The role of fertility in achieving Africa's schooling MDGs: Early evidence for sub-Saharan Africa

Abstract

In theory, declines in national fertility boost schooling by reducing age dependency, but questions remain about the size and catalysts of this dividend. We address these questions in sub-Saharan Africa (SSA) by using a detailed framework and decomposition methods. Results about catalysts suggest that, beyond policy, dividends depend on characteristics of fertility transitions and changes in employment, economic performance, and public commitment to education. Results about the size of Africa's schooling dividends are mixed. On one hand, the annual schooling resource per child grew on average by $73 between 1990 and 2005, with a third of this growth tied to trends in age dependency. Yet despite these nominal gains, Africa lost ground relative to the world partly because age dependency declined even more in other regions. Only after 2105, the Millennium Development Goals (MDGs) deadline, will Africa begin to narrow its gap vis-à-vis the world average. Also, dividends are predicted to accrue earlier among countries already having higher enrollments, suggesting that transitions may initially raise schooling inequality across sub-Saharan countries.

Keywords:

• fertility transition
• demographic dividend
• decomposition
• sub-Saharan Africa

Acknowledgments

We acknowledge suggestions and insights from conference participants at Yaoundé University (June 2009), Korea University (July 2009), Brown University (November 2009), and Cornell University (January 2010). Research was supported in part by resources and grants from the Hewlett Foundation, the Spencer Foundation, and the Polson Institute for Global Development.

Notes

1. Our focus here, which is on resources, not enrollments, facilitates international comparisons because resources may be a better indicator of the quality of education experience than mere enrollment (World Bank 2005).

2. The most basic approach to studying dividends is to rely on cross-country regressions of socioeconomic advances on age dependency. However, macro-level regressions have many statistical limitations, including outliers, measurement error, omitted variables, data comparability, endogeneity, auto-correlation, and regression to the mean. As a result, studies have focused on the micro-levels effects of sibsize on schooling (Cassen 1994).

3. Enrollment data are often more readily available in gross rather than the preferred net form. Also, net enrollments have an upper bound of 100%, meaning that progress is hard to measure in countries nearing universal enrollment. Finally, enrollments say little about school quality, a growing concern at a time when the pursuit of MDGs via free public schooling may overwhelm school systems, leading some countries to achieve nominal gains in enrollment at the expense of quality (UNESCO 2005). Focusing on schooling resources (rather than enrollments) thus improves international and historical comparison, if one is interested in quality of schooling.

4. More broadly, one can assess how global fertility transitions affect world inequality in schooling resources per child. For instance, if one takes the Mean Log Deviation (MLD) as a measure of inequality

where I=global inequality; w, represents the population of individual countries as shares of the world population.

In that case, the change in global inequality can be decomposed as in (4) below, a breakdown that isolates, among others, how age dependency in various world regions affect global inequality in children's schooling resources

5. This is not a hard and fast rule, however. A negative sign is problematic only when a positive change in an outcome is followed by negative change in the following outcomes. If the reverse is true (negative change followed by positive change), then there is less worry. Furthermore, the last step poses special problems when declines in gross enrollments (i.e., a priori a negative trend) can in fact be a positive development. For instance, countries with gross primary enrollments that exceed 100% may see their numbers decline as a result of declining repetition rates.

6. This value reflects the unweighted average for countries in the study sample. Note the difference with results for the weighted SSA average, which was a $13.80. That this increase was much smaller suggests some combination of sample selection and larger gains in r among small countries.

7. It is important to note here that, as an aggregate, the region had a public commitment to education 6% higher than that of the world average in 1990. This is likely due to the fact that large countries were able to allocate greater portions of their budgets during this period.

8. We chose the third rather than the top quartile since some countries in the top quartile had exceptionally high GNI growth during this period, e.g. Angola (21%) or Nigeria (12%).