Manpower Survey¶
165,583 rows of employment by sector, occupation, race and gender, 1965–1994, for the sectors the survey covered.
Use the summary panel for totals
The file is a copy of the Manpower Survey data tapes, not a transcription of the printed reports, and no printed report is held for a year-by-year check. It covers the surveyed sectors only (agriculture only in 1965) and has breaks at 1969–71 (coverage), 1985–87 (the first CSS sample) and 1988–89 (the cut to 12,800 firms). Sector 610 is all wholesale and retail trade (plus catering from 1987) and 622 is the motor trade. For employment totals use panels/manpower_survey_summary.csv (year × broad sector × race, with the tape totals, the totals after the ledger's adjustments, and coverage notes), built by 14_manpower_survey_panel.py with the corrections ledger corrections/manpower_survey.csv. Gold and manufacturing are benchmarked against the Chamber of Mines and the Census of Manufacturing in panels/manpower_survey_checks.csv.
Unit of Observation¶
Each row is a (sector, occupation, year) combination with employment counts broken down by race and gender.
File¶
| File | Rows | Description |
|---|---|---|
manpower_survey/manpower_survey_1965_1994.csv |
165,583 | Full panel |
Variables¶
| Column | Description |
|---|---|
year |
Year (65-94, i.e. 1965-1994) |
sector |
Economic sector (39 unique) |
occupation |
Occupational category (754 unique) |
white_men |
White male employment count |
white_women |
White female employment count |
coloured_men |
Coloured male employment count |
coloured_women |
Coloured female employment count |
Indian_men |
Indian male employment count |
Indian_women |
Indian female employment count |
black_men |
Black male employment count |
black_women |
Black female employment count |
vacancies |
Vacancy count |
Research Applications¶
- Labor market segmentation: Which sectors reserved jobs for whites? How did this change?
- Structural transformation: Shift from mining/agriculture to manufacturing/services
- Racial employment gaps: Trends in black employment access by sector and skill level
- Job reservation erosion: When did racial barriers break down in specific occupations?
Geographic Note¶
This dataset has no geographic disaggregation — it is national-level data by sector and occupation. It cannot be joined to the district panel. Use it for national-level time series.
Provenance¶
- Source: the Manpower Survey of the Department of Labour (later Department of Manpower) and the Central Statistical Service
- What the file is: a copy of the survey's data tapes, the machine-readable files Crankshaw (1996) used, held as a Stata file dated 28 November 2012. It was supplied by Andrew Kerr of DataFirst (University of Cape Town) on 8 January 2013, who had obtained it under a PEDL grant; he placed no restrictions on its use, and the survey was to be made public, as it later was. No script that produced it survives. It carries a variable
G9(Kerr thought it might be an "other" race-gender category) that DataFirst's public release deletes, so it is an earlier copy than that release - Public release: DataFirst (University of Cape Town) distributes the survey as Manpower Survey 1965–1994, version 1.2 (29 June 2015), a public-use file available after login. Researchers should take the microdata from DataFirst; this archive does not redistribute its copy
- Converted to CSV: 2026-03-28; summary panel and benchmarks built 2026-10-20