The spatial understanding of unemployment in South Africa is often limited to provinces. This
paper explores ways to integrate unemployment data from 1991 to 2007 to obtain a spatially more
detailed understanding of the phenomenon. Creating such a temporal GIS is challenging because
of the changing administrative boundaries and this phenomenon is part of the modifiable areal unit
problem (MAUP). The MAUP was overcome by using proportional aggregation to the 2005
municipal boundaries. The unemployment data were analysed spatially through mapping standard
deviation intervals and spatial autocorrelation. Unemployment showed a spatial pattern over time
in South Africa and municipalities with high unemployment figures over time were spatially
concentrated. Furthermore, it is recommended that census data should be released at a small area
level to allow more accurate comparison of socio-economic data over time.