How are Standard Industrial Classification (SIC) codes collected?
CRSP does not actively assign SIC codes. We obtain NASDAQ SIC codes directly from the NASDAQ exchange and obtain NYSE, Amex, and ARCA SIC codes from Interactive Data (formerly FT Interactive Data). All data providers have told CRSP that they reference SEC documents as SIC code sources.
SIC codes can be useful for rough groupings of industries. Beyond that they should be used with caution—they are not assigned or reviewed with a strict procedure by any government agency, most large companies belong in multiple SIC codes, and they change over time. After the initial SIC code assignment when a company goes public, no government agency ever looks at that code or the company again—quite often a company reports its initial SIC code forever. There are cases in which companies would have obsolete SIC codes from the 1972-coding scheme in their SEC filings from the early 1990s.
The U.S. Department of Labor provides information on the 1987 SIC coding scheme at http://www.osha.gov/oshstats/sicser.html.