Backyard and informal dwellings (2001-2016)

In a previous GCRO Map of the Month, published in October 2017, we showed a dramatic picture of different parts of Gauteng becoming either richer or poorer over time. Our analysis indicated that many township areas had started from a relatively low base in 2001 – with median incomes lower than the provincial median in 2001 – and had fallen behind median growth rates for Gauteng between 2001 and 2011, in effect becoming poorer over time. We speculated that one possible reason for this might be the large increase in backyard structures and informal dwellings in many township areas over this period. This Map of the Month picks up the theme with an analysis of change in the number of backyard and informal dwellings over time.

Settlements in the Gauteng City-Region (GCR) are transforming rapidly, in part due to ongoing urbanisation as well as endogenous household growth generating as yet unmet demand for housing. This transformation manifests spatially both as expansion and the internal restructuring of settlements. This February 2018 Map of the Month contributes to our understanding of the unique ways in which settlements are transforming by analysing growth and decline in the number of backyard and informal dwellings in Gauteng between 2001 and 2016. The analysis uses point data from GeoTerraImage’s (GTI) Building Based Land Use dataset, in which each structure in Gauteng is identified from satellite imagery. ‘Informal housing structures’ and ‘Backyard structures associated with formal housing that may be used for housing purposes (formal or informal)’ are two of 87 land use classes in the dataset. A related category in the dataset is ‘Transitional dwellings’, which is defined as ‘Housing structures that are difficult to classify as either informal or formal’. Though it is also important, we do not focus in detail on this type of dwelling here.

Table 1 summarises the total number of each of these dwelling types in 2001 and 2016, and their percentage growth or decline. In 2001 there were far fewer backyard structures than dwellings in informal settlements. However backyard dwellings grew at a much faster rate (205%) than informal settlement dwellings (51%) over the period, and by 2016 there were over 800 000 backyard dwellings in Gauteng compared to some 600 000 informal settlement dwellings. The growth of backyard dwellings was particularly strong in Tshwane, which experienced a remarkable 393% increase.

Table 1: Dwelling classifications and overall growth (Source: GeoTerraImage, 2016)

The overall numbers of backyard and informal dwellings, and the growth or decline of each over time, are unevenly distributed across the province.

Figure 1 maps the increase or decrease of backyard and informal dwellings per square kilometre between 2001 and 2016. According to the GTI dataset, many built-up urban areas had no recorded backyard or informal dwellings, either in 2001 or 2016. These areas are shown in white on the map. The wealthier core of the province stands out here. Many suburban homes do of course have backyard cottages but these are not classed as backyard dwellings in the dataset.

Other areas, marked in grey shades on the map, saw increases in the number of backyard and informal dwellings in the period 2001 to 2016. The darker the grey the higher the growth. The average increase in the grey areas was 626 dwellings per km2 but a few places saw growth as high as 10 000 dwellings per km2. Large changes in the number of dwellings per km2 can be seen concentrated within areas such as Mamelodi, Diepsloot and Tembisa. In some parts of the province these increases in backyard and informal dwellings have densified already existing neighbourhoods; in others, such as Soshanguve, they have been associated with an expanding footprint of settlements.

While there might be a popular impression of unstoppable growth in less formal housing, some parts of the province – marked in shades of pink on the map – saw a reduction in the number of backyard and informal dwellings between 2001 and 2016. Informal dwellings in particular disappear, resulting from processes such as informal settlement upgrading or removal. The average decline in the pink areas was 297 dwellings per km2, and up to 7 620 dwellings per km2less in one place. Interestingly, the City of Johannesburg saw an absolute decrease of 1.4% in the number of informal dwellings between 2001 and 2016.

Figure 1

Figures 2 and 3 show the changes of backyard and informal dwellings separately. In figure 2, the light pink dots represent informal dwellings in 2001 whilst the red dots represent informal dwellings in 2016. The expansion of informal settlement dwellings is evident in areas such as Soshanguve and Orange Farm near Evaton. The map shows that the relative location of informal dwellings has not changed over time even if they now occupy more land. Informal dwellings remain on the fringes of the urban area, straddling the line between affordability and proximity to economic opportunities, and are often bounded by formal developments and natural barriers.

Figure 2

In figure 3, the light blue dots represent backyard dwellings in 2001 whereas the dark blue dots mark backyard dwellings in 2016. It is particularly striking how the footprint of backyard dwellings ‘mushroomed’ between 2001 and 2016, especially in areas such as Soshanguve, Mamelodi, Diepsloot and Evaton. The footprint of backyard dwellings is invariably linked to the location of older townships and new public housing settlements. The phenomenon of backyard structures has confounded attempts to ‘formalise’ South African settlements. Many occupants of formal dwellings have, without planning permission, constructed other dwellings on their properties in order to derive income from rental. Most are made from informal materials but many are also built with brick and mortar. There may be many of these backyard dwellings on each stand. Therefore, while the replacement of an informal settlement with a formal settlement might initially reduce the number of informal settlement dwellings in an area, the appearance of backyard structures over time may bring the number of informal dwellings back up.

Figure 3

Figures 2 and 3 show the changing footprints of backyard and informal settlement dwellings. What they do not show is the areas where these dwellings disappeared or where there was a transition from informal dwellings to formal dwellings and subsequent addition of backyard dwellings. Zooming in to selected areas reveals dynamic processes of internal settlement restructuring shown by the data.

The following maps show formal dwellings (in green), informal settlement dwellings (in red) and backyard dwellings (in blue) for three different areas – Soshanguve (figure 4), Katlehong (figure 5) and Tembisa (figure 6) – in 2001 and 2016. The figures illuminate a fascinating combination of residential ‘formalisation’ and ‘informalisation’ re-shaping Gauteng’s settlements. Note: separate PDF files can be downloaded in order to flip between the various images (see the links below).

Reading across the maps one can see a number of key trends.

First, while there has been some growth in informal settlements in some parts, such as at the top right corner of the Katlehong map or the middle bottom quadrant of the Soshanguve map, the general trend is towards a reduction in the area covered by informal dwellings (those marked in red) between 2001 and 2016. Dramatic reductions in informal dwellings are most noticeable on the left side of the Soshanguve map, the bottom centre quadrant of the Katlehong map, and the top centre portion of the Tembisa map.

Second, all the maps show a very noticeable growth in formal dwellings (marked in green) both in areas where there were previously informal dwellings, and elsewhere. While some informal settlements have disappeared altogether, such as in the top right quadrant of Soshanguve, the more dominant trend has been the replacement of red dots with green dots. Large swathes of informal dwellings have been replaced with formal dwellings throughout each of the three areas. A notable example of this can be seen on the left side of the Soshanguve maps.

Third, and most significantly, there has been an extraordinary increase in blue dots across all the maps between 2001 and 2016. Backyard dwellings have appeared both in areas that were predominantly green in 2001, and in areas that were mostly red, that is largely informal, in 2001. While the building of backyard dwellings on the properties of historical formal dwellings is of course important in its own right, the more remarkable dynamic here is a double-movement of formalisation of informal areas, followed by the arrival of backyard dwellings. On the one hand, therefore, many formal settlements are becoming more ‘informal’ through the addition of backyard dwellings and the expansion of informal areas. On the other hand, informal settlements are giving way to formal dwellings, only to then be ‘re-informalised’ in a new way with the addition of backyard dwellings.

Figure 4: (Click for PDF)

Figure 5: (Click for PDF)

Figure 6: (Click for PDF)

Methods and notes:

• The analysis uses point data from the GeoTerraImage (GTI) Building Based Land Use dataset. The GTI analysis uses satellite imagery to place a dot on each built structure, including residential dwellings, and then classify according to land use.
• In figure 1, change is calculated per square kilometre in voting districts. There was no specific reason that the data is organised by voting district – for example sub-place boundaries could also have been used – other than that they are roughly equivalent in size in many parts of the province and give a useful level of detail and gradation. An analysis of change per square kilometre controls for the variation in size of the voting districts as the units of analysis, and helps to emphasise the significance of changes relative to other areas.
• It is not clear whether the reduction in transitional dwellings between 2001 and 2016 shown in table 1 is caused by sustained improvements to dwellings or due to enhancements in the techniques of better identifying dwelling types.