zero represents either a real zero or a field that was suppressed for confidentiality reasons.  If it is a real zero, the values for nested industry earnings categories will add up to the earning totals.  If not, there will be residuals.  In many instances, you will be able to reasonably estimate the missing values by a combination of summing to totals and interpolating over time.

A few notes about the file layout.  The SAS file contained sector numbers but no actual sector descriptions.  I was able to identify the descriptions using several different sources and inserted them into CSV file.

The column headings y29-y84 represent years.  That is to say, y29 is income, population and earnings data for 1929, y40 is income, population and earnings, . . .

Fipstate is the fips state number

Fipcount is the fips county number

Fipcode is a unique county code.  It is mathematically defined as 1,000Xfipstate+fipcount

Locality is the locality (i.e., county, city, parish) name.

Secnum is the sector number.  It is somewhat different from the current BEA line code.  Moreover, the industrial classification system is SIC (Standard Industrial Classification System), the predecessor to the NAICS (North American Industry Classification System) used now.  However, the BEA line code is unique to BEA.

Earnings and income figures are expressed in thousands of dollars.  The exceptions are per capita income (actual value) and population (hundreds of dollars).

Good luck.