Population Studies

A Journal of Demography

ISSN: 0032-4728 (Print) 1477-4747 (Online) Journal homepage: https://www.tandfonline.com/loi/rpst20

Trends in the age at reproductive transitions in the

developing world: The role of education

John Bongaarts, Barbara S. Mensch & Ann K. Blanc

To cite this article: John Bongaarts, Barbara S. Mensch & Ann K. Blanc (2017) Trends in the age
at reproductive transitions in the developing world: The role of education, Population Studies, 71:2,
139-154, DOI: 10.1080/00324728.2017.1291986

To link to this article: https://doi.org/10.1080/00324728.2017.1291986

© 2017 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group

Published online: 11 Apr 2017.

Submit your article to this journal

Article views: 7142

View related articles

View Crossmark data

Citing articles: 49 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=rpst20
Population Studies, 2017
Vol. 71, No. 2, 139–154, https://doi.org/10.1080/00324728.2017.1291986

Trends in the age at reproductive transitions in the

developing world: The role of education

John Bongaarts, Barbara S. Mensch and Ann K. Blanc

The Population Council

Girls’ school participation has expanded considerably in the developing world over the last few decades, a
phenomenon expected to have substantial consequences for reproductive behaviour. Using Demographic
and Health Survey data from 43 countries, this paper examines trends and differentials in the mean ages
at three critical life-cycle events for young women: first sexual intercourse, first marriage, and first birth.
We measure the extent to which trends in the timing of these events are driven either by the changing
educational composition of populations or by changes in behaviour within education groups. Mean ages
have risen over time in all regions for all three events, except age at first sex in Latin America and the
Caribbean. Results from a decomposition exercise indicate that increases in educational attainment, rather
than trends within education groups, are primarily responsible for the overall trends. Possible
explanations for these findings are discussed.

Keywords: trends; transition to adulthood; life cycle; first sex; first marriage; first birth; developing world;
decomposition; education; educational composition

[Submitted June 2016; Final version accepted December 2016]

Introduction higher likelihood of maternal death (Blanc et al.

2013; Nove et al. 2014), infant death, and other
A successful transition to adulthood includes ade- poor child health outcomes (Finlay et al. 2011).
quate preparation for adult roles and the capacity Moreover, the timing of births has an effect on
to make informed decisions about the timing of population growth. For all these reasons, there is
key life events (Lloyd 2005). First sexual inter- considerable interest on the part of researchers
course, first marriage, and first birth are among and policymakers in trends that are the focus of
the most consequential of these events. Taking on this study: ages at first sex, marriage/union, and
the roles of sexual partner, spouse, and parent too childbirth in the developing world.
early may undermine the success of this transition, The paper begins with a review of the literature
jeopardize a young person’s health and well-being, on the impact of women’s schooling on fertility
and have deleterious effects on the health and and on the timing of first sex, first marriage, and
well-being of her offspring (Lloyd 2005). In less- first birth. This is followed by a detailed empirical
developed-country settings, where HIV is endemic analysis of levels and trends in the ages at these
and where young women suffer disproportionately life-cycle events, based on data collected by Demo-
from sexually transmitted infections (STIs), the graphic and Health Surveys (DHS) since 1986 in 43
earlier sex is initiated, the longer the exposure to less developed countries in Asia and Northern
the risk of infection. On average, the younger the Africa, Latin America and the Caribbean, and
age of a woman at marriage, the larger the age sub-Saharan Africa. The mean ages at events are
difference between spouses, the less agency estimated with a method that allows period-based
women are reported to have within marriage, and measurements to provide more up-to-date results
the higher the likelihood of marital dissolution than the conventional cohort-based approach used
(Singh and Samara 1996; Mensch 2005). And in earlier studies. Girls’ school participation has
finally, early childbirth, before achieving full phys- expanded considerably in many of these countries
ical maturity around age 16, is associated with a over the last few decades, with substantial

© 2017 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License
(http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium,
provided the original work is properly cited, and is not altered, transformed, or built upon in any way.
140 John Bongaarts et al.

consequences for reproductive behaviour. We provide the tools for learning and acquiring knowl-
examine the extent to which trends in the timing edge (Jejeebhoy 1995), thus improving women’s
of events are driven by changing educational attain- ability to understand and absorb family planning
ment within countries—that is, the educational com- messages and to consider the desirability of a
position of the population—or by changes in smaller family (Bledsoe and Cohen 1993). Most of
behaviour within groups that share the same level the hypothesized effects are direct but some are
of education. The discussion and conclusion sum- indirect and work through, for example, improved
marize the main findings of this analysis. child survival and delayed marriage, for which
many of the mechanisms overlap with those listed
above (Ainsworth et al. 1996).
Background

The effect of girls’ schooling on subsequent Girls’ schooling and the timing of reproductive
fertility events

Over 35 years ago, Caldwell (1980) argued that Recently, with the unparalleled growth in the popu-
mass education—which, he asserted, altered tra- lation of young people, increasing attention has
ditional family production and family dynamics— been devoted to identifying conditions for a success-
was the dominant force behind fertility transitions. ful transition to adulthood and the role of girls’
Since then an extensive demographic literature schooling in the timing of reproductive events
has documented and analysed the inverse associ- (Lloyd 2005, 2010; Mensch et al. 2005; Clark and
ation in the developing world between years of Mathur 2012; Grant 2015). DHS data indicate that
schooling attained by women and the number of girls who are enrolled in school are less likely to
children ever-born (Jejeebhoy 1995; Martin 1995; report initiating sex than their same-age counter-
Ainsworth et al. 1996; Bledsoe et al. 1999). The parts who are not enrolled (Lloyd 2009). DHS
magnitude of education differentials among data also indicate later ages at sexual initiation,
women varies between countries and the relation- first marriage, and childbearing among those with
ship is not always monotonic (Jejeebhoy 1995; more years of schooling (Lloyd 2005). In the
Bledsoe et al. 1999); for example, some narrowing majority of less developed countries, schooling is
appears to emerge as countries reach the end of considered incompatible with marriage and child-
the transition (Bongaarts 2003). Nevertheless, bearing, so that those who remain in school
differences in fertility between the more and less during adolescence delay marriage and childbear-
educated are observed in nearly all countries. ing, and those who marry early drop out of school
Researchers have discussed a number of mechan- (Lloyd and Mensch 2008). These associations have
isms underlying the association between formal led many to argue that increased schooling is the
schooling of girls and subsequent fertility, fertility dominant force behind the observed rise in age at
preferences, or both, although data are often marriage (United Nations Dept. of Economic and
lacking to support these assertions. A number of Social Affairs 2002). However, a regression analysis
factors that underlie the negative gradient of the amount of change in early marriage that
between education and fertility have been men- might be expected, based on change in educational
tioned, including greater autonomy, exposure to attainment from older to younger cohorts in 39
new gender and childbearing norms, improved DHS countries, revealed that in 15 countries the
knowledge about behaviours that enhance the observed change was lower than the expected
health of women and children, development of change. In about half of these 15 countries, the
social capital to navigate health institutions, altera- probability of early marriage actually increased
tion of childbearing preferences, growth in between cohorts despite the increase in schooling
women’s earning potential, and the concomitant (Mensch et al. 2005). An earlier analysis of
rise in the opportunity cost of women’s time, assor- change in four Latin American countries from the
tative mating, and increased material aspirations mid-1970s to the mid-1980s also concluded that
(Jejeebhoy 1995; Subbarao and Raney 1995; Ains- increases in education during that period were not
worth et al. 1996; Appleton 1996; Bledsoe et al. associated with later marriage, although improve-
1999; Diamond et al. 1999; Lloyd and Mensch ments in education did account for a large pro-
1999; Basu 2002; Ferré 2009; Behrman 2015; portion of observed fertility decline (Weinberger
Grant 2015). Literacy has also been said to et al. 1989).
 Trends in age at reproductive transitions 141

The endogeneity of schooling and country schooling experience’. The question

reproductive behaviour remains, however, whether the effects of girls’ edu-
cation on reproductive behaviour that have been
The question of whether the association between observed historically are compromised by the
girls’ education and reproductive behaviour in the declining school quality that has accompanied the
developing world is causal has generated consider- recent expansion of schooling in many less devel-
able interest in the last few years. Are the observed oped countries (Pritchett 2013). Will schooling,
differentials in reproductive behaviour between the given low and perhaps declining quality, continue
better and less well educated a result of selectivity to have a dampening effect on marriage and ferti-
of those with higher grade attainment? Using data lity, and, if it does, are the observed effects a func-
from sub-Saharan Africa, researchers have tion of more women becoming educated or of a
addressed the potential endogeneity of schooling change in preferences, for example, diffusion of
(Ferré 2009; Behrman 2015; Glick et al. 2015; ideas regarding the negative consequences of child
Grant 2015). One study found that the failure to marriage and early childbearing?
address endogeneity led to ‘substantial overestima-
tion’ of the effect of schooling on the timing of mar-
riage and childbirth among young women in Contributions of this analysis to the literature
Madagascar, although a significant effect was still
observed (Glick et al. 2015). Three studies have The preceding summary of the current literature on
taken advantage of natural experiments in the the association between education and reproductive
form of policy reforms in order to ‘instrument’ for behaviour in the developing world sets the context
years of schooling, and two found that education for our analysis and raises important questions best
still had significant effects on fertility timing answered by in-depth country analyses. We contrib-
(Kenya: Ferré 2009) and desired family size ute to this extensive literature by investigating
(Malawi, Uganda, and Ethiopia: Behrman 2015). whether: (1) increases or decreases in age at the
The one exception is a study by Grant (2015) that time of reproductive events are observed in recent
investigated the effect of the introduction of free surveys and (2) if so, to what extent those changes
primary education and subsequent expansion of are associated with shifts in the educational compo-
secondary schooling in Malawi. These increases in sition of the population of females, and to what
girls’ education did not lead to a delay in age at extent they are associated with alteration in behav-
first birth, which she argued may be because of iour within education groups. Given that girls’ edu-
poor and declining school quality with the large cation has been promoted as the priority focus for
influx of students. twenty-first-century population policy (Lutz 2014),
an examination of the contribution of girls’ schooling
to recent changes in the timing of first sex, marriage,
Declining school quality and reproductive and childbearing is warranted.
behaviour

The situation in Malawi is unlikely to be unique and Data and methods

prompts consideration of whether the benefits of
education for reproductive behaviour will be rea- The primary sources of data for this study were DHS
lized in countries where expansion of school partici- conducted in less developed countries since 1986.
pation is not accompanied by significant One of the distinguishing characteristics of the
improvements in learning outcomes. Basu and Ste- DHS is the comparability of the questionnaires. We
phenson (2005) observed that child survival was included in the analysis data from countries with
higher for mothers in India who did not complete nationally representative surveys covering at least
primary school relative to mothers with no edu- two timepoints; this resulted in data from 43
cation; it would appear that even a small amount countries in Asia and Northern Africa, Latin
of schooling, likely insufficient to confer literacy, America and the Caribbean, and sub-Saharan
may yield advantages in providing skills to Africa. To minimize sampling error in the estimates,
improve health. As Basu (2002, p. 1780) has we excluded surveys with sample sizes of fewer
argued elsewhere, there may be benefits to school than 3,000 currently married women. The threshold
attendance ‘without there being much that is com- of 3,000 was selected (somewhat arbitrarily) to mini-
pellingly positive about the typical developing mize the loss of countries from the study while still
142 John Bongaarts et al.

excluding potential outliers. We also excluded advantage of being available for all DHS surveys
countries with a population size of less than 2 but there are also drawbacks: (1) the median
million in 2000 and countries with inter-survey inter- refers to past experience of cohorts and is therefore
vals of five or fewer years. Several of the smallest not as current as is preferable for many analytic
countries and island nations are outliers in our ana- purposes; (2) the retrospective reporting of the
lyses for reasons that are not entirely clear, but date of life-cycle events may suffer from recall
high migration among young adults could be a errors that are likely to increase as time since the
factor. Short inter-survey intervals produce unstable event increases; and (3) the cohort median as calcu-
trend estimates. If a country had a first survey that lated by DHS is not independent of the quantum of
did not collect information on the timing of first first events and can change over time, even if the
sex, the next available survey with such data was mean is constant. For these reasons, we prefer the
selected. period mean as our primary indicator of the
The following countries were included in the timing of first birth, marriage, and sex. Regional
analysis: means are unweighted averages of country means
Asia and Northern Africa: Bangladesh, Cambodia, so that each country is given equal weight.
Egypt, India, Indonesia, Jordan, Morocco, Nepal,
Pakistan, Philippines, Turkey.
Latin America and the Caribbean: Bolivia, Brazil, Limitations
Colombia, Dominican Republic, Guatemala, Haiti,
Honduras, Peru. Our analysis has several limitations. First, the
Eastern and Southern Africa: Ethiopia, Kenya, number of countries included in each region
Madagascar, Malawi, Mozambique, Rwanda, Tanza- ranged from eight in Latin America and the Carib-
nia, Uganda, Zambia, Zimbabwe. bean to 14 in Middle and Western Africa. The
Middle and Western Africa: Benin, Burkina Faso, countries for which we had data are not necessarily
Cameroon, Chad, Democratic Republic of the representative of their regions. The proportion of
Congo (DRC), Côte d’Ivoire, Ghana, Guinea, the regional population covered by these countries
Liberia, Mali, Niger, Nigeria, Senegal, Togo. is 90 per cent in Middle and Western Africa,
In parts of the results and discussion, the latter two 75 per cent in Eastern and Southern Africa, 74 per
groups are combined and referred to as sub-Saharan cent in Asia and Northern Africa, and 54 per cent
Africa. Details of the survey years are provided in in Latin America and the Caribbean. Second,
Appendix A. several countries included in the analysis had not
For each of the surveys we estimated the period had a very recent DHS survey. The most recent
mean ages at first sex, first marriage, and first survey took place on average in 2010 but the
birth for the five years before the date of the range is from 1996 to 2014. It is possible that if
survey, by highest level of education attended (i.e., we had more recent data, particularly for Latin
no schooling, primary, and secondary or higher). America, and could include more of the populous
(Throughout this study the term ‘marriage’ includes countries (e.g., Mexico and Argentina), the results
consensual union.) The method for calculating these would be somewhat different. Third, the number
mean ages is described in Bongaarts and Blanc of years of schooling represented by the primary
(2015) and summarized in Appendix B. The or secondary level varies among countries. Finally,
period mean age at a life-cycle event is defined as the aggregate analyses mask important country
the mean age at which women would experience trends and our general explanations for observed
the event if they lived through the reproductive aggregate trends, by design, may not reflect particu-
years having the age-specific event rates observed lar country experiences.
in a particular period. The calculation of the mean
uses five-year age-specific rates of events as
reported by female respondents in the DHS. This Results: mean ages at first sex, marriage, and
general approach is widely used to calculate birth
current mean ages at first birth in countries with
vital statistics (Human Fertility Database 2015) Levels
but is not used in country reports published by
the DHS. Instead, DHS reports provide median Figure 1 plots the unweighted regional averages for
ages at past events as reported by women of differ- the country-level means of each of the three life-
ent current ages. These cohort medians have the cycle events: first sex, first marriage, and first birth
 Trends in age at reproductive transitions 143

Figure 1 Mean ages at first sex, first marriage, and first birth for women, averages by world region, c.2010
Notes: Regional averages are unweighted. Data are from the most recent survey in each country (c.2010).
Average age at first sex was not collected in several countries in the Asia/Northern Africa region.
Source: Authors’ calculations from DHS data on 43 countries, 1996–2014.

as estimated from the most recent DHS survey The averages plotted in Figure 1 conceal con-
(c.2010) in 43 countries. Substantial regional differ- siderable variation among country estimates. The
ences are apparent, with countries in Middle and figure also gives no indication of the correlation
Western Africa having the lowest average values between means of successive events. To examine
(17.9, 20.3, and 20.9 years for these events, respect- these issues, Figure 2 plots the mean age at first
ively), and Eastern and Southern Africa, and Latin marriage against the mean age at first sex for indi-
America and the Caribbean, having intermediate vidual countries. A positive correlation (r = 0.58)
values for first sex and birth. Latin America and exists between these variables, with the diagonal
the Caribbean had the highest age at first marriage line representing the points at which the means
(21.7) and Asia and Northern Africa the highest are equal. It is noteworthy that in all but one
age at first birth (22.8). The average age at first sex country the mean age at marriage was either
in Asia and Northern Africa was not available equal to or greater than the mean age at first sex.
because in several of the countries in this region The difference between the mean age at sex and
the question about first sex was not included in marriage reached a high of five years in Haiti,
surveys. where premarital sex was common and marriage
The time between the average ages at first sex was late.
and first marriage was around two and a half Figure 3 shows a strong correlation (r = 0.84)
years in sub-Saharan Africa and in Latin America between country mean ages at first birth and first mar-
and the Caribbean, indicating a substantial preva- riage. The majority of countries fall above the diagonal
lence of premarital sex. The difference between line, which indicates that the mean age at birth is
the average ages at first marriage and first birth higher than the mean age at first marriage. However,
ranged from 0.1 years in Eastern and Southern a substantial number show the reverse, which is
Africa to 1.4 years in Asia and Northern Africa. It expected in countries with premarital childbearing.
should be noted that the differences in intervals
between average ages at successive events are not
equal to the average interval between these events Trends
at the individual level, because some women who
have sex never marry and some who marry never To examine trends in the various indicators, we com-
have a child. pared the values estimated for the most recent survey
144 John Bongaarts et al.

Figure 2 Mean age at first marriage by mean age at first sex, women in 40 countries, c.2010
Note: The diagonal line represents the points at which the two means are equal. Data are from the most recent
survey in each country (c.2010). This figure shows 40 countries rather than 43, because average age at first sex
was not collected in several countries in the Asia/Northern Africa region.
Source: As for Figure 1.

Figure 3 Mean age at first birth by mean age at first marriage, women in 43 countries, c.2010
Note: The diagonal line represents the points at which the two means are equal. Data are from the most recent
survey in each country (c.2010).
Source: As for Figure 1.
 Trends in age at reproductive transitions 145

Figure 4 Trends in mean ages at first sex, first marriage, and first birth for women, by world region, c.1993–2010
Notes: Regional averages are unweighted. Data are from the first available survey (c.1993, shown on the left-
hand side) and most recent survey (c.2010, shown on the right-hand side) in each country. Average age at
first sex was not collected in several countries in the Asia/Northern Africa region.
Source: Authors’ calculations from DHS data on 43 countries, 1986–2014.

with those observed at the first available survey. On remained very common in parts of Middle and
average the most recent surveys occurred in 2010 Western Africa in 2010, with the highest values in
and the first surveys in 1993, thus giving an average Niger, Chad, and Mali (Figure 5(a)). In Asia and
interval between surveys of 17 years. (The first and Northern Africa, child marriage was high in Bangla-
last surveys in some countries deviate significantly desh, Nepal, and India (Figure 5(d)). Second, the
from the average. For example, the first and last prevalence of child marriage declined in 34 countries,
surveys in Brazil were conducted in 1986 and 1996, but rose in nine countries (e.g., Malawi and Madagas-
respectively.) As shown in Figure 4, the regional car in Figure 5(b)). Third, the pace of decline between
mean ages at reproductive events generally rose c.1993 and c.2010 was typically rather slow. Unless this
between c.1993 and c.2010. In particular, the mean pace accelerates, child marriage will remain wide-
ages at first sex, first marriage, and first birth all spread in parts of sub-Saharan Africa and Southern
rose in sub-Saharan Africa, as they did for first mar- Asia for the next few decades.
riage and first birth in Asia and Northern Africa.
(This observation applies to the regional averages
of country values, and not necessarily to every indi- Results: the role of education
vidual country.) In contrast, in Latin America and
the Caribbean, the timing of first birth did not Trends in education
change significantly and the regional mean age at
first sex declined substantially. This decline occurred Since the early 1990s, substantial investments have
in seven out of the eight Latin American and Carib- been made in achieving universal primary education
bean countries. and gender parity in education, with investment accel-
The upward trends in all regional mean ages at first erating after the adoption of the Millennium Develop-
marriage suggest that the prevalence of child marriage ment Goals in 2000. Enrolment in primary school has
was declining. This inference is examined in Figure 5, increased rapidly in many regions of the world and the
which plots the trend between c.1993 and c.2010 in the gender gap has narrowed, although educational
percentage of women aged 20–24 who married before attainment levels still favour males, especially at sec-
age 18, that is, who were married as children. Several ondary and higher levels (United Nations Dept. of
conclusions can be drawn. First, child marriage Economic and Social Affairs 2010). Levels of adult
146 John Bongaarts et al.

Figure 5 Percentage of women married before age 18, by world region, c.1993–2010
Note: Data are from the first available survey (c.1993) and most recent survey (c.2010) in each country.
Source: As for Figure 4.

literacy have increased since 1990 in all developing women had achieved at least some secondary edu-
regions; for example, the percentage literate among cation c.2010, whereas less than 45 per cent had any
women in sub-Saharan Africa rose from 43 to 53 per secondary education in the early 1990s.
cent between 1990 and 2008 and from 34 to 51 per
cent in Southern and Western Asia. However, the pro-
portion of all illiterate adults that are women has Education differentials
remained the same at approximately two-thirds,
even as the illiterate population has declined Trends in mean age at event by region and by highest
(UNESCO Institute for Statistics 2010). level of education attended are plotted in Figures 7–9.
Figure 6 shows the shift in the educational distri- These results confirm the expected correlation with
bution of women at first birth between c.1993 and education; in every region, for each of the three
c.2010 in the 43 countries included in our analysis. events, and at both points in time, a higher level of
(The educational distribution by age is weighted by education was associated with a higher mean age.
the age distribution of first births in the five years Overall, the difference in mean age between those
preceding the survey, because education levels vary with secondary and those with primary education
considerably with age, younger cohorts having was much greater than the difference between
higher levels than older cohorts.) In all regions, the women with primary and those with no education.
change was strongly positive. By c.2010, in all Another key finding is that the trends for all edu-
regions except Middle and Western Africa, fewer cation groups together (represented by thicker lines
than one in five women had received no education; on the figures) had steeper positive slopes (or less
in Middle and Western Africa 39 per cent of negative slopes) than the slopes of the education-
women were still without any formal schooling specific trends. For example, the mean age at first
c.2010. In Latin America and the Caribbean and in birth changed little over this period for Latin
Asia and Northern Africa, almost two-thirds of America and the Caribbean in the aggregate, but
 Trends in age at reproductive transitions 147

Figure 6 Educational distribution of women at age of first birth, by world region, c.1993–2010
Notes: Data are from the first available survey (c.1993, shown in the left-hand bar) and most recent survey
(c.2010, shown in the right-hand bar) in each country. The educational distribution by age is weighted by the
age distribution of first births in the five years preceding the survey.
Source: As for Figure 4.

the trends for the three education-specific means in weight given to the more highly educated groups in
this region were all downward. The explanation for the calculation of the overall mean rose over time,
this finding is that education levels have risen over and the weight for the less well educated groups
time in all countries and regions. As a result, the declined.

Figure 7 Trends in mean age at first sex by world region, for all women and by education level, c.1993–2010
Note: Data are from the first available survey (c.1993, shown on the left-hand side) and most recent survey
(c.2010, shown on the right-hand side) in each country. Average age at first sex was not collected in several
countries in the Asia/Northern Africa region.
Source: As for Figure 4.
148 John Bongaarts et al.

Figure 8 Trends in mean age at first marriage by world region, for all women and by education level, c.1993–
2010
Note: Data are from the first available survey (c.1993, shown on the left-hand side) and most recent survey
(c.2010, shown on the right-hand side) in each country.
Source: As for Figure 4.

Figure 9 Trends in mean age at first birth by world region, for all women and by education level, c.1993–2010
Note: Data are from the first available survey (c.1993, shown on the left-hand side) and most recent survey
(c.2010, shown on the right-hand side) in each country.
Source: As for Figure 4.

Decomposition of trends in effect of education . The ‘rate’ effect refers to trends in the edu-
cation-specific means. As shown in Figures 7–
The overall trend in the mean age at an event has two 9, education-specific trends in the mean ages
distinct components: at events varied among regions; they tended
to rise in sub-Saharan Africa and in Asia and
Northern Africa, and to decline in Latin
. The ‘composition’ effect refers to the impact of
America and the Caribbean.
improvements in educational attainment.
These improvements increase the size of the
more highly educated groups and reduce the To quantify the role of these two factors, we under-
size of the less well educated groups thus took a decomposition exercise, which is the same as
raising the average pace of increase in the age the one used by Weinberger et al. (1989). The
at an event because higher levels of education composition effect was estimated by holding edu-
are associated with a higher age at the event. cation-specific means constant while allowing the
 Trends in age at reproductive transitions 149

Figure 10 Annual change in mean age at first sex by world region, decomposed, c.1993–2010
Note: Data are from the first available survey (c.1993) and most recent survey (c.2010) in each country. The two
components may not sum to the total change because of interaction effects. Average age at first sex was not col-
lected in several countries in the Asia/Northern Africa region.
Source: As for Figure 4.

educational composition to evolve as observed. The bar shows the observed increase or decrease in the
rate effect was estimated by holding educational mean. The main finding from this analysis is that the
composition constant while allowing the education- composition effect was positive for all three events
specific means to evolve as observed. The difference in all regions (except for age at first sex in Eastern/
between the sum of these two components and the Southern Africa) for which data were available. This
observed change is the ‘interaction’ effect, which is as expected, given the universal improvements in
will not be examined here. education levels in the large majority of less devel-
Results are presented in Figures 10–12. For each oped countries in recent decades. In contrast, the
event and for each of the regions, three bars are pre- rate effect varied in size and direction, as expected
sented. The white bar measures the composition from Figures 7–9. This effect was largely positive in
effect and the hatched bar the rate effect. The black sub-Saharan Africa and in Asia and Northern

Figure 11 Annual change in mean age at first marriage by world region, decomposed, c.1993–2010
Note: Data are from the first available survey (c.1993) and most recent survey (c.2010) in each country. The two
components may not sum to the total change because of interaction effects.
Source: As for Figure 4.
150 John Bongaarts et al.

Figure 12 Annual change in mean age at first birth by world region, decomposed, c.1993–2010
Note: Data are from the first available survey (c.1993) and most recent survey (c.2010) in each country. The two
components may not sum to the total change because of interaction effects.
Source: As for Figure 4.

Africa, but negative in Latin America and the Carib- premarital sex) are likely factors. As expected, the
bean. The composition effect dominated the rate upward trends in age at marriage in all regions
effect for changes in first birth and first marriage. were associated with declines in child marriage.
Although it remained high in some places, the preva-
lence of child marriage fell in all but nine of the 43
Discussion and conclusion countries in our analysis.
In all regions, the average age at first birth was
At the time of the most recent DHS surveys (c.2010), higher than the average age at first marriage, which
the mean ages at first birth, first marriage, and first in turn was higher than the average age at first sex.
sex varied significantly by region, with events occur- (This average ordering of life-cycle events does not
ring at an earlier age in sub-Saharan Africa than in necessarily hold for individual countries or for indi-
Latin America and the Caribbean or Asia and North- vidual women, because sexual initiation can occur
ern Africa. For example, the average age at first birth before, at the time of, or after marriage, and birth
was 20.9 years in Middle and Western Africa and in can precede marriage.) The substantial gap
Eastern and Southern Africa, while it was 22.2 and between age at first sex and age at first marriage in
22.8 years in Latin America and the Caribbean and sub-Saharan Africa and in Latin America and the
in Asia and Northern Africa, respectively. These Caribbean indicates a high prevalence of premarital
differences are expected, given that on most develop- sex. Furthermore, the gap grew over time, suggesting
ment indicators sub-Saharan African countries typi- a change in the context of sexual initiation. This trend
cally score lower than elsewhere. may reflect a rising period at risk of premarital sex, as
Our examination of trends in the mean ages at life- marriage is increasingly delayed, as well as to greater
cycle events between c.1993 and c.2010 also showed social tolerance of this behaviour and wider avail-
significant differences between regions. In sub- ability and use of contraception (Mensch et al.
Saharan Africa and in Asia and Northern Africa, 2006). In contrast, the gap between ages at first mar-
the mean ages at first sex, first marriage, and first riage and first birth in these regions narrowed over
birth all rose during this period. These findings are time, indirect evidence of an increase in premarital
as expected with rising levels of education. In con- conceptions, births, or both. It might also be the
trast, in Latin America and the Caribbean the case that women who marry later are under increas-
mean age at first birth remained constant and the ing pressure by family members to prove their ferti-
mean age at first sex declined. We can only speculate lity. But the fact that the increase in age at first
why this region differed, but the combination of birth paralleled the increase in age at marriage in
increasing globalization and the declining influence Asian and Northern African countries, where pre-
of the Catholic Church (which disapproves of marital sex is not thought to be common, suggests
 Trends in age at reproductive transitions 151

that premarital childbearing and not the pressure to improvement in education among women in Ethiopia, a
have a child explains the narrowing of the gap larger shift in age at first birth would be expected. In
between marriage and first birth in other regions. Bangladesh and Nepal large improvements in edu-
In all regions, the level of girls’ education was posi- cation levels were also accompanied by lower than
tively associated with the timing of reproductive expected increases in age at first birth.
events. Among women who had attended secondary Over the roughly 17 years covered by this analysis,
school, the mean ages at first birth and first marriage sizeable improvements in the educational attainment
were typically several years higher than among of women of reproductive age were observed. The
women with primary education alone. In contrast the impact of educational improvements on the timing of
differences between women who had attended life-cycle events was examined with a decomposition
primary school and women with no education were exercise. The results indicate that in sub-Saharan
mostly less than one year, indicating that secondary Africa and in Asia and Northern Africa the changes
education had a more powerful effect on the timing in timing of first marriage and first birth overall were
of events than primary education. This finding may primarily a result of changes in the educational compo-
be explained by the selectivity of those who attend sec- sition of the population of females (the ‘composition’
ondary school, as well as the fact that ages at first sex, effect) and to a lesser extent changes in timing within
marriage, and birth may overlap with, and therefore categories of education (the ‘rate’ effect). This
compete with, the age of participation in secondary suggests that, even in the absence of changes in behav-
education more than with primary education. iour within education groups, age at first marriage and
The preceding discussion focused on regional pat- first birth would have increased.
terns in the timing of reproductive events. Although In Latin America and the Caribbean (as rep-
a full analysis of country-specific estimates is beyond resented by the eight countries in the analysis) the
the scope of this paper, a few results are noteworthy results were less straightforward. For the time inter-
(while keeping in mind that trends in individual val we analysed, the average age at first sex
countries are affected more severely by sampling decreased by almost a year, the age at marriage
and reporting errors). The average pace of increase increased by less than a year, and the age at first
in the mean age at first birth was 0.04 years per birth remained constant. The positive effects of
annum, but country-level changes ranged from highs shifts in the education distribution on the average
of 0.12 in Senegal, 0.11 in Ghana, and 0.10 in Jordan ages at marriage and first birth were almost cancelled
and Morocco, to a slight decrease in Chad, the out by the negative effects of decreases within edu-
DRC, Malawi, the Philippines, Brazil, and Colombia. cation categories. For mean age at first sex, the nega-
For age at marriage the average pace of increase in all tive rate effect was substantially greater than the
countries in the analysis was 0.06 years per annum, positive composition effect. The Latin American
substantially higher than the pace for age at first experience suggests that it is difficult to speculate
birth. In twelve countries, the pace of increase in the how future increases in educational attainment
mean age at first marriage was 0.10 years per annum might affect the timing of early reproductive events.
or higher, which translates into a one-year increase The variability in trends across regions and
every decade. Declines in the average age at marriage countries and in the composition and rate effects of
were observed in five of the 43 countries. these trends is linked to differential changes in four
The degree of correlation between trends in girls’ factors. The first is the rapidity with which education
education and trends in the ages at reproductive levels have altered. In countries where education
events also varied substantially among countries. For levels have increased markedly over a short period,
example, there was a rapid expansion of education in one would expect more rapid change in the timing
Ethiopia between 2000 and 2011, with the proportion of early life-cycle events. Next is the effect of selectiv-
of women with at least primary education rising from ity; as education becomes more widespread, those
34 to 67 per cent (about three percentage points with higher levels of schooling are not as select a
annually). During the same period, the mean age at group as in the past and therefore may be more
first birth increased by only a modest 0.03 years per resistant to changes in reproductive behaviour. In
annum. The age at first birth increased by approxi- addition, the group with no education becomes
mately the same annual amount in Burkina Faso, increasingly selective of those with lower socio-econ-
Egypt, Mali, Mozambique, and Uganda, despite the omic status. Third, the rate effect is influenced by
proportion of women with at least primary education norms that are altered not just by increased schooling
rising only about a third as much as in Ethiopia (one but also as a result of other secular changes, such as
percentage point or less annually). Given the substantial the economy or the political context.
152 John Bongaarts et al.

Finally, there is the attenuating effect of declining Bledsoe, Caroline H. and Barney Cohen. 1993. Social
school quality (Grant 2015). As noted earlier, the Dynamics of Adolescent Fertility in Sub-Saharan
rapid expansion of girls’ education over the last few Africa. Washington, DC: National Academy Press.
decades has not translated into substantially improved Bledsoe, Caroline H., Jennifer A. Johnson-Kuhn, and John
learning outcomes in many settings (Riddell 2003; G. Haaga. 1999. Introduction, in C. H. Bledsoe, J. B.
Pritchett 2013), which may reduce the transformative Casterline, J. A. Johnson-Kuhn, and J. G. Haaga
effect of education on the timing of reproductive (Eds.), Critical Perspectives on Schooling and Fertility
events. If, however, the association between increased in the Developing World. Washington, DC: National
schooling and reproductive behaviour is in part Academy Press, pp. 1–22.
because of a modification of attitudes and autonomy Bongaarts, John. 2003. Completing the fertility transition in
resulting from exposure to school, then continued the developing world: the role of educational differ-
changes in the educational composition of the popu- ences and fertility preferences, Population Studies 57
lation, particularly in Africa and Asia, will likely (3): 321–335.
result in a continued rise in age at marriage and at Bongaarts, John and Ann K. Blanc. 2015. Estimating the
first birth, even if academic skills at a particular current mean age of mothers at the birth of their first
grade level deteriorate somewhat over time. child from household surveys, Population Health
Metrics 13(25): Open Access (6 pages).
Caldwell, John. 1980. Mass education as a determinant of
Notes the timing of fertility decline, Population and
Development Review 6(2): 225–255.
1 John Bongaarts, Barbara S. Mensch, and Ann K. Blanc Clark, Shelley and Rohini Mathur. 2012. Dating, sex, and
are all at the Population Council, New York. Please schooling in urban Kenya, Studies in Family Planning
direct all correspondence to John Bongaarts, Population 43(3): 161–174.
Council, One Dag Hammarskjold Plaza, New York, NY Diamond, Ian, Margaret Newby, and Sarah Varle. 1999.
10017, USA; or by e-mail at: [email protected] Female education and fertility: examining the links, in
2 This research was supported by a grant from the William C. Bledsoe, J. Casterline, J. Johnson-Kuhn, and J.
and Flora Hewlett Foundation to the Population Haaga (Eds.), Critical Perspectives on Schooling and
Council. The authors are grateful to Katharine Fertility in the Developing World. Washington, DC:
McCarthy and Barbara Miller for assistance in analysing National Academy Press, pp. 23–48.
data and preparing this manuscript. Ferré, Céline. 2009. Age at First Child: Does Education
delay Fertility Timing? Policy Research Working Paper
References 4833. Washington, DC: World Bank Human
Development Division, South Asia Region.
Ainsworth, Martha, Kathleen Beegle, and Andrew Finlay, Jocelyn E., Emre Özaltin, and David Canning. 2011.
Nyamete. 1996. The impact of women’s schooling on The association of maternal age with infant mortality,
fertility and contraceptive use: a study of fourteen sub- child anthropometric failure, diarrhoea, and anaemia
Saharan African countries, World Bank Economic for first births: evidence from 55 low- and middle-
Review 10(1): 85–122. income countries, BMJ Open 1(2): e000226.
Appleton, Simon. 1996. How does female education affect fer- Glick, Peter, Christopher Handy, and David E. Sahn. 2015.
tility? A structural model for the Côte d’Ivoire, Oxford Schooling, marriage, and age at first birth in
Bulletin of Economics and Statistics 58(1): 139–166. Madagascar, Population Studies 69(2): 219–236.
Basu, Alaka M. 2002. Why does education lead to lower Grant, Monica J. 2015. The demographic promise of
fertility? A critical review of some of the possibilities, expanded female education: trends in the age at first
World Development 30(10): 1779–1790. birth in Malawi, Population and Development Review
Basu, A. and R. Stephenson. 2005. Low levels of maternal 41(3): 409–438.
education and the proximate determinants of childhood Human Fertility Database. 2015. Max Planck Institute for
mortality: a little learning is not a dangerous thing, Demographic Research (Germany) and Vienna
Social Science and Medicine 60: 2011–2023. Institute of Demography (Austria). Available www.
Behrman, Julia Andrea. 2015. Does schooling affect humanfertility.org (accessed: 16 December 2015).
women’s desired fertility? Evidence from Malawi, Jejeebhoy, Shireen. 1995. Women’s Education, Autonomy,
Uganda, and Ethiopia, Demography 52: 787–809. and Reproductive Behaviour: Experience from
Blanc, Ann K., William Winfrey, and John Ross. 2013. New Developing Countries. Oxford: Clarendon Press.
findings for maternal mortality age patterns: aggregated Lloyd, Cynthia B. and Barbara S. Mensch. 1999.
results for 38 countries, PLoS ONE 8(4): E59864–00. Implications of formal schooling for girls’ transitions
 Trends in age at reproductive transitions 153

to adulthood in developing countries, in C. Bledsoe, J. B. Mensch, Barbara S., Monica J. Grant, and Ann K. Blanc.
Casterline, J. A. Johnson-Kuhn, and J. G. Haaga (Eds.), 2006. The changing context of sexual initiation in sub-
Critical Perspectives on Schooling and Fertility in the Saharan Africa, Population and Development Review
Developing World. Washington, DC: National 32(4): 699–727.
Academy Press, pp. 80–104. Nove, Andrea, Zoe Matthews, and Alma Virginia
Lloyd, Cynthia B. 2005. Growing Up Global: The Changing Camacho. 2014. Maternal mortality in adolescents com-
Transitions to Adulthood in Developing Countries. pared with women of other ages: evidence from 144
Panel on Transitions to Adulthood in Developing countries, Lancet Global Health 2: (E155).
Countries. Washington, DC: National Academies Press. Pritchett, Lant. 2013. The Rebirth of Education: Schooling
Lloyd, Cynthia B. and Barbara S. Mensch. 2008. Marriage ain’t Learning. Washington, DC: Center for Global
and childbirth as factors in dropping out from school: an Development.
analysis of DHS data from sub-Saharan Africa, Riddell, A. R. 2003. The introduction of free primary edu-
Population Studies 62(1): 1–13. cation in sub-Saharan Africa. Paper commissioned for
Lloyd, Cynthia B. 2009. New Lessons: The Power of edu- the EFA Global Monitoring Report 2003/4, The Leap
cating Adolescent Girls (A Girls Count Report on to Equality.
Adolescent Girls). New York: Population Council. Singh, Susheela and Renee Samara. 1996. Early marriage
Lloyd, Cynthia B. 2010. The role of schools in promoting among women in developing countries, International
sexual and reproductive health among adolescents in Family Planning Perspectives 22(4): 148–157.
developing countries, in S. Malarcher (Ed.), Social Subbarao, K. and Laura Raney. 1995. Social gains from
Determinants of Sexual and Reproductive Health: female education: a cross-national study, Economic
Informing Future Research and Programme Development and Cultural Change 44(1): 105–128.
Implementation. Geneva: World Health Organisation, UNESCO Institute for Statistics. 2010. Global Education
pp. 113–131. Digest 2010: Comparing Education Statistics across the
Lutz, Wolfgang. 2014. A population policy rationale for the World. Quebec: UNESCO Institute for Statistics.
twenty-first century, Population and Development United Nations Dept. of Economic and Social Affairs.
Review 40: 527–544. 2002. ‘Concise Report on World Population
Martin, Teresa Castro. 1995. Women’s education and ferti- Monitoring 2002: Reproductive Rights and
lity: results from 26 Demographic and Health Surveys, Reproductive Health with Special Reference to
Studies in Family Planning 26(4): 187–202. Human Immunodeficiency Virus/Acquired
Mensch, Barbara S. 2005. The transition to marriage, in Immunodeficiency Syndrome (HIV/AIDS). Report of
C. B. Lloyd (Ed.), Growing up Global: The Changing the Secretary General (E/CN.9/2002/2),’ Paper pre-
Transitions to Adulthood in Developing Countries. sented at 35th Session, Commission on Population and
Washington, DC: National Academies Press, pp. Devlopment. New York, 1–5 April.
416–505. United Nations Dept. of Economic and Social Affairs.
Mensch, Barbara S., Susheela Singh, and John B. 2010. The World’s Women 2010: Trends and Statistics.
Casterline. 2005. Trends in the timing of first marriage New York: United Nations.
among men and women in the developing world, in C. Weinberger, Mary Beth, Cynthia Lloyd, and Ann Klimas
B. Lloyd, J. R. Behrman, N. P. Stromquist, and B. Blanc. 1989. Women’s education and fertility: a
Cohen (Eds.), The Changing Transitions to Adulthood decade of change in four Latin American
in Developing Countries: Selected Studies. Washington, Countries, International Family Planning Perspectives
DC: National Academies Press, pp. 118–171. 15(1): 4–14.
154 John Bongaarts et al.

Appendix A: Demographic and health survey Appendix B: Estimating the period mean
countries and years ages at life-cycle events
The method for estimating the period mean ages at life-
N (married cycle events (such as first birth, first marriage, and first
Year women) sex) from DHS surveys follows Bongaarts and Blanc
(2015). In the absence of events before age 15 and above
First Last First Last age 44
Region/country survey survey survey survey
MIDDLE/WESTERN AFRICA
Benin 1996 2011 4,198 11,680

44
Burkina Faso 1993 2010 5,325 13,563 (a + 0.5)b(a, t)
Cameroon 1998 2011 3,676 9,791 M(t) = 15

44 (B1)
15 b(a, t)
Chad 1996–97 2004 5,832 4,663
Congo, 2007 2013–14 6,622 12,096
Democratic
Republic
where M(t) = average age at the event in period t, b(a,t) =
Côte d’Ivoire 1994 2011–12 5,271 6,308 age-specific event rate at age a in period t (events per 1,000
Ghana 1988 2014 3,156 5,321 women).
Guinea 1999 2012 5,561 6,726 In countries where large numbers of events are recorded
Liberia 1986 2013 3,538 5,385 (e.g., from vital registration), event rates are available by
Mali 1995–96 2012–13 8,222 8,819 single age and single year. As a result, annual estimates
Niger 1992 2012 5,560 9,881 of M(t) can be obtained. In contrast, in applications of
Nigeria 1990 2013 6,880 27,829 this equation to DHS surveys, samples of events in single
Senegal 1992–93 2014 4,450 5,500 years are relatively small. To obtain robust estimates of
Togo 1998 2013 5,819 6,282 the mean ages at events for a survey, we calculated b(a,t)
by single year of age for a period of five years before
EASTERN/SOUTHERN AFRICA each survey. In addition, we excluded surveys with
Ethiopia 2000 2011 9,789 10,287 sample sizes of currently married women below 3,000 to
Kenya 1989 2014 4,765 18,549 minimize sampling errors. The ages included in equation
Malawi 2000 2010 9,451 15,527 (B1) were from 15 to 44. The upper age was set at
Madagascar 1992 2008–09 3,736 12,038 44 years because in a few countries no data were available
Mozambique 1997 2011 6,529 9,331 for higher ages. Moreover, the incidence of first sex, mar-
Rwanda 1992 2010 3,785 6,897 riage, and births at higher ages was negligible in the
Tanzania 1991–92 2010 6,038 6,411 countries included in this analysis. Equation (B1) uses a
Uganda 1988–89 2011 3,180 5,417 minimum age of 15 years because respondents under age
Zambia 1992 2013 4,457 9,858 15 are not included in DHS surveys. Unfortunately, the
incidence of events under age 15 was not negligible in a
Zimbabwe 1994 2010–11 3,788 5,702
number of countries. An adjustment to equation (B1)
ASIA/NORTHERN AFRICA was therefore needed to account for the occurrence of
Bangladesh 1993–94 2011 8,980 16,635 events below age 15. This adjustment relied on the retro-
Cambodia 2000 2014 9,071 11,898 spective reporting of events before age 15 by women
India 1992–93 2005–06 84,678 93,089 aged 15–19 at the time of the survey. The adjusted equation
Indonesia 1987 2012 10,906 33,465 is as follows:
Jordan 1990 2012 6,168 10,801
Nepal 1996 2011 7,982 9,607
Pakistan 1990–91 2012–13 6,364 12,936
Philippines 1993 2013 8,961 9,726
44
14p(T) + (a + 0.5)b(a, t)
Turkey 1993 2003–04 6,270 7,671 M∗ (t) = 15

44 (B2)
Egypt 1988 2014 8,220 20,459 p(T) + 15 b(a, t)
Morocco 1987 2003–04 5,447 8,781
LATIN AMERICA/CARIBBEAN
Bolivia 1989 2008 4,941 10,161
Brazil 1986 1996 3,471 7,584 where M*(t) = adjusted average age at the event in period
Colombia 1990 2010 4,531 26,281 t, b(a,t) = age-specific event rate at age a in period t (events
per 1,000 women), p(T ) = proportion of women aged 15–
Dominican 1986 2013 4,133 5,064
19 reporting an event before age 15 at T, the time of the
Republic survey (early events per 1,000 women).
Guatemala 1987 1998–99 3,377 3,963 Events before age 15 were assumed to have occurred at
Haiti 1994–95 2012 3,113 7,808 exact age 14. Equation (B2) was used to estimate the
Honduras 2005–06 2011–12 11,612 12,846 period mean ages at first birth, first marriage, and first
Peru 1991–92 2011 8,740 12,672 sex throughout this study.