It is commonly claimed that the South African labour market is unusually inflexible owing to the strength of the unions and the system of centralized collective bargaining. One aspect of labour market inflexibility concerns the responsiveness of wages to local unemployment. This part of the research programme on the South African labour market is examining the relationship between unemployment and wages across space.
 

The wage curve has been the subject of intensive discussion, primarily using OECD data where unemployment rates are typically 5-12%. The robustness of the wage curve, both across countries and across ranges of unemployment rates, has become an interesting question. In this research we are looking for a wage curve in South Africa, where the average unemployment rate is very high. We test for the wage curve relationship. We find that the negative relationship between local unemployment rate and local wages is robust to the inclusion or exclusion of the regional dummy variables that might standardize for a long run positive reverse relationship. Even in the mid-1990s, after the relaxation of apartheid controls on mobility, the Harris-Todaro model (which posits a positive relationship between unemployment and wages) appeared not to be relevant to South Africa. This finding presents a challenge to one of the best accepted theoretical models of the way the labour market operates in developing countries with high unemployment.

 

 

The research shows that the elasticity of wages to local unemployment rates in South Africa is -0.1, similar to that found in other countries, including the US and the UK. This is striking because South Africa has a national unemployment rate of over 30%. We find that the wage curve elasticity persists over a much wider range of unemployment rates than in OECD countries, implying that unemployment in South Africa can have a large impact on wages.

South Africa has one of the most interesting labour markets in the world. Its powerful labour market institutions and strong labour unions would suggest lesser flexibility of wages to local unemployment than in OECD countries. Yet we find that the South African wage curve elasticity is no weaker than in OECD countries, and that it is robust over a wide range of unemployment. Tripling unemployment from 10% to 30% reduces wages by approximately 30%. The wage curve was subjected to a range of robustness tests - allowing for possible endogeneity, dividing the workers into high/low unemployment rate areas, dividing the sample into different groups of workers, and allowing for the correlation of errors across individuals within regions - but none of these reduces the wage curve elasticity below about -0.10 over the range of unemployment between 10% and 30%. At rates of unemployment above 30%, the wage unemployment elasticity falls to zero. The South African labour market displays considerable wage-flexibility in response to local labour market conditions across this large country.

The paper illuminates the important debate about the relevant definition of unemployment in South Africa. The South African government has adopted the narrow concept as the official definition and has thus implicitly de-emphasized estimates of broad unemployment. Yet, our work so far indicates that the broad definition of unemployment is the appropriate one for labour market analysis in South Africa since local wage determination takes discouraged workers into account as genuine labour force participants. Although even the narrow estimate of unemployment is high and justifies policy priority, the alliance between the ruling party and the trade union movement (representing the employed) might well lead to a playing down of the unemployment issue. Our argument in favour of the broad measure of unemployment therefore has policy relevance. Moreover, it is clear that unemployment contributes to poverty both directly and indirectly through its effect on the wage level.