Doctor Educationis / There is a wide-reaching consensus that data required for poverty analysis in
developing countries are inadequate. Concerns have been raised on the accuracy and
adequacy of household surveys, especially those emanating from Sub-Saharan Africa.
Part of the debate has hinted on the existence of a statistical tragedy, but caution has
also been voiced that African statistical offices are not similar and some statistical
offices having stronger statistical capacities than others. The use of generalizations
therefore fails to capture these variations. This thesis argues that African statistical
offices are facing data challenges but not necessarily to the extent insinuated.
In the post-1995 period, there has been an increase in the availability of household
surveys from developing countries. This has also been accompanied by an expansion
of poverty analyses efforts. Despite this surge in data availability, available household
survey data remain inadequate in meeting the demand to answer poverty related
enquiry. What is also evident is that cross sectional household surveys were conducted
more extensively than panel data. Resultantly the paucity of panel data in developing
counties is more pronounced. In South Africa, a country classified as ‘data rich’ in this
thesis, there exists inadequate panel surveys that are nationally representative and
covers a comprehensive period in the post-1995 period. Existing knowledge on poverty
dynamics in the country has relied mostly on the use of the National Income Dynamic
Study, KwaZulu Natal Dynamic Study and smaller cohort-based panels such as the
Birth to Twenty and Birth to Ten cohort studies that have rarely been used in the
analysis of poverty dynamics.
Using mixed methods, this thesis engages these data issues. The qualitative component
of this thesis uses key informants from Statistics South Africa and explores how the
organization has measured poverty over the years. A historical background on the
context of statistical conduct in the period before 1995 shows the shaky foundation that
characterised statistical conduct in the country at the inception of Statistics South Africa
in 1995. The organization since then has expanded its efforts in poverty measurement;
partly a result of the availability of more household survey data. Improvements within
the organization also are evidenced by the emergence of a fully-fledged Poverty and
Inequality division within the organization. The agency has managed to embrace the
measurement of multidimensional poverty. Nevertheless, there are issues surrounding
xv
available poverty related data. Issues of comparability affect poverty analysis, and these
are discussed in this thesis. The informants agreed that there is need for more analysis
of poverty using available surveys in South Africa.
Against this backdrop, the use of pseudo panels to analyse poverty dynamics becomes
an attractive option. Given the high costs associated with the conduct of panel surveys,
pseudo panels are not only cost effective, but they enable the analysis of new research
questions that would not be possible using existing data in its traditional forms.
Elsewhere, pseudo panels have been used in the analysis of poverty dynamics in the
absence of genuine panel data and the results have proved their importance.
The methodology used to generate the pseudo panel in this thesis borrows from
previous works including the work of Deaton and generates 13 birth cohorts using the
Living Conditions Surveys of 2008/9 and 2014/15 as well as the IES of 2010. The birth
cohorts under a set of given assumptions are ‘tracked’ in these three time periods.
The thesis then analysed the expenditure patterns and poverty rates of birth cohorts.
The findings suggested that in South Africa, expenditures are driven mostly with
incomes from the labour market and social grants. The data however did not have
adequate and comparative variables on the types of employment to further explore this
debate. It also emerged that birth cohorts with male headship as well as birth cohorts in
urban settlements and in White and Indian households have a higher percentage share
of their income coming from labour market sources. On the other hand, birth cohorts
with female headship and residing in rural, African and in Coloured households are
more reliant on social grants. The majority of recipients of social grants receive the
Child Social Grant and its minimalist value partly explains why birth cohorts reporting
social grants as their main source of income are more likely to be poor when compared
to birth cohorts who mostly earn their income from the labour market. Residing in a
female-headed household, or in a rural area as well as in Black African and Coloured
increases the chances of experiencing poverty. This supports existing knowledge on
poverty in South Africa and confirms that these groups are deprived. The results of the
pseudo panel analysis also show that poverty reduced between 2006 and 2011 for most
birth cohorts but increased in 2015. Policy recommendations to reduce poverty
therefore lie in the labour market. However, given the high levels of unemployment in
the country today, more rigorous labour incentives are required.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:uwc/oai:etd.uwc.ac.za:11394/7801 |
| Date | January 2019 |
| Creators | Mabhena, Rejoice |
| Contributors | May, Julian D |
| Publisher | University of the Western Cape |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
Page generated in 0.0022 seconds