United NationsDevelopment Programme

Human Development Report Office

Demography and Human

Development:
Education and Population Projections
by Wolfgang Lutz and Samir KC, International Institute for Applied Systems
Analysis (IIASA)

O C C A S I O N A L PA P E R 2 0 1 3 / 0 4
Wolfgang Lutz is Founding Director of the
Wittgenstein Centre for Demography and Global
Human Capital and Professorial Research Fellow
at the Oxford Martin School for 21st Century
Studies.  He holds PhDs in Demography from the
University of Pennsylvania and in Statistics from the
University of Vienna.  His work focuses on family
demography, fertility analysis, population projec-
tions and interactions between population and envi-
ronment. Dr Lutz is author or editor of 28 books
and more than 200 refereed articles. In 2010 he
received the highest science award in Austria, the
Wittgenstein-Preis.

Samir K.C. is leader of the “Modeling Human

Capital Formation” project of the World Population
Programme at the International Institute for Applied
Systems Analysis (IIASA). He joined the programme
as a Research Scholar in May 2005. His current
research focuses on population projections with
various levels of disaggregation, including education
and health. Dr. K.C. holds a PhD from the Faculty
of Spatial Sciences at the University of Groningen,
the Netherlands and a Master’s in statistics from
Tribhuvan University, Nepal. His major research
interest is in developing and applying multi-state
population models in demographic analyses and
projections.

UNDP Human Development Report Office

304 E. 45th Street, 12th Floor
New York, NY 10017, USA
Tel: +1 212-906-3661
Fax: +1 212-906-5161
http://hdr.undp.org/

Copyright © 2013
by the United Nations Development Programme
1 UN Plaza, New York, NY 10017, USA
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system or transmitted, in any form or by
any means, electronic, mechanical, photocopying, recording
or otherwise without prior permission. This paper does not
represent the official views of the United Nations Development
Programme, and any errors or omissions are the authors’ own.
 4

Demography and Human Development:

Education and Population Projections
WOLFGANG LUTZ AND SAMIR KC, INTERNATIONAL INSTITUTE FOR APPLIED SYSTEMS ANALYSIS (IIASA)

ABSTRACT

Changes in the size and structure of human populations are conventionally modeled and projected by stratifying the population by
age and sex. Here we present new approaches, which use multidimensional demographic methods to add educational attainment
as a third dimension in studying population dynamics. In virtually all societies, better-educated adults have lower mortality rates
and their children better chances of survival. Almost universally too, women with higher levels of education have fewer children,
through choice and higher access to birth control. Because of these pervasive differentials, scenarios of potential improvements
in education have significant implications for future population growth. The projection of future educational attainment distribu-
tions is of significant interest in its own right as well, as education has a great influence on almost every aspect of progress in
human development.

level of female education and the availability of reproduc-

INTRODUCTION
tive health services are the two most important ones that are
Human development can best be studied with models that open to policy interventions (Bongaarts and Sinding 2011,
have human beings rather than monetary or other units at the and Lutz and KC 2011). The effect of education on fertility is
core of their analysis. Demography, which can also be defined particularly strong in countries that still have relatively high
as the mathematics of people, specifies all of its models strictly overall fertility levels and hence are in the early phases of their
in terms of human beings according to different relevant char- demographic transitions. There are many reasons to assume
acteristics. Hence, it offers a most appropriate approach to the that these pervasive differentials are directly caused by educa-
study of human development across the world. Traditionally, tion, which enhances the level of information, changes the
demographic analysis has mostly focused on the changing motivations for behaviour, and empowers people to better
composition of populations by age and gender. But human pursue their own preferences, although strict causality can
beings have many observable and measurable characteristics only be proven for specific cases in which natural experiments
that distinguish one individual from another and that can be occurred. For the following projections by level of education,
considered highly relevant for human development; these char- it is sufficient to assume that systematic associations will
acteristics can also be assessed in aggregate and used to distin- continue to persist over the coming decades, as they have for
guish one sub-group of a population from another. Here we more than a century for all countries for which data exist.
will focus on the level of highest educational attainment and to Because of these fertility and mortality differentials by
a lesser extent also health status in addition to age and gender. education, future changes in the educational composition
In virtually all societies, better educated men and women of the population will greatly influence the future outlook
have lower mortality rates, and their children have better for overall population trends around the world. In addition,
chances of survival (KC and Lentzner 2010). Almost uni- education is not only an important source of population het-
versally, women with higher levels of education have fewer erogeneity that influences population dynamics, but it is also
children, presumably because they want fewer and find better an important influence on people’s capabilities and empow-
access to birth control. There are, of course, many factors erment, as will be discussed below. Hence there are many
influencing the level of fertility that range from the status of reasons for making investments in education in general and
women within the family, to female labour force participation, in female education in particular in programmes directed at
to general socio-economic development. However, extensive better global health, population stabilization, poverty reduc-
research has shown that among this myriad of factors, the tion and sustainable development.

UNDP Human Development Report Office

OCCASIONAL PAPER 2013 /04 1
 4 DEMOGRAPHY AND HUMAN DEVELOPMENT

consequences of education only refer to the length and level

1. MEASURING AND MODELLING EDUCATION1 of formal education, because these elements have the only
When measuring education, it is important to distinguish con- systematically available data. It is plausible to assume that
ceptually between education flows and stocks. the quality and content of education also matter for many of
Flows refer to the process of education – to schooling these consequences, although little empirical evidence exists
and, more generally, the production of human capital – and to date.
may consist of formal and informal education. The process The projections presented here are based on the demo-
of education is the central focus of pedagogy and educa- graphic method of multi-state population projection, which
tion science, where the usual statistical indicators are school was developed at IIASA during the 1970s, and is now a well
enrolment rates, student-teacher ratios, drop-out rates and accepted method among technical demographers. Our base-
repetition rates. line year, providing the empirical starting point, is 2000, the
Human capital refers to the stock of educated adults, same as in our reconstruction of education distribution in the
which is the result of past education flows for younger adults past. This allows the backward and forward projections to
in the more recent past and for older ones some decades ago. be connected in a gapless time series. We chose 2000 as the
This stock is usually measured in terms of the quantity of base year, since the data for 2005 were not available for a vast
formal education (highest level of attainment or mean years majority of countries.
of schooling) but the quality dimension (the general knowl- The basic idea of projection is straightforward: Assuming
edge and cognitive skills people actually have) and the content that the educational attainment of a person remains invari-
or direction of education also matter. For countries with data ant after a certain age, we can derive, e.g., the proportion of
on the cognitive skills of the adult population, the evidence women without any formal education aged 50-54 in 2005
has shown significant economic impacts of education qual- directly from the proportion of women without any formal
ity (Hanushek and Woessmann 2008) but the number of education aged 45-49 in 2000. Continuing to assume that this
these countries is still very limited. The content of education proportion is constant along cohort lines, the proportion of
matters more for higher education than for basic education, women without education aged 95-99 in 2050 for the same
where the main aim is the acquisition of literacy skills and cohort follows directly. In a similar manner, the proportions
basic numeracy. for each educational category and each age group of men and
The quantity of formal education is often measured by women can simply be moved to the next older five-year age
the mean years of schooling of the adult population above group as one moves forward in time in five-year steps.
either the age of 15 or 25. This has the advantage of captur- These proportions would be precisely correct if no indi-
ing the entire human capital of a population at one point in vidual moves up to the category with primary education after
time in one single number (or two when one distinguishes the age of 15, and if mortality and migration did not differ
between men and women). Yet, there is much to be gained by level of education. This follows directly from the fact that
from decomposing this highly aggregate indicator into the full the size of a birth cohort as it ages over time can only change
distribution of educational attainment categories (and thus through mortality and migration. However, strong links do in
capturing the differences in the composition of education fact exist between the education level and mortality, fertility
stocks across countries and regions) and into different age and migration behaviour. Accordingly, the above approach is
groups, and hence capturing inter-cohort changes that drive adjusted to correct for these effects. The size of birth cohorts
many of the consequences of improving human capital on is dependent on the levels of education of women of child-
society and the economy. bearing age, where a negative relationship is traditionally
There are many ways in which education and the result- observed. In projecting these cohorts forward, differential
ing human capital affect human well-being at the individual survival rates are applied to the education groups. The differ-
and society levels. A huge body of literature shows that better entials are based on a comprehensive literature review as well
educated people tend to have better health, higher income and as on modelling exercises based on past data. The details of
higher life satisfaction (Lutz 2009); better educated societies these adjustments are provided in later sections of this paper.
tend to have higher economic growth and better institutions The above treats the different education groups essentially
(Lutz et al. 2008, 2010). Most of these assessments of the as separate sub-populations. In addition, at younger years,
transitions between education categories may occur. These are
1 This sections draws heavily on Lutz et al. 2007; KC et al. 2010. described in detail in later sections. The analysis is simplified

UNDP Human Development Report Office

2 OCCASIONAL PAPER 2013 /04
 1. Measuring and Modelling Education

by the assumption that changes in educational attainment are the same time, a large number of detailed categories would be
uni-directional; in other words, individuals can only move unwieldy and limit the number of countries for which data are
from the ‘no education’ status to primary, and on to second- available. Consequently, a relatively small number of catego-
ary and possibly to tertiary, but never revert to a lower status. ries is used to cover the entire spectrum. This means that the
In reality, the likelihood of an individual making the tran- categories are relatively broad. Note, for instance, that ‘pri-
sition from one educational attainment level to the next high- mary’ does not refer to completed primary, but to having more
est is strongly dependent on the education of the parents. This than one year of primary schooling. Likewise, for the purposes
educational inheritance mechanism is not, however, modeled of this study, ‘secondary’ refers to lower secondary, not com-
explicitly here. Instead, the assumptions regarding the tran- pleted upper secondary. As a result, the ‘secondary’ category
sition rates and their future development are statistically is quite broad, encompassing ISCED levels 2-4. The reason
derived from the aggregate behaviour of education systems for not splitting off ISCED 4, however, is that the distinction
in the past. Since this expansion is partly the result of the between ISCED 3 and 4 is one of the least clear-cut. ISCED 4
inheritance mechanism, i.e., the fact that many parents aspire programmes “are often not significantly more advanced than
for their children to reach an education level at least as high programmes at ISCED 3” (UNESCO Institute for Statistics
as they themselves did, inheritance is implicitly reflected in the 2009, p. 258). As a result, attempting to distinguish between
projection, even though it is not formally part of the model. ISCED 3 and 4 would have undermined the straightforward
Such an approach appears preferable at this time, because hierarchical interpretation of our education categories.
data on the aggregate growth patterns of education systems, Our procedure for each country can be summarized as
on which assumptions for the future can be based, are much follows:
more readily available than robust data on the micro-process
of educational inheritance. • A baseline population distribution by five-year age group,
The starting point for the projection is data collected for sex and level of educational attainment is derived for the
each country (typically around the year 2000) that gives the year 2000.
total population by sex, five-year age groups, and four attain- • For each five-year time step, cohorts move to the next
ment categories based on the current International Standard highest five-year age group.
Classification of Education (ISCED 1997): no education, pri-
mary, secondary and tertiary (see Table 1). • Mortality rates are applied, specific to each age, sex and
education group, and to each period.

• Age- and sex-specific educational transition rates are

Table 1: Education categories
applied.
Category Definition • Age-, sex- and education-specific net migrants are added
No education (E1) No formal education or less than one year of primary
to or removed from the population. In the projections
presented here, the migration assumptions correspond to
Primary (E2) Uncompleted primary, completed primary (ISCED 1) and
uncompleted lower secondary those used in UN population projections, with additional
assumptions on the education profile of migrants. Since
Secondary (E3) Completed lower secondary (ISCED 2) uncompleted and
completed higher secondary (ISCED 3/4) and uncompleted the United Nations assumes an overall decrease in the
tertiary education volume of migration over time, these assumptions are not
Tertiary (E4) Completed tertiary education (ISCED 5/6) of critical importance.

• Fertility rates are applied, specific to each age, sex and

A single set of categories applied to all countries regardless education group, and to each period, to determine the size
of their state of educational development inevitably requires of the new 0-4 age group.
some compromises. Surveys used exclusively in developing
• The new population distribution by age, sex and level
countries have historically provided little differentiation at
of educational attainment is noted, and the above steps
higher education levels. Conversely, data collected in indus-
repeated for the next five-year time step.
trialized countries may not differentiate below completed
primary level. For present purposes, the entire spectrum from The aim of the projection is to obtain a dataset with the
no education to completed tertiary needs to be covered. At population distributed by five-year age groups, starting at age

UNDP Human Development Report Office

OCCASIONAL PAPER 2013 /04 3
 4 DEMOGRAPHY AND HUMAN DEVELOPMENT

Figure 1a: Structure by age, sex and level of education for Figure 1b: Projected structure for South Africa for the year
South Africa for the year 2000 2050
Age cohort
Age cohort
95 95
90 90
85 85
80 80
75 75
70 70
65 65
60 60
55 55
50 50
45 45
40 40
35 35
30 30
25 25
20 20
15 15
3000 2000 1000 0 1000 2000 3000 3000 2000 1000 0 1000 2000 3000
Males Population Females Males Population Females
(thousands) (thousands)
No Education Primary Education Secondary Education Tertiary Education

15 and with the highest age group 100+; by sex; and by four but as exercises in ‘what if’ reasoning. As such they serve the
levels of educational attainment over a period of 50  years, important purpose of illustrating the consequences of differ-
from 2000 (base year) to 2050, in five-year intervals. ent kinds of trends and policy environments on global human
To illustrate the kind of information that this projection capital. In any case, the notion that we can avoid making
method generates for 120 countries in the world, Figure  1 assumptions about future educational attainment trends is a
gives an example in terms of age pyramids by level of educa- fallacy. Since fertility is influenced by education levels, popu-
tion for South Africa. The first pyramid (Figure  1a) shows lation projections inevitably make implicit assumptions about
the structure by age, sex and level of education for the year the population’s future educational attainment, even if these
2000, which is the empirical baseline information used for the remain unstated. In our view, it is preferable to be explicit
reconstruction. The second pyramid (Figure 1b) gives the pro- about these assumptions. The changing educational composi-
jected structure for the year 2050, resulting from our method. tion of a population not only matters as a source of heteroge-
neity in population projections. Information about the likely
future educational composition is also very important in its
2. ALTERNATIVE EDUCATION SCENARIOS 2 1F
own right for charting progress in human development and
drawing inferences in terms of the likely benefits from such
While the fertility, mortality and migration assumptions under- different possible trends.
lying the projections presented here are discussed extensively The following four scenarios try to demarcate the broad
elsewhere (Lutz et al. 2004; KC et al. 2010), we will focus on range of possible future trends in schooling. Since the stock
the definition of four different scenarios reflecting different of human capital can only change through changing patterns
education policies in countries around the world. of school enrolment for the young generation (in addition
While making assumptions about future educational to minor possible changes through differential mortality and
development over the course of several decades is a difficult migration, and very rarely adult education), all four scenarios
task, it is not intrinsically more difficult than making assump- are defined in terms of future trends in school enrolment.
tions about reproductive behaviour or mortality. Like variant They range from the worst case of declining school enrolment
projections of demographic indicators, the education scenar- rates, to the case of no further improvements in enrolment
ios below are not to be interpreted as predictions or forecasts, rates that might result from political inaction, to a medium
policy scenario in which it is assumed that the future expan-
2 This section draws heavily on KC et al. 2010. sion of enrolment rates follows the collective experience of

UNDP Human Development Report Office

4 OCCASIONAL PAPER 2013 /04
 2. Alternative Education Scenarios

the countries that are already somewhat more advanced in GLOBAL EDUCATION TREND (GET) SCENARIO: MEDIUM
educational expansion, to the strong policy fast track sce- POLICY SCENARIO
nario. This last scenario assumes that the country manages This is a more complex scenario that is not derived from
to follow the experience of nations such as the Republic of a single, simple assumption. Informally, the GET scenario
Korea and Singapore, who experienced some of the most assumes that a country’s educational expansion will con-
rapid expansions in schooling in human history. While the verge on an expansion trajectory based on the historical
medium policy global education trend scenario may be con- global trend. The identification of the global trend is based
sidered the most likely trajectory for every country, the range on a data-driven judgemental analysis. This means it is nei-
from the fastest possible expansion to the decline of enrol- ther derived by mechanistically applying a statistical model,
ment rates spans all future trends that might be considered nor a mere ‘expert estimate’, but based on the application of
plausible (in the sense of not impossible), although they are domain knowledge to the empirical data.
not necessarily likely trajectories. This scenario, which resembles the collective experience
of countries already further advanced in education expansion,
CONSTANT ENROLMENT NUMBER (CEN) SCENARIO: needs to make some special assumptions for front-runners,
WORST CASE SCENARIO including the assumption of a possible maximum of educa-
In a sense, this is a worst-case scenario in that it assumes tion. From a theoretical perspective, the limiting constraints
zero expansion of schooling. In the context of population of educational expansion differ at different stages. Initially,
growth, this implies that the proportion of children enrolled expansion in enrolment is likely to be essentially limited by
in school is actually declining over time. Although some the available supply of school places. As long as only a small
African countries have shown such a trend during the 1980s, fraction of each cohort is enrolled in primary school, it seems
here this scenario is not presented as a likely future possibility, plausible that each additional school that is built can be filled
but serves reference purposes only. Its technical definition is with willing students. At this stage, enrolment is largely sup-
straightforward. ply-limited. Once the vast majority of each cohort is enrolled,
The assumption is that in each country, the number in say 90% or more, the fact that the remaining 10% are not
each cohort (by gender) making each educational transition at enrolled is unlikely to be the result of a lack of school places.
the appropriate age remains constant over time. Accordingly, In fact, by the time 90% are enrolled, cohort growth will
the relative share of the attainment levels can rise and fall, typically have fallen considerably, meaning that raising the
depending on changes in cohort size. enrolment ratio further does not require physical expansion.
Instead, enrolling the last few percent is typically a matter
CONSTANT ENROLMENT RATIO (CER) SCENARIO: NO of accessing hard-to-reach populations, such as children in
ACTION SCENARIO remote rural areas, working children, those suffering from
Like the previous scenario, the projection of constant transi- disabilities and so on. Complete enrolment of these groups in
tion rates between attainment levels (and as a result constant school requires not school expansion, but well-designed and
proportions in each level within each cohort) serves largely targeted demand-side interventions.
illustrative purposes. It demonstrates the implications of Both accelerating and decelerating phases of attainment
extending the status quo into the future, without regard for expansion can be modeled well by cubic splines at all attain-
contextual change. In its disregard of historical upward trends ment levels. The placement of the point at which the curve
and of the opportunity for ‘no-cost expansion’ when cohort switches from accelerating to decelerating expansion was
size declines, it is a somewhat pessimistic scenario. chosen to ensure the splines connect smoothly. The exact
The technical definition of the CER scenario is straight- placement is not critical, since the curve is approximately
forward. In each country, the proportion of each cohort (by linear for much of the central section. Fitting such bi-cubic
gender) making each educational transition at the appropri- models to each country shows good individual fits (in the vast
ate age remains constant over time. Note that these constant majority of cases with an adjusted R-squared greater than
proportions are applied not to cohorts at birth, but to cohorts 0.8), and the resulting parameters, indicating the ‘pace’ with
of survivors at the relevant age. This ensures that a decrease which different countries traverse the cubic curve, turn out to
in infant mortality by itself will not reduce the educational have a unimodal, fairly symmetric and tightly clustered dis-
transitions of surviving children under the assumption of con- tribution. The parameter means across the individual country
stant proportions. models may therefore reasonably be considered to constitute

UNDP Human Development Report Office

OCCASIONAL PAPER 2013 /04 5
 4 DEMOGRAPHY AND HUMAN DEVELOPMENT

the ‘typical’ global trend. Countries that had already achieved Development Goals (MDGs) and the Education for All (EFA)
99% or higher participation were excluded in determining the goals took a ‘milestones’ approach, and the EFA Fast Track
overall mean expansion parameter. The projected trajectories Initiative (FTI) shows that international support will not be
resulting from applying these global trend parameters were indiscriminate and thinly spread, but focused on those coun-
examined for their plausibility. tries lagging furthest behind.
In general, the comparison of the average growth patterns Our FT scenario illustrates the implications of a new
across phases and genders shows that while the schooling of round of similar policy initiatives focusing on secondary edu-
girls may have started later, it has been expanding at a much cation (the next logical step after the achievement of universal
faster pace. The gender difference is more or less the same at primary education). Since our category E3 is based on com-
primary and secondary levels, but dramatically greater at the pleted lower secondary, it is assumed that the target propor-
tertiary level. Between 1970 and 2000, female tertiary attain- tion will eventually be universal attainment of E3 by 2050.
ment growth has been closer to the pace of male primary or Since this target is very far off, however, a more achievable
secondary expansion in the past. intermediate target is assumed for 2030. While no interna-
The application of the GET scenario will normally result tional policy frameworks have so far recommended specific
in more optimistic trajectories of educational attainment than targets for tertiary participation, the rapid pace of tertiary
either the CEN or CER scenarios. However, it is important expansion in many parts of the world displays a great amount
to note that in the case of the latter, this is not by definition. of ambition in this regard, both individual and national. As
In a context of rapidly falling cohort sizes, the assumption of an explicitly optimistic scenario, the achievement of a propor-
constant enrolment may in principle translate into increases tion of 60% at attainment level E4 is assumed to be reached
in the proportions at higher attainment levels that are higher by 2050. While this is as high as some of the very highest
than the increases implied by the GET scenario. In fact, it levels observed today, it is already exceeded by the aspirations
could be argued that the GET scenario is rather conserva- of teenagers in some developing and transitional countries
tive. The above parameters imply, for instance, that it takes a (OECD 2007). Even if the assumptions turn out to be over-
typical country about 40 years to raise female participation optimistic (given that it appears increasingly unlikely that the
in primary schooling from 50% to 90%, and over 30 years EFA goals for enrolment will be achieved in 2015), the sce-
after that to reach 99%. The latter in particular may seem dis- nario serves as a useful comparison with actual developments
couragingly slow (moreover, female primary participation has to assess how much human capital has been ‘lost’ by missing
been the fastest process over the period studied), but reflects the targets.
the empirical average. While some countries have expanded For the primary level (E2), the attainment of 99% transi-
access much more quickly, others have stagnated and made tion to E2 by 2015 is assumed under the FT scenario, corre-
even less progress during the final decades of the 20th century, sponding approximately to the achievement of the EFA goals.
and this is reflected in the average slope. Lower secondary schooling (E3) is assumed to reach 50%
of each cohort by 2030, and 90% by 2050. Tertiary (E4) is
THE FAST-TRACK (FT) SCENARIO: STRONG POLICY assumed to reach 60% by 2050.
SCENARIO Finally, a word on intergenerational transmission of edu-
In addition to the above scenarios that define lower bounds, cation in the scenarios. This is implicit in the above assump-
worst-case or ‘no change’ environments, an optimistic sce- tions, which have been defined in terms of the proportion of
nario has been defined. This is based on the GET scenario a cohort that reaches a certain level of highest educational
defined above, but assumes the achievement of certain mile- attainment. These proportions are comprised of children who
stones. If stated targets in attainment are not reached by come from mothers with that level of education plus children
certain years (both of which are defined below) under the with mothers from lower education groups (upward mobility)
assumptions of the GET scenario, then an accelerated rate of and children whose mothers had a higher educational level
growth is applied that meets these targets. (downward mobility). Here we only refer to mothers’ edu-
As a result, the increases over the GET scenario are not cation, since there are more data available and the effect is
in the form of uniformly faster growth, but in the form of better documented. Clearly, fathers’ education also matters
lifting up the countries that are furthest behind. This reflects in this context. Since there is generally very little empirical
the actual political dynamics through which such accelera- information on educational upward and downward mobility
tion has been attempted in the past. Both the Millennium in most countries of the world, and since those movements

UNDP Human Development Report Office

6 OCCASIONAL PAPER 2013 /04
 3. Results

are greatly dependent on government policies with respect certainty that the elderly of the future will be much better
to the expansion of the school system, it seemed appropri- educated and therefore most likely also in better health. In
ate to define the scenarios in terms of overall proportions. part, because of the high education of young women, the birth
In the GET scenario, which is based on the continuation of rates in the Republic of Korea have also declined to very low
empirically given trends that already include such intergen- levels, and the population is expected to age very rapidly and
erational transmission mechanisms, a simple continuation of start declining over the longer run. Because today the younger
past transmission and mobility trends is assumed. In the CEN population is already among the best educated in the world,
and CER scenarios, upward mobility is assumed to be cur- the different scenarios as shown in Table 2 only differ slightly
tailed, while in the FT, it is assumed to be greatly enhanced. with respect to the proportion of women with tertiary educa-
Finally, it is important to stress that child mortality (up to age tion in 2050.
15) under all scenarios is derived from the mother’s educa- Figure 3 shows the trends since 1970 for the world’s most
tion, and hence different education scenarios will also result populous country, China, according to the medium policy
in different mortality trajectories, as will be discussed in the GET scenario. It clearly demonstrates the enormous educa-
last section. tional expansion that took place over the past decades, which
without doubt was a major engine of economic growth and
poverty reduction (Lutz et al. 2008). In the 1970s, more than
3. RESULTS a third of Chinese adults had no formal education at all, and
Numerical and graphical results for all countries in the world a significant part had only some primary education. It was
for which data are available (98% of the world population) during the 1980s and 1990s that rapid expansion of the school
have been collated for five-year age groups, five-year steps system even in remote rural areas resulted in a significant
in time up to 2050, four educational categories, and all four increase in the proportion of people who had completed junior
education scenarios described above. Due to the enormous secondary education, which was concentrated among younger
amount of data and figures, only trends for selected important adults. With the passage of time, these better educated cohorts
countries and scenarios will be presented here. will move up to older age groups and result in significant
For highlighting the results, we chose five countries from further improvements of the average educational attainment
different continents and at very different stages of socio- of the adult population. As Table 2 shows, this momentum
economic development. Table 2 gives selected indicators of of improvements in human capital will even under the worst
female educational attainment for these countries in 2010, scenarios result in further increases in the education of the
and as projected to 2050 according to the four alternative female adult population. While today around 10% of Chinese
scenarios. The four figures show the changing educational women of reproductive age have completed tertiary education,
compositions over time for each country. Only one scenario by 2050 this proportion will have increased to between 14%
per country is depicted for illustration. and 50%. Under the FT scenario and even under the GET
The Republic of Korea has seen one of the fastest expan- scenario, the Chinese population in 2050 will be much better
sions of education in human history. While in the 1950s it was educated than the populations of Europe today.
one of the poorest countries, with a vast majority of the adult The picture for India looks very different. Figure 4 shows
population never having been in school, today women there that up until 2000, more than half the adult population of
are among the best educated in the world. The table shows that India had no formal schooling whatsoever; this trend was
43% of all women aged 15-44 have completed tertiary educa- particularly pronounced among women (Table 2). Given great
tion, a value much higher than in most European countries. momentum in improvements in the educational attainment
Figure  2 shows the dramatically changing educational of adults, recent efforts by the Indian Government to greatly
composition of the Republic of Korea from 1970 to 2050 increase school enrolment in rural areas (which are already
resulting from the no-action constant enrolment scenario reflected in the GET scenario shown in the graph) will only
from 2010 onwards. Cohorts that were of school age in the very slowly reduce the uneducated proportion of the total adult
1950s have to a large proportion received no formal education population. Partly due to the lower level of female education,
at all, while today more than half of all young women have the total population of India is expected to grow more rapidly
completed tertiary education. As a consequence of the great than in China. At the same time, there is an impressive expan-
momentum described above, elderly people in the Republic of sion of the number of better educated people in India, which
Korea today still have very low education, but we know with is undoubtedly a key factor in India’s recent economic growth.

UNDP Human Development Report Office

OCCASIONAL PAPER 2013 /04 7
 4 DEMOGRAPHY AND HUMAN DEVELOPMENT

Table 2: Results of projections by age, gender and level of education for five selected countries and all four scenarios
focusing on the education of women of reproductive age
Population (millions) Proportion (women aged 15-44)
Country All Women aged 15-44 No education (%) Primary (%) Secondary (%) Tertiary (%)

2010 195 47 3 38 52 7

2050 FT 223 40 0 9 57 34

Brazil GET 224 41 0 9 71 20

CER 240 44 1 32 60 7

CEN 238 44 0 24 68 8

2010 GET 1,341 318 3 18 69 10

2050 FT 1,266 190 0 2 48 50

China GET 1,301 198 0 2 69 29

CER 1,328 202 1 10 76 14

CEN 1,294 195 0 0 76 24

2010 GET 1,225 284 35 23 35 6

2050 FT 1,604 323 0 16 45 38

India GET 1,624 328 6 12 61 21

CER 1,682 339 21 24 46 8

CEN 1,687 340 21 24 47 7

2010 GET 41 9 9 35 53 2

2050 FT 90 22 0 9 62 29

Kenya GET 90 22 1 8 82 9

CER 101 24 6 32 60 2

CEN 119 27 57 15 27 1

2010 GET 48 11 0 3 54 43

2050 FT 47 7 0 0 50 50

Republic of Korea GET 47 7 0 0 50 50

CER 47 7 0 2 44 53

CEN 46 7 0 0 45 55

However, even under this rather optimistic GET scenario, India The picture for Brazil lies between those for China and
will have highly unequal distribution of education by 2050. India. Brazil has not seen as rapid an education expansion
Under the two more pessimistic scenarios, more than one out as China, but also has a significantly lower proportion of
of five young Indian women—more than 70 million women of people without any formal education than India. As in the
childbearing age—will still be without any formal education. Asian Tiger states and many other countries, in Brazil, the
Even under the most optimistic FT scenario, which is current economic boom together with effective anti-poverty
depicted in Figure  4, and which assumes education expan- programmes linked to education are closely associated with a
sions similar to those in the Republic of Korea over the past rapid expansion of the better educated, younger work force.
decades, the slow change in the composition implies that The data by five-year age groups demonstrate this clearly.
by the middle of the century, there will be a sizable group Frequently given interpretations of the recent economic boom
of uneducated, mostly elderly adults in India. However, the as due mostly to the export of raw materials tend to disregard
assumed rapid expansion in tertiary education under this sce- this important human capital factor. Today, more than half
nario will result in a very well educated younger labour force. of reproductive age women have completed junior secondary

UNDP Human Development Report Office

8 OCCASIONAL PAPER 2013 /04
 3. Results

Figure 2: Population trends by level of education education, and only 3% of them have no formal schooling.
according to the constant enrollment ratio (CER) If Brazil chooses to pursue the strong policy FT trajectory,
scenario for the Republic of Korea (1970-2050) its proportion of young women with completed tertiary edu-
cation would increase from the current 7% to 34%, similar
Population (millions)
to what the Asian Tiger states recently experienced. Even the
50 medium policy GET scenario (depicted in Figure 5) shows an
TERTIARY improvement by a factor of three in the proportion of highly
40 educated women.
Some developing countries have not yet gained enough
educational momentum to assure continued further increases
30 SECONDARY
in the average education of the population. Kenya has
recently seen significant expansion of primary and junior
20 PRIMARY secondary education even among women. Although there are
some doubts about the quality of this education, it needs to
NO EDUCATION
be acknowledged as an achievement that already more than
10 90% of young women (a greater share than in India) have at
least had some schooling. However, fertility rates in Kenya are
AGES 0-14
0 still very high by any standard, with Kenyan women having
1970 1980 1990 2000 2010 2020 2030 2040 2050 on average 4-5 children.
YEAR While the FT scenario shows how Kenya could acceler-
ate its development, Figure 6 gives the most pessimistic CEN
scenario, which assumes that essentially no new schools
are being built and the number of students does not further
increase, despite rapid population growth due to rather high
Figure 3: Population trends by level of education
according to the global education trend (GET) birth rates. For this reason, it will be a major challenge for
scenario for China (1970-2050) the school system to keep pace with the rapidly expanding
number of children of school age. In the CEN scenario, the
Population (millions) country’s Government (for whatever reason) is assumed not
1400 to keep pace with population growth. Since women with low
education have particularly high fertility, this will result in
TERTIARY
1200 even more births with a still lower prospect for education. As
a consequence, there will be a decline in the average educa-
1000
tional attainment of the population, and an even more rapid
SECONDARY pace of population increase, which would make all efforts
towards social and economic development an even more dif-
800
ficult uphill battle.
PRIMARY Figure 7 finally shows the global impact of alternative edu-
600
cation trends on population growth paths until the middle of
NO EDUCATION
the century. It is based on identical education-specific fertility
400
rates; hence, all the differences are a function of the chang-
ing proportion of women in different education categories as
200
a consequence of the different education scenarios described
AGES 0-14
above. Population size in the FT scenario will be over 1 bil-
0
lion people, lower than under the CEN scenario by 2050. This
1970 1980 1990 2000 2010 2020 2030 2040 2050 implies that alternative education trajectories alone over the
YEAR
next 40 years will make a difference to global population size
that is bigger than the entire population of Africa today or
three times the current US population. This effect is strongest

UNDP Human Development Report Office

OCCASIONAL PAPER 2013 /04 9
 4 DEMOGRAPHY AND HUMAN DEVELOPMENT

Figure 4. Population trends by level of education Figure 5: Population trends by level of education
according to the fast track (FT) scenario for India according to the global education trend (GET)
(1970-2050) scenario for Brazil (1970-2050)

Population (millions) Population (millions)

TERTIARY
2,000 200

1,500 150 SECONDARY

TERTIARY

SECONDARY
1,000 100
PRIMARY
PRIMARY
NO EDUCATION
500 NO EDUCATION 50

AGES 0-14 AGES 0-14

0 0
1970 1980 1990 2000 2010 2020 2030 2040 2050
1970 1980 1990 2000 2010 2020 2030 2040 2050
YEAR
YEAR

for countries with currently high fertility rates and high educa- Figure 6: Population trends by level of education (1970-
tion differentials. In Kenya, as discussed above, the population 2050) according to the CEN scenario for Kenya
would increase from 31 million in 2000 to 84-85 million in
Population (millions)
2050 under the optimistic scenarios (FT, GET), but increase to
140 TERTIARY
an incredible 114 million if no new schools are built (CEN).
The difference between these extreme scenarios is 30 million—
120 SECONDARY
about the size of Kenya’s total population in 2000.
PRIMARY
100

4. EDUCATION, HEALTH AND MORTALITY3 80 NO

EDUCATION
Education has long been considered a key covariate of health
60
and mortality in both industrialized and developing coun-
tries; recently, more research has substantiated the causal
40
nature of this relationship.  In particular, a great deal of
research has focused on substantiating the independent and
20
causal relationship between maternal schooling and child
AGES 0-14
health (Caldwell 1979; Fuchs et al. 2010). Reviews of sur-
0
veys and census data from developing countries (Hobcraft
1970 1980 1990 2000 2010 2020 2030 2040 2050
et al. 1984; Mensch et al. 1985) and econometric analyses of YEAR
cross-country macro-data have confirmed women’s educa-
tion as the most significant determinant of child mortality,
where estimates show that each additional year of schooling
is associated with a 5-7% reduction in child death (Mensch
et al. 1985; Schultz 1993).

3 This section heavily draws on Lutz and KC 2011.

UNDP Human Development Report Office

10 OCCASIONAL PAPER 2013 /04
 4. Education, Health and Mortality

Figure 7: Alternative projections of total world population size by level of education according to the four different
education scenarios (with otherwise identical education-specific fertility and mortality rates)

GET CER CEN FT

10

9
8

7
6
Population (billions)

0
2000

2010

2020

2030

2040

2050

2000

2010

2020

2030

2040

2050

2000

2010

2020

2030

2040

2050

2000

2010

2020

2030

2040

2050
YEARS

Population 0-14 No education Primary Secondary Tertiary

A systematic assessment for a large number of developing world regions. The numbers provided are the absolute numbers
countries using recent Demographic and Health Survey (DHS) of child deaths (ages zero to four) resulting from projections
micro-level data finds that almost universally, a mother’s edu- that assume identical education-specific fertility and mortality
cation is more important for child survival than household rates, and only differ by the assumed future paths of education
income and wealth. This finding has significant implications expansion in the form of the scenarios described above.
for setting policy priorities, because it makes a big difference Looking at India, the country with the highest absolute
whether the emphasis is put primarily on girls’ education or number of child deaths in the world, the number of deaths
on household income generation. occurring in 2010-2015 of children born in this period is esti-
Table 3 shows the magnitude of empirically measured mated to be around 7.9 million. Under the most optimistic
child mortality differentials in a selection of developing coun- FT education scenario, this number will decline to almost half
tries. The indicators for the under-five mortality rate and the its level by 2025-2030, and to as low as 3.1 million by 2045-
total fertility rate are compiled from DHS for the most recent 2050. Even under the more likely GET scenario, it would
surveys (after 2004). The under-five mortality rate is high decline to 5.4 million and 3.5 million, respectively. Under
in many African countries, especially among the children of the pessimistic CEN and CER scenarios that assume no or
uneducated mothers. The worst situation among the countries only very limited future education expansion, some further
listed in the table is in Mali and Niger. In every country, the decline is likely to happen due to the momentum of educa-
mortality rates are lower for better educated mothers. In some tional expansion (the already better educated youth cohorts
countries, primary education is already associated with much will move up in the age pyramid) and the number of child
lower child mortality (e.g., in Bolivia and Nigeria); in others, deaths would slowly decline to around 5.6-6.1 million by the
the decisive difference only comes with secondary education middle of the century.
(e.g., in Liberia and Uganda). In China, although it currently still has a greater popula-
Given these strong differentials in child mortality by the tion than India, the absolute number of child deaths is esti-
level of education of mothers, it is not surprising that over mated to be only around 1.7 million, i.e., less than a quarter
time, alternative possible education trends will result in quite of those in India. Over time, further declines in the numbers
significant differences in child mortality. Table 4 gives some of child deaths according to the different education scenarios
numerical illustrations of this fact for selected countries and are much steeper in China. Under the FT scenario—which

UNDP Human Development Report Office

OCCASIONAL PAPER 2013 /04 11
 4 DEMOGRAPHY AND HUMAN DEVELOPMENT

Table 3: Under-five mortality rates and total fertility rates for selected countries for recent years,
by level of education of mothers
  Under-five child mortality* Total fertility rate
(per 1,000 live births) (number of live births per woman)
Period* No education Primary Secondary Overall No education Primary Secondary Overall
or higher or higher
Bangladesh 2007 93 73 52 74 3.0 2.9 2.5 2.7

Benin 2006 143 121 78 136 6.4 5.2 3.7 5.7

Bolivia 2008 134 87 44 76 6.1 4.7 2.6 3.5

Cambodia 2005 136 107 53 106 4.3 3.5 2.6 3.4

Congo (Brazzaville) 2005 202 134 101 123 6.2 6.3 4.0 4.8

Democratic Republic 2007 209 158 112 155 7.1 7.1 5.0 6.3
of Congo
Egypt 2008 44 38 26 33 3.4 3.2 3.0 3.0

Ethiopia 2005 139 111 54 132 6.1 5.1 2.0 5.4

Ghana 2008 103 88 67 85 6.0 4.9 3.0 4.0

Guatemala 2008 65 40 20 45 5.2 3.8 2.3 3.6

Guinea 2005 194 172 92 188 6.2 5.1 3.3 5.7

Haiti 2005-06 123 97 65 102 5.9 4.3 2.4 3.9

Honduras 2005-06 55 40 20 37 4.9 3.8 2.2 3.3

India 2005-06 106 78 49 85 3.6 2.6 2.1 2.7

Indonesia 2007 94 60 38 51 2.4 2.8 2.6 2.6

Liberia 2009 164 162 131 158 7.1 6.2 3.9 5.9

Mali 2006 223 176 102 215 7.0 6.3 3.8 6.6

Nepal 2006 93 67 32 79 3.9 2.8 2.2 3.1

Nicaragua 2006 67 40 26 41 4.4 3.2 2.0 2.7

Niger 2006 222 209 92 218 7.2 7.0 4.8 7.0

Nigeria 2008 210 159 107 171 7.3 6.5 4.2 5.7

Pakistan 2006-07 102 85 62 93 4.8 4.0 2.8 4.1

Philippines 2008 136 47 30 37 4.5 4.5 3.0 3.3

Rwanda 2007-08 174 127 43 135 6.1 5.7 3.8 5.5

Senegal 2008-09 112 74 33 100 5.6 4.5 3.1 4.9

Sierra Leone 2008 170 187 130 168 5.8 5.1 3.1 5.1

Swaziland 2006-07 150 106 95 106 4.9 4.5 3.4 3.9

Uganda 2006 164 145 91 144 7.7 7.2 4.4 6.7

Zambia 2007 144 146 105 137 8.2 7.1 3.9 6.2

Note: * Under-five child mortality refers to the period of 10 years prior. Source: DHS.

actually is relatively close to the path China has recently CER scenario because under conditions of declining absolute
embarked on—the number of child deaths will decline to only numbers of children, as will be the case in China, the assump-
around 0.5 million by mid-century, less than a third its cur- tion of constant absolute numbers of students implies a fur-
rent level. Even under the most pessimistic scenarios, it would ther expansion of the proportion of people who are educated.
only be slightly higher at 0.6-0.8 million. Here it is interesting By the middle of the century, under the CEN scenario, the
to note that the CEN scenario results in fewer deaths than the number of child deaths in India will be almost 10 times higher

UNDP Human Development Report Office

12 OCCASIONAL PAPER 2013 /04
 Conclusion

than in China, despite the fact that China today still has a
CONCLUSION
significantly higher total population size.
In Kenya, the child mortality differentials due to educa- The countries of the world are currently at different stages
tion are particularly impressive. Currently, it is estimated that of their demographic transition processes. It confuses many
there are 0.58 million child deaths, as defined above. Under people that in some countries the concern is already with
the most optimistic FT scenario, which is rather unlikely population shrinking (such as in Eastern Europe), while in
given Kenya’s current school system, this number would only others (mostly in Africa) it is still primarily with population
slightly decline by the middle of the century to 0.37 million. growth. This pattern is primarily the consequence of the dif-
In contrast, the CER scenario, assuming constant enrolment ferent timing of the phases of demographic transition in dif-
rates, would see an increase in the number of child deaths to ferent parts of the world. In Europe, death rates started to
0.69 million, and the most pessimistic CEN scenario would decline in the late 19th century due to better nutrition and
even see an explosion to 1.5 million. These significant dif- hygiene. Population growth was high during this period due
ference are only due to different future education trends. to continued high birth rates, which only started to fall (with
Otherwise, education-specific fertility and mortality trends the exception of an earlier fertility decline in France) after
are identical, showing assumed secular declines over time cor- World War I. Since all traditional societies had cultures and
responding to the fertility and mortality assumptions made in normative systems that were in favour of high fertility, in
the UN population projections. order to assure the survival of the population under high-
When interpreting these numbers, it is important to mortality conditions, it typically takes societies several dec-
understand that the projected decline in the absolute numbers ades to adjust to the new lower child mortality conditions and
of child deaths reflects the combined effects of more educated change the normative system. In much of East and South-East
women having fewer children, and fewer of these children Asia, a massive mortality decline after World War II was fol-
dying due to lower child mortality rates. One can, of course, lowed by an equally rapid fertility decline during the last two
also separately study these two trends and give relative meas- decades of the past century. It can be argued that the time lag
ures of child mortality, but under the view that every death of between the mortality and fertility declines (also interpreted
a child is one too many, it is appropriate to actually compare as a perception lag) was shortened due to rapidly increasing
the absolute numbers of child deaths. education in those countries. In Africa, the mortality decline

Table 4: Estimated and projected numbers of child deaths among children born in the stated period according to the
different education scenarios (thousands of deaths)
2010-15 2025-30 2045-50

GET CEN CER FT GET CEN CER FT GET

Brazil 328 224 232 177 182 161 185 102 106

China 1,716 897 1,067 871 969 625 809 526 633

India 7,872 6,707 6,479 4,806 5,393 6,096 5,566 3,064 3,510

Kenya 582 920 644 482 495 1,552 693 371 392

Republic of Korea 9 8 9 9 9 7 7 7 7

Mali 488 519 497 318 448 541 490 150 337

Pakistan 1,927 1,641 1,610 1,225 1,398 1,676 1,517 773 976

South Africa 288 198 204 165 171 134 137 93 102

Africa 16,552 18,964 16,381 12,095 13,803 24,185 16,807 7,495 10,380

Asia 15,029 11,715 11,544 8,924 9,845 10,561 9,780 5,681 6,493

Europe 276 209 218 204 207 196 207 187 186

Latin America and the 1,192 963 911 704 724 950 774 413 435
Caribbean
Northern America 162 160 162 155 154 165 164 152 151

Oceania 11 11 12 11 11 12 12 10 10

UNDP Human Development Report Office

OCCASIONAL PAPER 2013 /04 13
 4 DEMOGRAPHY AND HUMAN DEVELOPMENT

has started still later and has been more gradual—in some REFERENCES
countries even interrupted by HIV/AIDS. The perception lag
Bongaarts, J., and S. Sinding. 2011. “Population policy
seems to be particularly long for various reasons that range
in transition in the developing world.” Science
from exceptionally strong pro-natalist traditional values to
333(6042), pp. 574-576.
little progress in terms of education. Although in most African
societies the fertility transition has already started, the gap Caldwell, J. 1979. “Education as a factor in mortality
between low death rates and high birth rates is still great, decline: An examination of Nigerian data.”
resulting in very rapid population growth. Population Studies 33, pp. 395-413.
The demographic transition not only affects the pace of
DHS (Demographic and Health Surveys). www.measuredhs.com.
population growth, but also comes with an age-structural
transition. When birth rates start to decline, the proportion Fuchs, R., E. Pamuk and W. Lutz. 2010. “Education or
of children to adults also falls, which results in a decreas- wealth: Which matters more for reducing child
ing young-age dependency ratio. A given number of adults mortality in developing countries?” Vienna
need to support a smaller number of children, which is a force Yearbook of Population Research 2010, pp. 175-
that can drive economic growth and improvements in infra- 199. Verlag der Österreichischen Akademie der
structure and education, and is hence called the demographic Wissenschaften, Vienna.
dividend or demographic bonus. While much of the literature
Hanushek, E. A., and L. Woessmann. 2008. “The role of
in this field only points to the changing age dependency pat-
cognitive skills in economic development.” Journal
tern, it is also worth giving more attention to the improving
of Economic Literature 46, pp. 607-668.
education profile and the age structure as factors enhancing
economic growth. It has been argued that investments in Hobcraft, J. N., J. W. McDonald and S. O. Rutstein.
education actually have a double effect on economic growth 1984. “Socio-economic factors in infant and
and poverty reduction: They tend to bring down birth rates child mortality: A cross-national comparison.”
and trigger the demographic dividend, and at the same time Population Studies 38(2), pp. 193-223.
increase the economic productivity of the young adult popu- KC, S., B. Barakat, A. Goujon, V. Skirbekk, W. Sanderson
lation, which has an independent and possibly even stronger and W. Lutz. 2010. “Projection of populations by
direct effect on economic growth (Lutz 2009). level of educational attainment, age and sex for 120
Education matters greatly for almost every aspect of countries for 2005-2050.” Demographic Research
progress in human development. The empowering function 22, article 15, pp. 383-472. www.demographic-
of education is considered a major goal in its own right. In research.org/Volumes/Vol22/15/, DOI: 10.4054/
particular, universal basic education is a significant factor DemRes.2010.22.15.
for enhancing the health and empowerment of men and
women alike; this is also why it figures prominently among KC, S., and H. Lentzner. 2010. “The effect of education
the MDGs. The Human Development Index, one of the most on adult mortality and disability: A global
widely used indicators of desirable social and economic pro- perspective.” Vienna Yearbook of Population
gress, consists of three components, one measuring progress Research 2010(8), pp. 201-235.
in education itself and the two others capturing progress in Lutz, W. 2009. “Sola schola et sanitate: Human capital
health and material well-being (UNDP). As discussed above, as the root cause and priority for international
there is reason to assume that these latter two aspects are development?” Philosophical Transactions of the
to some extent driven by progress in the education of the Royal Society B: Biological Sciences 364(1532),
population, and that indeed human capital may be viewed as pp. 3,031-3,047.
the root cause of human development (Lutz 2009). Human
capital data differentiated by age and gender help to shed Lutz, W., J. Crespo Cuaresma and M. J. Abbasi-Shavazi.
some light on this assertion. They can open the door to many 2010. “Demography, education, and democracy:
further studies that go beyond the cruder previous indicators Global trends and the case of Iran.” Population and
of education, which lacked age or distributional detail, to pro- Development Review 36(2), pp. 253-281.
vide a richer understanding of the many longer term returns
to investments in education.

UNDP Human Development Report Office

14 OCCASIONAL PAPER 2013 /04
 References

Lutz, W., J. Crespo Cuaresma and W. Sanderson. 2008.

“The demography of educational attainment and
economic growth.” Science 319, pp. 1,047-1,048.

Lutz, W., A. Goujon, S. KC and W. Sanderson. 2007.

“Reconstruction of populations by age, sex and
level of educational attainment for 120 countries
for 1970-2000.” Vienna Yearbook of Population
Research 2007, pp. 193-235. Verlag der Österrei-
chischen Akademie der Wissenschaften, Vienna.

Lutz, W., and S. KC. 2011. “Global human capital:

Integrating education and population.” Science 333,
pp. 587-592.

Lutz, W., W. C. Sanderson and S. Scherbov, eds. 2004.

The End of World Population Growth in the
21st Century: New Challenges for Human
Capital Formation and Sustainable Development.
Earthscan, London.

Mensch, B., H. Lentzner and S. Preston. 1985. “Socio-

economic differentials in child mortality in
developing countries.” ST/ESA/SER.A/97. United
Nations, New York.

OECD (Organisation for Economic Co-operation and

Development). 2007. Education at a Glance. Paris.

Schultz, T. P. 1993. “Mortality decline in the low-income

world: Causes and consequences.” American
Economic Review 83(2), pp. 337-42.

UNESCO Institute for Statistics. 2009. Global Education

Digest 2009: Comparing Education Statistics
Across the World. Montreal, Canada.

UNDP, United Nations Development Program. Human Devel-

opment Index. http://hdr.undp.org/en/statistics/hdi/.

UNDP Human Development Report Office

OCCASIONAL PAPER 2013 /04 15
 The Human Development Report is an independent
pub­lication commissioned by the United Nations
Development Programme (UNDP). Its editorial autonomy
is guaranteed by a special resolution of the General
Assembly (A/RES/57/264), which recognizes the Human
Development Report as “an independent intellectual
Human Development Report 2013 exercise” and “an important tool for raising awareness
about human development around the world.”

The Rise of the South: From the beginning, the Report has been a pioneer of
Human Progress in a Diverse World methodological innovation and development thinking.
Often provocative, the Report was launched in 1990 with
the goal of putting people at the center of development,
going beyond income to assess people’s long-term well-
being. The Reports’ messages — and the tools to implement
them — have been embraced by people around the world,
as shown by the publication of autonomous National Human
Development Reports by more than 140  countries over
the past two decades. The Human Development Report is
translated into more than a dozen languages and launched
in more than 100 countries annually.
Contributors to the Report include leading development
scholars and practitioners, working under the coordination
of UNDP’s Human Development Report Office

Human Development Report 2013:

The Rise of the South – Human Progress in a Diverse World

The 2013 Human Development Report examines the

causes and consequences of the “Rise of the South,” and
identifies policies emerging from this new reality that can
promote greater sustainability, social cohesion and human
development progress throughout the world in the decades
to come.

The report may be downloaded at http://hdr.undp.org/

Printed copies may be ordered at https://unp.un.org/hdr/

United Nations Development Programme

Human Development Report Office
304 E 45th Street, FF-1204
New York, NY, 10017 USA

Copyright 2013, UNDP