Too few people?

After years of worrying about overpopulation, some demographers now predict that global population will peak in 2040 and then start a headlong dive. Nicholas Eberstadt looks at some of the curious consequences of a world without siblings
December 20, 1997

Over the past few years, some of the world's best demographers have begun a dramatic reassessment of the world's future. They are now seriously proposing that the world's population, rather than continuing to increase, might peak in our lifetimes and then go into an indefi-nite decline in the generations immedia- tely ahead. This scenario is reflected in, among other studies, the UN Population Division's biennial compendium, World Population Prospects-the oldest and largest contemporary attempt to outline future demographic trends. The current 1996 Revision includes "low variant" projections which anticipate zero population growth for the world as a whole by the year 2040 and negative growth-that is to say, depopu-lation-thereafter.

Like the two alternative projections ("medium" and "high") also offered, this low variant is "thought to provide reasonable and plausible future trends." The eventual global depopulation visualised in this variant, it must be emphasised, is not calamitous-it does not result from Malthusian, environmental or any other variety of disaster. On the contrary: the possible stabilisation and ultimate decline of world population-within the lifetime of most of the earth's current inhabitants-are assumed to occur in what World Population Prospects terms "conditions of orderly progress."

Of course, the UN's new low variant projections do not provide a sure vision of the future. But they do offer a glimpse of one particular, and by no means fantastic, version of it-a version whose outlines have rarely been described and whose ramifications have scarcely been pondered. At a time when all manner of potential "population problems" are regularly accorded official attention by national and international authorities, such neglect is striking.

This is not the first time that population specialists (or others) have raised the prospect of long-term population decline. About 60 years ago, expectations of imminent depopulation were widespread in the western world. We now know that those predictions were wide of the mark. Indeed, at the very time when we were supposed to be entering into permanent negative growth as a result of sub-replacement fertility-in the 1950s and 1960s-western countries were actually in the middle of a demographic surge driven by a postwar baby boom.

The paradox of long-term demographic forecasting is that its methods combine superb technique with an almost complete lack of predictive theory. Mathematical demography is an elegant and sophisticated construct; supplied with the necessary assumptions, it can generate detailed and internally consistent population projections. Those assumptions, unfortunately, are precisely the sticking point.

The defining characteristic of population change in the modern era has been the emergence and spread of sustained fertility decline. Yet this phenomenon has always posed unanswerable questions. The first country in the world to embark upon long-term fertility decline did so in the late 18th century. That country was not industrialising England, as modernisation theories would lead us to expect, but France-then impoverished, overwhelmingly agrarian, predominantly illiterate and devoutly catholic.

More recent fertility trends have proved no less nettlesome. Despite the wealth of information on social and economic conditions in contemporary industrial societies, demographers were unable to predict either the transnational postwar baby boom or the subsequent shift to a sub-replacement fertility level in every country of the Organisation for Economic Co-operation and Development (OECD). As for developing countries, demographers have been unable to forecast either the onset of fertility decline or the trajectory it follows, once it begins.

For better or worse, our only recourse in addressing these issues is to consult the empirical record. From this we know that the fertility level of a fairly large population under conditions of orderly progress can be very low indeed. Eastern Germany's post-unification fertility level, if continued, would be less than one birth per woman per lifetime. We also know that a country's fertility level can drop dramatically once the process of secular fertility decline begins. Between the early 1960s and the early 1990s Thailand's estimated total fertility rate (TFR) plunged from more than six births per woman per lifetime to fewer than two. We know, further, that sub-replacement fertility can characterise fairly poor contemporary societies-China, Cuba and possibly Sri Lanka, among others. Moreover, fertility levels can remain below replacement for prolonged periods. Japan entered into sub-replacement fertility over 40 years ago and has been moving further away from net replacement for the past quarter century. But all these particulars offer little gui- dance for long-range forecasts of a country's population total-much less for the world as a whole.

explaining the un low variant

The UN low variant model uses estimates of the world's current population composition (by country, age and sex) and calculates future populations based on three sets of assumptions: migration, mortality and fertility. For the period 1995 to 2000, the model envisions a net migration of about 1.6m people a year to the more developed regions (the OECD countries of the early 1990s, eastern Europe and the European parts of the former Soviet Union) from the less developed regions (everywhere else). This stream gradually diminishes, ceasing in 2025. These assumptions are clearly arbitrary; there is no particular reason to think they will be correct. But, given their magnitude, they exert only a slight influence on population trends in the more developed regions-and, of course, none at all on global trends.

On mortality, the UN model assumes that life expectancy at birth will rise from roughly 75 years today to 81 years in 2050 in the more developed regions. For the less developed regions, life expectancy is assumed to increase from the current estimate of 64 to 76 years by 2050; in the least developed countries (mainly in sub-Saharan Africa), it is seen as rising from 52 to 72 years. By the benchmarks of the immediate past, such improvements in longevity seem feasible-between 1950 and 1995, global life expectancy is thought to have risen by 20 years (from about 45 to about 65 years).

The model's most important assumptions concern future fertility trends. By the UN's estimate, total fertility rates (TFRs) in the more developed regions averaged about 1.7 births per woman per lifetime in the early 1990s; the low variant assumes that these will stabilise in another decade at about 1.4 (about the EU's level today). In the less developed regions, TFRs averaged 3.3 in the early 1990s; they are now estimated at just below three, declining further to about two in 2020 and to 1.6 in 2050. For the least developed countries, where the average number of births per woman per lifetime in the 1990s is estimated to have exceeded five, TFRs are posited to drop below four by 2010, below three by 2020 and below two by 2035.

Another way to look at fertility assumptions is from the viewpoint of net replacement. In the more developed regions, the net reproduction rate (NRR) is already down to about 0.7-the next generation, following present patterns of childbearing, will be about 30 per cent smaller than the current one. The UN low variant assumes that the NRR will stay close to its present level for the next half century, registering just under 0.7 in 2050. In this picture the less developed regions drop below replacement by about 2010, the least developed regions fall below replacement by about 2030. For the world as a whole, the NRR today is placed at over 1.2; global sub-replacement begins in about 2010 and it is stipulated at 0.74 by 2050-about the same as in today's industrial democracies.

The arithmetical consequence of this bundle of assumptions-none of them outlandish-is a world in which population peaks and then declines forever. On these computations, the human population would reach its apogee in about the year 2040, at over 7.7 billion-about one third more than the 5.8 billion thought to be alive today. Between 2040 and 2050, the world's population would fall by about 85m. Thereafter, world population would shrink by roughly 25 per cent with each successive generation.

The trends which would result in an ultimate global population decline would also bring about a significant redistribution of world population. In 1995, the ratio of population between less developed and more developed regions stood at about four to one; in 2050, by these projections, it would be seven to one. The balance of population would shift dramatically-not only between countries but between entire continents. In 1995, for example, the estimated populations of Europe (including Russia) and Africa (including Egypt and the Maghreb states) were almost exactly equal. In 2050, by these projections, Africans would outnumber Europeans by over three to one.

In a world of negative population growth, the rankings of the 12 most populous countries would also look rather different from those with which we have become familiar. Only half the largest countries of 1950 would remain on the list for the year 2050. Nigeria-which did not even make the list for 1950-will be the world's fourth largest country in 2050, just edging out the US. New additions to the big 12 between now and the year 2050 would include Ethiopia, Zaire and Iran. Whereas six of the 12 largest countries in 1950, and four in 1995, come from the more developed regions, only one-the US-would remain in 2050. Just how demographically negligible today's industrial democracies would be in this vision of the year 2050 may be illustrated with one comparison: not a single European state-including Russia-would match the Philippines in total population. Other things being equal (admittedly, in world politics they seldom are) these trends presage a tremendous shift in the balance of global power.

These projected demographic forces of longer lives and falling fertility would inexorably pave the way for a radical ageing of the human population-a shift whose magnitude would be without historical precedent. In 1900 or so, the median age of the world's population was about 20 years-not far from what it had been in all earlier eras. By 1995, it had reached about 25 years. By 2050, in this low variant world, the median age would be over 42.

the gerontological drift

In the less developed regions, the median age would almost double between 1995 and 2050, jumping from 23 to 41 years. To put this in perspective, the average population from these regions would be more aged than the "greyest" populations in the world today. (In Germany and Japan median age is now just under 40 years.) But people in the more developed regions of 2050 would be older still. By then, the median age in these regions would exceed 51. Japan's median age would be 53; Germany's, 55; Italy's 58.

In the UN low variant, the number of children would sharply decline in all regions, while there would be a population boom among the elderly (or groups currently considered elderly). In 1950, children under the age of five were just under one seventh of the global population. Today, they are about one ninth of the total. In 2050, they would account for less than one twentieth. Conversely, persons aged 65 and over made up about 5 per cent of the world's population in 1950, 6.5 per cent in 1995; they would account for over 18 per cent of the total in 2050. Where there were 2.5 young children for every elderly person in the world in 1950, by 2050 there would be almost four elderly persons for every child. In these projections Italy serves as the extreme instance of demographic ageing: scarcely 2 per cent of the population in 2050 would be under the age of five-but 40 per cent would be 65 or older.

This dramatic worldwide ageing would especially affect the female population. For the first time in the modern era, possibly the first time in human experience, "women of reproductive age" would no longer constitute the norm for humanity. In 1995, an estimated 51 per cent of the world's women were between 15 and 49 years old. Under low variant assumptions, however, by 2050 over 55 per cent of all women will be outside their childbearing years. In the more developed regions, nearly two thirds of all women would not be of reproductive age.

Now consider people between the ages of 15 and 24-the vigorous and exuberant young adults who influence fashion and style, exemplify physical beauty and do most of the fighting in times of war. In the low variant version of the future, the size of this youthful group shrinks significantly in both relative and absolute terms. In the world as a whole, there would be 100m fewer youths in 2050 than in 1995. Comprising 18 per cent of the world's population in 1995, they would account for less than 12 per cent by 2050. More developed regions would especially lack young people: less than 9 per cent of their population would be 15 to 24 years of age. In fact, barely half as many young people would be living in these countries as live there now.

Such a gerontological drift raises questions about the health of the societies in this projected future. Would a depopulating planet be a planet of wheelchairs-of increasingly infirm senior citizens whose escalating demands for medical services and care burden the rest of society? Or would the revolution in longevity be accompanied by a revolution in health, extending the boundaries of middle age-and thereby the scope for active, vigorous and productive existence?

The available literature on research into health and ageing, as it happens, is inconclusive-it points in opposite, mutually exclusive, directions. According to one school of thought, the risk of illness and mortality changes are inversely related: longer lives mean worse health for the survivors. The other school holds that improvements in life expectancy translate into greater life expectancy free from disability, even for persons in their 70s and early 80s.

Reviewing the points of controversy in these studies, the ambiguity of the term "healthy life" is striking. Mortality is easy to define and thus (in theory) to measure. Health is subjective and has many gra-dations. It is possible, indeed likely, that data on self-perceived health status are confounded by the higher expectations of those who are better off. In the US and elsewhere, despite physical evidence to the contrary, affluent people seem more inclined than the less well off to rate their own health as unsatisfactory.

But the international data seem to sup-port the argument that improvements in "disability-free" life expectancy occur near-ly as rapidly as improvements in life ex- pectancy itself-at least for people under 85 years of age and so long as "disability" is carefully and objectively defined. Proper health habits and appropriate medical help can already offer most people an active and independent life well into their 80s. To this extent, anxieties about a coming era of dependent invalids would appear to be misplaced.

Yet we should remember that at times the quality of life for older persons may hinge on discrete but expensive medical treatments. Such services would be more available in rich countries than in poor ones, even in the year 2050, and differences in the health status of the elderly might in the future replace the summary index of "development" that the infant mortality rate provides today.

the economics of ageing

In the 1930s, when the spectre of depopulation haunted western intellectuals, many of the most eminent economists of the day, including John Maynard Keynes, Gunnar Myrdal and Joan Robinson, argued that low fertility and stagnant or declining population could compromise economic performance. By stifling demand, sluggish or negative population growth could exacerbate or even precipitate "under-consumption" and a crisis of unemployment. At the very least, low fertility would press down the investment rate or slow the allocation of new labour into promising and productive areas.

With the benefit of hindsight, most of these arguments now look surprisingly weak. Depression era economists were too ready to explain the great international slump of the 1930s-which was essentially non-demographic in nature-in terms of the fertility patterns and population trends of the day. (Ironically, scarcely a generation later, eminent economists were attributing the same ills to overly rapid rates of population growth.)

Careful review of the empirical record suggests that demographic forces, of whatever kind, need be no more than a secondary factor in overall economic performance. This empirical record also suggests that well thought out public policies, in tandem with suitable private arrangements, can capitalise on the opportunities inherent in a country's population trends. In fact, during the modern era nations have prospered even in the wake of seemingly calamitous "population problems." West Germany, Taiwan and South Korea each flourished after their sudden, forced and tumultuous absorption of millions of refugees; Japan enjoyed rapid development after the second world war, even though its male life expectancy had dropped to Neolithic levels as a result of the war.

By comparison with such trials, the demographic challenges posed by gradual population ageing and eventual population decline seem decidedly modest. Indeed, there are reasons to be guardedly optimistic about the macroeconomic consequences of these trends. Surveying demographic prospects in the US, Harvard economist David M Cutler and his colleagues have made the point that prolonged sub-replacement fertility would actually lower that country's investment needs and increase its living standards (consumption levels), because less capital would be required by new entrants into the labour force. Although expenditures on the care and support of the elderly would rise, these costs would be substantially offset, Cutler and his colleagues reckon, by a reduced need for spending on the young. In all, they conclude, the optimal savings rate in 2050 would probably be slightly lower than the optimal savings rate today.

However, the demographics of depopulation might pose one important and so far novel problem for the education and training of the workforce. In a world where nearly half the population lives to the age of 80 or beyond, the ordinary person's "economically active life expectancy" could quite conceivably be as long as 50 working years-or longer. Given the arithmetic of sustained below-replacement fertility, it is not difficult to imagine circumstances half a century from now in which the majority of a country's workers were over the age of 50.

If future education systems operated on today's principles, most people at work would have received their final formal training over a generation earlier; they would be functioning with the knowledge and techniques of an increasingly distant past. We should not overstate the problem: on-the-job training, refresher courses and the like are already familiar in the modern workplace. But the age structure changes which negative population growth would bring would considerably intensify the mismatch between an education system designed to train people when they are young, and their desire to enjoy a long and worthwhile career in an increasingly complex economy. The institutions and routines of higher education would have to be fundamentally re-examined and recast.

The possible cessation and decline of population growth in coming decades may pose no insuperable macroeconomic problems for future generations, but it stands to make enormous difficulties for the state. In a world like the one suggested in the UN's low variant projections, governments would be subject to intense budgetary and political pressure to overhaul their welfare systems. Negative population growth would especially threaten the central feature of the modern welfare state-the nation-wide, tax-financed, pay-as-you-go pension programme. Weighed down by unalleviable demographic burdens, it is hard to see how these programmes could remain viable.

The government-run social security and pension programmes in almost all of today's industrial democracies finance these operations by taxing today's workers in order to fund today's retirees. Because these systems were established in periods of relatively high fertility and relatively rapid population growth, pay-as-you-go pension systems had the political allure of promising generous benefits on the cheap. In an unguarded moment 30 years ago, the economist Paul Samuelson captured the reasoning which underpinned this approach to public finance: "The beauty about social insurance is that it is actuarially unsound. Everyone who reaches retirement age is given benefit privileges that far exceed anything he has paid in... social security is squarely based on what has been called the eighth wonder of the world-compound interest."

With below-replacement fertility and increasing longevity, however, the arithmetic of pay-as-you-go retirement programmes changes unforgivingly. As the ratio of employees to retirees falls, a universal pay-as-you-go retirement system has only three options for preventing bankruptcy: reduce pension benefits; raise taxes; restrict eligibility. There are no alternatives.

Although populations in the less developed regions would not, in these projections, be so very "grey," those countries would probably be less capable of maintaining state-based pay-as-you-go retirement systems in the year 2050 than OECD countries are today. First, the dependency ratio of elderly to working age population would be higher in the less developed regions, on average, than it is in any OECD country today. Second, the less developed regions half a century from now may not, on average, be nearly as affluent as OECD countries today. The calculations of the economic historian Angus Maddison suggest that (adjusting for international differences in purchasing power of local currencies) GDP per capita for what the UN describes as less developed regions was about one fifth of the more developed countries' per capita GDP in the early 1990s-and less than one sixth of the OECD countries. If these regions enjoy long-term per capita growth rates of 3 per cent a year for the next half century, their average output level would still be nearly 40 per cent lower than the OECD's today. (To get a sense of what this would mean, imagine financing western Europe's pension burden in the coming decade with western European incomes of the late 1960s.)

Already the actuarial status of state-run retirement systems in most OECD countries looks unsustainable. In the US, according to economists at the OECD, the net present value of the unfunded deficit in the social security system amounts to only 23 per cent of GDP. I say "only"-because the unweighted average of that deficit for the 20 OECD countries examined came to 95 per cent of GDP. Even if the implicit social contract underlying these systems were gutted-by restricting pension eligibility to cover less than a third of the retirement age population, for instance-over half of these pension systems would still remain underfunded for the foreseeable future.

One reform to deal with the problem is likely to be later retirement ages, as populations make greater productive use of their extended active life spans. More fundamentally, pay-as-you-go financing will probably be replaced by self-financed retirement benefits. Although such a change could involve a full privatisation of social insurance, it is also possible to imagine the reformed pension systems operating under the aegis of government. Even under government supervision, however, it is hard to see how self-financed pensions (which explicitly acknowledge the beneficiary's creation of his or her retirement account) could lend themselves as readily to redistributive or other non-market objectives as pay-as-you-go arrangements have done. Thus declining population growth might not suppress the appetite of the state, but it might well check the voting public's willingness to feed it.

a world without siblings

Nearly 40 years ago, Jean Fourastie, the French sociologist, wrote a vivid and penetrating essay on how family and social life change under the influence of the modern decline in mortality. The revolution in survival chances, he asserted, had transformed marriage from a binding but temporary contract to a much lengthier, possibly more tenuous, commitment; it had reduced old age from an almost mystical status to a common and often pitiable physical condition; and it had all but banished the procession of death and suffering which had previously conditioned family life. Fourastie also noted that the modern revolution in mortality schedules had entirely altered the ordinary person's chances of participating in "intellectual life" (which he took to begin at age 12) and "independent life" (which began, in his view, at about age 20). The scope for "creative intellectual life," he observed, had been hugely expanded by improvements in survival chances. By Fourastie's calculations, modern man could expect to experience between three and six times as many years of life in his 40s and 50s (which Fourastie designated the peak period of creativity) as the "traditional man" of the 17th century. (This great extension of "creative intellec-tual life," I would add, may have contributed to modern economic growth, which has been so strongly driven by applied advances in knowledge.)

If a revolution in mortality has already recast social rhythms and relations within the family, a revolution in fertility may have a similar impact in the future. More particularly, the fertility decline envisioned in the UN's low variant projections would set the stage for a world never before experienced: a world in which the only biological relatives for many people-perhaps most people-will be their ancestors.

Paradoxically, the great reduction in fertility witnessed in western societies over the past two centuries has been accompanied by a parallel reduction in childlessness. In the modern world, as the demographer Laurent Toulemon has observed, "very few couples remain childless voluntarily." Under the modern regimen of sub-replacement fertility, it seems, very few parents seek a third child, but almost everyone chooses to have that first baby if they can.

In such circumstances, prolonged bouts of fertility far below the replacement level would profoundly alter the composition of the typical family. Consider the possibilities for Italy, currently the country with the world's lowest fertility level. At present, Italy's TFR is estimated at less than 1.2; the UN's low variant projections anticipate the continuation of this pattern to the year 2050. If Italy's current fertility regimen is extended for two generations, the Italian family will be completely redefined. For in that future world, under reasonable assumptions about the incidence of childlessness and larger families, almost three fifths of Italy's children will have no siblings, cousins, aunts or uncles. They will have only parents, grandparents and perhaps great-grandparents.

Italy's position today is at an extreme within the fertility continuum among contemporary nations. But projecting the fertility rates for the entire EU forward over two generations only slightly alters the Italian scenario: about 40 per cent of European children would have no collateral blood relatives; less than one sixth would have a brother or a sister and a cousin.

While it is possible to describe this new kind of family, we can hardly imagine what it would portend. Throughout the memory of human experience, the family has been the primary and indispensable instrument for socialising a people. Within the family the individual found extended bonds of obligation and reciprocal resources-including emotional resources-to draw upon. Under the demographic projections considered here, this would change momentously. For many people, "family" would come to be understood as a unit without biological contemporaries or peers.

How will each person's little tribe be formed in such a future? Who will we play with, learn from, love unthinkingly and fight with ferociously, knowing all the while that we can do these things because we are linked by an indissoluble common tie? To paraphrase the poet Robert Frost: if "family," means "the people who must take you when no one else will," and blood relatives our own age are no longer the norm, who then will take us in?

The nuclear family may have marked a radical departure from previous kinds of family arrangement. But as we have seen, the nuclear family does not begin to approach the limits of social atomisation which may await us in a depopulating world. Difficult as the implications of these changes may be to comprehend today, we may yet manage to assess them very carefully. It is not impossible that we will eventually experience them firsthand.