Syndicate content

Social Mobility and Education

Servaas van der Berg's picture

In his post on this blog, Augusto Lopez-Claros correctly identifies illiteracy as an important factor in global inequality, and places the blame for much of the illiteracy that exists squarely at the feet of government choices. A perspective from South Africa – a country with extreme inequality – confirms that education may be the key to reducing inequality.

TS36-12 World Bank Not surprisingly, given their history, South Africans are obsessed with inequality. Income distribution features prominently in all political debates, in government policies and in the National Development Plan. Yet there is little understanding that the roots of this inequality lie in the labour market, particularly in the wage distribution, and that changing this distribution requires a dramatic improvement in the weak quality of most of South Africa’s schools.

Two relationships explain the central role of education in social mobility. The first, known as the Mincerian earnings function, is that, always and everywhere, more education is on average associated with higher income.  As in Latin America, in South Africa, this relationship is strongly convex, i.e. incremental returns to education are greater for higher levels of education (Figure 1).

The second regularity is that children who have richer parents generally perform better at school. Again, the South African relationship is strongly convex (Figure 2). As richer children get a better quality of education, they are more likely to progress to higher levels of education, which in turn bring them much higher rewards in the labour market. Since these two lines have a strong slope in South Africa, inequality is likely to perpetuate itself from generation to generation.  Social mobility will be limited.

Why do the returns to education rise so fast with the level of education? In most middle-income countries, it is because of a mismatch between the demand and the supply of skills. Growing economies of have an almost insatiable need for skills, but usually the education system fails to provide these skills in adequate numbers. Thus the skilled earn a large wage premium, one that will remain large as long as these skills are under-supplied and as long as there is an over-supply of unskilled labour. In South Africa someone with a university degree earns three times as much as a person with a school-leaving certificate (Grade 12), who in turn earns more than those with only primary education. This leads to a Gini coefficient of wages – considering only those employed – of 0.60.

Almost all other factors that influence overall household level inequality would tend to raise the Gini coefficient above even this high level: Household size and structure tend to exacerbate income inequalities between the rich and the poor-- higher fertility rates amongst the poor lead to large household sizes and many members too young to work; unemployment affects the poor more than the rich; and property income and dividends largely go to the rich. Thus only transfer incomes (social grants) favour the poor. These social grants are of great importance to reduce poverty in the South African context, but even though South Africa spends about 3% of GDP in the form of social grants (far more than any other developing country), this cannot make a significant dent in inequality. So one may conclude that in such circumstances wage inequality places a floor under overall inequality. That means the Gini for household income would be above 0.60.

To reduce overall income inequality, therefore, we need to contend with the causes of wage inequality. Minimum wage legislation and strong unions do have some effect, but these come at the cost of unemployment. We have to address the root cause of labour market inequality--the quality of education. A school system as weak as South Africa’s, where test scores in most schools are lower than in some far poorer countries (that also spend far less on education), cannot provide the rapid growth of skills that would reduce the wage premium. Thus the focus of policy should be on raising the quality of education, but this takes more than spending money, and is likely to be a slow process. Moreover, it takes even longer – a generation – to fully work its way through the labour market.

This is not a message that policymakers who are in a hurry to see results like to hear. Indeed, the racial dimension of South African inequality has been reduced substantially since the end of apartheid. But partly as a consequence, income inequality within the black population has risen to a magnitude of some of the most unequal countries in the world, while overall income inequality has remained stubbornly high. Non-income aspects of poverty have also been reduced through the provision of housing, water, sanitation and electricity to large numbers of people. But service delivery in health, education, welfare and police has remained dismal, despite massive allocations of public funds. Thus the prospects for reducing income inequalities systemically are not auspicious, and the time it takes for any successes to work through to the labour market is likely to be long.

Unlike the global context that Augusto refers to, the South African problem is not that inadequate fiscal resources are being devoted to education. The question is how to improve the quality of education that has remained weak despite fiscal resources. Reducing income inequality in South Africa may be even more difficult.


Submitted by Alan Gelb on

A very interesting piece. If the wage Gini is 0.6 for those actually employed one wonders what it would look like including those not employed and probably earning very little. If unemployment is in the range of 25%, this is equivalent to 33% relative to those actually employed. One can graft this increment on to the labor axis and estimate the new Gini. It comes out at about 0.7. This is getting up to the estimate of the South Africa Transformation Audit, which puts the Gini before taxes and transfers (that is the market income Gini) at 0.77. This is very high, but most is accounted for by the combination of wage dispersion and unemployment.

Submitted by Servaas on

You are right, once one accounts for unemployment, the Gini coefficient increases even more. In fact, South African broad unemployment (including discouraged work seekers) is closer to one-third of the labour force than the one-quarter you assumed, which means that the Gini amongst the full labour force is even higher still than you mention. So the very high South African Gini (values obtained differ greatly according to the data set used) is indeed largely accounted for by the combination of wage dispersion and unemployment. Servaas

Add new comment