What is poverty and how do we measure it?
"To be poor is to be hungry, to lack shelter and
to be sick and not cared for, to be illiterate
and not schooled. But for poor people,
in poverty is more than this. Poor people are
vulnerable to adverse events outside their
control. They are often treated badly by the institutions
of state and society and excluded from voice
and power in those institutions”(World Bank 2000
From a human development perspective, poverty
the denial of choices and opportunities for
tolerable life”(UNDP 1997, p.5).
Policy debates have indeed been distorted by
on income poverty and income
to the neglect of deprivations that
to other variables, such as unemployment, ill
lack of education, and social exclusion”
Poverty takes on multiple dimensions and in essence describes a state of deprivation that prevents an individual from attaining some minimum “socially acceptable” standard of living. This state of deprivation can therefore be
measured in a number of ways and according to various approaches.
The most commonly used method of profiling poverty in a society involves firstly establishing the minimum amount of money required to meet the cost of an individual’s (or household’s) basic needs, which would include a food and non-food component. This poverty line is then utilised in conjunction with specific measures of poverty to develop an appropriate description of indigence
in the society. However, poverty lines are difficult to measure and the choice of line often imprecise. Furthermore, indices are often subjective and
arbitrary in the choice of variables or weights used in their construction. Yet another way of comparing relative wellbeing of pre-defined groups is the poverty dominance approach. In this type of analysis, no poverty lines are used, but groups are measured against each other in terms of chosen indicators such as income levels or access to certain assets or services. Using income as an example, the dominance method graphically depicts the cumulative proportion
of those with access to each and every income level, for each group being considered. If it is shown that one group’s cumulative distribution always falls above or below anothers’, strong statements can be made about relative
wellbeing. For example, in the case of income, it could then be stated that one group is better off than another for each and every income poverty line chosen.
Measuring the number of poor according to a poverty line approach and considering changes over time requires accurate income or expenditure data. In South Africa the income and expenditure surveys (1995 and 2000) collect
the relevant data. At present, there are a number of questions on the quality and comparability of these datasets, with the data quality of the 2000 survey currently under review. Given this, there is no recent dataset with which one can confidently carry out poverty counts for the country.
An alternative approach to measuring poverty, in the absence of data on income and consumption, is to use asset-based indicators. An asset index can be constructed using data on household durables (for example, owning a radio, refrigerator etc) and household characteristics (for example, the number of
rooms in a house or sanitation facilities). Provided a sufficiently broad class of asset indicators is used, the index should reflect differentiation of living standards across households1. One of the advantages of using an asset-based index is that it avoids the problems of recall bias, seasonality and mismeasurement that can occur with income and consumption based measures of poverty2.
A well-known example of an index used to reflect differences in wellbeing is the United Nations Development Programme’s Human Development Index (HDI). This measure is used to compare countries in terms of their achievements in attaining a certain standard of living proxied for by indicators of life expectancy at birth, adult literacy, a combined primary, secondary and tertiary schooling enrolment ratio and Gross Domestic Product per capita.
With the recent release of census 2001, we have at our disposal a wealth of data on a range of living standards indicators other than income/expenditure. Through comparisons of these indicators with those reported in census 1996,
we can develop a picture of deprivation in 2001 and of changes in wellbeing over time. In this report we use mainly an asset-based approach combined in some instances with poverty dominance analysis.
Our choice of poverty indicators includes those commonly used in the literature on poverty analysis, the relevance of which to South Africa is
strongly supported by the findings of the South African Participatory Poverty Assessment which contextualised the experience of poverty through the voices of South Africans themselves. Key themes to emerge from this piece of work
revealed that in South Africa to be poor means to:
“Be alienated from your community, to be unable to sufficiently feed your family, to live in overcrowded conditions, use basic forms of energy, lack adequately paid and secure jobs and to have fragmented families” (May 2000, p5).
Furthermore, we consider vulnerability a key dimension of poverty and include indicators of low adequacy to deal with shocks in our analysis.
Ultimately though, while the income poverty approach is widely used and extremely powerful, it does exclude both assets possessed and services accessed by individuals in the society. These rates of ownership and access invariably,
are critical, additional descriptors of poverty in a society, and they will be the dominant analytical descriptor of poverty in this report.
The aim of this report is to provide a picture of asset and services deprivation, economic activity, and health and safety, and to illustrate
the changes in these indicators from 1996 to 2001. We use the census data (1996 and 2001) to analyse shifts in these indicators during the 5 year period, and in cases where data on shifts is not available, we attempt to create a broader
picture for 2001 using more detailed data from the Labour Force Survey, September 2001. We analyse the indicators of economic and social wellbeing on the national level and for three provinces, namely, Gauteng, KwaZulu-Natal,
and Limpopo. These provinces were chosen to illuminate the regional discrepancies in wellbeing that exist in South Africa, even on a provincial level. Gauteng was chosen as a proxy for the richest provinces (Gauteng and the
Western Cape), Limpopo as representative of the poorest provinces (Limpopo, the Free State and the Eastern Cape) and KwaZulu-Natal as the more average performer3.
The report begins with a brief overview of the demographic and location-specific characteristics of the country in 1996 and 2001.
Section 2 then considers key indicators of deprivation such as access to basic goods and services. Section 3 focuses on economic indicators of wellbeing and includes data on education, employment and unemployment. Sections 4 and 5 provide additional coverage of fundamental poverty indicators not included
in the previous sections such as health status and crime. Finally, section 6 ends the report with some concluding remarks.
McKenzie 2003, p 9
McKenzie 2003, p 3
Prior research on income poverty has found that the Eastern Cape, Free State and Limpopo rank as the poorest provinces, with Gauteng and the Western Cape being the least poor. The remaining provinces are found between these extremes. The poverty rates and rankings for the provinces vary with the survey data and poverty lines used (see Bhorat et.al 2001; May 2000; Statistics South Africa 2000.)