Sustainable Development Banner
Knowledge2 banner

October 2004

Magnitude of social disparities in primary education in Africa: Gender, geographical location, and family income in the context of Education for All (EFA)

by Alain Mingat
Department of Human Development
Africa Region, World Bank

The paper was presented at a joint UNICEF-World Bank Conference organized in June 2003 Ouagadougou, Burkina Faso

Also available in French

I. Background on the Millennium Education Goals

Progress towards the millennium education goals will imply actions of a magnitude, which, for many African countries, will need to be much more intense than it has been in the past1. The indicator retained to identify the quantitative goal is universal completion of a primary education cycle of at least six years by 2015. A joint goal is also that the children will have acquired the intended knowledge. Six years of schooling are necessary in African contexts to ensure the retention of literacy skills at the adult age, as demonstrated by the following Graph 1.

Graph 1: Proportion of adults able to read easily against highest level of schooling achieved in Togo and Niger

Source: Data from MICS (UNICEF), 2000 and authors’ estimations

That being said, the road towards the fulfillment of this goal, whereby all the children of appropriate age on the continent will have accomplished at least six years of schooling in 2015, is still very long. The data contained in the following table demonstrate this reality.

Table 1: Proportion of children reaching the sixth year of schooling; regional averages and data from selected African countries, 1990 and 2000

Based on the information contained in Table 1, it is clear that the road towards universal completion of six years of primary schooling will be difficult for developing countries as a whole, but that it will be especially challenging for low-income countries from sub-Saharan Africa. At least three principal complementary reasons are responsible for this situation:

1. The first reason is that while it is certainly relevant to aim for universal completion of at least six years of schooling, this is much more ambitious that aiming to achieve a 100 percent Gross Enrollment ratio by 2015. Indeed, if access to the first year of primary schooling is a necessary condition for the realization of this objective, reaching this goal would imply that students admitted to school remain enrolled for the entire cycle. Data from 2000 demonstrate that the regional average (country weighted) of the gross enrollment ratio in low-income African countries is evaluated at approximately 77 percent; this figure suggests that only 23 percent are missing to reach 100 percent. Since the average completion rate only comes to 46 percent, it implies that 54 percentage points are short from the goal. The global challenge for African countries is therefore of an entirely different magnitude. Obviously, the road for some countries (Niger, Chad, Ethiopia, Malawi, Madagascar) is longer than for others, which means that policies for reaching universal primary schooling completion must be analyzed on a country-by-country basis.

2. The second reason is linked to the dynamics of movement towards the goal. This point is first appreciable when one notes that the countries will need to achieve much more progress in their rates of primary schooling completion in 15 years than they have in the past forty. The dynamics of movement can then also be appreciated by comparing the situation observed in 2000 with that occurring some years before. From a long-term perspective, African countries have gained between 25 and 30 points in completion rates between 1960 and 2000, which amounts to an average progress rhythm of between 0.6 and 0.8 points per year for the given time period. Looking at the dynamics during the last decade, the data in Table 1 indicate that the rate of primary education completion in low-income sub-Saharan African countries grew from 41 percent in 1990 to 46 percent in 2000. This gain of 5 percentage points in ten years corresponds to an average rhythm (averaged over the time period and over the set of countries in question) of only 0.5 points per year. If an annual rhythm of 1% per year was adopted as a reference for all the countries on the continent, a completion rate of only 60 percent would be achieved by 2015 and the goal of universal completion for the continent would be postponed until beyond 2050.

The rhythm noted during the last decade has however varied from one country to the next, being negative in some and positive in others. Various conflicts have made progress impossible in a number of countries on the continent. Indeed an examination of certain countries (such as the Gambia and Guinea, for example) reveals that remarkable progress in primary education completion rates has been achieved during the last decade. That being said, even for countries with the highest level of performance, it is difficult to envisage annual gains in primary education completion rates greater than 2.5 or 3.0 percent per year over a period of 10 years. Assuming the adoption of appropriate policies at the national level and the mobilization of international financial and technical support at the desirable scale, a number of African countries could reach universal primary education completion by 2015 if they could adopt a progress rhythm of about 3 percentage points per year. This will still not be sufficient for approximately 15 countries on the continent however. For these countries, the international community will either have to de facto accept that the formally defined goal will not be reached in the pre-determined timeframe, or instead consent to very particular efforts in favor of these countries to help them to reach the goal.

3. The third reason is in a way social and qualitative. Until now, we have examined numbers and implicitly accepted proportionality-based logical reasoning. Thus we have assumed that countries that have achieved gains of 2 percent per year in their primary education completion rates over the past several years will be able to maintain this rhythm towards universal completion for their entire school-age population.

To reason thus is to not recognize that the characteristics of those currently unable to complete primary schooling (either because they did not have access to a school or because they left it prematurely) are essentially different from those of students who complete it. Very clearly, education systems evolve by first providing services to the easiest populations to enroll (those that have a high level of demand for education and that are located in areas easy to serve). The systems must progressively include populations progressively more difficult to enroll in school (low level of demand, at-risk populations, areas with difficult accessibility) in order to improve their quantitative performance with regards to coverage. The data in the section of Table 1 on individual African countries during 2000 illustrates this situation very clearly.

On average, for the eight countries presented, the rate of primary education completion comes to 37 percent. This means that 63 percent of the age group has a level of schooling lower than a complete primary education (6 years for seven of the countries considered, 5 years for Mozambique). For girls, this proportion increases to 72 percent and to 81 percent if they live in rural areas. This means that there are 133 girls for every 100 boys who do not complete primary education schooling and who will have to be included in order to reach the goal of universal primary education completion. This difference is not only quantitative. Indeed, its existence indicates that obstacles to the schooling of girls are more significant in some countries, which will imply specifically aimed and occasionally difficult measures. The same applies to the contrast between urban and rural areas. In all countries, chances of having a complete primary education are significantly higher for urban dwellers than for rural dwellers. Policies must evidently be developed to improve the schooling of young people in urban areas (as much to improve access and retention as to improve the quality of services provided), but the bulk of the efforts to improve the rate of primary education completion must consist in including rural dwellers, and girls in particular. We will explore these issues of disparity and of the characteristics of the populations that will need to be included in order to achieve the goal universal primary education completion in the second part of this document. Suffices to remember at this point that reaching this goal will essentially imply including those who are readily excluded and that this population will be more difficult to enroll in schooling than those who are already enrolled.

II. Social characteristics that make a difference and that should be targeted

What are the personal and social characteristics of the children who will need to be included for the goal to be reached? Who are they and where are they? The answer to these questions must of course be provided for each individual country in order to identify the necessary aims and then determine the most appropriate strategies for including these children. Here we address these questions in a more general fashion, even if the evaluations we conducted were based on national data for a number of countries on the continent. The statistical basis we employed was established through an analysis of household surveys undertaken in twenty African countries2 around 2000 (MICS, EDS or consumption surveys for poverty analyses). Household surveys actually have three advantages over administrative statistics:

  1. Risks of under-evaluating and over-estimating school enrollments are averted. Under-estimation is possible when some schools do not send back the survey forms or when school principals in some cases have an interest in reported enrollment numbers that are lower than actual numbers. Over-estimation can occur when school principals have incentives to include students that they do not actually have because monetary management incentives or the resources allocated by the Ministry to the schools depend on the reported numbers.

  2. It is possible to avoid having to resort to demographic projections. Indeed, these projections are occasionally linked to significant uncertainties, notably when the population census on which they are based are themselves uncertain (due to population displacements or imperfect counting for random or specific reasons) or relatively dated while temporal evolutions of demographic parameters (demographic transition, impact of AIDS, changes in infant mortality indicators) could not be well integrated.

  3. Interesting personal and social variables that help to understand school enrollment are available in household surveys. These variables are not always available in school statistics. Information on gender or region is available through both school statistics and household surveys; but this is not the case in most cases for distinctions between urban and rural areas. Indeed, household surveys provide information on the family’s place of residence but only sometimes the area in which the school (but not the family) is located. At the primary education level, assimilating the school’s location to the area where the children live does not generally pose difficulties, but the same can not be done at the secondary level or later. In these cases, urban schools often enroll children who live in rural areas. Furthermore, information regarding family income, or the fact that a child is orphaned or disabled for example, is evidently extremely important both with regards to poverty reduction strategies and to the achievement of the goal of universal primary education completion. Such information is only available in household surveys.

The analyses carried out in the twenty countries of our sample notably facilitated an evaluation of the magnitude of the global disparities among population groups within the country along the lines of gender, urban/rural distinction, and income quintiles3. Furthermore, the statistical analysis shed some light on indicators such as gross rate of primary enrollment, rate of access to first year of primary schooling, retention rate throughout the primary schooling cycle, and rate of completion of this same cycle (the reference criterion for the millennium education goal).

The analysis results may be read in a variety of complementary manners: i) we may seek to evaluate the respective weight of each of the three social factors considered (gender, geographic location and income) globally for the sample of studied countries; ii) we may then seek to find the extent to which overall differences exist in the levels of social disparity in primary education from one country to the next; iii) we may finally seek to cross these two perspectives and to identify, country by country, the social characteristics around which the most significant differences occur. We shall limit ourselves to the first two aspects in this document.

II.1 Relative weight of disparities according to gender, geographic location, and income

The disparity analysis was conducted on the four main indicators for primary education: Gross Rate of Enrollment, Rate of Access to the first year of schooling, transversal Retention Rate between the first and last years of the cycle, and finally, the Rate of Completion of Primary Education. Moreover, the measure of disparities was based on the difference of observed rates as well as that of their ratio (this ratio is often preferable) within the categories built according to gender (boys and girls), geographic location (urban and rural), and household income (first and fifth quintiles). Table 2 presents a summary of the main results.

Table 2: Social disparities according to various indicators for primary education for the 21-country sample

We may first observe that the overall averages for our 21-country sample are reasonably close to the numbers proposed by Bruns, Mingat, and Rakotomalala (2003) from administrative statistics for the 33 low-income countries of the region. Indeed, the Gross Enrollment Rate is 77 percent in that study, compared to 78 percent in our sample, and the Completion Rate is 44 percent, as compared to 42 percent in our sample.

The data in Table 2 provide an overall idea of the relative magnitude of disparities with regards to gender, geographic location, and family income in primary education for the set of countries in the sample. First, we may note that the figures about the different outcomes are relatively close to one another, whether we look at access, retention, or completion. We may also observe that the measure of relative magnitude of disparities in the three population groups is approximately the same regardless of whether the analysis is based on the difference between rates or on their ratio. For practical purposes, the commentary in this document is made based on the average of the four rates and on the measure of disparities based on the ratios of rates. The results are unambiguous: i) disparities in primary schooling between urban and rural areas (in favor of urban dwellers) are on average two to three times greater than those existing between boys and girls (in favor of boys); ii) the gaps in primary schooling between children belonging to the first and fifth income quintiles (in favor of wealthier children) are three to four times greater than those observed on average with regards to gender.

Although these results are probably very strong by themselves, they still call for three types of remarks:

In terms of education policy, activities established to fight against disparities have often focused on gender. Indeed, programs to improve the enrollment of girls exist in almost all aid agencies. These programs may have had a positive impact since the disparities against girls in primary education have significantly decreased over the past 15 years. We do not have relevant data to help us determine to what extent the same trend applied to disparities linked to geographic location and income. Regardless, there is no doubt that today, in order for the millennium goal of universal primary education completion to be achieved, the challenge is to help those young people who do not currently have a complete primary education to have one tomorrow, knowing that the people in this category are for the most part rural dwellers and the poorest. Here lies the priority for action.

On a practical level, targeted actions can be easier if the targeted group is clearly identifiable. Distinguishing boys from girls is easy; however, it is difficult to conceive of specific actions for girls that would exclude boys (it may actually be better this way). Even if the distinction is sometimes a little conventional, urban and rural areas are also easily identifiable without targeting or exclusion difficulties. Income level is harder to determine, notably on an individual basis. Yet income is the variable to which the most significant disparities are attached. A possible means of establishing concrete actions could be to construct a poverty map and to target areas with a high proportion of families living in the deepest poverty. In many cases, these areas are located in rural areas. However, this does not apply to all rural areas within a country, while some urban areas do have a significant proportion of families living in very difficult conditions.

Until now, we have considered the country sample as a whole, implying that it may be possible to hold a generic discourse regarding low-income countries in sub-Saharan Africa. As we will now see, this might not be the case.

II.2 Overall level of social disparities in the various countries studied

We can calculate an aggregate indicator for each country, just as we did for the four indicators for primary education for the sample as a whole. Furthermore, in order to obtain an overall measure, we also aggregated the results over the three population groups to identify a global indicator of the magnitude of social disparities in each country. Because basic data were not available for all parameters characterizing student flow (access, retention, completion, and GER) in all the countries and for the three population groups, we had to resort to intermediary estimations in order to obtain decent internationally compatible indicators. These circumstances also led to limit the number of countries somewhat by giving up on the countries for which the number of raw data was insufficient. Table 3 below offers the results of the analysis. Graph 2 provides a graphic illustration of differences among the overall indicator of social disparities in primary education for all the countries (calculated here with differences in rates, although the image would have been almost identical if we had used ratios).

Table 3: Overall indicator of social disparities in primary education in 21 countries

The numbers clearly indicate that the overall indicator of social disparities (aggregated according to gender, geographic location and income level) in primary education is very much at variance among the countries of Sub-Saharan Africa in this sample. The indicator is on average lower in Anglophone countries (20.4 points) than in Francophone countries (29.9 points) and in Lusophone countries (34.0 points). The overall indicator is relatively on the low side in Uganda, Mauritania, and Rwanda, but very much on the high side in Benin, Guinea, and Burkina Faso.

If we analyze the level of disparities within one country, we must expect it to be linked to the level of system coverage. If we take the extreme case of universal coverage, by definition, the entire population is in school (the wealthy and the poor; boys and girls; urban and rural dwellers) and no disparity exists among the groups. However, if coverage is only partial, disparities can occur. The expansion route for a system begins with the better-off groups of population, easy to reach segments that have a strong demand for schooling, followed by harder to reach groups with lesser demand for educational services. The final route towards universal coverage will then have to turn to those living in the most remote areas and/or who have the weakest demand for schooling. With a “sequential” structure of this type, it can be expected that disparities, which are high when school coverage is low, will progressively decrease as coverage increases. This is what Graph 3 displays.

The graph demonstrates the existence of i) a notable structural relationship implying that disparities tend to decrease “mechanically” as coverage improves, and ii) substantial variations on either end of the average relationship, implying that for a given level of coverage, some countries are more inequitable than others.

1For an analysis of the necessary conditions for reaching the millennium education goals, the reader may consult the recent study by Bruns, Mingat and Rakotomalala, “Achieving universal primary schooling in 2015; A chance for all of the world’s children”; World Bank, 2003.
2These countries are Angola, Benin, Burundi, Cameroon, the Central African Republic, Chad, Côte-d’Ivoire, Guinea, Guinea-Bissau, Madagascar, Malawi, Mauritania, Niger, Nigeria, Uganda, Rwanda, Sierra-Leone, Togo, and Zambia.
3When this information were not directly available in the survey, methods of factorial analysis were used to evaluate the income quintiles based on the number of active people and the living conditions of households.


SD Homepage l Back to Top l FAO Homepage