South africa: New survey still has flaws
Thématique :
sud afrique
By Ethel Hazelhurst, June 25, 2008
South Africa's failure to create jobs is a controversial political issue. And official data from Statistics SA on employment and unemployment is often disputed.
Among other things, the official unemployment rate of 23 percent is considered misleading by many because it excludes people who are too discouraged to look for work. When disheartened work seekers were included, the unemployment rate was 35.8 percent.
In an attempt to upgrade the quality of its information, Stats SA has made wide-ranging changes to the labour force survey (LFS). It says these changes will have ''very important consequences for the country's employment statistics".
The agency posted a report on the changes on its website this week. Data users have mixed views on their value.
Azar Jammine, the chief economist at Econometrix, said the agency's decision to recruit permanent staff - capable of conducting first-hand interviews rather than posing questions through a translator - was a major advance.
Carel van Aardt, a professor of research at Unisa's Bureau of Market Research, said the decision to simplify the questionnaire would make respondents more co-operative.
The use of simple questions to elicit information about employment status was positive, because people responded to "aggressive questions" about being out of work by claiming that they were not looking for work.
But Van Aardt, who specialises in demographic and econometric modelling, said his most serious concerns about the data, such as the official definition of unemployment, had not been addressed.One of the problems he identified was that people who were not available for work within a certain period were not considered work seekers.But they might have specific reasons for not being available within that period and could well be work seekers, he said.
He was concerned that the approach would lead to an understatement of unemployment.The announcement of the new quarterly labour force survey and the release of quarterly employment statistics (QES) yesterday have focused attention on the quality of data from Stats SA.
In the past, economists and other researchers had voiced concern about conflicting information generated by the two surveys.QES figures are based on a survey of more than 22 000 formal businesses in the non-agricultural sector and include data from the Unemployment Insurance Fund and the SA Revenue Service.
The labour force survey is derived from face-to-face interviews with the members of more than 30 000 households.Two measures of employment are a common practice among statistical agencies, according to Pali Lehohla, the statistician-general.
Economists agree that they both serve a purpose, but they argue that the differences between the results need to be addressed.When the two sets of figures were released in March, Jammine identified anomalies between the two data releases that were not fully explained by the different methodologies and samples. And he discovered abrupt changes in the trends revealed by the QES that seemed unlikely.
Van Aardt said the two surveys should be part of the same process.Previously published twice a year, the re-engineered LFS would be published quarterly from August 28.
The changes to the methodology were made after recommendations by the International Monetary Fund three years ago.Stats SA set up a project team of both local and international experts on "labour force content and questionnaire design, household survey sampling and weighting, household survey field operations and systems development".
South Africa's failure to create jobs is a controversial political issue. And official data from Statistics SA on employment and unemployment is often disputed.
Among other things, the official unemployment rate of 23 percent is considered misleading by many because it excludes people who are too discouraged to look for work. When disheartened work seekers were included, the unemployment rate was 35.8 percent.
In an attempt to upgrade the quality of its information, Stats SA has made wide-ranging changes to the labour force survey (LFS). It says these changes will have ''very important consequences for the country's employment statistics".
The agency posted a report on the changes on its website this week. Data users have mixed views on their value.
Azar Jammine, the chief economist at Econometrix, said the agency's decision to recruit permanent staff - capable of conducting first-hand interviews rather than posing questions through a translator - was a major advance.
Carel van Aardt, a professor of research at Unisa's Bureau of Market Research, said the decision to simplify the questionnaire would make respondents more co-operative.
The use of simple questions to elicit information about employment status was positive, because people responded to "aggressive questions" about being out of work by claiming that they were not looking for work.
But Van Aardt, who specialises in demographic and econometric modelling, said his most serious concerns about the data, such as the official definition of unemployment, had not been addressed.One of the problems he identified was that people who were not available for work within a certain period were not considered work seekers.But they might have specific reasons for not being available within that period and could well be work seekers, he said.
He was concerned that the approach would lead to an understatement of unemployment.The announcement of the new quarterly labour force survey and the release of quarterly employment statistics (QES) yesterday have focused attention on the quality of data from Stats SA.
In the past, economists and other researchers had voiced concern about conflicting information generated by the two surveys.QES figures are based on a survey of more than 22 000 formal businesses in the non-agricultural sector and include data from the Unemployment Insurance Fund and the SA Revenue Service.
The labour force survey is derived from face-to-face interviews with the members of more than 30 000 households.Two measures of employment are a common practice among statistical agencies, according to Pali Lehohla, the statistician-general.
Economists agree that they both serve a purpose, but they argue that the differences between the results need to be addressed.When the two sets of figures were released in March, Jammine identified anomalies between the two data releases that were not fully explained by the different methodologies and samples. And he discovered abrupt changes in the trends revealed by the QES that seemed unlikely.
Van Aardt said the two surveys should be part of the same process.Previously published twice a year, the re-engineered LFS would be published quarterly from August 28.
The changes to the methodology were made after recommendations by the International Monetary Fund three years ago.Stats SA set up a project team of both local and international experts on "labour force content and questionnaire design, household survey sampling and weighting, household survey field operations and systems development".
