Educational mobility

It seems parental education matters a great deal for children’s educational outcomes (Figure 3.1).¹⁴ Measured as the proportion of the variation in a child’s educational attainment that is explained by variation in parental educational attainment, there is a significant degree of transmission from one generation to the next. ¹⁵ Furthermore, there is no downward trend – even among younger cohorts parental education explains more than 60% of the variation.¹⁶ In general, these results are consistent with those obtained from those household surveys that contain information on parental education.¹⁷

Breaking this regional result down reveals considerable differences at the country level (Figure 3.2). Guatemala exhibits the highest coefficients for all indicators, implying the lowest mobility. At the other end of the scale, Costa Rica, Honduras, El Salvador and Colombia present considerably higher levels of mobility. Chile’s position is surprising, showing low levels of mobility on this measure.

These differences are economically significant. For example, the underlying elasticities imply that a 4-year difference in parental education would on average imply 1.6 years more of education for the next generation in Costa Rica, while in Guatemala the equivalent figure would be 3.4 years. Given a year of additional education is worth 12% – the average return to education in Latin America¹⁸ – these extra years could translate into a differential in wage earnings of 19% and 41%, respectively.¹⁹

Latin America in the global context

Latin American countries are well down the world rankings in terms of educational mobility. They rank below not only OECD countries but also their developing peers (Figure 3.3). To the region’s high level of static income inequality can, it seems, be added very unequal access to opportunities to progress.²⁰

Mobility and the middle sectors

Is this bleak picture repeated across all levels of education? The answer can be explored from two viewpoints.

The first is the correlation between parental and child education for different levels of child education (Figure 3.4). For women and men alike, the importance of parental education decreases at higher levels of educational outcomes. Thus, for those with low or medium levels of education, parental background is more important than for those at the higher ends of the distribution. How do the middle sectors perform within this? Combining the household data from Table 3.1 with the data used in Figure 3.3 suggests that middle-sector children will typically lie in the fifth and sixth deciles of Figure 3.4. The importance of parental education in these deciles is not significantly different from that at the lower tail of the distribution, while it is significantly higher than for the ninth decile (where people on average have 15 years of education).

The other way of looking at educational mobility is to compute transition matrices between the highest level of education reached by the parents and the highest degree reached by the child, differentiating by gender (Figure 3.5). For very low levels of parental education there is a high likelihood that children will perform better. A person whose parents were illiterate, for example, has an almost 80% probability that they will achieve at least some primary education. This is the same general trend identified in Table 3.1 of faster increase in educational attainment at the bottom of the distribution. However, at levels of education linked to the middle sectors ("some secondary education" and up) mobility is much lower, while at the upper end the positive influence of parental achievement again rises. Table 3.A1 in the statistical annex presents the entire transition matrices.

The overall conclusions are the same. At low levels of parental education ("illiterate" to "complete primary"), the child generally performs better. At the middle of the distribution ("incomplete secondary" and "complete secondary"), the level of education attained by the offspring tends to peak around complete secondary education. Even though this group has better access to tertiary studies, the gap with those whose parents have tertiary studies remains large. For example, out of every 100 children who have parents with incomplete secondary education roughly 10 finish tertiary studies, while for those who have parents with completed tertiary education the equivalent figures are 58 for women and 47 for men. To put this in context, about 80% of the 25- to 44-year-old cohort have parents with incomplete secondary education or less.²¹ The good news is that for those with the most unfavourable family background there seems to be upward mobility, and for those at the top downward mobility is very unlikely. But the middle sectors seem to remain trapped, unable to break into tertiary education.²² In this regard, the U-shape of the graph is striking.

Younger cohorts

The data used so far to measure mobility are based on people who have already completed their educational cycle (at least 25-years old in 2009). The analysis is therefore open to the criticism that more recent policy changes may not be captured. From a policy perspective, it is interesting to focus on the population still in the educational system, since they would be the target of any interventions made today.

A number of researchers have pursued this idea in Latin America.²³ These studies have analysed the importance of parental background (education and income, among other variables) in explaining variations in the schooling gap between households – the difference between the highest grade the child has achieved and where it should be according to its age. The thinking behind this is that when family background is an important explanatory factor these characteristics are more likely to persist across generations and therefore mobility will be lower.

We can test this by looking at the evolution of a suitably constructed social-mobility index (Figure 3.6). For 11 out of the 16 countries considered, mobility has increased (though the change is only statistically significant for Brazil, Chile, Peru and Venezuela), while mobility has declined significantly only in Colombia and Uruguay. The picture painted supports the view that some countries have improved mobility in recent times. Chile and Peru, for example, which seem low-mobility countries when analysed using older cohorts, appear much more mobile here. In the case of Chile, this is consistent with evidence that the importance of family background in explaining test scores in mathematics has diminished significantly over the last decade.²⁴