Skip to content

The Geographical Economy of South Africa

Literature › Works

Krugell, Waldo F. (2005). The Geographical Economy of South Africa. PhD thesis in Economics, Potchefstroom Campus, North-West University (promoter: W.A. Naude). North-West University, Potchefstroom.
Type thesis
Scope Post-1994 South African economy
Method econometric
Period studied 1998–2002
Regions South Africa, 354 magisterial districts, six metropolitan areas (Johannesburg, East Rand, Pretoria, Cape Town, Durban, Port Elizabeth)
Themes demography urbanisation, comparative theory, methods data sources, inequality poverty unemployment, manufacturing industrial policy, transition post apartheid, bantustans removals influx, origins segregation
Citation key krugell2005

Argument

The thesis asks whether geography, institutions or policy drive growth across South African localities. It tests the hypothesis that geography matters independently of institutions. It reviews the geographical economics (new economic geography) framework, sketches the historical origins of spatial inequality, and describes city size and growth patterns. It then estimates GLS random-effects and Arellano-Bond dynamic panel growth regressions for 354 magisterial districts over 1998-2002. Distance from internal markets (Johannesburg), human capital, export share and the local capital stock significantly affect growth. Distance to ports and urban density work indirectly by raising export propensity. There is no absolute convergence, only tentative evidence of very slow conditional convergence, and sigma-divergence on one measure.

Evidence and method

Regional Economic Focus database (Global Insight Southern Africa) of GVA by magisterial district, built from Statistics SA, the 1996 census, October Household Surveys and Labour Force Surveys. Capital stock estimates from Econometrics, distances calculated with SA Explorer software, rainfall from the South African Weather Service. Secondary literature on spatial policy (SDIs, IDZs, local government demarcation) and city-size statistics (primacy ratio, rank-size, H-measure). Stata 8.0 estimation.

Findings

  • Economic activity is highly concentrated in a minority of places (the introduction gives 70 per cent of GDP; Chapter 4 gives 82 per cent, an unreconciled inconsistency). (p. 11)

    In South Africa, 70 per cent of GDP is produced in only 20 per cent of places.

  • Spatial inequality in income per capita across places is large. (p. 62)

    The richest 20 per cent of places had an average per capita income in 2000 of R25277 compared to an average per capita income of R5452 of the poorest 20 per cent of places.

  • Apartheid reinforced the spatial pattern established by the mineral-energy complex through its homeland and urban segregation policies. (p. 65)

    In the 20th century Apartheid reinforced the historical regional development patterns induced by the emerging mineral-energy complex of the 19th century with its homeland policies and Group Areas Act.

  • Apartheid spatial policy segmented markets and raised costs. (p. 65)

    It also resulted in segmented labour and consumption markets and created artificial internal barriers to trade.

  • The six metros hold about 31 per cent of the population but produce 55 per cent of GDP. (p. 79)

    Table 4.1 shows that South Africa's six major cities, with 31 per cent of South Africa's total population, contribute, in total, 55 per cent of South Africa's GDP.

  • The rank-size rule fits South African cities but Zipf's Law is rejected, indicating relatively even city sizes. (p. 82)

    Thus, whilst the rank-size rule applies to South Africa (its explains 98 per cent of the variance in city sizes) Zipf's Law does not hold for South Africa.

  • Growth was very dispersed across districts, with the slowest-growing places clustered in former gold-mining centres and rural areas. (p. 86)

    Table 4.3 shows that the 25 slowest growing cities and towns in South Africa contracted at an average rate of -5.66 per cent per annum over the period 1998 to 2002.

  • Unconditional regression shows no absolute convergence across districts, though the fit is poor. (p. 93)

    The positive slope of the regression line is a clear sign of divergence.

  • Denser, more urbanised districts are more export-oriented, so density and distance to ports matter indirectly for growth. (p. 98)

    It was also found that population density was significantly and positively correlated with export shares, and entered with a coefficient of 0.20.

  • The dynamic GMM model identifies human capital, capital stock, exports and distance from Johannesburg as significant growth determinants. (p. 99)

    Education, the initial capital stock, export share, and distance from Johannesburg (access to markets / transport costs) are shown to be significant determinants of the growth rate.

  • There is conditional convergence in the dynamic model, but at a very low rate. (p. 99)

    The coefficient of the initial income per capita shows evidence of conditional convergence, albeit at a very low rate.

  • Sigma-convergence tests show slight divergence in income dispersion across districts. (p. 100)

    When using the coefficient of variation as a measure it indicates that the variation of income per capita among magisterial districts increased slightly over the period - in other words, divergence occurred.