Many recent studies compared the 1995 October Household Survey (OHS) with the latest available
Labour Force Survey (LFS) to derive the unemployment ‘trends’ in South Africa since the
transition, but this approach only gives a snapshot of unemployment at two points in time.
Although the better approach is to examine all available labour surveys to derive the real
unemployment trends during the period, this does not mean these trends are fully reliable and
comparable, as the sampling method, weighting technique, questionnaire design and labour
market status derivation methodology to define the unemployed are different across the surveys.
In particular, the unemployment estimates in OHS 1995–99 and during the changeover between
OHS and LFS in both narrow and broad terms increased rapidly. This paper aimed to address
these issues, if possible, in order to improve the comparability and reliability of unemployment
aggregates across the surveys.