A multidimensional analysis of poverty in a former South African Homeland
Abstract
The study involved a multidimensional analysis of poverty in the former South African homeland of QwaQwa. It was carried out by determining various dimensions of poverty in which the residents of the area are deprived. Seven dimensions of deprivation, constructed from 18 indicators, were identified. These dimensions include education (educational achievement and literacy), housing and clothing (floor material, walls material, roof material and condition, rooms and clothing), water and sanitation (sources of water and sanitation type), assets, (TV or radio, cell phone or landline telephone and other assets) air quality (fuel for lighting, cooking and heating), income and health (proximity to health facilities and food security). The dimensions in the study are equally weighted. The analysis was achieved through a number of theoretical and empirical objectives. The theoretical objectives of the study were related to poverty theories, which included the analysis of unidimensional and multidimensional approaches, as well as a review of previous studies in South Africa, with special attention to the former South African homelands, including QwaQwa, from which the study’s data was collected. In general, the former South African homelands have been characterised as having poor infrastructural development and welfare services, coupled with high levels of poverty, which have resulted in very low standards of living compared to most parts of South Africa. The study considers the historical nature and the poverty dynamics of these areas in providing a comprehensive point of departure to understand the multidimensional nature of poverty, based on Alkire and Foster’s family of measures.
Empirically, the study has determined various deprivations and their levels, as defined by union (deprived in one or more indicators), intersection (deprived in all dimensions) and the dual cut-off (cut-off between union and intersection criteria) method of identification. It is important to note that the identification study is mostly built upon Alkire and Foster’s dual cut-off method of identification. The aggregate deprivation by dimension represented by the approximate headcount in each indicator of seven dimensions used in the study indicated that the former homeland A multidimensional analysis of poverty in a former South African homeland is more deprived than the whole of South Africa in almost all dimensions, regularly by a wide margin. By identifying the poor, any poverty index based on the intersection approach would judge the former homeland of QwaQwa as almost poverty free (0.002%) while 100% of all households would live in poverty if identification of the poor is established on the union approach. However, this study considered a k-threshold (k=33%) which is very similar to the k threshold chosen by other researchers in similar frameworks. At k=33% of all indicators, about 61% of the households are deprived in 53.4% of total dimensions, on average, with the intensity of poverty M0 being 0.32. The study has analysed the poverty dimension through a breakdown of their intensity, in order to establish their specific contribution to overall poverty. Given the cut-off k=33%, health, most importantly, is the highest contributor to the breadth of poverty (19.06%), followed by water and sanitation (18.46%), education (15.80%), income (15.10%), housing and clothing (14.16%), air quality (12.53%) and assets (4.89%). The study established different levels of poverty by comparing income poverty to multidimensional poverty measures. Multidimensional poverty estimates showed higher estimates of poor, with 61%, compared to 40% of households which fall below the income poverty line. The analysis showed that the use of the income measure is bound to underestimate the level of poverty in an area. For example, a certain number of the households, identified non-poor by income, were discovered to have even 11 and 12 deprivations out of a total of 18 dimensions, thus showing the limitation of the monetary measure of poverty. Based on the Spearman correlation, the relationship between the level of income of the household, and the number of deprivations associated with it, indicated a lack of overlap between the group of poor acknowledged by the monetary measure and the group of the poor detected by the multidimensional approach. All pairs were not correlated, and even the income deprivation is not correlated with all other aspects of deprivation, while it is claimed to be the measure of poverty that reflects deprivation in other dimensions. This emphasises the arguments this study is built on, that deprivation in one dimension does not satisfactorily justify the deprivation in any other dimension. A multidimensional analysis of poverty in a former South African homeland The study utilised the aggregate measures of poverty suggested by Alkire and Foster in estimating the change in level of deprivation in QwaQwa, thus establishing the depth and level of inequality in the area. The study performed poverty decomposition, using sub-groups within the area, by examining villages and household characteristics. Concerning the depth and inequality among the poor, the study considered the multidimensional poverty gap M1 and multidimensional squared poverty gap M2 for the analysis. The study found that with respective cut-offs there are large poverty gaps and high inequality among deprived states of the poor.
Based on the subgroup decomposability property of Alkire and Foster’s poverty measures, the analysis of poverty in eight villages within QwaQwa was undertaken according to household size. At k=33% cut-off, the highest multidimensional headcount ratio and intensity of poverty was found in Boitekong followed by Kudumane, Mabolela, Qholaqhoe, Mphatlalatsane, Matsikeng, Ntshehele and Bochabela villages. The study established the robustness of cut-off variations which provides village ranking in terms of areas that are worse off than others. It was confirmed that Kudumane and Boitekong are dominated by other villages both in poverty headcount ratio and intensity of poverty. Combining the inequality and robustness, the study has established that the villages with the highest multidimensional intensity of poverty M0 and multidimensional poverty gap M1 have the highest estimates of multidimensional squared poverty gap M2. This implies that, in those villages, the multidimensionally poor households are far from ceasing to be so and their state is accompanied by high inequality. This pattern was also found when the robustness of village rankings to changes in the poverty cut-off k was used.
By exploring the multidimensional poverty based on the former homeland of QwaQwa, using the Alkire and Foster family of poverty measures, the study provides a shift in attention from solely income measures to include other intrinsically important dimensions when analysing poverty. By using the survey data, the study identified the share of poverty in terms of dimensions that hold the highest share of poverty. The study has, in general, provided an analysis of, and insights into, the broader perspective of multidimensional poverty as a necessary shift from the traditional unidimensional perspective of poverty, which basically centres on income. A multidimensional analysis of poverty in a former South African homeland Through the breakdown of dimensions and subgroup decomposability of poverty measures used, the study has provided a methodology which not only produces meaningful results, but can also potentially provide useful tools for budget allocation among areas and within a region, by applying, among other things, the use of different dimensions. It should, however, be noted that the dimensions used in the study are not exhaustive and other dimensions could also be incorporated, subsequently leading to alternative deprivation cut-off values for respective analysis. The case of QwaQwa has demonstrated a classical example of the need to clearly specify the setting area developmental goals on the basis of holistic understanding of households challenges and, in this case, poverty status and levels, if meaningful development is to be achieved. Such an in-depth understanding in setting developmental goals has been enriched from the proposed methodology of the study and can be used as a progress-monitoring instrument. This analysis is particularly relevant in connection to South Africa’s bold adoption of the poverty-related Millennium Development Goals (MDGs), keeping in mind the spatial legacy of apartheid. The formulation of effective interventions to combat poverty requires a clear grasp of its manifestations in a particular area. The study’s special contribution is hence on poverty analysis methodology, as guided by the multidimensional poverty measure, which goes beyond the identification of geographical deprivation by setting a threshold level in order to define who the multidimensional poor are, the degree of the deprivations and the nature of their challenges. On the basis of evidence established in this study, five areas of policy adaptation have been recommended. They are the adoption of Alkire and Foster’s methodology in identifying the needs of poor household, the priotisation of the needs/dimensions, prioritisation of target area, the household targeting as determined by the household size, the involvement of local leaders and a suggested model for identifying needs, both at the macro and micro level.