Multidimensional poverty in the GCR (2015/16 data)
Following our earlier multidimensional poverty map based on 2013 data, we present here an update based on Quality of Life Survey IV (2015/16) survey data. Poverty is often measured in terms of income. For example, many studies make use of a ‘poverty line’ in which those who earn less than $2.00 a day are defined as poor. An alternative approach is to recognise that poverty is multidimensional and can be measured across a variety of indicators such as housing, water, sanitation, energy, ownership of communication assets, food, employment and education. In terms of this approach, poverty is measured not according to an income threshold but according to the number of indicators in which a household* is deprived (known as the ‘headcount ratio’).
In addition to measuring the proportion of households that are multidimensionally poor we can also measure the intensity of their poverty. Some households are deprived in terms of more indicators than others. The greater the number of dimensions in which they are deprived, the greater the intensity of their poverty.
There is a dummy example at the end of this write-up to illustrate how the calculations are made.
Map 1 below shows the percentage of respondents per ward who were disadvantage on three or more of nine indicators of poverty. According to our calculations, 13.3% of households in Gauteng are multidimensionally poor in that they were deprived on three or more of nine indicators of poverty. Out of the 508 wards in Gauteng, 65 were not multidimensionally poor in that no respondents in the ward were lacking in three or more indicators. The percentages of respondents per ward who were multi-dimensionally poor range from 1,1% to 66,7%. The mapping shows that many of the poorer wards were located on the edges of the city-region.
Map 1: Headcount ratio
For those households indentified as poor it is also necessary to assess just how poor they are, that is, the depth or intensity of their poverty. Intensity is measured by asking, of those respondents who are multidimensionally poor, what was the average percentage of indicators that they were deprived on. The higher the percentage, the deeper the level of poverty for those who are poor, as illustrated in the visualisation at the end of this write-up.
The analysis of this measure gave a rather striking result for Gauteng, as shown in Map 2. While a headcount ratio picture alone suggests that poverty is concentrated on the edges of the city-region, an intensity analysis indicates that there are a number of wards in the core where poverty is much deeper than that in wards on the outskirts. This most likely has to with high levels of informality in areas such as Alexandra, Diepsloot and Daveyton. Informality in housing poses challenges in terms of accessing other amenities such as piped water, electricity and flush toilets.
Map 2: Intensity
Multidimensional poverty index
That said, when an overall Multidimensional Poverty Index (MPI) is computed by combining headcount and intensity (the proportion of households that are lacking in three or more dimensions is multiplied by the average percentage of indicators in which poor households are lacking), outlying wards do feature as relatively poor compared to wards located in the interior of the province. This is essentially because of the larger proportions of households that are multidimensionally poor in these areas.
Map 3: Multidimensional poverty index
- Poverty remains prevalent in peripheral areas of Gauteng, in particular the west, south and north-western parts of the province. Low economic activity in these areas limits access to income opportunities, hence compromising households’ ability to access better livelihood opportunities and amenities. For example, Soshanguve in the north served predominantly as a dormitory township during the apartheid era, and consequently today has fewer economic opportunities and lower levels of service.
- However, even in economically stronger municipalities, such as the three metros of Johannesburg, Ekurhuleni, and Tshwane, there is evidence of multidimensional poverty. Wards in areas such as Alexandra (in Johannesburg), Tembisa and Daveyton (in Ekurhuleni), and Mamelodi (in Tshwane) exhibit high levels of poverty in spite of their relatively central locations. In some of these areas the intensity of poverty is higher than on the periphery.
- Unemployment was the most prevalent dimension of poverty on the bank of indicators used in this analysis. Lack of access to jobs limits the ability of a household to access other amenities and opportunities, trapping the household in poverty.
Change over time
A comparison of multidimensional poverty calculations for QoL III 2013/14 data and QoL IV 2015/16 data suggests that poverty levels in the GCR have worsened since 2013. The headcount ratio and the intensity of poverty rose by 0.3 and 1.0 percentage points respectively, while the overall multidimensional poverty index rose by 0.003 percentage points. Several municipalities such as Johannesburg, Lesedi, Merafong, Mogale City and Tshwane experienced an increase in the proportion of households that are multidimensionally poor between 2013 and 2015.
These results suggest that poverty remains a major challenge for government. The finding that more areas in the outskirts are multidimensional poverty is worrying as this could be a self-perpetuating phenomenon. Areas with low economic activity typically struggle to raise enough revenue to expand or improve service delivery, hence compromising their ability to reduce multidimensional poverty among their residents. This is further reinforced by the finding that income-poor households are also multidimensionally poor. Unemployment features as a key factor underpinning multidimensional poverty since unemployed households face challenges in accessing services, and also tend to locate themselves in informal areas where service provision is limited.
Calculating multidimensional poverty
How did we do the calculations for this analysis? Suppose Ward X has 10 households each with varying levels of access to adequate housing, services, assets for communications, and so on. Suppose also that there are 9 indicators of poverty against which the households are assessed. In the table, a dot means that a household is deprived in that indicator. For instance, Household 1 is deprived in 1 out of the 9 indicators (or 11.1%), while Household 2 is deprived in 2 out of the 9 indicators (or 22.2%) and so on. If a household is deprived in 33,3% or more of the indicators, it is regarded as multidimensionally-poor. Therefore, in this ward, households 3 to 8 are multidimensionally-poor and households 1, 2, 9, and 10 are not. Put differently, 6 out of 10 households (or 60%) in this ward are multidimensionally poor. This measure is called the headcount ratio which is the proportion of households that are multidimensionally poor in that ward.
Furthermore, we can measure intensity of poverty. This assesses the number of indicators in which respondents are lacking (in our case, Households 3 to 8). Intensity is defined as the average level of poverty for those households that are poor. Household 8 is the poorest since it is lacking in 8 of the 9 indicators. When all the multidimensionally poor households (3 to 8), are taken together the average proportion of the indicators in which they lack is 61,1%. This figure defines the intensity of poverty for this ward.
A Multidimensional Poverty Index (MPI) is calculated by multiplying the two percentages, for Headcount and Intensity, together. It gives a combined view of both the breadth and depth of poverty in an area.
* Although the Quality of Life Survey data reflects the responses of individual respondents, these respondents speak to the condition of their households on many of the indicators, especially with regard to housing and service deficits. As a shorthand we therefore refer to households in this analysis.