Skip to content

Post-Apartheid Patterns of Internal Migration in South Africa

Literature › Works

Pieter Kok; Michael O'Donovan; Oumar Bouare; Johan van Zyl (2003). Post-Apartheid Patterns of Internal Migration in South Africa. HSRC Publishers (Human Sciences Research Council), Cape Town.
Type book
Scope Post-1994 South African economy
Method mixed
Period studied 1975–1997
Regions South Africa, Gauteng, Eastern Cape, Limpopo, Western Cape, former homelands (Transkei, Ciskei, Lebowa etc.)
Themes demography urbanisation, bantustans removals influx, transition post apartheid, methods data sources, mining migrant labour, inequality poverty unemployment, comparative theory
Citation key kok2003

Argument

Using Census '96 (the first census with national internal-migration data) the book shows that post-apartheid migration patterns largely continue pre-1994 patterns: overall migration rates barely changed between 1975-80 and 1992-96 despite the abolition of influx control and group areas. Only about a quarter of the population has ever migrated, and 'migrants proper' tend to be better-educated, urban and better-off, whereas labour migrants are low-income, low-education, male and African. Former homelands remain labour reservoirs, and their residents are less deterred by distance than others. Economistic theories (neo-classical, dual labour market, world systems) fit poorly; the authors review non-economic models (value-expectancy, networks) and conclude that census data and their own gravity, MCA and MNA models explain little, so purpose-made survey data are needed.

Evidence and method

Census '96 Migration Community Profile (district-level) and the 10 per cent sample of Census '96; 5 per cent sample of the 1980 census (excluding Transkei, Bophuthatswana, Venda); 1997 October Household Survey (c. 30 000 households); 1996 GGP and SAPS crime statistics for the gravity model; Reserve Bank employment figures; secondary literature on migration theory.

Findings

  • Census '96 was the first national source of internal migration data; the earlier census data on migration were partial. (p. 19)

    Only Census ’96 covered migration data for the entire country and provided useful data on movement patterns.

  • Migration levels barely changed between the late 1970s and early 1990s despite the abolition of apartheid's migration controls. (p. 72)

    the overall level of migration between the late 1970s and early 1990s did not change significantly. In fact, the period saw a slight overall decline in rates from about 13 per cent to 11 per cent.

  • Only about a quarter of South Africans have ever migrated across district boundaries. (p. 71)

    on average, only about a quarter (24 per cent) of the population has ever migrated across district boundaries.

  • Gauteng dominated inter-provincial migration, with the Eastern Cape the main loser. (p. 56)

    Gauteng experienced a net gain of more than a quarter of a million people (262 000), while the Eastern Cape lost a roughly similar number of people (206 000).

  • Eight largest sources of labour migrants are metropolitan areas, challenging the rural stereotype. (p. 58)

    the eight largest sources of migrant workers are in metropolitan areas – bringing into question the stereotyping of labour migrants as predominantly rural residents.

  • Distance deters migration strongly outside the former homelands but is only loosely related to migration from the former homelands. (p. 65)

    The dispersion is indicated by an r2 of 0,03 – only 3 per cent of the difference in one variable can be explained by changes in the other

  • Who migrates contradicts popular and economic-model expectations: migrants proper are better educated and urban; only labour migrants fit the stereotype. (p. 81)

    The analysis above indicates that while this may be true for ‘labour migrants’, the profile of ‘migrants’ is a diametrical opposite.

  • High unemployment districts have low in- and out-migration, suggesting people in poor rural areas are trapped. (p. 59)

    high unemployment rates are associated with low rates of migration both into and out of a district, i.e. the level of population movement in respect of low-employment areas is low.

  • Formal employment fell sharply in the 1990s, especially in mining, construction and manufacturing, hitting the unskilled hardest. (p. 75)

    from the high in the employment cycle in 1989 to 1997, the average annual level of employment has now declined by 7,5 per cent, indicating a loss of about 422 000 formal sector jobs.

  • The multivariate MCA model explains little of the variation in migrant status, so important explanatory factors are missing from census data. (p. 82)

    this multivariate analysis explains only 12 per cent of the variation, which means that more than seven-eighths of the variation in the population is not explained by this model.

  • Labour migrants are probably severely undercounted in the Census '96 sample relative to the October Household Survey. (p. 110)

    Whereas the 10 per cent sample indicated that there were approximately 1,16 million labour migrants in 1996, the 1997 October Household Survey revealed that there were then over 2,17 million labour migrants.

  • In the 1997 OHS, about half of moves were work-related, but the reasons model explains under 5 per cent of variation. (p. 122)

    the majority (50 per cent or more) of migratory moves are reportedly caused by work-related reasons. However, other (non-work-related) reasons make up the other half of the reasons for moving.