Multidimensional Impacts of Disability on Poverty: Evidence from Large-scale Data in South Africa
Igei, Kengo (2016). 'Multidimensional Impacts of Disability on Poverty: Evidence from Large-scale Data in South Africa' Paper presented at the annual conference of the HDCA, Tokyo 2016.
Disability has being examined from the perspective of the capability approach since about a decade ago. Sen by himself has never thoroughly worked on disability, but sometimes cited persons with disabilities (PWDs) as the instance of the diversity of human beings in his explanation of the capability approach. In the recent theoretical literature, disability is considered as the deprivation of basic capabilities resulting from impairments (Mitra, 2006; Trani et al. 2011). Therefore, disability and poverty in the sense of the capability approach are closely related to each other, and, in fact, the interdependence between them has been indicated by existing empirical studies (Groce et al., 2011; Palmer, 2011).
Considering the deprivation of capabilities in various aspects of well-being, some empirical studies have explored the multidimensional poverty of PWDs in several developing countries (e.g., Mitra, Posarac, and Vick, 2013; Trani and Cannings, 2013; Trani et al., 2015). But, there are still room to analyze the poverty of PWDs with higher attention to the two issues emphasized by Groce et al. (2011) and Palmer (2011): the interdependence of disability and poverty, and the existence of the confounding factors such as sex, population group, and type and severity of disabilities. This study attempts to investigate the multidimensional impacts of disability on poverty by controlling for these two issues as much as possible. In order to isolate the effects of poverty on disability, the method of exact covariate matching is adopted, i.e., each PWD is matched with a person with almost the same personal and household characteristics expect for disability. In addition, this study disaggregates the impacts of disability into those for the subgroups categorized by the type and severity of disabilities, sex, population group, and type of residence (urban, tribal, or farm area).
Conventionally, the sample size of the data has hindered these analyses. For example, it is often difficult to construct a pair of persons who shares the exactly same characteristics in relatively higher dimensions. It is also difficult to secure a number of PWDs required for the reliable analysis of the poverty for the subgroups divided by type and severity of disabilities because of the small number of persons with disabilities in the data. In order to deal with these challenges, this study utilizes a large-scale data in South Africa, where precise empirical analyses about the poverty of PWDs have not been conducted yet.
The data used by this study is the 10% sample data of South African census in 2011 (Statistics South Africa, 2015).
The sample size is about four million of individuals within about a million of households. The census survey asked about the existence of disabilities, based on the short set of disability questions developed by the Washington group. It covered six domains of functioning (seeing, hearing, communication, walking or climbing stairs, remembering or concentrating, and self-care such as washing, dressing, and feeding), and asks household heads to report each condition of all household members older than five years old using the following four choices: “no difficulty,” “some difficulty, “a lot difficulty,” and “cannot do at all.” In this study, a person is defined to have a disability in a domain if he/she has “a lot difficulty” or a difficulty to the degree of “cannot do at all.” In addition to six types of disabilities, this study takes into account the persons who have disabilities in multiple domains.
The unit of analysis of is an individual, and only individuals aged 5 to 64 years are incorporated in the analysis. Applying the definition of disability above, the overall sample is divided into 127,134 PWDs and 3,397,090 non-PWDs. The most frequently observed disability is about seeing (27% among the PWDs group), and the second is about self-care (23%), and the third is about remembering (9.8%). The persons with multiple difficulties account for 20% among the PWDs group. The prevalence of disability does not differ much by sex, but slightly differ by population group: the ratios of black African and white PWDs are respectively 86% and 4% in the PWDs group, whereas those ratios are 80% and 8% in the non-PWDs group. This difference may indicate the causal relationship from poverty possibly owing to the past racial segregation to the occurrence of disabilities.
The variables used to match PWDs with non-PWDs differ by children and adults. Disabled children aged 5 to 14 years were matched with non-PWDs who are identical to them in terms of age, sex, population group, municipality of residence, type of residence area (urban, tribal, or farm area), and the education level of parents. For disabled adults aged 15 to 64 years, the matching was based on age, sex, population group, municipality of residence, type of residence area, main language, and province of birth. For the adults, I admitted within two years difference of age to increase the rate of matching. As a result, 96.3% of children with disabilities and 94.6% of adults with disabilities were matched with non-disabled counterparts. After matching, this study compares PWDs with matched non-PWDs in terms of education, employment, wealth conditions, household income, access to basic services, and multidimensional poverty index developed by Alkire and Foster (2011). In addition, the differences between disabled and non-disabled are examined by type and severity of disabilities, sex, population group, and type of residence.
The main findings of this study are as follows: (1) PWDs in South Africa are more multidimensionally deprived than non-PWDs, (2) disparities between PWDs and non-PWDs are different according to the type of disability, with the disparity being particularly large for those with difficulties in walking, remembering, and with multiple difficulties, (3) racial and regional inequality still remains in South Africa, with the disadvantages to Black African and living in the tribal area being larger than having disabilities.