Democracy has increased employment but, ironically, also unemployment as more people seek work.

Research by UCT academic Martin Wittenberg has shown that an increase in unemployment has been one result of democracy.

Working under the auspices of Economic Research Southern Africa (ERSA), Wittenberg says the freeing up of local labour market conditions and the improved quality of education have made black South Africans more employable. In official statistics this is reflected in a modest pick-up in employment levels since 1995.

Ironically, since improved labour market conditions have encouraged more people to seek work, there has also been an increase in the number of unemployed South Africans in this period.

The official definition of unemployment measures the number of people who are willing and able to work, but who cannot find employment. Improved education and better prospects of finding work are drawing greater numbers into the realm of those willing and able to work; hence the rise in measured unemployment.

Wittenberg suggests that the lifting of all restrictions to job access might have released a pent-up demand for participation in the labour market that only became evident with the demise of apartheid.

"As more people join the labour force, competition for jobs will become heightened, unless overall economic demand increases even more.

"One might say that while the human and social capital of African males has improved over time, the returns attached to that capital have declined."

His findings therefore offer fresh insight in the debate around the limited reduction in South Africa's unemployment levels.

Participants in the debate have suggested that we have reached some sort of equilibrium level of unemployment, or that democratisation has strengthened the position of organised labour relative to the unemployed - and hence the persistence of unemployment.

There have also been many who challenge the reliability of the official statistics in giving a clear picture of trends in employment and unemployment levels.

Wittenberg's research confirms that there have been some inconsistencies in the regular surveys of household activities and labour market participation and says: "Regrettably, changes in the sampling design by Statistics South Africa have made attempts to compare trends over time very difficult."

In particular, the 1995 national household survey seems to give an exaggerated reading of employment levels.

"The distortion is unfortunate, given that this survey usually is used as the benchmark against which labour market trends since the advent of democracy are measured."

However, Wittenberg uses a new statistical technique, the Lemieux-style decomposition, which sidesteps the concerns about the comparability of the various data sets. He explains that the standardising method disentangles data changes that are due to measurement errors from those that are attributable to real change.