Unemployment in South Africa : in search of a spatial model.
Consistent high unemployment perpetuates inequalities in the South African society. The 2014 growth expectation for the South African economy is 1.5 per cent and this will most certainly not be enough to reduce unemployment. This research aimed to create an understanding of the spatial intricacies related to unemployment and to create a longitudinal dataset since 1991. The challenge with such a dataset is that boundaries of the enumeration and administrative areas have changed continually in the past and makes it difficult to compare unemployment spatially over time. These particular problems were addressed by aggregating data for 1991 and 1996 census from magisterial districts to the 2005 municipal boundaries. Area based weighted areal interpolation was used and it assumes that data is distributed homogeneously across the area of each source unit. The 2001 census and 2007 community survey data was available at the 2005 municipal level, and therefore a longitudinal socio-economic dataset of four time points could be created. The results showed that unemployment has been spatially persistent in a number of areas. Furthermore, a spatial grouping of unemployment by municipality showed that metropolitan municipalities had unique unemployment characteristics whereas the remainder of the country could be clustered into five distinct groups. A spatial comparison between unemployment and poverty at municipality level revealed that people can be poor and unemployed, but also poor and employed. Finally, the longitudinal data was used to do spatial forecasting of future unemployment trends and these accounted for up to 60 per cent of change in unemployment. These national and provincial spatial unemployment models consisted of coefficients like the percentage of people employed in mining and agriculture. This research added new knowledge in terms of the spatio-temporal understanding of unemployment in South Africa. It created a methodology to overcome modifiable areal unit problems (MAUP) and a longitudinal dataset of unemployment and related socio-economic variables. Refined spatial data was this research’s main challenge and it recommends that unemployment data should be released at the most detailed spatial level possible - like sub-place or enumeration areas. The quality and timeliness of data remain obstacles for policy-making. Therefore, labour market data at a sub-place level would provide a more meaningful analysis. The results from census 2011 will allow the creation of longitudinal socio-economic trends at a spatially detailed level in South Africa in the future.
Ph. D. University of KwaZulu-Natal, Durban 2015.
Unemployment--South Africa., Space perception--South Africa., Structural unemployment--South Africa., Theses--Geography and environmental management.