Despite the comparatively high level of institutional development for data collection and analysis in South Africa, an important adjunct of apartheid
has been the absence of credible and comprehensive data on which policies can be grounded. The apartheid era governments showed little interest in
collecting information of this nature and often suppressed data that described conditions in the former bantustan areas. It was not until the 1993 Project for Statistics on Living Standards and Development (PSLSD) that a comprehensive quantitative database of household conditions was created. Since then, a number of similar data gathering exercises have been undertaken by Statistics South Africa (the former Central Statistical Services). By design, the samples in these studies are representative 'snap-shots' of their respective time periods. Nonetheless, they can be used to analyse changes over time for aggregate indicators such as poverty or unemployment rates for the general population. Repeated cross-sectional surveys cannot,
however, be used to address a variety of other important dynamic questions. To understand better what is happening to individual households over
time, and the dynamics of their situation, a different type of survey is required in which the same households interviewed in the first period are reinterviewed in subsequent surveys. These are usually referred to as longitudinal or panel surveys, and can be used to determine whether the
same or different households are in poverty in two or more periods, as well as to examine the processes underlying any observed transitions.

The objective of this paper is to consider the advantages gained from using panel data for policy analysis. While costly, international experience
shows that such studies have tremendous potential not only in terms of offering opportunities to analyse social and economic dynamics but also in terms of improving the accuracy of data and analysis