Re-measuring Twenty-first Century Population Ageing

Probably the most famous demographic riddle of all time is the one that the Sphinx was said to have posed to travellers outside the Greek city of Thebes: ‘Which creature walks on four legs in the morning, two at noon, and three in the evening?’ Unfortunate travellers who could not answer the riddle correctly were immediately devoured. Oedipus, fresh from killing his father, was the first to have got the answer right. The correct answer was ‘humans’. People crawl on their hands and knees as infants, walk on two feet in adulthood, and walk with a cane in old age. We easily recognize the three ages of humans. Humans are born dependent on the care of others. As they grow, their capacities and productivities generally increase, but eventually these reach a peak. After a while, capacities and productivities decline and, eventually, if they are lucky enough to survive, people become elderly, often again requiring transfers and care from others. The human life cycle is the basis of all studies of population ageing, and so we cannot begin to study population ageing without first answering the Sphinx’s riddle. However, answering the Sphinx’s riddle is not enough to get us started on a study of population ageing. We must take two more steps before we begin. First, we must recognize that not all people age at the same rate. As seen in Chapter 5, nowadays more educated people tend to have longer life expectancies than less educated people. Second, we must realize that there is no natural generalization of the Sphinx’s riddle to whole populations. Populations cannot be categorized into the stages of infancy, adulthood, and old age. Indeed, if the Sphinx was reborn today, we might find her sitting near another city and posing an equally perplexing riddle, one especially relevant for our times: ‘What can grow younger as it grows older?’ Answering this riddle correctly is the central challenge of this chapter and the key to understanding population ageing in the twenty-first century.

Future Fertility in Low Fertility Countries

The ongoing transition to low fertility is, alongside the long-term expansion of life expectancy, the key force reshaping populations around the world. It has sweeping economic and social repercussions as it affects labour markets, intergenerational ties, gender relations, and public policies. Many middle-income countries, including China, Brazil, Iran, and Turkey, have joined the expanding list of low fertility countries. Consequently, low fertility is no longer an exclusive feature of rich Western societies. As close to half of the global population now lives in regions with below replacement fertility, low fertility has become a truly global phenomenon. What are the key ingredients of this ‘revolutionary’ change? Expanding education, rising income, the rise of gender equality, female labour force participation, ideational changes, consumerism, urbanization, family disintegration, economic uncertainty, globalization, modern contraception, and many other complementary or contrasting forces are often highlighted. But how will these drivers shape the long-term future of fertility? Will fertility in most countries stabilize at around the replacement level threshold, as implied by the demographic transition theory, or will it decline below this level? Is very low fertility merely a ‘passing phenomenon’, a sign of a temporary imbalance between rapid social and economic changes and opportunities on the one hand, and family, gender relations, and reproduction on the other? This chapter aims to present both a comprehensive overview of the forces shaping contemporary reproductive behaviour in low fertility countries and an exploration of possible future scenarios based upon a new IIASA–Oxford survey of international experts introduced in Chapter 2 of this volume. We begin with a presentation of recent trends in fertility in low fertility settings followed by a review of the particular recent histories of fertility change in North America, Europe, and the emerging low fertility settings in East Asia, Latin America, and the Middle East. We then explore the theoretical and empirical evidence that has been cited in the literature as underpinning these past trends and possible future scenarios. As well as ‘meta-theories’ such as the Second Demographic Transition (SDT), section 3.2 considers the roles played by cultural, biomedical, and economic factors, family policies, economic uncertainty, education, and the contribution of migrants’ fertility.

The Rise of Global Human Capital and the End of World Population Growth

This is the first of three chapters that present the population projections by age, sex, and level of educational attainment for all countries in the world with a time horizon of 2060, and extensions to 2100. Before discussing the Wittgenstein Centre for Demography and Global Human Capital (WIC) projections, however, it is worth stepping back to consider how social structures change over time. While understanding the evolution of social structures is important under the conventional demographic approach that breaks down populations by age and sex, a more in-depth understanding of the changes in human capital requires that the interplay between different levels of schooling over time (the flow variable), and the changing educational attainment composition of the adult population (the stock variable) be taken into account. Societies can be stratified along several dimensions. In conventional social science the divisions studied refer to social class, race, or ethnicity. Demographers routinely break down populations by age and sex. Another important demographic dimension is that of birth cohorts or generations, that is, persons born and socialized during the same historical period. Particularly during periods of rapid social change, young cohorts tend to differ from older ones in important respects, and the demographic process of generational replacement is a powerful driver of socio-economic change. This process is analytically described by the theory of ‘Demographic Metabolism’, recently introduced as a generalized predictive demographic theory of socio-economic change by the first author (Lutz, 2013), building on earlier work by Mannheim (1952) and Ryder (1965). Ryder, who introduced the notion of Demographic Metabolism in a qualitative way, saw it as the main force of social change. While this theory applies to many stable human characteristics that are acquired at young age and remain invariant over a lifetime, it is particularly appropriate for studying and modelling the dynamics of the change in the distributions of highest educational attainment by age and sex over time. This perspective on human capital formation is the main focus of this book. This first of the three results chapters will highlight the results with respect to future population numbers by level of education in different parts of the world.

Future Fertility in High Fertility Countries

Beginning in 1960, a phenomenon occurred that John Caldwell named the ‘global fertility transition’ (Caldwell, 1997), in which fertility declines have become the general rule throughout the world, including in the majority of the less developed countries. This is important partly because fertility is in many circumstances negatively associated with socio-economic development (Bryant, 2007). From 1970–75 to 2005–10, the average total fertility rate (TFR) for the developing world fell by half, from 5.4 to 2.7 births per woman on average (United Nations, 2011). However, global figures hide important differences in fertility levels among the different regions. In Asia and Latin America, the reproductive behaviour of women reflected the pattern of change noted by Caldwell, halving the TFR in the last 35 years. In Africa, on the contrary, fertility stagnated at 6.2–6.4 from 1950 to 1985, and then began a decline that was much slower than in other developing regions. As a whole, the TFR of sub-Saharan Africa has, for decades, been higher than the fertility levels elsewhere. This was the case in 1950 and 1975, and remains so today. Fertility differences among countries are now larger than ever because transitions to replacement fertility have not yet started in some subpopulations of Western and Middle Africa, but have already been completed in others (e.g. in the economically most advanced countries of Asia, especially East Asia, as well as in many countries in Latin America and the Caribbean). As a result, the observed TFRs of (former) developing countries in 2005–10 range from a high of 7.1 in Niger to a low of 1.0 in Hong Kong. All regions of the world experience wide variations in their TFRs. For instance, East Asia has experienced a faster fertility decline than countries like Pakistan in south-central Asia. Moreover, fertility levels can show significant variations within a single country. This is the case in India, where Northern and Southern patterns of fertility are very different. Overall, regional variations are most apparent in sub-Saharan Africa.

Future Mortality in Low Mortality Countries

This chapter provides an overview of past and expected future trends in life expectancy in populations with low levels of mortality. High and low mortality populations were separated on the basis of the level of child mortality in the year 2010 according to the revised estimates of the United Nations Inter-agency Group for Child Mortality Estimation (2011), with the threshold being 40 deaths per 1,000 children below the age of 5 years. The low mortality population is comprised of 132 countries including Europe, North America, most of Oceania and Latin America, large parts of Asia (excluding the high mortality area in Central and Southern Asia), and Northern Africa. The populations of these countries are already engaged in an advanced phase of the demographic and ‘epidemiologic transition’. Because they previously experienced strong decreases in infant mortality, the future mortality trends are driven mainly by mortality in adult ages, primarily the old and oldest-old. Although the data sources on which the existing estimates of life expectancy for these populations are based vary considerably (owing to differences in the death registration systems and the estimation techniques, see, e.g., Luy, 2010), we have relatively good knowledge of past and current mortality levels and trends and their causes. Despite the similar general trends, today’s low mortality countries are very heterogeneous in various aspects, including medical standards, access to health care, and behavioural risk factors, such as smoking prevalence. These diversities are strongly related to the populations’ stages of economic development and contribute to a broad variance of life expectancy levels. Among men, life expectancy at birth for the years 2005–10 ranges between 60.2 in Kazakhstan and 79.5 in Iceland. Among women, the range is between 67.8 in the Solomon Islands and 86.1 in Japan. To demonstrate this relationship between economic development and life expectancy we classified countries according to their current per capita income as an indicator of the economic development level of the populations. We used the World Bank classification, which groups countries into high income (≥$12,276 annually), upper middle income ($3,976–$12,275), lower middle income ($1,006–$3,975), and low income (≤$1,005).

Data and Methods

The preceding chapters have all contributed to building the knowledge base for the actual Wittgenstein Centre for Demography and Global Human Capital (WIC) projections that will be presented and discussed in the second part of this book. This chapter stands as a bridge between the two parts. Its focus is the translation and operationalization of the empirical evidence and the substantive arguments presented so far into specific population projections by age, sex, and level of educational attainment for all countries in the world. This is a complex exercise in which data and methodology play the crucial roles. The cohort–component multidimensional projections presented in this volume require a large amount of information, ranging from base-year data on population disaggregated by levels of educational attainment by age and sex, to data on fertility, mortality, and migration by age, sex, and education for the base year, and, finally, to the assumed numerical values of these determinants according to the different scenarios. This new set of expert argument-based projections by age, sex, and educational attainment presents an important new step at the forefront of international population projections. As discussed in Chapter 1, this is a logical next step in the tradition of international population projections by the World Population Program of the International Institute for Applied Systems Analysis (IIASA). This effort also goes beyond what the United Nations (UN) and other agencies have been doing in two important ways: it provides the most comprehensive and systematic summary of expert knowledge on future fertility, mortality, and migration to date—including the input of hundreds of demographers from around the world—and it translates this into the most comprehensive set of human capital projections for 195 countries. The WIC projections cover all countries in the world with more than 100,000 inhabitants. In this effort, the study builds on and significantly expands earlier IIASA reconstructions and projections of the population by age, sex, and educational attainment for 120 countries published in 2007 and 2010 (KC et al., 2010; Lutz et al., 2007).

Introduction

This book addresses systematically and quantitatively the role of educational attainment in global population trends and models. By adding education to the traditional demographic characteristics of age and sex, this distinguishing feature substantially alters the way we look at changes in populations and how we project them into the future. In most societies, particularly during the process of demographic transition, women with more education have fewer children, both because they want fewer and because they find better access to birth control. And better educated men and women in virtually all societies have lower mortality rates and their children have a better chance of survival. Migration flows also differ by level of education, and better educated migrants integrate more easily into receiving societies. These pervasive demographic differentials by level of education matter greatly for population dynamics. When we explicitly address this important source of population heterogeneity the projected future population trends are different from those based on the conventional stratifications that include only age and sex. In addition, the future educational attainment levels of the adult population are of great interest in their own right as a key determinant of outcomes ranging across economic growth, quality of governance, and adaptive capacity to environmental change. Traditionally in demography, the sex of a person is considered the most fundamental characteristic because it is essential for studying the process of reproduction. Mortality and migration also show significant variation by gender. Age is another key characteristic because it is the main driver of biological maturation at an early age and is directly related to school attendance, labour force entry, and retirement, all landmarks that are important for social institutions. Because there are distinct age-related patterns of fertility, mortality, and migration intensities, gender and age are considered the most fundamental demographic dimensions. In addition, demographers frequently take into account other biological, social, and economic characteristics, including place of residence (especially urban or rural), citizenship, marital status, race, migration status, employment status, health/disability status, and educational attainment. These additional characteristics are not systematically considered in every study, but tend to appear in corresponding topical studies.

Future Education Trends

Education is an inherently long-term endeavour. Not only in the sense that formal schooling alone may last a significant part of a lifetime, but also because part of the reason for spending this time in school is the promise of benefits for decades to come. This is true at both the individual and societal levels. For the underlying educational systems, the long-term nature of education is felt more keenly than at the individual level. Schools are built to serve multiple generations of students, and teachers are often hired for life as civil servants. A newly trained teacher of today will, towards the end of her/his career, teach students who, in turn, may well still be in the active labour force 100 years from now. Educators themselves also hold expectations about the long-term future. Part of why we care whether a Dalit boy gets some form of early childhood education is because we expect that, as a consequence, his increased chances to complete school will benefit him not only for the next 10 or 15 years, to the end of our programme intervention or planning horizon, but for the rest of his life. His own education may even make it more likely that he will send his own children to school. If he sends a daughter to school, her education will possibly lead her to wait until she is in her late 20s to bear her first child, when she is better able to provide care. That ‘delayed benefit’ of the Dalit boy going to school now might not occur until sometime in the 2050s. To insist that we ‘learn for life, not for school’ is a cliché, yet there are strikingly few attempts to look ahead—much less project quantitatively—how today’s students will contribute to the educational composition of tomorrow’s population, and the implications for their personal life course and the challenges of their generation. Some of the key contemporary policy debates concern very long-term issues. Among these are the sustainability of pensions, the provision of health care, and response to environmental degradation. For all of these, the education of the concerned populations matters.

How Education Drives Demography and Knowledge Informs Projections

This chapter provides the background necessary for understanding our approach to projecting population and human capital. First, we investigate the proper place of education in demographic analysis and the evidence for an underlying causal relationship between education and demographic outcomes. Second, we emphasize the importance of explicit assumptions undergirding population projections and detail our procedures for incorporating the views of hundreds of experts into sets of assumptions that drive the Wittgenstein Centre (WIC) projections. Subsequent chapters build on this background in their detailed discussions of trends and arguments in fertility, mortality, migration, and education. A major innovative feature of this volume is the systematic addition of educational attainment as a standard demographic dimension in addition to age and sex for demographic analyses, particularly for projections. The underlying assumption is that educational attainment is not just one of many socio-economic factors that matter for population, as it is often viewed in conventional demographic analysis, but is the single most important source of empirically observable population heterogeneity next to age and sex. The suggestion of routinely adding educational attainment as a dimension of demographic analysis is not new. It was first proposed in a Population and Development Review article by Lutz et al. (1998), entitled ‘Adding Education to Age and Sex’. More recently, the idea of adding the education factor to demographic analysis was discussed by Lutz (2010) in a commentary entitled, ‘Education Will be at the Heart of 21st Century Demography’. It has also been the focus of two recent articles by Lutz and KC, one published in Philosophical Transactions entitled, ‘Dimensions of Global Population Projections’ (2010), and a review article published in Science entitled, ‘Global Human Capital: Integrating Education and Population’ (2011). In the latter paper they argue that an additional demographic dimension should be added routinely to age and sex in population analyses and projections according to three criteria: (i) its explicit consideration should be feasible in terms of available data and methodology; (ii) it should matter substantially in terms of altering population dynamics; and (iii) it should be of interest in its own right in terms of its social and economic implications.

Alternative Scenarios in the Context of Sustainable Development

The number of people inhabiting the earth has fluctuated significantly over the course of human history, in response to both natural changes in the environment and stresses to local habitats created by the populations themselves. From the first appearance of Homo sapiens some 200,000 years ago in Africa until about 35,000 years ago, the world’s human population was well under one million, a number that meant the threat of extinction was always looming (Biraben, 2002). Only after the Neolithic Revolution, which introduced agriculture, did the world population increase significantly, surpassing 100 million about 7,000 years ago. But it was in the nineteenth century that population growth began to accelerate in what are now the world’s most industrialized countries. This rapid population increase was a consequence of declining death rates owing to better nutrition and hygiene, more accessible fresh water supplies, and advances in preventive medicine. Immediately after World War II, death rates began a precipitous fall, owing primarily to the development of antibiotics and other medical advances. For several decades after the war, birth rates remained very high (and in some cases increased owing to the better health of women) because high fertility norms had been deeply imbedded in most traditional cultures and religions. Such norms tend to change only slowly, and, as a consequence, the world population experienced a dramatic increase, from 2.5 billion in 1950 to more than 7 billion today. The previous chapters have detailed what we now expect about the future of global population trends over the course of the twenty-first century. But in presenting this analysis, we have only occasionally made explicit reference to the fact that all human life depends on functioning environmental life support systems. This ‘support system’ aspect of human life was included in the expert meeting and IIASA–Oxford survey described in Chapters 2–7, with some of the predefined arguments explicitly referring to the possible effects of environmental change on future trends in health, mortality, and migration. With a time horizon of up to 2050, these possible effects of environmental change on the population outlook were seen by the experts as having only marginal impacts.