Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Dimensions of global population projections: what do we know about future population trends and structures?

Dimensions of global population projections: what do we know about future population trends and... Phil. Trans. R. Soc. B (2010) 365, 2779–2791 doi:10.1098/rstb.2010.0133 Review Dimensions of global population projections: what do we know about future population trends and structures? 1,2,3, 1 Wolfgang Lutz and Samir KC International Institute for Applied Systems Analysis (IIASA), Schlossplatz 1, A-2361 Laxenburg, Austria Austrian Academy of Sciences, Dr Ignaz Seipel-Platz 2, 1010 Vienna, Austria Vienna-University of Economics and Business 2–6 Augasse, 1090 Vienna, Austria The total size of the world population is likely to increase from its current 7 billion to 8 – 10 billion by 2050. This uncertainty is because of unknown future fertility and mortality trends in different parts of the world. But the young age structure of the population and the fact that in much of Africa and Western Asia, fertility is still very high makes an increase by at least one more billion almost certain. Virtually, all the increase will happen in the developing world. For the second half of the century, population stabilization and the onset of a decline are likely. In addition to the future size of the population, its distribution by age, sex, level of educational attainment and place of residence are of specific importance for studying future food security. The paper provides a detailed discussion of different relevant dimensions in population projections and an evaluation of the methods and assumptions used in current global population projections and in particular those produced by the United Nations and by IIASA. Keywords: world population; population increase; population decline; dimensions of population projections; age structure; level of educational attainment 1. INTRODUCTION (called assessments and published in their World While future trends in the number and composition of population prospects series) have been made every humans on this planet has been a topic of scientific 2 years. The most recent UN assessments have a enquiry and discussion for centuries and at least since time horizon to 2050. Bongaarts (2009) gives a Thomas Malthus entered the field of structured quanti- concise summary of the main results and implications tative analysis, the first modern global population of the UN projections. At irregular intervals the UN projection, which explicitly considered the age and sex also publishes long-term population projections with structure of the population (the so-called cohort- time horizons from 2150 to 2300. component method), was carried out by Frank The World Bank started to produce independent Notestein of the Princeton Office of Population population projections in 1978. These were always Research in 1945 (Notestein 1945). At the national meant primarily for internal use in the Bank’s develop- level, several population projections precede this first ment planning and were published as part of the comprehensive global projection. Hajnal (1955) provides series of World development reports. After 1984, the a good overview of these early population projections. World Bank projections were revised approximately Frank Notestein subsequently became the first every two years and in most cases only one variant director of the then newly established United Nations was published but with a long time horizon to 2150. Population Division. This unit began producing regu- Around 1995, the World Bank stopped publishing lar global population projections in the early 1950s. separate projections but presumably continued to use Between 1951 and 2008, the UN published 21 sets them for internal purposes for a number of years. of estimates (past and current conditions) and projec- The Washington-based Population Reference Bureau tions (future) for all countries and territories of the (PRB) publishes independent world population pro- world. Before 1978 these projections were revised jections (population size only and a single scenario) approximately every 5 years; since then new revisions every year as part of its annual World population data sheet. Since 2000 it has published projected population sizes for all countries and territories of the world * Author for correspondence (lutz@iiasa.ac.at). for 2025 and 2050. In some cases the projections are identical to those of the UN and the US Census While the Government Office for Science commissioned this review, Bureau (USCB), but in some cases different the views are those of the author(s), are independent of Government, and do not constitute Government policy. country-specific information is used. The USCB produces single scenario projections for One contribution of 23 to a Theme Issue ‘Food security: feeding the world in 2050’. all countries in the world as of 1985 with a varying 2779 This journal is q 2010 The Royal Society 2780 W. Lutz & K. C. Samir Review. Global population projections time horizon. The World Population Programme of that has recently received the greatest public attention, the International Institute for Applied Systems Analy- namely, the questions if and when we will see an end of sis (IIASA) based outside Vienna (Austria) began population growth followed by a possible population producing global population projections at the level decline. In the final section of the paper we propose of 13 world regions in 1994. One of the purposes a specific existing population projection (the UN was to produce population projections as part of scenario of the IIASA Education Projections) for use the Special Report on Emissions Scenarios (SRES) in the Foresight Project. We present the results for (Nakicenovic et al. 2000) that underlie the global emis- the specified regions and countries in tabular and sion scenarios used by the Intergovernmental Panel for graphical form. Climate Change (IPCC). This was followed by three rounds of probabilistic projections at the level of 13 2. DIMENSIONS CONSIDERED IN POPULATION regions (which were all published in the pages of PROJECTIONS Nature: Lutz et al. 1997, 2001, 2008b). The IIASA For many users the most important piece of infor- projections come from a distinctly scientific back- mation expected from population projections is the ground, which is illustrated by the fact that in its future total size of the population; for many policy publication (e.g. Lutz et al. 2004a) much more space and research questions other dimensions of the com- is used for justifying and discussing methods and position of the population are also of interest. The assumptions than for presenting results. Recently, dimensions that will be explicitly discussed here are IIASA developed (in collaboration with Eurostat and age, sex, rural/urban place of residence, educational the UK Office of National Statistics) a new approach attainment, labour force participation, parity status, for evaluating substantive arguments about alternative household status and health status. All these dimen- assumptions on possible future trends based on large sions have in common that they are properties of numbers of questionnaires ascertaining the evaluation each individual member of the population which are, of alternative arguments by experts. This is currently at least in theory, observable and measurable. They being translated (together with Oxford University) can hence be called observable sources of population into a new set of science-based population projections heterogeneity. These dimensions are not only interest- for all countries in the world by age, sex and level of ing in their own right for helping to answer specific education for publication in 2012. questions relating to their future distributions in the O’Neill et al. (2001) published a comprehensive population but to the extent that they are associated 85-page review entitled A guide to global population with different fertility, mortality and migration intensi- projections. In this article, they describe in considerable ties, their changing distributions in the total detail the projections of all five agencies discussed population also impact on the projected future popu- above. The paper includes a critical review of the lation size. In the following, we will discuss each of methods used and of the ways that assumptions are these dimensions individually, assess to what degree defined by those five agencies. In addition, the results they are likely to impact on the course of overall popu- are compared in quite some detail. For this reason, we lation, discuss how they may be important in their own do not want to repeat the exercise of systematically right in the context of studying food security, and comparing the approaches and results of the different review existing global level projections, which explicitly projections, but rather focus on a few key challenges address this dimension. that point to the future. It is important to note that in addition to these In the next section we will discuss the important observable sources of population heterogeneity, there questions of what we call ‘dimensions’ of population is still unobserved heterogeneity in every population projections. The basic idea is that any population is which is hard to capture empirically. Theoretical con- an assembly of individuals and each individual is siderations suggest that such unobserved heterogeneity different from any other individual. But which of the can significantly impact future population dynamics many relevant properties of individuals should we (Vaupel & Yashin 1985), but there is little one can do explicitly consider in population projections? The about it except to be aware of the problem and be lowest dimensionality is chosen for the case that all cautious about the validity of the conclusions drawn. individuals are taken as equal (homogeneous) and Given this problem associated with hidden heterogen- only a total growth rate is assumed for projecting eity, it is even more important to explicitly measure population size. The cohort-component method and incorporate the observable sources of population breaks down by age and sex; this is a first step in the heterogeneity wherever feasible and thus try to mini- direction of increasing dimensionality by explicitly mize the possible biases caused by overall heterogeneity. considering these individual properties. But there are many further properties that deserve consideration in population projections and which will be discussed in (a) Population size the following section. Most users of population projections seem to be pri- We will also address the tedious but very important marily interested in the total size of the population. issue of how to deal with uncertainty in population From a substantive point of view, however, it is diffi- projections. While a single best-guess population pro- cult to think of specific policy or research questions, jection is likely to satisfy the expectations of a where only changes in absolute size matter and majority of users, disregarding the issues of uncer- where it is supposedly irrelevant whether one refers tainty can be rather dangerous and therefore deserves to newborn babies, young adults or frail elderly serious attention. We will then turn to the one issue people. Presumably, the great interest in data on the Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2781 sheer number of people has two main reasons. First, on fixed age categories and does not consider the total population size can be seen as a first-order fact that healthy life-expectancy tends to increase in approximation of the scale of the population-related parallel with total life-expectancy. Hence, a 65-year- issues in particular, if it is assumed that the age pat- old person today is typically quite different from one terns of the two populations to be compared do not of the same age who lived several decades ago and differ greatly. Second, total population size is an who had a much shorter remaining life-expectancy. important denominator of many frequently used This idea has recently been translated into a redefini- indicators ranging from gross domestic product tion of age, where age is not just measured as the (GDP) per capita to food consumption per person to time since birth but can alternatively be viewed as greenhouse gas emissions per person. In all those the expected time to death, in which case the popu- cases a separately derived total quantity is divided by lation dynamics looks quite different (Lutz et al. total population size in order to produce an indicator 2008b; Sanderson & Scherbov 2008). that can be compared across populations. The distinction between men and women as a key Traditionally, population projections have been pro- dimension of population dynamics has been a feature duced by simply taking total population size and of demography almost from the beginning. It has to making assumptions on the future growth rates of do with the fact that fertility rates are almost exclu- the population. This has been the standard method sively measured with women. One of the reasons for until the age-specific, so-called cohort-component this is that not all fathers are known and that the repro- method became the widely used standard after World ductive age range for women is shorter than for men War II. But for some specific applications, where the and shows a clearer and partly biologically determined age structure is difficult to assess or where the popu- age pattern. Also, mortality rates tend to differ signifi- lation is simply too small to have meaningful age cantly between men and women. Except for some groups (as is sometimes the case with small area cases of extreme female discrimination, male mortality population projections), simple growth rate-based rates are typically higher at every single age. In projections are still being used. But for national addition to these demographic reasons there is, population projections, age-group-wise projections of course, significant substantive interest in the age- have clearly become the state-of-the-art because specific proportions of men and women in the they allow differentiation between the behavioural population which matters for issues ranging from the components (fertility, mortality and migration) and marriage market to labour market to consumer embedded changes that are merely owing to age- preferences. structural effects. This is particularly important for The formal model for doing population projections countries that have gone through significant fluctu- by age and sex was introduced by Cannan (1895) ations in fertility and mortality trends and hence but it took until after World War II to have this have irregular age structures, such as most European cohort-component method spread around the world countries after the two world wars. But even for high as the dominant way for doing population projections. fertility countries that traditionally had very stable Today, it is used by all national and international and regular age pyramids, the assumed future fertility statistical agencies. decline will be partly compensated in terms of popu- lation growth through the age-structural momentum, i.e. the fact that in the future larger numbers of poten- (c) Rural–urban place of residence tial mothers will enter the reproductive age and the Place of residence has always been an important number of births will continue to increase even if dimension since the very origin of demography when period fertility is at replacement level. This is another statistical information on households was collected good reason for explicitly considering the age structure for taxation purposes and conscriptions for the army. wherever possible. Associating people with a specific locality, however, requires that people have a more or less permanent residence and are not permanently on the move. (b) Age and sex This is one problem facing statisticians when trying Particularly in the context of population ageing, the to distinguish between urban and rural populations explicit interest in the population’s changing age struc- in countries (such as China), where millions of ture has recently increased significantly. While there people (the so-called floating population) are con- has always been an interest by planners in the likely stantly on the move between their home villages and future number of school-age children or young the urban centres where they work but have no legal adults trying to enter the labour market, the currently right of residence. Another problem with comparing dominating questions relate to the sustainability of reported proportions urban across countries and over pension systems under conditions of rapid increase in time is that there is a multitude of changing national the number of elderly. Closely related questions deal sets of definitions based partly on the size of the with the likely impact of ageing on healthcare cost or agglomeration, its statutory status, its population care for the elderly in general. There is significant con- density or the proportion of the labour force that cern that an increase in the mean age of the labour is working in agriculture. Since urbanization will be force may be associated with decreasing productivity discussed elsewhere, it will not be covered here. in the future (Skirbekk 2008). These are all questions From a population projections point of view, the for which a projection of the population by age is question whether a person lives in a rural or an essential. Much of this discussion, however, is based urban community is not only seen in its own right Phil. Trans. R. Soc. B (2010) 2782 W. Lutz & K. C. Samir Review. Global population projections but is also viewed as a relevant source of population populations that are still in the process of demographic heterogeneity. Almost universally women in urban transition, more educated women have lower fertility. areas have lower fertility than women in rural areas. These educational differentials can be very significant. Also in terms of international migration, new immi- The Demographic and Health Survey for Ethiopia, for grants today generally move to urban areas except for instance, shows that women without any formal edu- some special cases where immigrant labour is concen- cation have on average six children, whereas those trated in agriculture. With respect to mortality, the with secondary education have only two (see http:// pattern is more complex. While the inhabitants of www.measuredhs.com). Significant differentials can cities traditionally had higher mortality owing to a be found in most populations of all cultural traditions. higher risk of infectious disease, this pattern virtually Only in a few modern societies does the strongly nega- disappeared over the course of the twentieth century tive association give way to a U-shaped pattern in because healthcare and education were better in which the most educated women have a somewhat urban areas. But most recently there are indications higher fertility than those with intermediate education. of a reversal, with urban slums in some African cities But globally, the education differentials are so perva- having worse mortality conditions than even remote sive that education may well be called the single villages in the same country. most important observable source of population het- From a methodological point of view, the best erogeneity after age and sex (Lutz et al. 1999). There model to project a population by age and sex for are good reasons to assume that during the course of urban and rural areas separately is the so-called a demographic transition the fact that higher education multi-state model, which was developed at IIASA in leads to lower fertility is a true causal mechanism, the 1980s (Rogers & Land 1982; Keyfitz 1985). It where education facilitates better access to and infor- has become the state-of-the-art for multi-dimensional mation about family planning and most importantly population projections in which separate fertility, mor- leads to a change in attitude in which ‘quantity’ of chil- tality and migration rates are assumed for the different dren is replaced by ‘quality’, i.e. couples want to have sub-populations as well as age- and sex-specific tran- fewer children with better life chances. sition rates between sub-populations. Because of data In terms of measuring education, it is important to limitations, such true multi-state projections by distinguish between stock and flow variables. Most urban – rural place of residence exist only for selected education programmes and studies are concerned groups of countries. with the process of schooling itself, including many The United Nations Population Division publishes important topics that range from the construction of the only urbanization projections on a global level, schools and the training of teachers to the organization with updates every two years. The most recent revision of schools, the quality of teaching and the content of (United Nations 2008) estimates the population living curricula. The extent to which these schooling efforts in urban areas in 229 countries and territories of cover the entire population are usually measured the world for the period 1950 – 2005 and projects through enrolment rates as published for most the same for 2010 – 2050 in five year intervals. The countries by UNESCO and other agencies. But for basic method of estimation and projection is still the many of the social and economic benefits of education same as when it was developed in the 1970s with it does not directly matter how many people are in only a few minor modifications. The main source of school at any given point in time, but rather how information used in the projection is the urban – rural many have completed their education and use their growth difference at the two latest available time acquired skills in the labour force. Hence, the main points (mostly inter-census). The difference is then interest is not on the flow (school enrolment) but assumed to follow a logistic path and is estimated rather on the changing stock (human capital). The based on the past experience of many countries and stock can be measured by categories of highest the expectation that the urbanization process will educational attainment (here the UNESCO-ISCED slow down as the level of urbanization grows. In categories have become the standard) as well as by other words, this is not a multi-state model, but only mean years of schooling. Since the original empirical the application of assumed overall proportions urban data mostly come in terms of attainment and the to national level projections. This model by definition calculation of mean years of schooling requires assumes continuous increases in the proportion urban additional country-specific assumptions, the former and is unable to model possible ups and downs. This is often preferred. Also, considering the full attainment model is isomorphic to the literacy projection model distribution provides information about inequalities that has been used by UNESCO for decades, but that is hidden in the case of only using an average which was recently replaced by a cohort-component such as the mean years of schooling. model. Because of the strong association between female education and fertility, future changes in the compo- sition of the female population by educational (d) Educational attainment attainment make a big difference. Since many develop- Like the other dimensions discussed above, education ing countries have seen major improvements in school is also an important source of population heterogen- enrolment rates of both girls and boys in those eity and bears a significant weight of its own. Almost countries, the future women of reproductive age are universally more educated people have lower mortality bound to be more educated than today’s. This great and there is sufficient evidence that this is a real effect momentum and rather easy predictability of the edu- and not just owing to selectivity. Also for all cational attainment distribution is a consequence of Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2783 the fact that education is typically acquired early in life distribution of women by the number of children and then remains virtually constant along cohort lines. they have already given birth to, is an important deter- If we know how many 10-year-old girls with some pri- minant of fertility in the near-term future. Highly mary education there are today, we know (when important for economic and labour force consider- considering differential mortality and migration) how ation is the distribution of the population by age, sex many 50-year-olds with at least primary education and labour force participation rates. While there are there will be in 40 years’ time. Given the fact that in several national projections of future labour force par- most countries the younger cohorts are better edu- ticipation, consistent global projections do not yet cated than the older ones, further declines in fertility exist. Another very relevant dimension in the context as well as mortality are virtually pre-programmed. In of global population ageing and the associated needs a few African countries where school enrolment rates for care for the elderly is the projection of the popu- have actually declined over the past two decades, lation by health status. While significant progress has this has led to a stalled fertility decline and partly been made in producing internationally comparable worsening mortality conditions. indicators of disability and its impact on activities of Aside from its effects on population dynamics, there daily life, up to now this has not yet led to the pro- is also significant interest in knowing the educational duction of global projections which cross-classify structure of the population for a broad range of health status with age and sex. Alternatively, one social and economic development concerns. Based could try to classify people according to certain disease on a newly reconstructed set of educational attainment risks or infection statuses (such as HIV-positive) if the distributions by age and sex (Lutz et al. 2007a) for data are available, and a certain disease is of specific most countries back to 1970, it has recently been interest. Finally, an important field of analysis and pro- shown that indeed the improvement of educational jections deals with the living arrangements/household attainment in the working age population has been composition of people. Here, some progress in the the most consistent and significant driver of economic direction of global projections has recently been growth around the world (Lutz et al. 2008a). But made in the context of trying to project future green- improving education by age and sex has also been house gas emissions and the importance of shown to matter for countries transitioning to democ- demographic/household factors in climate change racy and more liberal rights (Abbasi-Shavazi et al. mitigation (van Imhoff & Keilman 1991; Jennings 2008; Lutz 2009b). It has already been mentioned et al. 2004; Jiang & O’Neill 2009). that education is perhaps the single most important health determinant. For the question of food security, it has long been shown that the basic education 3. TREATMENT OF UNCERTAINTY of the agricultural labour force is a key factor in In the field of social and economic trends—as in all agricultural production (Hayami & Ruttan 1971). fields that deal with human behaviour—the future As the set of population – education – development – can never be projected with certainty. In the context agriculture (PEDA) models commissioned by the of the cohort-component model of population projec- UN Economic Commission for Africa for a number tion that differentiates by age and sex, all three of African countries shows, when including education components of population change (i.e. future fertility, in an agricultural production function, it turns out mortality and migration trends) are uncertain. For to be one of the key determinants in reducing each of these three components of change, the levels malnutrition and food insecurity (Lutz et al. 2004b). as well as the specific age patterns (by sex) are uncer- IIASA has recently produced population projec- tain, but the level is considered to be of primary tions by age, sex and four levels of educational importance. attainment for 120 countries following different scen- How to deal with uncertainty in population forecasting ario assumptions with respect to alternative future was the topic of a recent special issue of the Inter- education trends and different education-specific ferti- national statistical review (Lutz & Goldstein 2004). It lity trends (KC et al. 2010). These projections are summarizes the state-of-the-art in a field of research based on a true multi-state model, i.e. individuals are that has recently gained more interest both among assumed to be subjected to different mortality and demographers as well as users. Based on this analysis, fertility levels depending on their educational attain- we will distinguish between four different ways that are ment level. Hence, even identical trends in commonly used to deal with the difficult issue of education-specific fertility can lead to different overall uncertainty: (i) ignore uncertainty and publish only fertility levels in the case of differing education one projection; (ii) define alternative probability-free trends. We will return to this issue when discussing scenarios; (iii) publish high, medium and low variants the recommended combination of these education that are supposed to cover a ‘plausible range’; and (iv) projections with the UN population projections by produce fully probabilistic projections that give age and sex. quantitative information about the range of uncertainty. For many users it is sufficient to have only one ‘best-guess’ forecast. This is the projection which is (e) Other relevant dimensions seen as the most likely population trajectory from There are several other dimensions of population that today’s perspective, based on the best knowledge avail- are considered relevant either in their own right or as able today. Such a projection can give some orientation further important sources of heterogeneity. The about the direction into which things are moving. It is parity distribution of the female population, i.e. the sufficient for many applications with a relatively short Phil. Trans. R. Soc. B (2010) 2784 W. Lutz & K. C. Samir Review. Global population projections time horizon and where the costs associated with a growth, this can be viewed as an acceptable simplifica- deviation from the projected mean are negligible. But tion when interested primarily in total population size. for many planning purposes these costs can be rather When the interest is in population ageing, however, or significant. For example, if there is a plan to build a age-specific indicators such as the future proportion of new primary school and the projection should tell the population above age 80, this can result in a major the local authorities how many 5 – 10-year-olds there underestimation of uncertainty. A final problem with are likely to be in the next 10 – 20 years in a certain this approach arises when aggregating national projec- school district, such projections are associated with tions to regional or global ones. The global high significant uncertainties and any deviations from the variant of the UN projections is calculated as the best-guess projection can turn out to be rather costly. sum of all national level high variants, hence assuming This is even more so for projections of the number that there is a perfect correlation among all national of people with pension entitlements in a given country. trends. Each country simultaneously experiences the In terms of global level projections, those presented by highest plausible fertility trajectory. Such perfect cor- the USCB, the World Bank and the Population Refer- relation is highly unlikely, however, and a higher ence Bureau provide only single projections for all than expected trajectory in one part of the world countries in the world. may be partly offset by diverging trends in other The definition of ‘alternative scenarios’ is a fre- parts of the world. Hence, what is considered as a quently chosen approach that goes beyond a single plausible range at the national level is not necessarily projection and presents the user with several possible plausible at the global level. future trajectories. The terminology is somewhat Only ‘fully probabilistic’ projections can avoid such inconsistent and variants (see next paragraph) are problems. These are based on pre-defined uncertainty often called scenarios. Generally, scenarios are defined distributions over time for each of the three com- as consistent sets of assumptions. The notion orig- ponents which are stochastically combined in large inally came from the world of theatre, where scenes numbers (typically several thousands) of cohort-com- show consistent pictures of our possible setting. In ponent projections with the individual random draws demography, this is typically done by combining being subject to assumed autocorrelations and, in the different assumptions about future trends in fertility, case of multi-regional projections, correlations mortality and migration in a way that should ‘not be among the regions. In most projections, the com- impossible’. Hence, scenarios are generally considered ponents are assumed to be uncorrelated between free of probabilities with the only criterion being each other. As the field of probabilistic population pro- internal consistency. The user is given a broad range jections has grown, there have been essentially three of what could be theoretically possible without being traditions in defining the uncertainty distributions: told what is likely to happen and what probability one based on time series analysis (which is only appli- range is covered by the interval between the highest cable to countries that have long empirical series of and lowest scenario considered. At the global level, data and where no structural discontinuities are the most frequently used population scenarios are expected); one based on the analysis of the errors of those produced in the context of the SRES scenarios old projections (ex post error analysis combined with of the IPCC (Nakicenovic et al. 2000), which in the assumption that future errors will essentially be turn are based on combinations of projections pro- the same as past projection errors); and one based duced by IIASA and the United Nations Population on the evaluation of expert arguments resulting in Division. Eurostat also presents its population sets of subjective probability distributions. While projections in the form of different scenarios and an more experts as well as national and international increasing number of European national statistical agencies have entered this field of probabilistic projec- agencies follow this practice. tions, some convergence among these different The presentation of ‘high, medium and low var- approaches seems to be ongoing. It is also interesting iants’ goes a step further in providing the user with that the initiative for doing such projections does not some range of uncertainty that can be interpreted always come from the producers but rather from the intuitively. For many decades, this approach has users. For example, in the case of the UK Office of been the common practice of the UN population pro- National Statistics, such projections were recently pro- jections as well as a large majority of national statistical duced upon request by the financial planning offices. The variants are mostly generated by assuming authorities. At the global level, IIASA (Lutz et al. three alternative fertility trajectories, which are then 1997, 2001, 2008b) has produced the only probabilis- combined with identical mortality and migration tra- tic population projections at the level of 13 world jectories. They are typically presented to the user as regions. providing a ‘plausible range’ of future population From a theoretical perspective, fully probabilistic trends, hence providing a first approach to uncertainty projections are superior in avoiding the problems of although the user is not specifically told what is meant other approaches as discussed above. They provide by ‘plausible’. Does it mean 100 per cent of all poss- the users with the most comprehensive and detailed ible cases, 80 per cent or only 50 per cent? A further information in terms of the demographic ‘risks’ that shortcoming of this approach is that it only covers can be combined with a cost function. But this the uncertainty associated with future fertility trends approach is criticized by some as conveying a false and does not include uncertainties associated with sense of precision. It is indeed true that this approach future mortality or migration. Since fertility is the requires more detailed explicit assumptions in terms of most important determinant of long-term population the full distributions and correlations than the other Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2785 high end: TFR = 2.5 TFR = 2.5, LEMAX = 120 TFR = 2.0, LEMAX = 120 6 6 TFR = 1.7, LEMAX = 120 TFR = 1.5, LEMAX = 120 2 TFR = 1.0, LEMAX = 120 low end: TFR = 1.0 year Figure 1. Total world population in billions: probabilistic projections until 2100. Yellow, 95% interval; green, 60%; blue, 20% and extensions to 2200. The scenarios shown combine different levels of total fertility rate as indicated with the assumption that life-expectancy continues to increase up to a maximum level of 120 years. Source: Lutz & Scherbov (2008). common approaches. Yet, the proponents of probabil- conditions in addition to the uncertain future paths istic projections argue that they only try to make for the three components. This is even relevant at otherwise implicit assumptions explicit and extensive the level of world population projections because of sensitivity analyses show that the overall results tend the great uncertainty of current fertility levels in to be very robust to changes in the specific shapes of the world’s most populous country, China, as will be the assumed uncertainty distributions. But at this discussed in more detail below. stage it is clear that a universally accepted approach to quantitatively describing the uncertainty of popu- lation projections has not yet been developed and 4. THE PROSPECT OF WORLD POPULATION more research in this field is needed. STABILIZATION OR DECLINE AFTER PEAK A final source of uncertainty that often tends to be There has been much discussion recently about the disregarded in population projections is the uncer- long-term outlook for world population and in par- tainty about current conditions, i.e. the population ticular whether the global population is likely to size and structure as well as the levels of fertility and stabilize or even decline in size after a peak during mortality in a given country in the starting year of the second half of the twenty-first century. This discus- the projections. While this tends to be well documen- sion was, in part, triggered by an article published in ted in industrialized countries, there are major gaps of Nature in 2001 entitled The end of world population information, in particular, in Africa and parts of Asia. growth (Lutz et al. 2001), which indicated that there For many countries the information is only based on was an 80 – 90% chance that the world population sample surveys and for some countries even such would reach a peak before 2100 and start to decline information is not available. While the UN Population thereafter. Whether or not such a peak will actually Division makes heroic efforts to come up with popu- occur and at what level and how steep the following lation estimates for all countries in the world, the decline will be is primarily a function of the assump- fact that many numbers referring to the recent past tions made about long-term fertility levels in have to be adjusted in every new assessment round different parts of the world. as new information becomes available shows how dif- When it comes to global population projections ficult this task is. But demographers have become so beyond 2050, there are only the occasionally published used to having a precise point estimate for any demo- UN long-term projections as well as the IIASA world graphic indicator, even in countries with very population projections for comparison. In 2004, the fragmentary empirical information, that they often UN Population Division published a study entitled forget about the uncertainty of those indicators refer- World population to 2300 (United Nations 2004)in ring to the starting conditions when making which different long-term scenarios were presented projections. A recent study on past projections in five for all countries in the world by extending their usual southeast Asian countries (Khan & Lutz 2008) projections to 2050 by another 250 years. While life- showed that the errors introduced by incorrect infor- expectancy was assumed to continue to increase over mation about starting conditions was in some cases the entire period although at a decelerating rate, even higher than the error owing to incorrect assump- the fertility scenarios defined were all very close to tions about the future. Lutz et al. (2007b) expanded replacement level. Of the three long-term fertility the concept of probabilistic population projections assumptions, the low one assumed universal conver- to explicitly include the uncertainty about starting gence to 1.85, the high one to 2.35 and the medium Phil. Trans. R. Soc. B (2010) total population 2200 2786 W. Lutz & K. C. Samir Review. Global population projections one to whatever the replacement level might be (some- and projections (from 1950 to 2050). Also, for all what below 2.1 under good mortality conditions). It is international agencies that are part of the UN family interesting to note that the fertility level of 1.85, which of agencies (including the Food and Agriculture in the projections to 2050 is assumed to be the conver- Organization), there is an institutional agreement to gence level of the medium variant (with the low variant only and consistently use the UN projections in being 1.35) for the long-range projections, is the order to avoid embarrassment arising from using lowest fertility scenario presented. But even these different numbers in different parts of the UN. very small differences in the long-term fertility levels Hence, despite the above described shortcomings of produce significantly different long-term global popu- the UN projections, particularly in the way they deal lation sizes: by 2100, the resulting populations are 5.5 with uncertainty, there is no doubt that virtually all (low), 9.1 (medium) and 14.0 (high) billion, and by groups and agencies dealing with food security and 2200, the differences further increase to 3.2, 8.5 and agriculture expect to see the UN medium variant as 21.2 billion, respectively. Hence, the medium scenario their population projection of choice. For this reason results in some sort of population stabilization in the and with the only exception of China—as discussed very long run, but only because global long-term below—this driver review also recommends using the fertility is assumed to remain constant exactly at the UN medium variant in terms of population size and replacement level of fertility—a level that is defined the age and sex structures that come with it. as the one producing long-term stationarity. Minor But this review can do better than just considering deviations to the lower side will produce significant the population by age and sex. As discussed above, population decline and to the higher side will result education is a key dimension in the study of develop- in substantial long-term increases. ment and food security. IIASA has recently produced Lutz & Scherbov (2008) recently published another population projections for most countries in the set of long-term global projections, which extends the world by age, sex and four levels of educational attain- IIASA probabilistic population projections—which go ment up to 2050. While the baseline scenario in these to 2100—further into the future by defining scenarios projections deviates from the UN medium variant in covering a wider range of possible future fertility levels. assuming different future fertility levels throughout Selected findings are shown in figure 1 and were dis- Europe (following the Eurostat projections that cussed in a recently published editorial in the Journal have been worked out in collaboration with national of the Royal Statistical Society entitled Towards a world statistical offices) as well as in some east Asian of 2–6 billion well-educated and therefore healthy and countries, IIASA has also calculated a so-called ‘UN wealthy people (Lutz 2009a). The extension scenarios Scenario’ for comparative purposes. Since in the pro- start in 2080 (the year until which the assumptions jections by level of education assumptions are for the probabilistic projections were defined); some defined in terms of future education-specific fertility scenarios continue from the level of the median of and mortality trends, which then have to be weighted the projected distribution; others from the upper and by the size of the respective education categories if lower bounds of the projected 90 per cent uncertainty overall fertility and mortality rates are to be calculated, range (figure 1). the changing educational composition over time The figure clearly illustrates what has been dis- induces differences to the UN projections even if edu- cussed above with regard to the UN long-range cation-specific fertility and mortality assumptions are projections. In the long run global fertility levels defined to be as close as possible to the UN assump- below 2.0 will result in population decline and above tions. A full congruence of the two sets of 2.0 in long-term population increase. But the figure projections can only be achieved through an iterative also illustrates that there is a real chance that global procedure in which for every country and every point population could fall below its current size by the in time, the education-specific rates are modified in a middle of the next century, even if global fertility way that their weighted average becomes identical to levels were somewhat higher than what is being the overall rates assumed by the UN (2009). This pro- experienced in Europe today. cedure has been performed for all countries for the ‘UN-Scenario’ and results in a projected age, sex and education structure that is identical to the age 5. PROPOSAL FOR USE OF POPULATION and sex structure of the UN medium variant, but PROJECTIONS IN FOOD SECURITY also gives the education distribution as well. The ASSESSMENTS user can then choose whether to use this scenario, The criteria for recommending a specific population which is perfectly in line with the widely used UN pro- projection to be used for the assessment and planning jections, or alternatively take the projections that result of global food security include both the widespread from the independently defined education-specific fer- use and acceptance by major international agencies tility trends. Since for the coming 40 years (time as well as substantive dimensions, such as the detail horizon 2050), the differences between these two scen- of relevant information, its scientific basis and the arios are minimal, in the following we will illustrate plausibility of the assumptions made. only the ‘UN Scenario’. In terms of the use of international population Only for China do we recommend a deviation from projections, there is no doubt that the UN population the UN medium variant because China has such sig- projections dominate the field. This is mostly owing to nificant weight when studying the world population their long and well-established tradition and the easy and there are convincing arguments that fertility in availability of country-specific data for both estimates China is currently significantly lower than given by Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2787 the UN and is likely to stay so for the coming decades. Table 1. Projections of total population size for continents as well as selected countries and regions (UN Scenario of The 2008 UN assessment gives a total fertility of 1.77 IIASA education projections). for China for the periods 2000 – 2005 and 2005 – 2010. This is subsequently assumed to increase to 1.79 in area 2000 2010 2020 2030 2040 2050 2010 – 2015, 1.84 in 2015 – 2020 and stay constant at 1.85 thereafter. But the level of fertility in China world 6124 6885 7617 8233 8699 9021 around the 2000 census and thereafter has become a Africa 821 1032 1271 1518 1765 1998 topic of intense scientific analysis and discussion. Asia 3705 4145 4546 4846 5024 5095 While the National Statistical Agency published the Europe 729 730 722 707 687 664 figure of 1.22 for the census year, most scholars Latin America and 523 594 660 713 750 769 agree that this number reflects an undercount with Caribbean the only question being how much of an undercount. North America 316 349 379 405 427 445 In an effort to produce probabilistic population projec- Oceania 31 35 39 43 46 49 tions for China that also assume uncertain starting Brazil 174 199 220 236 248 254 conditions, Lutz et al. (2007b) reviewed more than China 1270 1330 1371 1374 1324 1238 20 different estimates based partly on different India 1046 1220 1379 1506 1597 1658 methods and different data sources. They concluded UK 59 62 64 66 68 69 that the best guess for 2000 was a level around 1.5 European Union 482 495 498 496 489 479 with significant uncertainty bounds. Most recently, a Former Soviet 289 284 279 271 261 249 new study by Morgan et al. (2009) convincingly Union demonstrates that total fertility is currently slightly NW Europe 246 253 258 262 262 261 below 1.5 and is expected to remain there at least for Nile catchment 225 285 354 424 492 555 the coming two decades. Based on this strong scientific sub-Saharan Africa 680 867 1081 1308 1540 1761 reasoning, we suggest that the best-guess projections used for China should be based on a constant total fertility assumption of 1.5. This has been implemented Of the countries that are individually listed, China— in the output tables and graphs given below. currently the most populous country in the world with 1.3 billion inhabitants—will continue to grow until around 2030 owing to population momentum (i.e. 6. DESCRIPTION OF RESULTS more young women entering the reproductive ages) The following section presents and discusses selected even though fertility is assumed to be well below the results of the chosen UN Scenario of the IIASA replacement level. By 2050, China’s population size is global projections by level of education by using expected to be lower than it is today and 420 million the regional definitions given as a standard for all the lower than India’s, which is likely to surpass China as driver reviews. These projections are available at the the most populous country shortly before 2020. level of individual countries on the website of While the picture of future population growth is IIASA’s World Population Programme (www.iiasa.ac. quite differentiated, with some countries and regions at/Research/POP). In addition to six continents they expected to grow substantially, whereas others are list the data for four key countries and five regions of expected to shrink in terms of future population special interest. This scenario was calculated on the ageing, all countries and regions are moving in the basis of the 2006 UN assessment before the 2008 same direction. Currently about 8 per cent of the assessment came out, and the UN made minor adjust- total world population is above the age of 65. This pro- ments between the 2006 and the 2008 revisions. portion is likely to double over the coming 20 years Therefore, small discrepancies might appear between and by 2040 reach the level of 16 per cent, which is the data listed here and those currently available the level currently experienced in Europe. Asia is the from the website of the UN Population Division. most rapidly ageing continent where the current pro- Table 1 gives the results in terms of total population portion above age 65 is likely to increase by a factor size. It shows the total world population increasing from of three from currently 7 to 21 per cent in 2050. an estimated 6.885 billion in 2010 to 7.6 billion in China will rapidly catch up with Europe and reach 2020, 8.2 billion in 2030 and around 9 billion by some 27 per cent above the age of 65 by the middle 2050. These numbers clearly indicate the projected of the century, although currently its proportion decelerating speed of world population growth. While elderly is only half of the European one. Even in the decadal increase in world population is estimated Africa where the population structures are still very by the UN to be 760 million between 2000 and 2010, young (only 3% of the population are above age 65), it is projected to decline to 616 million for the decade the projected increase in life-expectancy together 2020 – 2030 and 322 million for 2040 – 2050. The dis- with declines in fertility will result in significant tribution of this growth over continents shows that the ageing in the longer run (table 2). population of Africa is still expected to roughly When interpreting these numbers on projected double, whereas that for Europe is already on a declin- proportions elderly, we also need to consider that ing trajectory. It is worthwhile noting that the disability-free life-expectancy so far tends to devastating AIDS pandemic, which lowered life-expect- increase at roughly the same speed as total life- ancy in the worst-hit countries and also had a minor expectancy and that the 65-year-olds of the future depressing effect on population growth, does not can be expected to be in better health conditions really influence this big picture of population growth. than the 65-year-olds today. It has recently been Phil. Trans. R. Soc. B (2010) 2788 W. Lutz & K. C. Samir Review. Global population projections Table 2. Projections of the proportions of the population above age 65 for continents as well as selected countries and regions (UN Scenario of IIASA education projections). area 2000 (%) 2010 (%) 2020 (%) 2030 (%) 2040 (%) 2050 (%) world 7 8 10 13 16 19 Africa 3 3 4 5 5 7 Asia 6 7 9 13 17 21 Europe 15 16 19 23 25 28 Latin America and Caribbean 6 7 9 12 15 19 North America 12 13 16 20 21 21 Oceania 10 11 14 16 18 19 Brazil 5 7 9 13 16 19 China 7 8 12 17 24 27 India 5 5 7 9 11 14 UK 16 17 19 22 24 24 European Union 16 17 20 24 27 29 Former Soviet Union 11 11 13 16 18 21 Nile catchment 3 4 4 5 6 8 NW Europe 16 18 21 24 26 26 sub-Saharan Africa 3 3 3 4 5 6 Table 3. Projections of the proportions of the population (above age 15) that have junior secondary or higher education for continents as well as selected countries and regions (UN Scenario of IIASA education projections). proportion with at least secondary education sex area 2000 (%) 2010 (%) 2020 (%) 2030 (%) 2040 (%) 2050 (%) female world 53 59 65 71 77 82 Africa 26 35 43 52 60 67 Asia 45 54 62 69 76 82 Europe 85 89 92 94 95 96 Latin America and Caribbean 53 62 70 78 84 89 North America 95 94 94 94 94 95 Oceania 96 99 100 100 100 100 Brazil 52 62 71 80 86 91 China 56 66 75 82 89 93 India 28 38 48 58 67 75 UK 73 81 87 90 92 94 European Union 80 85 90 93 94 95 Former Soviet Union 96 98 99 99 99 99 Nile catchment 25 34 42 52 60 67 NW Europe 83 87 91 93 94 95 sub-Saharan Africa 21 29 38 47 55 63 male world 62 67 72 76 79 83 Africa 38 45 52 58 63 68 Asia 59 66 72 76 81 84 Europe 86 89 92 93 94 96 Latin America and Caribbean 52 60 67 73 79 83 North America 94 94 94 94 95 95 Oceania 96 98 100 100 100 100 Brazil 48 56 64 72 78 83 China 71 78 84 87 91 94 India 47 55 63 70 75 80 UK 73 81 86 89 91 92 European Union 82 86 89 92 93 94 Former Soviet Union 96 98 98 99 99 99 Nile catchment 37 43 50 56 61 66 NW Europe 84 87 90 93 94 95 sub-Saharan Africa 33 39 47 54 59 65 Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2789 (a)( 1800 b) 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 year year (c) 1800 tertiary 20+ secondary 15+ primary 15+ no edu15+ 0–14 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 year Figure 2. (a) China, (b) India and (c) sub-Saharan Africa: projected trends in the total population by level of highest educational attainment (children below age 15 in grey at the bottom). demonstrated (Lutz et al. 2008b) that when age is not worse-educated populations in Africa which hence defined as the time since birth but alternatively as the might be viewed as a rather optimistic scenario. expected time to death, then the coming speed of Table 3 also shows that while today almost universally population ageing (i.e. people moving closer to their adult men are better educated than adult women, this death) will be much more moderate. The future is likely to change in the future because of the fact that elderly are also likely to be better educated and more female school enrolment rates in most countries are likely to continue to be gainfully employed depending approaching those of men and in many countries on the incentive structures that will be in place. The even surpassing them. greatest challenge associated with population ageing Figure 2 shows the projected trends in the absolute is probably in the poorest countries where often no numbers of the population by four educational attain- old-age support systems exist aside from one’s own ment categories for India and China as well as the family. This also needs to be considered in the context region of sub-Saharan Africa. The population below of studying future rural populations and the agricultural the age of 15 is indicated as a separate group at the work force. bottom of the graph. While for China the picture As discussed above, almost universally more educated shows a peaking in the size of the population followed people are in better health and are more productive. by a decline over the coming decades, the number of Recent studies have shown that there are some people with secondary or tertiary education will con- thresholds both with respect to health and to economic tinue to increase. The trend in sub-Saharan Africa growth in the sense that universal primary education shows a very different picture characterized by contin- (one of the key Millennium Development Goals) is not ued rapid population growth. India is in an sufficient but that it requires high proportions of the intermediate position with decelerating population population with at least completed junior secondary growth associated with a rapid expansion of the more education (to age 15) to help bring countries out of educated segments of population. But even in Africa the vicious circle of poverty, high population growth this projection of populations by level of education andfoodinsecurity(Lutz et al. 2008a). For this reason gives rise to more optimism for the future than the table 3 focuses on the proportion of the population usual focus on population size alone, because it with junior secondary or higher education. shows that the most rapidly growing segment of the The trends shown in table 3 are based on the Global population is that with secondary or tertiary education Education Trend Scenario (KC et al. 2010), which under this admittedly rather optimistic scenario. assumes that in terms of the proportions of cohorts The comparison between India and China in ending up in the different educational attainment figure 2 is particularly interesting since they are the categories, the countries later in the process follow two population billionaires frequently mentioned the trend of the more advanced countries. Since the together as the two great economic powers of the 1970s, this trend has been dominated by the speed future. But the figures illustrate quite clearly that in of educational expansion of many Asian countries. terms of the human capital of their populations, the Similar expansions are also assumed for the currently two countries are very different. Over the past decades, Phil. Trans. R. Soc. B (2010) population in millions population in millions 2790 W. Lutz & K. C. Samir Review. Global population projections China has heavily invested in universal primary and by level of educational attainment, age and sex for 120 countries for 2005 – 2050. Demograph. Res. 22, 383 – near universal secondary education. Although still 472. See http://www.demographic-research.org/Volumes/ less than half of the total population today has second- Vol22/15/. (doi: 10.4054/DemRes.2010.22.15) ary or higher education, this is certain to change as the Keyfitz, N. 1985 Applied mathematical demography, 2nd edn. better educated younger cohorts move up to higher age New York, NY: Springer. groups. In contrast, India suffers from the fact that Khan, H. T. A. & Lutz, W. 2008 How well did past UN currently still about half of all adult women have population projections anticipate demographic trends in never been to school. This is also the main reason six South-east Asian countries? Asian Popul. Stud. 4, why fertility in India is still rather high and as a conse- 77 – 95. (doi:10.1080/17441730801966964) quence the population will experience significant Lutz, W. 2009a Editorial: towards a world of 2 – 6 billion growth over the coming decades. Recently, school well-education and therefore healthy and wealthy people. J. R. Stat. Soc. A 172, 701 – 705. (doi:10.1111/j. enrolment rates in India have increased at all levels 1467-985X.2009.00612.x) but it will take many decades until India will be able Lutz, W. 2009b Sola schola et sanitate: human capital as the to match the level of schooling of the average Chinese. root cause and priority for international development? This will have implications for food security, health, Phil. Trans. R. Soc. B 364, 3031 – 3047. (doi:10.1098/ economic growth and adaptive capacity to climate rstb.2009.0156) change. Lutz, W. & Goldstein, J. (Guest editors) 2004 Special issue In conclusion, this review has attempted to high- on ‘how to deal with uncertainty in population forecast- light some recent developments in the methodology ing?’. Int. Stat. Rev. 72, 1 – 106, 157 – 208. and the content of global population projections and Lutz, W. & Scherbov, S. 2008 Exploratory extension of in particular place emphasis on the different dimen- IIASA’s world population projections: scenarios to 2300. Interim Report IR-08-022. International Institute sions of population change that should be explicitly for Applied Systems Analysis, Laxenburg, Austria. addressed in population projections. The educational Lutz, W., Sanderson, W. & Scherbov, S. 1997 Doubling of attainment distribution has been singled out as a world population unlikely. Nature 387, 803 – 805. key dimension, which perhaps should be routinely (doi:10.1038/42935) added to age and sex in our studies of the trends Lutz, W., Goujon, A. & Doblhammer-Reiter, G. 1999 and consequences of human population size and Demographic dimensions in forecasting: adding edu- structures. cation to age and sex. In Frontiers of population forecasting. Supplement to vol. 24, 1998 Population and This paper was prepared for the Driver Review DR1 of the Development Review (eds W. Lutz, J. W. Vaupel & D. Foresight Project on Global Food and Farming Futures, A. Ahlburg), pp. 42 – 58. New York, NY: The Population funded by the UK Government Office for Science. It is Council. also based on research carried out in IIASA’s World Lutz, W., Sanderson, W. & Scherbov, S. 2001 The end of Population Programme, which is partially funded by the European Research Council (ERC) Advanced Investigator world population growth. Nature 412, 543 – 545. Grant focusing on ‘Forecasting Societies’ Adaptive (doi:10.1038/35087589) Capacities to Climate Change’ (ERC-2008-AdG 230195- Lutz, W., Sanderson, W. C. & Scherbov, S. (eds) 2004a The FutureSoc). end of world population growth in the 21st century: new chal- lenges for human capital formation and sustainable development. London, UK: Earthscan. REFERENCES Lutz, W., Scherbov, S., Makinwa-Adebusoye, P. K. & Reniers, G. 2004b Population-environment-development- Abbasi-Shavazi, M. J., Lutz, W., Hosseini-Chavoshi, M. & Samir, K. C. 2008 Education and the world’s most agriculture interactions in Africa: a case study on Ethiopia. In The end of world population growth in the rapid fertility decline in Iran. Interim Report IR-08- 010. International Institute for Applied Systems Analysis, 21st century: new challenges for human capital formation and sustainable development (eds W. Lutz, Laxenburg, Austria. Bongaarts, J. 2009 Human population growth and the W. C. Sanderson & S. Scherbov), pp. 187 – 225. London, UK: Earthscan. demographic transition. Phil. Trans. R. Soc. B 364, 2985 – 2990. (doi:10.1098/rstb.2009.0137) Lutz, W., Goujon, A., Samir, K. C. & Sanderson, W. 2007a Reconstruction of population by age, sex and level of edu- Cannan, E. 1895 The probability of a cessation of the growth of population in England and Wales during the next cational attainment of 120 countries for 1970 – 2000. Vienna Yearbook Popul. Res. 5, 193 – 235. (doi:10.1553/ century. Econ. J. 5, 505 – 515. (doi:10.2307/2956626) Hajnal, J. 1955 Prospects for population forecasts. J. Am. populationyearbook2007s193) Lutz, W., Scherbov, S., Cao, G. Y., Ren, Q. & Zheng, X. Y. Stat. Assoc. 50, 309 – 322. (doi:10.2307/2280963) Hayami, Y. & Ruttan, V. 1971 Agricultural development: an 2007b China’s uncertain demographic present and future. Vienna Yearbook Popul. Res. 5, 37 – 59. (doi:10. international perspective. Baltimore, MD: Johns Hopkins University Press. 1553/populationyearbook2007s37) Lutz, W., Crespo Cuaresma, J. & Sanderson, W. 2008a The Jennings, V. E., Lloyd-Smith, C. W. & Ironmonger, D. S. 2004 Global projections of household numbers using demography of educational attainment and economic growth. Science 319, 1047 – 1048. (doi:10.1126/science. age determined ratios. Working Papers Series, No. 914. Department of Economics, University of Melbourne, 1151753) Lutz, W., Sanderson, W. & Scherbov, S. 2008b The coming Melbourne, Australia. Jiang, L. & O’Neill, B. C. 2009 Household projections for acceleration of global population ageing. Nature 451, 716 – 719. (doi:10.1038/nature06516) rural and urban areas of major regions of the world. Interim Report IR-09-026. International Institute for Morgan, P., Zhigang, G. & Hayford, S. 2009 China’s below- replacement fertility: recent trends and future prospects. Applied Systems Analysis, Laxenburg, Austria. Samir, K. C., Barakat, B., Goujon, A., Skirbekk, V., Popul. Dev. Rev. 35, 605 – 630. (doi:10.1111/j.1728- 4457.2009.00298.x) Sanderson, W. & Lutz, W. 2010 Projection of populations Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2791 Nakicenovic, N. et al. 2000 Emissions scenarios.In A special United Nations 2004 World population to 2300. New York, report of working group III of the Intergovernmental Panel NY: United Nations, Population Division. on Climate Change. Cambridge, UK: Cambridge Univer- United Nations. 2008 World urbanization prospects: the 2007 sity Press. revision. New York, NY: Department of Economic and Notestein, F. W. 1945 Population: the long view. In Food for Social Affairs, Population Division, United Nations. the world (ed. T. W. Schultz), pp. 36 – 69. Chicago, IL: United Nations. 2009 World population prospects: the 2008 University of Chicago Press. revision. New York, NY: Department of Economic and O’Neill, B. C., Balk, D., Brickman, M. & Ezra, M. 2001 A Social Affairs, Population Division, United Nations. guide to global population projections. Demogr. Res. 4, van Imhoff, E. & Keilman, N. 1991 LIPRO 2.0: an appli- 203 – 288. cation of a dynamic demographic projection model to Rogers, A. & Land, K. (eds) 1982 Multidimensional math- household structure in the Netherlands. Amsterdam/Lisse, ematical demography. London, UK: Academic Press. The Netherlands: Swets and Zeitlinger, NIDI/CBGS Sanderson, W. & Scherbov, S. 2008 Rethinking age and Publications No. 23. aging. Popul. Bull. 63, 3 – 16. Vaupel, J. W. & Yashin, A. I. 1985 Heterogeneity’s ruses: Skirbekk, V. 2008 Age and productivity capacity: descrip- some surprising effects of selection on population tions, causes and policy options. In Ageing horizons, dynamics. Am. Stat. 39, 176 – 185. (doi:10.2307/ vol. 8, pp. 4 – 12. Oxford, UK: Oxford Institute of Ageing. 2683925) Phil. Trans. R. Soc. B (2010) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Philosophical Transactions of the Royal Society B: Biological Sciences Pubmed Central

Dimensions of global population projections: what do we know about future population trends and structures?

Philosophical Transactions of the Royal Society B: Biological Sciences , Volume 365 (1554) – Sep 27, 2010

Loading next page...
 
/lp/pubmed-central/dimensions-of-global-population-projections-what-do-we-know-about-BgA7vB9CEl

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Pubmed Central
Copyright
© 2010 The Royal Society
ISSN
0962-8436
eISSN
1471-2970
DOI
10.1098/rstb.2010.0133
Publisher site
See Article on Publisher Site

Abstract

Phil. Trans. R. Soc. B (2010) 365, 2779–2791 doi:10.1098/rstb.2010.0133 Review Dimensions of global population projections: what do we know about future population trends and structures? 1,2,3, 1 Wolfgang Lutz and Samir KC International Institute for Applied Systems Analysis (IIASA), Schlossplatz 1, A-2361 Laxenburg, Austria Austrian Academy of Sciences, Dr Ignaz Seipel-Platz 2, 1010 Vienna, Austria Vienna-University of Economics and Business 2–6 Augasse, 1090 Vienna, Austria The total size of the world population is likely to increase from its current 7 billion to 8 – 10 billion by 2050. This uncertainty is because of unknown future fertility and mortality trends in different parts of the world. But the young age structure of the population and the fact that in much of Africa and Western Asia, fertility is still very high makes an increase by at least one more billion almost certain. Virtually, all the increase will happen in the developing world. For the second half of the century, population stabilization and the onset of a decline are likely. In addition to the future size of the population, its distribution by age, sex, level of educational attainment and place of residence are of specific importance for studying future food security. The paper provides a detailed discussion of different relevant dimensions in population projections and an evaluation of the methods and assumptions used in current global population projections and in particular those produced by the United Nations and by IIASA. Keywords: world population; population increase; population decline; dimensions of population projections; age structure; level of educational attainment 1. INTRODUCTION (called assessments and published in their World While future trends in the number and composition of population prospects series) have been made every humans on this planet has been a topic of scientific 2 years. The most recent UN assessments have a enquiry and discussion for centuries and at least since time horizon to 2050. Bongaarts (2009) gives a Thomas Malthus entered the field of structured quanti- concise summary of the main results and implications tative analysis, the first modern global population of the UN projections. At irregular intervals the UN projection, which explicitly considered the age and sex also publishes long-term population projections with structure of the population (the so-called cohort- time horizons from 2150 to 2300. component method), was carried out by Frank The World Bank started to produce independent Notestein of the Princeton Office of Population population projections in 1978. These were always Research in 1945 (Notestein 1945). At the national meant primarily for internal use in the Bank’s develop- level, several population projections precede this first ment planning and were published as part of the comprehensive global projection. Hajnal (1955) provides series of World development reports. After 1984, the a good overview of these early population projections. World Bank projections were revised approximately Frank Notestein subsequently became the first every two years and in most cases only one variant director of the then newly established United Nations was published but with a long time horizon to 2150. Population Division. This unit began producing regu- Around 1995, the World Bank stopped publishing lar global population projections in the early 1950s. separate projections but presumably continued to use Between 1951 and 2008, the UN published 21 sets them for internal purposes for a number of years. of estimates (past and current conditions) and projec- The Washington-based Population Reference Bureau tions (future) for all countries and territories of the (PRB) publishes independent world population pro- world. Before 1978 these projections were revised jections (population size only and a single scenario) approximately every 5 years; since then new revisions every year as part of its annual World population data sheet. Since 2000 it has published projected population sizes for all countries and territories of the world * Author for correspondence (lutz@iiasa.ac.at). for 2025 and 2050. In some cases the projections are identical to those of the UN and the US Census While the Government Office for Science commissioned this review, Bureau (USCB), but in some cases different the views are those of the author(s), are independent of Government, and do not constitute Government policy. country-specific information is used. The USCB produces single scenario projections for One contribution of 23 to a Theme Issue ‘Food security: feeding the world in 2050’. all countries in the world as of 1985 with a varying 2779 This journal is q 2010 The Royal Society 2780 W. Lutz & K. C. Samir Review. Global population projections time horizon. The World Population Programme of that has recently received the greatest public attention, the International Institute for Applied Systems Analy- namely, the questions if and when we will see an end of sis (IIASA) based outside Vienna (Austria) began population growth followed by a possible population producing global population projections at the level decline. In the final section of the paper we propose of 13 world regions in 1994. One of the purposes a specific existing population projection (the UN was to produce population projections as part of scenario of the IIASA Education Projections) for use the Special Report on Emissions Scenarios (SRES) in the Foresight Project. We present the results for (Nakicenovic et al. 2000) that underlie the global emis- the specified regions and countries in tabular and sion scenarios used by the Intergovernmental Panel for graphical form. Climate Change (IPCC). This was followed by three rounds of probabilistic projections at the level of 13 2. DIMENSIONS CONSIDERED IN POPULATION regions (which were all published in the pages of PROJECTIONS Nature: Lutz et al. 1997, 2001, 2008b). The IIASA For many users the most important piece of infor- projections come from a distinctly scientific back- mation expected from population projections is the ground, which is illustrated by the fact that in its future total size of the population; for many policy publication (e.g. Lutz et al. 2004a) much more space and research questions other dimensions of the com- is used for justifying and discussing methods and position of the population are also of interest. The assumptions than for presenting results. Recently, dimensions that will be explicitly discussed here are IIASA developed (in collaboration with Eurostat and age, sex, rural/urban place of residence, educational the UK Office of National Statistics) a new approach attainment, labour force participation, parity status, for evaluating substantive arguments about alternative household status and health status. All these dimen- assumptions on possible future trends based on large sions have in common that they are properties of numbers of questionnaires ascertaining the evaluation each individual member of the population which are, of alternative arguments by experts. This is currently at least in theory, observable and measurable. They being translated (together with Oxford University) can hence be called observable sources of population into a new set of science-based population projections heterogeneity. These dimensions are not only interest- for all countries in the world by age, sex and level of ing in their own right for helping to answer specific education for publication in 2012. questions relating to their future distributions in the O’Neill et al. (2001) published a comprehensive population but to the extent that they are associated 85-page review entitled A guide to global population with different fertility, mortality and migration intensi- projections. In this article, they describe in considerable ties, their changing distributions in the total detail the projections of all five agencies discussed population also impact on the projected future popu- above. The paper includes a critical review of the lation size. In the following, we will discuss each of methods used and of the ways that assumptions are these dimensions individually, assess to what degree defined by those five agencies. In addition, the results they are likely to impact on the course of overall popu- are compared in quite some detail. For this reason, we lation, discuss how they may be important in their own do not want to repeat the exercise of systematically right in the context of studying food security, and comparing the approaches and results of the different review existing global level projections, which explicitly projections, but rather focus on a few key challenges address this dimension. that point to the future. It is important to note that in addition to these In the next section we will discuss the important observable sources of population heterogeneity, there questions of what we call ‘dimensions’ of population is still unobserved heterogeneity in every population projections. The basic idea is that any population is which is hard to capture empirically. Theoretical con- an assembly of individuals and each individual is siderations suggest that such unobserved heterogeneity different from any other individual. But which of the can significantly impact future population dynamics many relevant properties of individuals should we (Vaupel & Yashin 1985), but there is little one can do explicitly consider in population projections? The about it except to be aware of the problem and be lowest dimensionality is chosen for the case that all cautious about the validity of the conclusions drawn. individuals are taken as equal (homogeneous) and Given this problem associated with hidden heterogen- only a total growth rate is assumed for projecting eity, it is even more important to explicitly measure population size. The cohort-component method and incorporate the observable sources of population breaks down by age and sex; this is a first step in the heterogeneity wherever feasible and thus try to mini- direction of increasing dimensionality by explicitly mize the possible biases caused by overall heterogeneity. considering these individual properties. But there are many further properties that deserve consideration in population projections and which will be discussed in (a) Population size the following section. Most users of population projections seem to be pri- We will also address the tedious but very important marily interested in the total size of the population. issue of how to deal with uncertainty in population From a substantive point of view, however, it is diffi- projections. While a single best-guess population pro- cult to think of specific policy or research questions, jection is likely to satisfy the expectations of a where only changes in absolute size matter and majority of users, disregarding the issues of uncer- where it is supposedly irrelevant whether one refers tainty can be rather dangerous and therefore deserves to newborn babies, young adults or frail elderly serious attention. We will then turn to the one issue people. Presumably, the great interest in data on the Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2781 sheer number of people has two main reasons. First, on fixed age categories and does not consider the total population size can be seen as a first-order fact that healthy life-expectancy tends to increase in approximation of the scale of the population-related parallel with total life-expectancy. Hence, a 65-year- issues in particular, if it is assumed that the age pat- old person today is typically quite different from one terns of the two populations to be compared do not of the same age who lived several decades ago and differ greatly. Second, total population size is an who had a much shorter remaining life-expectancy. important denominator of many frequently used This idea has recently been translated into a redefini- indicators ranging from gross domestic product tion of age, where age is not just measured as the (GDP) per capita to food consumption per person to time since birth but can alternatively be viewed as greenhouse gas emissions per person. In all those the expected time to death, in which case the popu- cases a separately derived total quantity is divided by lation dynamics looks quite different (Lutz et al. total population size in order to produce an indicator 2008b; Sanderson & Scherbov 2008). that can be compared across populations. The distinction between men and women as a key Traditionally, population projections have been pro- dimension of population dynamics has been a feature duced by simply taking total population size and of demography almost from the beginning. It has to making assumptions on the future growth rates of do with the fact that fertility rates are almost exclu- the population. This has been the standard method sively measured with women. One of the reasons for until the age-specific, so-called cohort-component this is that not all fathers are known and that the repro- method became the widely used standard after World ductive age range for women is shorter than for men War II. But for some specific applications, where the and shows a clearer and partly biologically determined age structure is difficult to assess or where the popu- age pattern. Also, mortality rates tend to differ signifi- lation is simply too small to have meaningful age cantly between men and women. Except for some groups (as is sometimes the case with small area cases of extreme female discrimination, male mortality population projections), simple growth rate-based rates are typically higher at every single age. In projections are still being used. But for national addition to these demographic reasons there is, population projections, age-group-wise projections of course, significant substantive interest in the age- have clearly become the state-of-the-art because specific proportions of men and women in the they allow differentiation between the behavioural population which matters for issues ranging from the components (fertility, mortality and migration) and marriage market to labour market to consumer embedded changes that are merely owing to age- preferences. structural effects. This is particularly important for The formal model for doing population projections countries that have gone through significant fluctu- by age and sex was introduced by Cannan (1895) ations in fertility and mortality trends and hence but it took until after World War II to have this have irregular age structures, such as most European cohort-component method spread around the world countries after the two world wars. But even for high as the dominant way for doing population projections. fertility countries that traditionally had very stable Today, it is used by all national and international and regular age pyramids, the assumed future fertility statistical agencies. decline will be partly compensated in terms of popu- lation growth through the age-structural momentum, i.e. the fact that in the future larger numbers of poten- (c) Rural–urban place of residence tial mothers will enter the reproductive age and the Place of residence has always been an important number of births will continue to increase even if dimension since the very origin of demography when period fertility is at replacement level. This is another statistical information on households was collected good reason for explicitly considering the age structure for taxation purposes and conscriptions for the army. wherever possible. Associating people with a specific locality, however, requires that people have a more or less permanent residence and are not permanently on the move. (b) Age and sex This is one problem facing statisticians when trying Particularly in the context of population ageing, the to distinguish between urban and rural populations explicit interest in the population’s changing age struc- in countries (such as China), where millions of ture has recently increased significantly. While there people (the so-called floating population) are con- has always been an interest by planners in the likely stantly on the move between their home villages and future number of school-age children or young the urban centres where they work but have no legal adults trying to enter the labour market, the currently right of residence. Another problem with comparing dominating questions relate to the sustainability of reported proportions urban across countries and over pension systems under conditions of rapid increase in time is that there is a multitude of changing national the number of elderly. Closely related questions deal sets of definitions based partly on the size of the with the likely impact of ageing on healthcare cost or agglomeration, its statutory status, its population care for the elderly in general. There is significant con- density or the proportion of the labour force that cern that an increase in the mean age of the labour is working in agriculture. Since urbanization will be force may be associated with decreasing productivity discussed elsewhere, it will not be covered here. in the future (Skirbekk 2008). These are all questions From a population projections point of view, the for which a projection of the population by age is question whether a person lives in a rural or an essential. Much of this discussion, however, is based urban community is not only seen in its own right Phil. Trans. R. Soc. B (2010) 2782 W. Lutz & K. C. Samir Review. Global population projections but is also viewed as a relevant source of population populations that are still in the process of demographic heterogeneity. Almost universally women in urban transition, more educated women have lower fertility. areas have lower fertility than women in rural areas. These educational differentials can be very significant. Also in terms of international migration, new immi- The Demographic and Health Survey for Ethiopia, for grants today generally move to urban areas except for instance, shows that women without any formal edu- some special cases where immigrant labour is concen- cation have on average six children, whereas those trated in agriculture. With respect to mortality, the with secondary education have only two (see http:// pattern is more complex. While the inhabitants of www.measuredhs.com). Significant differentials can cities traditionally had higher mortality owing to a be found in most populations of all cultural traditions. higher risk of infectious disease, this pattern virtually Only in a few modern societies does the strongly nega- disappeared over the course of the twentieth century tive association give way to a U-shaped pattern in because healthcare and education were better in which the most educated women have a somewhat urban areas. But most recently there are indications higher fertility than those with intermediate education. of a reversal, with urban slums in some African cities But globally, the education differentials are so perva- having worse mortality conditions than even remote sive that education may well be called the single villages in the same country. most important observable source of population het- From a methodological point of view, the best erogeneity after age and sex (Lutz et al. 1999). There model to project a population by age and sex for are good reasons to assume that during the course of urban and rural areas separately is the so-called a demographic transition the fact that higher education multi-state model, which was developed at IIASA in leads to lower fertility is a true causal mechanism, the 1980s (Rogers & Land 1982; Keyfitz 1985). It where education facilitates better access to and infor- has become the state-of-the-art for multi-dimensional mation about family planning and most importantly population projections in which separate fertility, mor- leads to a change in attitude in which ‘quantity’ of chil- tality and migration rates are assumed for the different dren is replaced by ‘quality’, i.e. couples want to have sub-populations as well as age- and sex-specific tran- fewer children with better life chances. sition rates between sub-populations. Because of data In terms of measuring education, it is important to limitations, such true multi-state projections by distinguish between stock and flow variables. Most urban – rural place of residence exist only for selected education programmes and studies are concerned groups of countries. with the process of schooling itself, including many The United Nations Population Division publishes important topics that range from the construction of the only urbanization projections on a global level, schools and the training of teachers to the organization with updates every two years. The most recent revision of schools, the quality of teaching and the content of (United Nations 2008) estimates the population living curricula. The extent to which these schooling efforts in urban areas in 229 countries and territories of cover the entire population are usually measured the world for the period 1950 – 2005 and projects through enrolment rates as published for most the same for 2010 – 2050 in five year intervals. The countries by UNESCO and other agencies. But for basic method of estimation and projection is still the many of the social and economic benefits of education same as when it was developed in the 1970s with it does not directly matter how many people are in only a few minor modifications. The main source of school at any given point in time, but rather how information used in the projection is the urban – rural many have completed their education and use their growth difference at the two latest available time acquired skills in the labour force. Hence, the main points (mostly inter-census). The difference is then interest is not on the flow (school enrolment) but assumed to follow a logistic path and is estimated rather on the changing stock (human capital). The based on the past experience of many countries and stock can be measured by categories of highest the expectation that the urbanization process will educational attainment (here the UNESCO-ISCED slow down as the level of urbanization grows. In categories have become the standard) as well as by other words, this is not a multi-state model, but only mean years of schooling. Since the original empirical the application of assumed overall proportions urban data mostly come in terms of attainment and the to national level projections. This model by definition calculation of mean years of schooling requires assumes continuous increases in the proportion urban additional country-specific assumptions, the former and is unable to model possible ups and downs. This is often preferred. Also, considering the full attainment model is isomorphic to the literacy projection model distribution provides information about inequalities that has been used by UNESCO for decades, but that is hidden in the case of only using an average which was recently replaced by a cohort-component such as the mean years of schooling. model. Because of the strong association between female education and fertility, future changes in the compo- sition of the female population by educational (d) Educational attainment attainment make a big difference. Since many develop- Like the other dimensions discussed above, education ing countries have seen major improvements in school is also an important source of population heterogen- enrolment rates of both girls and boys in those eity and bears a significant weight of its own. Almost countries, the future women of reproductive age are universally more educated people have lower mortality bound to be more educated than today’s. This great and there is sufficient evidence that this is a real effect momentum and rather easy predictability of the edu- and not just owing to selectivity. Also for all cational attainment distribution is a consequence of Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2783 the fact that education is typically acquired early in life distribution of women by the number of children and then remains virtually constant along cohort lines. they have already given birth to, is an important deter- If we know how many 10-year-old girls with some pri- minant of fertility in the near-term future. Highly mary education there are today, we know (when important for economic and labour force consider- considering differential mortality and migration) how ation is the distribution of the population by age, sex many 50-year-olds with at least primary education and labour force participation rates. While there are there will be in 40 years’ time. Given the fact that in several national projections of future labour force par- most countries the younger cohorts are better edu- ticipation, consistent global projections do not yet cated than the older ones, further declines in fertility exist. Another very relevant dimension in the context as well as mortality are virtually pre-programmed. In of global population ageing and the associated needs a few African countries where school enrolment rates for care for the elderly is the projection of the popu- have actually declined over the past two decades, lation by health status. While significant progress has this has led to a stalled fertility decline and partly been made in producing internationally comparable worsening mortality conditions. indicators of disability and its impact on activities of Aside from its effects on population dynamics, there daily life, up to now this has not yet led to the pro- is also significant interest in knowing the educational duction of global projections which cross-classify structure of the population for a broad range of health status with age and sex. Alternatively, one social and economic development concerns. Based could try to classify people according to certain disease on a newly reconstructed set of educational attainment risks or infection statuses (such as HIV-positive) if the distributions by age and sex (Lutz et al. 2007a) for data are available, and a certain disease is of specific most countries back to 1970, it has recently been interest. Finally, an important field of analysis and pro- shown that indeed the improvement of educational jections deals with the living arrangements/household attainment in the working age population has been composition of people. Here, some progress in the the most consistent and significant driver of economic direction of global projections has recently been growth around the world (Lutz et al. 2008a). But made in the context of trying to project future green- improving education by age and sex has also been house gas emissions and the importance of shown to matter for countries transitioning to democ- demographic/household factors in climate change racy and more liberal rights (Abbasi-Shavazi et al. mitigation (van Imhoff & Keilman 1991; Jennings 2008; Lutz 2009b). It has already been mentioned et al. 2004; Jiang & O’Neill 2009). that education is perhaps the single most important health determinant. For the question of food security, it has long been shown that the basic education 3. TREATMENT OF UNCERTAINTY of the agricultural labour force is a key factor in In the field of social and economic trends—as in all agricultural production (Hayami & Ruttan 1971). fields that deal with human behaviour—the future As the set of population – education – development – can never be projected with certainty. In the context agriculture (PEDA) models commissioned by the of the cohort-component model of population projec- UN Economic Commission for Africa for a number tion that differentiates by age and sex, all three of African countries shows, when including education components of population change (i.e. future fertility, in an agricultural production function, it turns out mortality and migration trends) are uncertain. For to be one of the key determinants in reducing each of these three components of change, the levels malnutrition and food insecurity (Lutz et al. 2004b). as well as the specific age patterns (by sex) are uncer- IIASA has recently produced population projec- tain, but the level is considered to be of primary tions by age, sex and four levels of educational importance. attainment for 120 countries following different scen- How to deal with uncertainty in population forecasting ario assumptions with respect to alternative future was the topic of a recent special issue of the Inter- education trends and different education-specific ferti- national statistical review (Lutz & Goldstein 2004). It lity trends (KC et al. 2010). These projections are summarizes the state-of-the-art in a field of research based on a true multi-state model, i.e. individuals are that has recently gained more interest both among assumed to be subjected to different mortality and demographers as well as users. Based on this analysis, fertility levels depending on their educational attain- we will distinguish between four different ways that are ment level. Hence, even identical trends in commonly used to deal with the difficult issue of education-specific fertility can lead to different overall uncertainty: (i) ignore uncertainty and publish only fertility levels in the case of differing education one projection; (ii) define alternative probability-free trends. We will return to this issue when discussing scenarios; (iii) publish high, medium and low variants the recommended combination of these education that are supposed to cover a ‘plausible range’; and (iv) projections with the UN population projections by produce fully probabilistic projections that give age and sex. quantitative information about the range of uncertainty. For many users it is sufficient to have only one ‘best-guess’ forecast. This is the projection which is (e) Other relevant dimensions seen as the most likely population trajectory from There are several other dimensions of population that today’s perspective, based on the best knowledge avail- are considered relevant either in their own right or as able today. Such a projection can give some orientation further important sources of heterogeneity. The about the direction into which things are moving. It is parity distribution of the female population, i.e. the sufficient for many applications with a relatively short Phil. Trans. R. Soc. B (2010) 2784 W. Lutz & K. C. Samir Review. Global population projections time horizon and where the costs associated with a growth, this can be viewed as an acceptable simplifica- deviation from the projected mean are negligible. But tion when interested primarily in total population size. for many planning purposes these costs can be rather When the interest is in population ageing, however, or significant. For example, if there is a plan to build a age-specific indicators such as the future proportion of new primary school and the projection should tell the population above age 80, this can result in a major the local authorities how many 5 – 10-year-olds there underestimation of uncertainty. A final problem with are likely to be in the next 10 – 20 years in a certain this approach arises when aggregating national projec- school district, such projections are associated with tions to regional or global ones. The global high significant uncertainties and any deviations from the variant of the UN projections is calculated as the best-guess projection can turn out to be rather costly. sum of all national level high variants, hence assuming This is even more so for projections of the number that there is a perfect correlation among all national of people with pension entitlements in a given country. trends. Each country simultaneously experiences the In terms of global level projections, those presented by highest plausible fertility trajectory. Such perfect cor- the USCB, the World Bank and the Population Refer- relation is highly unlikely, however, and a higher ence Bureau provide only single projections for all than expected trajectory in one part of the world countries in the world. may be partly offset by diverging trends in other The definition of ‘alternative scenarios’ is a fre- parts of the world. Hence, what is considered as a quently chosen approach that goes beyond a single plausible range at the national level is not necessarily projection and presents the user with several possible plausible at the global level. future trajectories. The terminology is somewhat Only ‘fully probabilistic’ projections can avoid such inconsistent and variants (see next paragraph) are problems. These are based on pre-defined uncertainty often called scenarios. Generally, scenarios are defined distributions over time for each of the three com- as consistent sets of assumptions. The notion orig- ponents which are stochastically combined in large inally came from the world of theatre, where scenes numbers (typically several thousands) of cohort-com- show consistent pictures of our possible setting. In ponent projections with the individual random draws demography, this is typically done by combining being subject to assumed autocorrelations and, in the different assumptions about future trends in fertility, case of multi-regional projections, correlations mortality and migration in a way that should ‘not be among the regions. In most projections, the com- impossible’. Hence, scenarios are generally considered ponents are assumed to be uncorrelated between free of probabilities with the only criterion being each other. As the field of probabilistic population pro- internal consistency. The user is given a broad range jections has grown, there have been essentially three of what could be theoretically possible without being traditions in defining the uncertainty distributions: told what is likely to happen and what probability one based on time series analysis (which is only appli- range is covered by the interval between the highest cable to countries that have long empirical series of and lowest scenario considered. At the global level, data and where no structural discontinuities are the most frequently used population scenarios are expected); one based on the analysis of the errors of those produced in the context of the SRES scenarios old projections (ex post error analysis combined with of the IPCC (Nakicenovic et al. 2000), which in the assumption that future errors will essentially be turn are based on combinations of projections pro- the same as past projection errors); and one based duced by IIASA and the United Nations Population on the evaluation of expert arguments resulting in Division. Eurostat also presents its population sets of subjective probability distributions. While projections in the form of different scenarios and an more experts as well as national and international increasing number of European national statistical agencies have entered this field of probabilistic projec- agencies follow this practice. tions, some convergence among these different The presentation of ‘high, medium and low var- approaches seems to be ongoing. It is also interesting iants’ goes a step further in providing the user with that the initiative for doing such projections does not some range of uncertainty that can be interpreted always come from the producers but rather from the intuitively. For many decades, this approach has users. For example, in the case of the UK Office of been the common practice of the UN population pro- National Statistics, such projections were recently pro- jections as well as a large majority of national statistical duced upon request by the financial planning offices. The variants are mostly generated by assuming authorities. At the global level, IIASA (Lutz et al. three alternative fertility trajectories, which are then 1997, 2001, 2008b) has produced the only probabilis- combined with identical mortality and migration tra- tic population projections at the level of 13 world jectories. They are typically presented to the user as regions. providing a ‘plausible range’ of future population From a theoretical perspective, fully probabilistic trends, hence providing a first approach to uncertainty projections are superior in avoiding the problems of although the user is not specifically told what is meant other approaches as discussed above. They provide by ‘plausible’. Does it mean 100 per cent of all poss- the users with the most comprehensive and detailed ible cases, 80 per cent or only 50 per cent? A further information in terms of the demographic ‘risks’ that shortcoming of this approach is that it only covers can be combined with a cost function. But this the uncertainty associated with future fertility trends approach is criticized by some as conveying a false and does not include uncertainties associated with sense of precision. It is indeed true that this approach future mortality or migration. Since fertility is the requires more detailed explicit assumptions in terms of most important determinant of long-term population the full distributions and correlations than the other Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2785 high end: TFR = 2.5 TFR = 2.5, LEMAX = 120 TFR = 2.0, LEMAX = 120 6 6 TFR = 1.7, LEMAX = 120 TFR = 1.5, LEMAX = 120 2 TFR = 1.0, LEMAX = 120 low end: TFR = 1.0 year Figure 1. Total world population in billions: probabilistic projections until 2100. Yellow, 95% interval; green, 60%; blue, 20% and extensions to 2200. The scenarios shown combine different levels of total fertility rate as indicated with the assumption that life-expectancy continues to increase up to a maximum level of 120 years. Source: Lutz & Scherbov (2008). common approaches. Yet, the proponents of probabil- conditions in addition to the uncertain future paths istic projections argue that they only try to make for the three components. This is even relevant at otherwise implicit assumptions explicit and extensive the level of world population projections because of sensitivity analyses show that the overall results tend the great uncertainty of current fertility levels in to be very robust to changes in the specific shapes of the world’s most populous country, China, as will be the assumed uncertainty distributions. But at this discussed in more detail below. stage it is clear that a universally accepted approach to quantitatively describing the uncertainty of popu- lation projections has not yet been developed and 4. THE PROSPECT OF WORLD POPULATION more research in this field is needed. STABILIZATION OR DECLINE AFTER PEAK A final source of uncertainty that often tends to be There has been much discussion recently about the disregarded in population projections is the uncer- long-term outlook for world population and in par- tainty about current conditions, i.e. the population ticular whether the global population is likely to size and structure as well as the levels of fertility and stabilize or even decline in size after a peak during mortality in a given country in the starting year of the second half of the twenty-first century. This discus- the projections. While this tends to be well documen- sion was, in part, triggered by an article published in ted in industrialized countries, there are major gaps of Nature in 2001 entitled The end of world population information, in particular, in Africa and parts of Asia. growth (Lutz et al. 2001), which indicated that there For many countries the information is only based on was an 80 – 90% chance that the world population sample surveys and for some countries even such would reach a peak before 2100 and start to decline information is not available. While the UN Population thereafter. Whether or not such a peak will actually Division makes heroic efforts to come up with popu- occur and at what level and how steep the following lation estimates for all countries in the world, the decline will be is primarily a function of the assump- fact that many numbers referring to the recent past tions made about long-term fertility levels in have to be adjusted in every new assessment round different parts of the world. as new information becomes available shows how dif- When it comes to global population projections ficult this task is. But demographers have become so beyond 2050, there are only the occasionally published used to having a precise point estimate for any demo- UN long-term projections as well as the IIASA world graphic indicator, even in countries with very population projections for comparison. In 2004, the fragmentary empirical information, that they often UN Population Division published a study entitled forget about the uncertainty of those indicators refer- World population to 2300 (United Nations 2004)in ring to the starting conditions when making which different long-term scenarios were presented projections. A recent study on past projections in five for all countries in the world by extending their usual southeast Asian countries (Khan & Lutz 2008) projections to 2050 by another 250 years. While life- showed that the errors introduced by incorrect infor- expectancy was assumed to continue to increase over mation about starting conditions was in some cases the entire period although at a decelerating rate, even higher than the error owing to incorrect assump- the fertility scenarios defined were all very close to tions about the future. Lutz et al. (2007b) expanded replacement level. Of the three long-term fertility the concept of probabilistic population projections assumptions, the low one assumed universal conver- to explicitly include the uncertainty about starting gence to 1.85, the high one to 2.35 and the medium Phil. Trans. R. Soc. B (2010) total population 2200 2786 W. Lutz & K. C. Samir Review. Global population projections one to whatever the replacement level might be (some- and projections (from 1950 to 2050). Also, for all what below 2.1 under good mortality conditions). It is international agencies that are part of the UN family interesting to note that the fertility level of 1.85, which of agencies (including the Food and Agriculture in the projections to 2050 is assumed to be the conver- Organization), there is an institutional agreement to gence level of the medium variant (with the low variant only and consistently use the UN projections in being 1.35) for the long-range projections, is the order to avoid embarrassment arising from using lowest fertility scenario presented. But even these different numbers in different parts of the UN. very small differences in the long-term fertility levels Hence, despite the above described shortcomings of produce significantly different long-term global popu- the UN projections, particularly in the way they deal lation sizes: by 2100, the resulting populations are 5.5 with uncertainty, there is no doubt that virtually all (low), 9.1 (medium) and 14.0 (high) billion, and by groups and agencies dealing with food security and 2200, the differences further increase to 3.2, 8.5 and agriculture expect to see the UN medium variant as 21.2 billion, respectively. Hence, the medium scenario their population projection of choice. For this reason results in some sort of population stabilization in the and with the only exception of China—as discussed very long run, but only because global long-term below—this driver review also recommends using the fertility is assumed to remain constant exactly at the UN medium variant in terms of population size and replacement level of fertility—a level that is defined the age and sex structures that come with it. as the one producing long-term stationarity. Minor But this review can do better than just considering deviations to the lower side will produce significant the population by age and sex. As discussed above, population decline and to the higher side will result education is a key dimension in the study of develop- in substantial long-term increases. ment and food security. IIASA has recently produced Lutz & Scherbov (2008) recently published another population projections for most countries in the set of long-term global projections, which extends the world by age, sex and four levels of educational attain- IIASA probabilistic population projections—which go ment up to 2050. While the baseline scenario in these to 2100—further into the future by defining scenarios projections deviates from the UN medium variant in covering a wider range of possible future fertility levels. assuming different future fertility levels throughout Selected findings are shown in figure 1 and were dis- Europe (following the Eurostat projections that cussed in a recently published editorial in the Journal have been worked out in collaboration with national of the Royal Statistical Society entitled Towards a world statistical offices) as well as in some east Asian of 2–6 billion well-educated and therefore healthy and countries, IIASA has also calculated a so-called ‘UN wealthy people (Lutz 2009a). The extension scenarios Scenario’ for comparative purposes. Since in the pro- start in 2080 (the year until which the assumptions jections by level of education assumptions are for the probabilistic projections were defined); some defined in terms of future education-specific fertility scenarios continue from the level of the median of and mortality trends, which then have to be weighted the projected distribution; others from the upper and by the size of the respective education categories if lower bounds of the projected 90 per cent uncertainty overall fertility and mortality rates are to be calculated, range (figure 1). the changing educational composition over time The figure clearly illustrates what has been dis- induces differences to the UN projections even if edu- cussed above with regard to the UN long-range cation-specific fertility and mortality assumptions are projections. In the long run global fertility levels defined to be as close as possible to the UN assump- below 2.0 will result in population decline and above tions. A full congruence of the two sets of 2.0 in long-term population increase. But the figure projections can only be achieved through an iterative also illustrates that there is a real chance that global procedure in which for every country and every point population could fall below its current size by the in time, the education-specific rates are modified in a middle of the next century, even if global fertility way that their weighted average becomes identical to levels were somewhat higher than what is being the overall rates assumed by the UN (2009). This pro- experienced in Europe today. cedure has been performed for all countries for the ‘UN-Scenario’ and results in a projected age, sex and education structure that is identical to the age 5. PROPOSAL FOR USE OF POPULATION and sex structure of the UN medium variant, but PROJECTIONS IN FOOD SECURITY also gives the education distribution as well. The ASSESSMENTS user can then choose whether to use this scenario, The criteria for recommending a specific population which is perfectly in line with the widely used UN pro- projection to be used for the assessment and planning jections, or alternatively take the projections that result of global food security include both the widespread from the independently defined education-specific fer- use and acceptance by major international agencies tility trends. Since for the coming 40 years (time as well as substantive dimensions, such as the detail horizon 2050), the differences between these two scen- of relevant information, its scientific basis and the arios are minimal, in the following we will illustrate plausibility of the assumptions made. only the ‘UN Scenario’. In terms of the use of international population Only for China do we recommend a deviation from projections, there is no doubt that the UN population the UN medium variant because China has such sig- projections dominate the field. This is mostly owing to nificant weight when studying the world population their long and well-established tradition and the easy and there are convincing arguments that fertility in availability of country-specific data for both estimates China is currently significantly lower than given by Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2787 the UN and is likely to stay so for the coming decades. Table 1. Projections of total population size for continents as well as selected countries and regions (UN Scenario of The 2008 UN assessment gives a total fertility of 1.77 IIASA education projections). for China for the periods 2000 – 2005 and 2005 – 2010. This is subsequently assumed to increase to 1.79 in area 2000 2010 2020 2030 2040 2050 2010 – 2015, 1.84 in 2015 – 2020 and stay constant at 1.85 thereafter. But the level of fertility in China world 6124 6885 7617 8233 8699 9021 around the 2000 census and thereafter has become a Africa 821 1032 1271 1518 1765 1998 topic of intense scientific analysis and discussion. Asia 3705 4145 4546 4846 5024 5095 While the National Statistical Agency published the Europe 729 730 722 707 687 664 figure of 1.22 for the census year, most scholars Latin America and 523 594 660 713 750 769 agree that this number reflects an undercount with Caribbean the only question being how much of an undercount. North America 316 349 379 405 427 445 In an effort to produce probabilistic population projec- Oceania 31 35 39 43 46 49 tions for China that also assume uncertain starting Brazil 174 199 220 236 248 254 conditions, Lutz et al. (2007b) reviewed more than China 1270 1330 1371 1374 1324 1238 20 different estimates based partly on different India 1046 1220 1379 1506 1597 1658 methods and different data sources. They concluded UK 59 62 64 66 68 69 that the best guess for 2000 was a level around 1.5 European Union 482 495 498 496 489 479 with significant uncertainty bounds. Most recently, a Former Soviet 289 284 279 271 261 249 new study by Morgan et al. (2009) convincingly Union demonstrates that total fertility is currently slightly NW Europe 246 253 258 262 262 261 below 1.5 and is expected to remain there at least for Nile catchment 225 285 354 424 492 555 the coming two decades. Based on this strong scientific sub-Saharan Africa 680 867 1081 1308 1540 1761 reasoning, we suggest that the best-guess projections used for China should be based on a constant total fertility assumption of 1.5. This has been implemented Of the countries that are individually listed, China— in the output tables and graphs given below. currently the most populous country in the world with 1.3 billion inhabitants—will continue to grow until around 2030 owing to population momentum (i.e. 6. DESCRIPTION OF RESULTS more young women entering the reproductive ages) The following section presents and discusses selected even though fertility is assumed to be well below the results of the chosen UN Scenario of the IIASA replacement level. By 2050, China’s population size is global projections by level of education by using expected to be lower than it is today and 420 million the regional definitions given as a standard for all the lower than India’s, which is likely to surpass China as driver reviews. These projections are available at the the most populous country shortly before 2020. level of individual countries on the website of While the picture of future population growth is IIASA’s World Population Programme (www.iiasa.ac. quite differentiated, with some countries and regions at/Research/POP). In addition to six continents they expected to grow substantially, whereas others are list the data for four key countries and five regions of expected to shrink in terms of future population special interest. This scenario was calculated on the ageing, all countries and regions are moving in the basis of the 2006 UN assessment before the 2008 same direction. Currently about 8 per cent of the assessment came out, and the UN made minor adjust- total world population is above the age of 65. This pro- ments between the 2006 and the 2008 revisions. portion is likely to double over the coming 20 years Therefore, small discrepancies might appear between and by 2040 reach the level of 16 per cent, which is the data listed here and those currently available the level currently experienced in Europe. Asia is the from the website of the UN Population Division. most rapidly ageing continent where the current pro- Table 1 gives the results in terms of total population portion above age 65 is likely to increase by a factor size. It shows the total world population increasing from of three from currently 7 to 21 per cent in 2050. an estimated 6.885 billion in 2010 to 7.6 billion in China will rapidly catch up with Europe and reach 2020, 8.2 billion in 2030 and around 9 billion by some 27 per cent above the age of 65 by the middle 2050. These numbers clearly indicate the projected of the century, although currently its proportion decelerating speed of world population growth. While elderly is only half of the European one. Even in the decadal increase in world population is estimated Africa where the population structures are still very by the UN to be 760 million between 2000 and 2010, young (only 3% of the population are above age 65), it is projected to decline to 616 million for the decade the projected increase in life-expectancy together 2020 – 2030 and 322 million for 2040 – 2050. The dis- with declines in fertility will result in significant tribution of this growth over continents shows that the ageing in the longer run (table 2). population of Africa is still expected to roughly When interpreting these numbers on projected double, whereas that for Europe is already on a declin- proportions elderly, we also need to consider that ing trajectory. It is worthwhile noting that the disability-free life-expectancy so far tends to devastating AIDS pandemic, which lowered life-expect- increase at roughly the same speed as total life- ancy in the worst-hit countries and also had a minor expectancy and that the 65-year-olds of the future depressing effect on population growth, does not can be expected to be in better health conditions really influence this big picture of population growth. than the 65-year-olds today. It has recently been Phil. Trans. R. Soc. B (2010) 2788 W. Lutz & K. C. Samir Review. Global population projections Table 2. Projections of the proportions of the population above age 65 for continents as well as selected countries and regions (UN Scenario of IIASA education projections). area 2000 (%) 2010 (%) 2020 (%) 2030 (%) 2040 (%) 2050 (%) world 7 8 10 13 16 19 Africa 3 3 4 5 5 7 Asia 6 7 9 13 17 21 Europe 15 16 19 23 25 28 Latin America and Caribbean 6 7 9 12 15 19 North America 12 13 16 20 21 21 Oceania 10 11 14 16 18 19 Brazil 5 7 9 13 16 19 China 7 8 12 17 24 27 India 5 5 7 9 11 14 UK 16 17 19 22 24 24 European Union 16 17 20 24 27 29 Former Soviet Union 11 11 13 16 18 21 Nile catchment 3 4 4 5 6 8 NW Europe 16 18 21 24 26 26 sub-Saharan Africa 3 3 3 4 5 6 Table 3. Projections of the proportions of the population (above age 15) that have junior secondary or higher education for continents as well as selected countries and regions (UN Scenario of IIASA education projections). proportion with at least secondary education sex area 2000 (%) 2010 (%) 2020 (%) 2030 (%) 2040 (%) 2050 (%) female world 53 59 65 71 77 82 Africa 26 35 43 52 60 67 Asia 45 54 62 69 76 82 Europe 85 89 92 94 95 96 Latin America and Caribbean 53 62 70 78 84 89 North America 95 94 94 94 94 95 Oceania 96 99 100 100 100 100 Brazil 52 62 71 80 86 91 China 56 66 75 82 89 93 India 28 38 48 58 67 75 UK 73 81 87 90 92 94 European Union 80 85 90 93 94 95 Former Soviet Union 96 98 99 99 99 99 Nile catchment 25 34 42 52 60 67 NW Europe 83 87 91 93 94 95 sub-Saharan Africa 21 29 38 47 55 63 male world 62 67 72 76 79 83 Africa 38 45 52 58 63 68 Asia 59 66 72 76 81 84 Europe 86 89 92 93 94 96 Latin America and Caribbean 52 60 67 73 79 83 North America 94 94 94 94 95 95 Oceania 96 98 100 100 100 100 Brazil 48 56 64 72 78 83 China 71 78 84 87 91 94 India 47 55 63 70 75 80 UK 73 81 86 89 91 92 European Union 82 86 89 92 93 94 Former Soviet Union 96 98 98 99 99 99 Nile catchment 37 43 50 56 61 66 NW Europe 84 87 90 93 94 95 sub-Saharan Africa 33 39 47 54 59 65 Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2789 (a)( 1800 b) 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 year year (c) 1800 tertiary 20+ secondary 15+ primary 15+ no edu15+ 0–14 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 year Figure 2. (a) China, (b) India and (c) sub-Saharan Africa: projected trends in the total population by level of highest educational attainment (children below age 15 in grey at the bottom). demonstrated (Lutz et al. 2008b) that when age is not worse-educated populations in Africa which hence defined as the time since birth but alternatively as the might be viewed as a rather optimistic scenario. expected time to death, then the coming speed of Table 3 also shows that while today almost universally population ageing (i.e. people moving closer to their adult men are better educated than adult women, this death) will be much more moderate. The future is likely to change in the future because of the fact that elderly are also likely to be better educated and more female school enrolment rates in most countries are likely to continue to be gainfully employed depending approaching those of men and in many countries on the incentive structures that will be in place. The even surpassing them. greatest challenge associated with population ageing Figure 2 shows the projected trends in the absolute is probably in the poorest countries where often no numbers of the population by four educational attain- old-age support systems exist aside from one’s own ment categories for India and China as well as the family. This also needs to be considered in the context region of sub-Saharan Africa. The population below of studying future rural populations and the agricultural the age of 15 is indicated as a separate group at the work force. bottom of the graph. While for China the picture As discussed above, almost universally more educated shows a peaking in the size of the population followed people are in better health and are more productive. by a decline over the coming decades, the number of Recent studies have shown that there are some people with secondary or tertiary education will con- thresholds both with respect to health and to economic tinue to increase. The trend in sub-Saharan Africa growth in the sense that universal primary education shows a very different picture characterized by contin- (one of the key Millennium Development Goals) is not ued rapid population growth. India is in an sufficient but that it requires high proportions of the intermediate position with decelerating population population with at least completed junior secondary growth associated with a rapid expansion of the more education (to age 15) to help bring countries out of educated segments of population. But even in Africa the vicious circle of poverty, high population growth this projection of populations by level of education andfoodinsecurity(Lutz et al. 2008a). For this reason gives rise to more optimism for the future than the table 3 focuses on the proportion of the population usual focus on population size alone, because it with junior secondary or higher education. shows that the most rapidly growing segment of the The trends shown in table 3 are based on the Global population is that with secondary or tertiary education Education Trend Scenario (KC et al. 2010), which under this admittedly rather optimistic scenario. assumes that in terms of the proportions of cohorts The comparison between India and China in ending up in the different educational attainment figure 2 is particularly interesting since they are the categories, the countries later in the process follow two population billionaires frequently mentioned the trend of the more advanced countries. Since the together as the two great economic powers of the 1970s, this trend has been dominated by the speed future. But the figures illustrate quite clearly that in of educational expansion of many Asian countries. terms of the human capital of their populations, the Similar expansions are also assumed for the currently two countries are very different. Over the past decades, Phil. Trans. R. Soc. B (2010) population in millions population in millions 2790 W. Lutz & K. C. Samir Review. Global population projections China has heavily invested in universal primary and by level of educational attainment, age and sex for 120 countries for 2005 – 2050. Demograph. Res. 22, 383 – near universal secondary education. Although still 472. See http://www.demographic-research.org/Volumes/ less than half of the total population today has second- Vol22/15/. (doi: 10.4054/DemRes.2010.22.15) ary or higher education, this is certain to change as the Keyfitz, N. 1985 Applied mathematical demography, 2nd edn. better educated younger cohorts move up to higher age New York, NY: Springer. groups. In contrast, India suffers from the fact that Khan, H. T. A. & Lutz, W. 2008 How well did past UN currently still about half of all adult women have population projections anticipate demographic trends in never been to school. This is also the main reason six South-east Asian countries? Asian Popul. Stud. 4, why fertility in India is still rather high and as a conse- 77 – 95. (doi:10.1080/17441730801966964) quence the population will experience significant Lutz, W. 2009a Editorial: towards a world of 2 – 6 billion growth over the coming decades. Recently, school well-education and therefore healthy and wealthy people. J. R. Stat. Soc. A 172, 701 – 705. (doi:10.1111/j. enrolment rates in India have increased at all levels 1467-985X.2009.00612.x) but it will take many decades until India will be able Lutz, W. 2009b Sola schola et sanitate: human capital as the to match the level of schooling of the average Chinese. root cause and priority for international development? This will have implications for food security, health, Phil. Trans. R. Soc. B 364, 3031 – 3047. (doi:10.1098/ economic growth and adaptive capacity to climate rstb.2009.0156) change. Lutz, W. & Goldstein, J. (Guest editors) 2004 Special issue In conclusion, this review has attempted to high- on ‘how to deal with uncertainty in population forecast- light some recent developments in the methodology ing?’. Int. Stat. Rev. 72, 1 – 106, 157 – 208. and the content of global population projections and Lutz, W. & Scherbov, S. 2008 Exploratory extension of in particular place emphasis on the different dimen- IIASA’s world population projections: scenarios to 2300. Interim Report IR-08-022. International Institute sions of population change that should be explicitly for Applied Systems Analysis, Laxenburg, Austria. addressed in population projections. The educational Lutz, W., Sanderson, W. & Scherbov, S. 1997 Doubling of attainment distribution has been singled out as a world population unlikely. Nature 387, 803 – 805. key dimension, which perhaps should be routinely (doi:10.1038/42935) added to age and sex in our studies of the trends Lutz, W., Goujon, A. & Doblhammer-Reiter, G. 1999 and consequences of human population size and Demographic dimensions in forecasting: adding edu- structures. cation to age and sex. In Frontiers of population forecasting. Supplement to vol. 24, 1998 Population and This paper was prepared for the Driver Review DR1 of the Development Review (eds W. Lutz, J. W. Vaupel & D. Foresight Project on Global Food and Farming Futures, A. Ahlburg), pp. 42 – 58. New York, NY: The Population funded by the UK Government Office for Science. It is Council. also based on research carried out in IIASA’s World Lutz, W., Sanderson, W. & Scherbov, S. 2001 The end of Population Programme, which is partially funded by the European Research Council (ERC) Advanced Investigator world population growth. Nature 412, 543 – 545. Grant focusing on ‘Forecasting Societies’ Adaptive (doi:10.1038/35087589) Capacities to Climate Change’ (ERC-2008-AdG 230195- Lutz, W., Sanderson, W. C. & Scherbov, S. (eds) 2004a The FutureSoc). end of world population growth in the 21st century: new chal- lenges for human capital formation and sustainable development. London, UK: Earthscan. REFERENCES Lutz, W., Scherbov, S., Makinwa-Adebusoye, P. K. & Reniers, G. 2004b Population-environment-development- Abbasi-Shavazi, M. J., Lutz, W., Hosseini-Chavoshi, M. & Samir, K. C. 2008 Education and the world’s most agriculture interactions in Africa: a case study on Ethiopia. In The end of world population growth in the rapid fertility decline in Iran. Interim Report IR-08- 010. International Institute for Applied Systems Analysis, 21st century: new challenges for human capital formation and sustainable development (eds W. Lutz, Laxenburg, Austria. Bongaarts, J. 2009 Human population growth and the W. C. Sanderson & S. Scherbov), pp. 187 – 225. London, UK: Earthscan. demographic transition. Phil. Trans. R. Soc. B 364, 2985 – 2990. (doi:10.1098/rstb.2009.0137) Lutz, W., Goujon, A., Samir, K. C. & Sanderson, W. 2007a Reconstruction of population by age, sex and level of edu- Cannan, E. 1895 The probability of a cessation of the growth of population in England and Wales during the next cational attainment of 120 countries for 1970 – 2000. Vienna Yearbook Popul. Res. 5, 193 – 235. (doi:10.1553/ century. Econ. J. 5, 505 – 515. (doi:10.2307/2956626) Hajnal, J. 1955 Prospects for population forecasts. J. Am. populationyearbook2007s193) Lutz, W., Scherbov, S., Cao, G. Y., Ren, Q. & Zheng, X. Y. Stat. Assoc. 50, 309 – 322. (doi:10.2307/2280963) Hayami, Y. & Ruttan, V. 1971 Agricultural development: an 2007b China’s uncertain demographic present and future. Vienna Yearbook Popul. Res. 5, 37 – 59. (doi:10. international perspective. Baltimore, MD: Johns Hopkins University Press. 1553/populationyearbook2007s37) Lutz, W., Crespo Cuaresma, J. & Sanderson, W. 2008a The Jennings, V. E., Lloyd-Smith, C. W. & Ironmonger, D. S. 2004 Global projections of household numbers using demography of educational attainment and economic growth. Science 319, 1047 – 1048. (doi:10.1126/science. age determined ratios. Working Papers Series, No. 914. Department of Economics, University of Melbourne, 1151753) Lutz, W., Sanderson, W. & Scherbov, S. 2008b The coming Melbourne, Australia. Jiang, L. & O’Neill, B. C. 2009 Household projections for acceleration of global population ageing. Nature 451, 716 – 719. (doi:10.1038/nature06516) rural and urban areas of major regions of the world. Interim Report IR-09-026. International Institute for Morgan, P., Zhigang, G. & Hayford, S. 2009 China’s below- replacement fertility: recent trends and future prospects. Applied Systems Analysis, Laxenburg, Austria. Samir, K. C., Barakat, B., Goujon, A., Skirbekk, V., Popul. Dev. Rev. 35, 605 – 630. (doi:10.1111/j.1728- 4457.2009.00298.x) Sanderson, W. & Lutz, W. 2010 Projection of populations Phil. Trans. R. Soc. B (2010) Review. Global population projections W. Lutz & K. C. Samir 2791 Nakicenovic, N. et al. 2000 Emissions scenarios.In A special United Nations 2004 World population to 2300. New York, report of working group III of the Intergovernmental Panel NY: United Nations, Population Division. on Climate Change. Cambridge, UK: Cambridge Univer- United Nations. 2008 World urbanization prospects: the 2007 sity Press. revision. New York, NY: Department of Economic and Notestein, F. W. 1945 Population: the long view. In Food for Social Affairs, Population Division, United Nations. the world (ed. T. W. Schultz), pp. 36 – 69. Chicago, IL: United Nations. 2009 World population prospects: the 2008 University of Chicago Press. revision. New York, NY: Department of Economic and O’Neill, B. C., Balk, D., Brickman, M. & Ezra, M. 2001 A Social Affairs, Population Division, United Nations. guide to global population projections. Demogr. Res. 4, van Imhoff, E. & Keilman, N. 1991 LIPRO 2.0: an appli- 203 – 288. cation of a dynamic demographic projection model to Rogers, A. & Land, K. (eds) 1982 Multidimensional math- household structure in the Netherlands. Amsterdam/Lisse, ematical demography. London, UK: Academic Press. The Netherlands: Swets and Zeitlinger, NIDI/CBGS Sanderson, W. & Scherbov, S. 2008 Rethinking age and Publications No. 23. aging. Popul. Bull. 63, 3 – 16. Vaupel, J. W. & Yashin, A. I. 1985 Heterogeneity’s ruses: Skirbekk, V. 2008 Age and productivity capacity: descrip- some surprising effects of selection on population tions, causes and policy options. In Ageing horizons, dynamics. Am. Stat. 39, 176 – 185. (doi:10.2307/ vol. 8, pp. 4 – 12. Oxford, UK: Oxford Institute of Ageing. 2683925) Phil. Trans. R. Soc. B (2010)

Journal

Philosophical Transactions of the Royal Society B: Biological SciencesPubmed Central

Published: Sep 27, 2010

References