Differentiating household income growth in Gauteng 2001-2011

This map of the month considers whether spatial inequality in Gauteng is improving or deteriorating over time. Our first step in investigating this issue was to map ward-level median household income growth in Gauteng over the period 2001 to 2011 (Figure 1). Each ward is shaded according to the ratio of its 2011 median household income to its 2001 median household income. By way of example, a figure such as 1.66 means that the ward’s median income in 2011 was 1.66 times its median income in 2001.

The map shows that median household income grew in nominal terms in 504 wards (in all but three wards) over this 10-year period, but it did so at different rates. Wards with a darker shading grew at a faster rate than wards with lighter shading. The lowest rate of growth was 0.85 (in a ward just north of Westonaria) and the highest rate of growth was 14.74 (in a ward on the eastern edge of Tshwane).

If one knows the economic geography of the province it is possible to see that some rich wards grew faster than some poor wards. For example, in 2011, the median income for a ward near Sandton was 3.8 times larger than its 2001 median household income. By contrast, the 2011 median household income of a ward in Ennerdale was 1.2 larger than its 2001 median. The following analysis shows more systematically whether wards that were rich or poor in 2001 were growing fast or slowly.

Figure 1. Income growth patterns between 2001 and 2011 in selected wards

The ‘convergence’ or ‘catch up’ hypothesis in economics predicts that income inequality will reduce over time because low-income regions will grow at a faster rate than high-income regions. This ‘convergence’ or ‘catch up’ is meant to happen in part because poorer regions have lower labour and other costs than already developed regions, thereby attracting businesses seeking lower-cost locations. Did ward level median incomes converge between 2001 and 2011? In Figure 2, we use Gauteng’s provincial median household income in 2001 (R 27 080 per annum) as the line to distinguish between ‘poor’ and ‘rich’ wards, and we use Gauteng’s provincial growth rate between 2001 and 2011 (a ratio of 1.95) to distinguish between ‘low’ and ‘high’ ward growth rates. We call wards with a median income below the province-wide median in 2001 ‘poor’ wards; wards with a growth rate lower than the 2001-11 provincial median growth rate ‘low-growth’ wards; and so on.

Figure 2 shows that wards in the top right quadrant are pulling away from other wards due to the fact that they started from a base median income higher than the provincial median in 2001, and had much growth rates higher than the provincial median rate between 2001 and 2011. Wards in the bottom right are cases of poorer wards – with median incomes lower than the province-wide median in 2001 – which had good growth rates over 2001-2011. They are therefore catching up.


Figure 2. The relationship between wards’ median annual income in 2001 and their median income growth rate between 2001 and 2011.

Figure 3 presents summary data for the entire province. It shows that 122 wards – about a quarter of those in the province – occupied the upper right quadrant of Figure 2 as rich wards with a high growth rate. The average growth rate across these wards was 2.51. By contrast 199 wards, 39%, occupied the bottom left quadrant as poor wards with a low growth rate of 1.71 on average.

Figure 3. Summary relationship between wards’ median income in 2001 and their median income growth rate between 2001 and 2011.

Figure 4 presents the growth rates of one sample ward from each of Figure 2’s quadrants. It shows how a rich ward with a high growth rate and a poor ward with a low growth rate diverge over time, while a rich ward with a low growth rate and a poor ward with a high growth rate converge over time.

Figure 4. Income growth patterns between 2001 and 2011 in selected wards

In Figure 5, we can see how these income growth patterns map spatially onto the province. The map shows where some wards grew fast and others grew slowly regardless of their initial position in 2001. Wards coloured orange – interestingly mostly those on the outer periphery – are those that started from a low base (they had median incomes lower than the province-wide median income) but grew faster than the province-wide median growth rate. Pale green wards are those that started from a higher base but grew slowly. Wards that are coloured dark green are those that were already rich in 2001 and nevertheless grew faster than the province-wide median growth rate. Green wards can be seen mostly in the urban core, confirming analyses presented in previous maps of the month, here and here.

Figure 5: Spatial distribution of differential growth rates per ward, 2001-2011

Wards coloured red are those that started from a low base in 2001 and whose median income grew more slowly than the median growth rate. The location of red wards is startling, and has profound implications. Red wards – those that were already poor and are failing to catch up – are mostly in parts of the GCR designated as African townships under apartheid. Most of Soweto, Tembisa and Ivory Park, Katlehong, Thokoza, Vosloorus, KwaThema, Orange Farm, Sebokeng, Mabopane, and so on, are all red. This suggests the ongoing adverse incorporation of township areas even during a period which saw strong economic growth and the massive expansion of social grants.

A number of factors may account for this:

  • Backyard dwellings have become a common and large-scale phenomenon in most townships. Many hundreds of thousands of these less formal dwellings have been inserted into township settings over the last decade and a half. Households in these dwellings would typically be poorer than those in the primary dwelling on the stand, driving down aggregate median incomes.
  • Many individuals in townships have indeed benefitted from rising incomes over the relevant period. In many cases improvements in circumstances triggered a lifestyle response to leave township areas for new homes in other parts of the city. This meant that even though many township residents may well have benefitted from economic development and improved prospects for social mobility, the township areas from which they moved suffered as they left.
  • More generally, background shifts in the economy – including relative de-industrialisation and a widening gulf between available job opportunities and skills, and in turn rising unemployment – have meant that overall gains in the economy have not been adequately shared.

The number and geographical location of red wards is deeply concerning. It is worth emphasizing the point by noting that while only three wards did not grow in nominal terms between 2001 and 2011, adjusting median incomes by inflation reveals a much more worrying picture. Adjusting for inflation, the 2001 median income of any ward should have been at least 1.789 times higher in 2011. We compared the actual 2011 median incomes of each ward to what their 2001 incomes would be in 2011 when adjusted for inflation. An alarming 168 wards fell below 0, which means they got materially poorer in real terms over the 2001-2011 period. Again, with a few exceptions, most of these wards were in township areas.

Lastly, we applied a spatial econometric model (Sergion & Montouri 1999) to test whether there is convergence or divergence overall. A negative result would mean that there is convergence over time and therefore a reduction of inequality, while a positive number would mean divergence and an increase in inequality. The result was a positive number (0.7%), confirming that at an overall level, there was no catch up. In effect, inequality within the province is widening with the growth of many already affluent wards outpacing the growth of poorer wards.

Further analysis of these issues is part of a forthcoming GCRO book, edited by Dr Koech Cheruiyot, The Changing Space Economy of City-Regions: The Gauteng City-Region, in production with Springer.

References

Sergion, J. R. & Montouri, B. D. (1999). U.S. regional income convergence: A spatial econometric approach. Regional Studies, 33(2), 143-156.

StatsSA (Statistics South Africa). (2001). Census. Pretoria: Statistics South Africa.

StatsSA (Statistics South Africa). (2011). Census. Pretoria: Statistics South Africa.

Link to projects: Economic geography of the GCR

Subscribe

The GCRO sends out regular news to update subscribers on our research and events.