Skip to content

Are Searching and Non-searching Unemployment Distinct States when Unemployment is High? The Case of South Africa

Literature › Works

Geeta Kingdon; John Knight (2000). Are Searching and Non-searching Unemployment Distinct States when Unemployment is High? The Case of South Africa. Centre for the Study of African Economies Working Paper WPS/2000-2. Centre for the Study of African Economies, Department of Economics, University of Oxford.
Type working paper
Scope Post-1994 South African economy
Method econometric
Period studied 1993–1997
Regions South Africa, former homelands, urban and rural South Africa
Themes inequality poverty unemployment, labour markets wages, methods data sources, transition post apartheid
Citation key kingdon2000

Argument

The paper asks whether jobless South Africans who say they want work but are not searching are a distinct state from the searching unemployed, or are discouraged workers who belong in the labour force. It tests this with four approaches. The first compares poverty and deprivation across the two groups. The second is a logit model of job search. The third compares perceived quality of life. The fourth is a wage-curve comparison of broad and narrow local unemployment. All four point the same way: non-searching reflects constraints (poverty, remoteness, high local unemployment, passive recruitment channels), not a taste for idleness. The authors conclude that the broad definition is the appropriate one in South Africa, so the 1997 rate should be read as 38% rather than 23%.

Evidence and method

Two nationally representative household surveys: SALDRU93 (about 9,000 households, 360 clusters, with income and community data) and the Statistics South Africa October Household Survey 1994 (33,000 households). Also StatsSA published unemployment series for 1993-97, the StatsSA 1997 Special Retrospective Survey, and the authors' earlier wage-unemployment study (Kingdon and Knight 1999b).

Findings

  • The official unemployment rate for 1997 differs sharply by definition, which makes the choice of measure consequential. (p. 4)

    According to Statistics South Africa, South Africa had an unemployment rate in 1997 of 23% or 38%, depending on whether the narrow or broad definition is used

  • Unemployment is long-lasting as well as widespread. (p. 4)

    in 1997, 37% of the narrowly defined unemployed had been out of work for more than 3 years and another 30% had been unemployed for between 1 and 3 years.

  • StatsSA's switch to the narrow measure in 1998 reversed its earlier choice of the broad measure. (p. 4)

    This reversed its earlier decision of 1993 - based on extensive consultation - to use the broad measure as the official definition.

  • Non-searchers report discouragement and cost barriers as the main reasons for having stopped searching. (p. 6)

    the main reasons given by the unemployed for having stopped searching are a loss of hope of finding work (33%), a lack of jobs in the areas in which they live (25%)

  • The non-searching unemployed are poorer than the searching unemployed. (p. 7)

    49% of the passive unemployed but only 38% of the active unemployed are in households where per capita income is below the international poverty line of $1 a day

  • Non-searchers are more concentrated in the poorest income decile than the searching unemployed are. (p. 7)

    the searching unemployed are 4.5 times as likely to be in the lowest per capita income decile as the employed (31.2 divided by 6.8), the non-searching unemployed are 6.5 times as likely to be so (28.2/4.4).

  • The job-search logit points to constraints (poverty, remoteness, high local unemployment) rather than tastes. (p. 10)

    It suggests that lack of job-search is not the outcome of tastes but an indication of constraints.

  • Quality-of-life regressions show no behavioural distinction between non-searching and searching unemployment. (p. 13)

    Ceteris paribus, the non-searching unemployed are no happier than the searching unemployed.

  • Wage-curve evidence shows local wages respond to broad unemployment, so non-searchers count in wage-setting. (p. 17)

    broadly measured unemployment has a far greater impact on wages than narrowly measured unemployment.

  • The authors conclude that the broad rate is the right headline measure for 1997. (p. 18)

    the appropriate measure of unemployment in 1997 was 38% rather than 23%.