The end of population growth in Asia

Journal of Population Research, May, 2003 by Wolfgang Lutz, Sergei Scherbov, Warren Sanderson

This paper presents probabilistic population projections for five regions of Asia (South Asia, Central Asia, China region, Pacific OECD and Pacific Asia) and Asia as a whole. Over this century, Asia will experience very heterogeneous demographic development: Central Asia is expected to almost double in population and South Asia will become by far the world's most populous region, rapidly surpassing the China region. Simultaneously, the Pacific OECD countries are likely to shrink in population size and experience extreme population ageing. The proportion of the population aged 60 and above in these countries (with Japan having the greatest weight) is expected to reach 50 per cent of the total population (with the 95 per cent uncertainty interval ranging from 35 to 61 per cent). The China region will experience a more rapid speed of ageing, with the proportion aged 60 and above expected to increase by a factor of four from 10 per cent in 2000 to 39 per cent in 2100.

Keywords: population projections, probability, Asia, fertility trends, mortality trends, ageing, population composition, population growth.

**********

Future books on demographic history are likely to refer to the twentieth century as the century of world population growth and the twenty-first century as that of population ageing. There is now a high probability that we will see the end of world population growth during this century (Lutz, Sanderson and Scherbov 2001). Asia, which is home to more than half of the world's population, is a vital force in this development. This is the topic of this paper.

The outlook for Asia today is very different from the one that demographers had some decades ago when rapid population growth was the overriding concern. Today, while the population is still growing rapidly in many parts of Asia, growth has significantly slowed in other parts, and the possibility that population growth in Asia may come to an end over the course of this century is in sight.

Many studies, including the contributions in this issue, clearly show that the future level of fertility in Asia is uncertain. Fertility is not the only uncertainty in population projections; the future trends in life expectancy and migration are equally uncertain. While it was once widely assumed that life expectancy would increase in all countries to maximum levels of 80-85 years and then stay constant, recent studies show tremendous scientific uncertainty on this issue. Some people think that humankind is already close to the limit of its lifespan, while others stress that if there is a limit, it is beyond 120 years, and that already today in low-mortality countries half of the female babies born can expect to live beyond age 100 (Vaupel and Lundstrom 1996; Bucht 1996). The volume of international migration is even less predictable, being heavily influenced by unforeseeable political developments in different parts of the world.

How should population forecasters deal with these major uncertainties? One possibility is simply to refuse to make forecasts beyond two or three decades. The problem with this attitude is that there is a demand for longer-range forecasts, especially in the context of assessing the effects of other long-term changes such as global climate change. In addition, completely probability-free 'if--then' scenarios, even if long-term, do not fully satisfy this demand. Demographers have more than sufficient knowledge indicating which future trends are more likely than others. The challenge is to produce forecasts that explicitly address and even quantify this uncertainty.

To deal with such uncertainty, the currently most popular way practised by the United Nations Population Division, as well as many national statistical offices, is to produce a high and low variant as well as the medium variant, which represents the future path considered most likely. The high-low range is said to cover a 'plausible range' of possible future paths. Unfortunately, there are three serious problems with this 'variants approach'. First, the variants usually only show the effects of alternative future fertility paths, since they typically assume identical mortality and migration paths. Given the high uncertainty in these two other factors, this approach is clearly deficient. Second, it is not clear what 'plausible range' exactly means. For the user it makes a big difference whether the given range covers 100 per cent or only 60 per cent of all paths considered possible. Third, when national-level high and low variants are aggregated to regional or global-level variants, this approach becomes proba bilistically inconsistent (National Research Council 2000) because it is far from certain that all countries simultaneously follow the high or low paths. In reality, trends will deviate from the expectation in different ways in different countries. Fully probabilistic projections are the best way to deal with these problems. This paper presents the as yet unpublished results of recent probabilistic world population projections for five Asian regions, which are defined in Appendix Table A1.