Population in East Asia
Griffith Feeney and
Andrew Mason
1998-08-20
During the second half of the twentieth century the world's developing countries experienced population growth at unprecedented rates. The populations of Africa and Latin America essentially tripled. Asia's population increased from 1.4 billion in 1950 to 3.5 billion in 1995. To many observers, Malthus' warnings about uncontrolled population growth two centuries earlier seemed prescient. In South Korea, Japan, Taiwan, and the island of Java, home to most of Indonesia's population, fears of rapid population growth were exacerbated by high population densities and limited natural resources. Asian leaders, among others, were gravely concerned about the implications of rapid population growth for the development of their own countries (Tsui, C xx).
The era of rapid population growth seems almost certain to end during the 21st Century. Population growth rates in Asia, Latin America, and the world peaked during the 1960s and are declining slowly but steadily. Change came somewhat later to Africa but population growth has slowed there, as well. Within major world regions there is considerable variation, but in most countries population growth is gradually declining (Table 1). Note 1
Declining population growth rates do not, of course, imply declining population. Every country in Asia is experiencing some population growth. Japan will be the first to experience zero population growth and then population decline, probably within a decade. All other countries in Asia will experience population growth for many decades. By the middle of the twenty-first century, Asia's population is expected to exceed 5 billion and Latin America and Africa will experience even more growth, in percentage terms (United Nations, 1995).
The study countries are experiencing, in broad terms, the same demographic changes occurring elsewhere in the developing world, but they are on the leading edge. Population growth rates peaked relatively early and have declined relatively fast. Change in Thailand and Indonesia has come somewhat later than in the more developed economies of Japan, South Korea, Taiwan, and Singapore. But compared with countries at similar development levels, Thailand and Indonesia have progressed rapidly toward an end to substantial population growth.
The demographic experience of these countries is explored below with two objectives in mind. The first is to examine the underlying processes that led to the era of rapid population growth and are leading to its end. The second is to explore in more detail the demographic changes that have occurred and will continue. The first four sections of the paper emphasize the macrodemographic perspective, examining changes in the size, growth, and age structure of the population. The three subsequent sections emphasize the household or family level, examining how demographic changes are influencing the lives of three key groups: prime-age adult women, children, and the elderly.
DEMOGRAPHIC TRANSITION
In traditional societies the risk of death is high throughout life. Women bear many children, but many die in infancy and childhood. Population grows slowly or not at all. High mortality risks and low growth rates produce a “young” age distribution, with relatively many young people and relatively few old people. Fertility, mortality and age distribution are in approximate long term equilibrium. Numbers of births and deaths may fluctuate sharply from year to year, but long term averages are reasonably stable.
In modern societies, women average about two children each and almost everyone survives to old age. Population again grows slowly or not at all. Under modern demographic conditions, however, slow growth combines with low mortality to produce an “old” age distribution, with much larger numbers of elderly persons. Fertility, mortality, and age distribution are again in approximate equilibrium. Note 2
Between the traditional and modern regimes comes the demographic transition. In the simplest account, death rates begin to decline while birth rates remain high, resulting in a surge in population growth. After a time, birth rates decline as well, and population growth eventually subsides.
The process is complicated, however, by changing population age distribution. Figure 1 shows birth, death and growth rates in a simulated demographic transition modelled loosely on the experience of Taiwan. At the turn of the century, birth and death rates were nearly equal, giving a population growth rate near zero. As is conventional among demographers, rates are expressed as annual births, deaths, and net increase per thousand population.
The transition begins with declining death rates. Birth rates do not change initially, so population growth rates rise. Birth rates in Figure 1 begin to decline after about 40 years, but growth rates continue to rise for another 10 years or so until declining birth rates outpace declining death rates.
In the simulation depicted here, expectation of life at birth rises from 25 to 73 years after 90 years and remains constant thereafter. The level of fertility declines from 5.5 children per woman after 55 years to 2.0 children per woman after 90 years and remains constant thereafter.
As is apparent from Figure 1, however, the transition does not end with the stabilization of mortality risks and average number of children born per woman. Note 3
Though the population growth rate has declined substantially from it's peak 90 years after the beginning of the transition, it continues to decline for another 40 years, during which time population growth continues. The birth rate continues to decline during this period, but the death rate rises.
This unexpected rise in the death rate is explained by changes in population age distribution. Ninety years into the transition, changes in fertility and mortality have resulted in an age distribution that is much younger than the equilibrium age distribution for the current levels of fertility and mortality. The age distribution has not “caught up” with changes in fertility and mortality. The catching up process takes about 40 years, during which the population age distribution grows steadily older. This aging pushes more and more people into the oldest age groups, where mortality risks are highest, and the death rate therefore rises. Population aging lowers the birth rate as well, but the effect here is less pronounced. Note 4
The population growth that occurs during the last 40 years of this model demographic transition is a manifestation of “population momentum.” The term, adopted by analogy to the physics of moving bodies, refers most often to a tendency for population to continue to increase after fertility rates have reached replacement level. Note 5
In the demographic literature, momentum often refers only to the impact of age-structure on birth rates. The impact of age-structure on death rates has been neglected. Here and in a similar simulation for India (Lee et al., Chapter X), the impact of age structure on death rates is as important as the impact on birth rates.
MORTALITY DECLINE
From the perspective of the modern world, mortality risks in the pre-modern world were almost unimaginably high. Reducing mortality risks and improving health surely rank among humanity’s greatest accomplishments. Improvements in mortality came first to the West. Life expectancy increased gradually during the 19th century and more rapidly during the 20th century as living standards improved and scientific advance increased our understanding of health and infectious disease. The gains in the US during the first half of the 20th century illustrate the success—average life-expectancy increased from 47 years in 1900 to 68 years in 1950.
Improvements in mortality were by no means confined to the more developed countries, but in the early 1950s the gap between the developed and the developing countries was enormous. Those born in more developed countries could expect to live to their mid-sixties; those born in less developed countries could expect to reach age 40. Mortality conditions were worst in Africa and little better in Asia. Latin American countries had a decided advantage within the developing world (Table 2).
The post-World War II era has seen a rapid decline in mortality risks throughout the developing world as well as continued decline in the developed world. The gap in life-expectancy between the developing and developed world remains large, 12.1 years for 1990-94, but this is less than half the 25.5 year gap of 1950-54. The greatest gains have been in Asia, where life-expectancy has increased by 23.2 years and the infant mortality rate has dropped from 180 to 65 infant deaths per 1000 live births between 1950-54 and 1990-94.
Two survival curves, one for Taiwan in 1906 and the other for Japan in 1996, help to illustrate the significance of the extremes of East Asia's twentieth century experience (Figure 2). Both curves show the proportion of a group of persons just born that will survive to any age given prevailing age-specific mortality risks.
In early 20th century Taiwan some 35 percent of all children died before reaching their 5th birthday. From age 5, proportions surviving declined nearly linearly through age 80, indicating an approximately uniform distribution of deaths from late childhood through very old age. Very few people survived long enough to reach old age.
In contrast, virtually all children born in contemporary Japan survive to adulthood. Few people die until well into the 50s, and half of all persons survive to age 80.
Japan's 1996 life-expectancy of 80 years was the highest of any nation in the world, but no country, no matter how impoverished or deprived, had a life-expectancy in the 1990s as low as the 28 years of 1906 Taiwan. Advances in life-expectancy have been sufficiently rapid during the twentieth century that the great majority of countries have mortality conditions much closer to Japan's than to turn-of-the-century Taiwan's.
For their level of development, the study countries began the high growth era with relatively high expectation of life. In a scatter plot of life-expectancy against per capita GNP (Figure 3), Japan, Taiwan, Singapore, South Korea, and Thailand are clearly along the upper envelope. Indonesia is the only country with life-expectancy that can be characterized as typical for its income level. Note 6
The study countries are not sharply distinguished from other countries by the speed of their mortality decline. For the world, as a whole, life expectancy increased by 18 years between 1950-54 and 1990-94. Japan, Taiwan, and Singapore experienced smaller gains, South Korea, Thailand, and Indonesia larger gains. However, mortality conditions did improve more rapidly in East Asian countries than was typical of countries at similar initial levels of mortality. Japan illustrates the point with a life-expectancy equal to Europe’s in 1950-54, but almost seven years greater in 1990-94 (Table 2).
That mortality risks fell rapidly in the study countries is hardly surprising. They were among the most rapidly developing in the world, and one would expect gains in life expectancy to be among the benefits of development. Yet the same plot of life-expectancy against per capita GNP for 1990 reveals an interesting phenomenon (Figure 4). With the exception of Japan, expectation of life for the study countries is clearly low in relation to income. The gains in life-expectancy have been substantial, but they have not kept pace with the gains in income.
As the study countries were entering their high economic growth era, then, they were further along in their mortality transition than other developing countries, and they proceeded more rapidly through their mortality transition than have other countries. The gains in life-expectancy did not quite keep pace, however, with the gains in per capita income.
FERTILITY DECLINE
In 1960, the countries of the world belonged to one of two groups. Of the 113 countries with a population of 2 million or more, 81 were high fertility countries in which women averaged 6.7 births over their reproductive life span. In the remaining 32 countries women averaged only 2.8 births per woman, and many of these countries had experienced lower levels of fertility in their recent pasts but they were in the midst of the post-World War II baby boom. Note 7
Most low income countries belonged to the high fertility group, most high income countries to the low fertility group. Per capita income of the high fertility countries averaged $1260 as compared with $5100 for the low fertility countries (1985 prices). There was substantial overlap, however. Venezuela, for example, a member of the high fertility group, had a per capita income of $6300; and Romania, a member of the low fertility group, had a per capita income of $480. Within the high and low fertility groups, the level of per capita income bears strikingly little relationship to the level of fertility (Figure 5).
Locating the study countries on Figure 5 is a simple method of identifying their distinctive features at this point in history. Japan was the only Asian country to have joined the low fertility group by the 1960s. Indeed, Japan was one of only three countries with replacement level fertility. Japan, however, was a relatively new member. Its TFR had dropped below 4 births per woman by 1940, rebounded to 4.5 births per woman in the aftermath of World War II, and then dropped precipitously, reaching replacement level in the late 1950s.
The other five study countries all belonged to the high fertility group, but in 1960-64 their fertility was low in relation to other high fertility countries. In Singapore, women were averaging just under five births. South Korea, Indonesia, and Taiwan belonged to a relatively small group of countries in which women were averaging around five and one-half births. Only women in Thailand were averaging six or more births. By contrast, women in 67 of the 81 high fertility countries were averaging six or more births.
Why were fertility rates relatively low among the study countries? In Taiwan, Singapore, and South Korea, fertility decline was underway by the 1960s. During the 1950s, women in each of these countries had been bearing children at about the same rate as women in Thailand. In contrast, Indonesian fertility was at or near its peak in the early 1960s. So far as has been documented, Indonesian women never averaged six births over the reproductive span. In any event, the high fertility study countries, as a group, had less distance to travel in order to complete the transition to replacement fertility.
By the early 1990s, six of the high fertility countries of the 1960s had achieved replacement fertility (Figure 6). All are located in East or Southeast Asia: Taiwan, South Korea, Thailand, Singapore, Hong Kong, and China. Indonesian women were averaging just under three births, a level quite low as compared with countries at similar development levels in 1960. Of the thirty-six countries with a per capita income of $1000 or less in 1960, only four had completed the fertility transition by the early 1990s, China, South Korea, Thailand, and Romania, and only one of the remaining countries, Indonesia, had a TFR below 3 births per woman. After Indonesia, India had achieved the lowest fertility with a TFR of 3.7. Note 8
The difference between the fertility-income relationship and the mortality-income relationship is striking. In 1960, the study countries were, relative to their level of development, far along in their demographic transitions both with respect to fertility and mortality. Mortality changes, though impressive, did not keep pace with economic growth; fertility changes did.
Although few of the high fertility countries of the 1960s had completed their fertility transitions by the 1990s, many experienced large and rapid declines in their rates of childbearing. Nearly half of these countries experienced a drop in the TFR by 2 or more births. Evidently rapid fertility decline has been a widespread phenomenon.
To analyze speed of fertility decline further we estimate the time at which fertility decline began and the rate of decline for 42 countries. All countries which had experienced sufficient fertility decline by 1990 for a rate of decline to be reliably estimated were included. The median rate of fertility decline for these countries is slightly over one child per woman per decade. Among countries with rates of decline over the median, the countries of East and Southeast Asia had slightly more rapid rates of decline (Table 3), but the differences are very small. The comparisons show that rapid rates of fertility decline are by no means unique to Asia. Note 9
Examination of these estimates leads to two conclusions. First, while fertility declined rapidly in East and Southeast Asia, it declined nearly as quickly in many other parts of the world as well. The medians shown in the last panel in Table 3 give a quick summary comparison.
Second, the speed of fertility decline does not bear a close relationship to economic growth. The countries of Latin America experiencing a substantial drop in fertility did not have high rates of economic growth. None grew as rapidly as the high growth economies of Asia nor was their economic performance, as a group, notably distinguished from other countries of the world.
The high-performing economies of East Asia are distinctive, however, with respect to the duration of their fertility transition—the number of years required to reach replacement fertility once decline has begun. For the East and Southeast Asian countries, the median duration of the transition is only 25 years (Table 3). Elsewhere, the transition is projected to take twice as long. Only in Jamaica is the shift from high to replacement fertility projected to occur as quickly as in East and Southeast Asia.
Several factors account for the shorter duration of the fertility transition in East Asia. First, initial fertility levels were lower than elsewhere by almost one birth per woman. Second, the rate of decline was somewhat faster than in Latin America. Third, East Asian countries sustained rapid fertility decline until replacement or below replacement fertility was achieved. The pattern for many countries and for UN projections is that fertility decline slows before replacement fertility is reached. In Chile, for example, fertility declined rapidly during the 1960s and 1970s, but much more gradually since. Note 10
Despite the considerable variation in the duration of fertility decline among the developing countries they are proceeding through the transition much more rapidly than countries in the West. In the United States, for example, nearly 120 years were required to bring the total fertility rate from near 7 children per woman in 1820 to 2.1 children per woman in 1937, a rate of decline of about 0.4 children per woman per decade. Fertility in France began in the late 18th century and continued well into the 19th century, with an average rate of decline of only about 0.2 children per woman per decade. In the late developing Scandinavian countries declines were somewhat more rapid, but in no case higher than 0.7 children per woman per decade. Note 11
Achieving a long-life expectancy and replacement fertility are important milestones, but the transitions in the study countries are far from over. All continue to experience population growth, and all are experiencing substantial changes in their age structures. The next section explains why population growth is continuing, how long it is likely to last, how much growth is likely to occur, and how population age structures will change.
POPULATION MOMENTUM AND CHANGING AGE STRUCTURE
The changes in population age distribution that occur during the demographic transition are a key feature of the demographic landscape in the study countries, arguably as important as the changes in mortality, fertility, and population growth. They are important demographically because, as we show in this section, age structure is responsible for continued rapid population growth even when mortality risks and childbearing behavior are consistent with zero population growth. They are important economically because economic activities vary so substantially with age.
To describe what happens to age distribution during the demographic transition we introduce a simple model. We begin with a pre-transition population in equilibrium. The number of births each year (B) is constant and equal to the number of deaths. The population growth rate is zero. Life expectancy is 40 years, with deaths uniformly distributed over an 80 year life span. Given these fertility and mortality conditions, the age distribution declines linearly from the number of births B at age 0 to zero at age 80, as shown by the line labeled “initial” in Figure 7. The total population is given by 40B.
Suppose that in year t mortality declines instantaneously to a post-transition regime in which everyone survives to age 80. Suppose further that childbearing declines in such a way as to maintain the annual number of births at B. After 20 years, in year t+20, the age distribution of persons under age 20 will be uniform and numbers of persons over age 20 will decline linearly to zero at age 80 (the “+20 years” lines in Figure 7). After 60 years the age distribution of all persons under 60 will be uniform, and after 80 years there will be equal numbers of persons at every age. At the end of 80 years, the transition is complete. Population growth is zero; the number of births and deaths are equal; and, the age distribution does not change from the uniform distribution achieved at year t+80.
This representation abstracts from many of the details of any particular country's demographic transition, but it demonstrates several key features of the transition. First, the shift to a modern mortality regime produces a substantial, but circumscribed, increase in the total population. The population increases to “fill out” the older ages. In the particular illustration provided here, the population doubles, increasing from 40B to 80B.
Second, the duration of the transition reflects the human life span. In the stylization, the number of years the transition lasts is determined exactly by the life-span. If the shift from a traditional to a modern mortality regime is not instantaneous, as of course it is not, the transition's duration will exceed the life span by an amount that depends on how rapidly mortality and fertility decline.
Third, the transition is accompanied by enormous changes in the population age distribution. The population under age twenty increases by only 14 percent. The population aged 20-60 doubles. The population 60 and older increases in size by 600 percent.
Finally, the shifts in age structure during the transition are sequenced in a particular manner. Population growth slows from the youngest age groups up. At the outset, the population under age 20 grows more slowly than the adult population and ceases to grow entirely 20 years into the transition. Only after 20 years, does growth in the adult population begin to slow. Sixty years into the transition, the population under 60 has stopped growing and the population 60 and older is just beginning to slow.
The changes described by this highly stylized model are surprisingly consistent with the experience of the study countries. In every case we see large increases in population numbers, declining population growth rates, a filling out of the population age distribution sequenced as described above, and a prolonged period during which approximate equilibrium between population age distribution and levels of fertility and mortality is re-established.
There is however one important aspect of age distribution change during the demographic transition that the model does not capture. As mortality risks fall and life expectancy increases at the beginning of the transition, population age distributions become younger—not older. Because mortality risks in infancy and childhood are extremely high, as illustrated in the Taiwan survival schedule in Figure 2, survivorship of children tends to increase more than survivorship of older persons.
The initial impact of mortality decline is thus to make the population younger. In the simulated demographic transition of Figure 1, the percentage of population aged 0-14 rises from 35.2 percent at the beginning of the transition to 41.5 percent 55 years into the transition. Only when fertility decline begins does youth dependency turn down, though the decline proceeds much more rapidly than the increase. The youth dependency burden thus rises in the early stages of the demographic transition, with important implications for economic growth and development (Williamson and Higgens, Chapter XX).
The example of Thailand shows how closely our highly stylized model of changing age distribution approximates the empirical experience of the study countries (Figure 8). The age distribution in 1970 was very young, the consequence both of rapid population growth in earlier decades and high mortality in more remote decades. Fertility fell sharply in the 1970s and 1980s, however, so that numbers of births were roughly constant during these decades and the number of persons aged 0-4 in 1990 was not much different (slightly less, in fact) from the number 0-4 in 1970. Note 12
At present and in future decades, population growth in the study countries, Indonesia excepted, will be driven primarily by this population momentum inherent in current age distributions. Replacement fertility has been achieved, but birth rates and death rates have not yet converged because of the age structures of the populations. Japan's population is currently growing very slowly and its population is projected to peak early in the next century. At that point, Japan will enter a period of de-population. By 2050, Japan's population is projected to be about 10 percent smaller than in 1990. (Table 4).
Populations of the other high-performing economies will be substantially larger in 2050 than in 1990. The populations of South Korea, Taiwan, and Singapore are projected to increase by twenty to thirty percent, Thailand's by more than fifty percent, and Indonesia's by almost 75 percent. By the middle of the twenty-first century, population growth is projected to be small, zero, or negative depending on the country.
Changing age structure has important implications for economic analyses. Of particular interest to broad assessments is the dependency structure of the population. Any human population consists of three broad age groups: those of working age whose efforts are primarily responsible for economic production; and, children and elderly who are economically dependent on prime-age adults. The evolution of the dependency structure over the demographic transition is illustrated by South Korea's experience (Figure 9).
Two important changes in dependency structure characterize South Korea and other countries going through the demographic transition. The first is the long swing in the relative size of the working age population. Over the demographic transition, the percentage of Korea's population in the working ages increases from 44 percent in the mid-sixties to an anticipated peak of 64 percent in 2005. This change represents an enormous increase in the "productive potential" of the South Korean population over a period of four decades. This very high level is not permanent, but the projected level in 2040 is still well above the pre-transition level. Were the projection extended beyond 2050 the decline in the working age population would continue until about half of all members of the population were in the working ages.
The second change apparent in Figure 9 is a shift in the dependency structure from one in which youth dominate to one in which youth and the elderly are roughly in balance. In 1960, over 94 percent of dependents, were under the age of 20. By 1990, this figure had fallen to 88 percent. By 2050, the number of elderly dependents will nearly equal the number of young dependents.
The changes in age structure in the study countries are summarized in Table 5. Between 1960 and 1990, four of the high-performing economies experienced substantial increases in the concentration of their populations at the working ages. The increases for Singapore, South Korea, Taiwan, and Thailand ranged from ten to seventeen percentage points. The smaller increase for Japan reflects the particular time span selected. Japan experienced a substantial increase during the 1950s. The increase in Indonesia has been modest to this point, because Indonesia's transition is occurring somewhat later than in the other countries.
Values for the world's regions provide some comparative context for the East Asian experience. Many countries in Africa are sufficiently early in their demographic transitions that the proportion of the population under 20 is rising and the proportion of working age declining. Other countries of the world, on average, experienced modest increases in the percentages of working age. Even in Europe, the growth of the elderly population was insufficient to reverse the trend. The increase in North America was relatively large and can be traced to the entry of baby-boomers into the working ages.
Comparing the country values of East Asia with regional averages overstates the extent to which the large increases in working age populations were confined to East Asia. Thirty-three countries, about 20% of the 150 with a 1990 population of 1 million or more, experienced an increase in their working age population of 5 percentage points or more. Nine increased by 10 percentage points or more, including several non-Asian cases.
The shift in dependency away from one dominated by youth is readily apparent in Table 5. The percentages under the age of 20 have declined steeply in all study group countries, except Indonesia. The percentages 65 and older have begun to increase, but, as noted above, most of the relative growth in the older population comes late in the demographic transition. Japan is much further along with twelve percent 65 and older in 1990. Moreover, the increase in the percentage elderly in Japan between 1960 and 1990 was greater than in any country in the world.
The macrodemographic perspective maintained in this and the preceding sections identifies important changes in the size, growth, and age structure of the population with clear and potentially important links to the macroeconomy. Other important aspects of the demographic transition are not easily described at the aggregate level. The following sections undertake a more micro-oriented approach looking at demographic changes as they relate to three important groups: prime-age adult women, children, and the elderly.
WOMEN
Demographic changes over the transition are of fundamental importance to the lives of women. Little more than a generation ago, women in many East Asian countries were averaging six or more births during their lifetime. Childbearing began in the teens and early twenties and extended into the late thirties and forties. Given the short life-expectancies, women could expect to devote a major portion of their adult lives to bearing and rearing children.
Today, young women in many East Asian countries are averaging two or fewer births during their lifetime. Often they are delaying marriage and childbearing to their late twenties and thirties. Instead of early marriage they are opting for more schooling and for work, changes explored more fully in Bauer (chapter x) and Phananiramai (chapter y).
The decline in childbearing during the transition was accompanied by a substantial delay in marriage and childbearing in all of the study countries. Women, on average, are marrying about 2 years later in Japan and Thailand, 3 years later in South Korea and Indonesia, and 5 years later in Singapore and Taiwan ( Table 6). Note 13 Japan, Taiwan, and Singapore are now among the latest marrying populations in the world. There is persuasive evidence that increasing percentages of East Asian women may never marry (Jones, 1997).
Consensual unions and childbearing outside of marriage are relatively rare in East Asia, so the delay in marriage has been closely matched by delayed childbearing. Because women are bearing fewer children, childbearing has become increasingly concentrated. In Japan, Taiwan, and South Korea, most women are bearing children in their late twenties and early thirties (Table 7). Of women aged 25-29, over one-third in South Korea and Taiwan and over one-half in Japan had not begun childbearing. Indonesian women commence bearing children at a younger age—half of all women in their early twenties were mothers in 1990. None the less, the shift to later childbearing ages is unmistakable even in Indonesia.
All of the study countries have experienced rapid decline in the number of births, but there are considerable differences among the countries in the trends in the number of surviving children. In the early stages of the transition, child mortality rates are so high that women have few surviving children. Depending on the speed of mortality and fertility decline, the number of surviving children may increase substantially. Only when child mortality rates reach low levels, will the trend in the number of surviving children necessarily follow the trend in childbearing.
The number of surviving children in a family rises in the early childbearing years, during which children are born more rapidly than than they die. As the mother passes through her 30s, however, deaths of children already born tend to outnumber new births, and the number of surviving children declines (see Bongaarts, 1987, for simulation results). There is no fixed “number of surviving children” for a woman and it is necessary to devise a suitable index.
We use the average number of surviving children for women aged 35-39 and have data for Indonesia, Japan, and South Korea ( Figure 10). In contrast to the downward trend in fertility emphasized above, all three countries experienced an extended period during which the average number of surviving children was increasing.
In Japan, for example, the number of surviving children increased during the first half of the twentieth century reaching a peak for women who were completing their childbearing around 1940. Over a period of thirty years, family size dropped from over 4 children per woman to 2 children per woman. Since 1970, five successive cohorts of women have averaged close to 2 children each.
The trends in South Korea and Indonesia are similar. Both countries experienced an extended period during which the number of surviving children increased followed relatively recently by a decline. The timing varies considerably from country to country. In South Korea, family size began to decline from about 4.5 children per woman in the 1960s. But even as late as 1975, women who were completing their childbearing average 4 children. Only women who were about to complete their childbearing in 1990 had close to 2 surviving children.
The Indonesian experience shows an even longer period during which family size was rising, although the increase was more gradual than in either South Korea or Japan. The decline in family size among women who have completed their childbearing is in its early stages in Indonesia, having dropped only 0.5 children from the 1980 peak.
These figures belie simple characterizations of changes in childrearing responsibilities of women during the era of high economic growth. This is clearer still if we use data from Figure 10 to calculate the average number of surviving children to women of working age (20-64) from 1960 to 1990 Table 8.
In Japan, women of working age in 1960 averaged fewer than three children. The average dropped to 2 children in 1990. South Korean women of working age were near their peak in 1960, with a completed family size of 4.2 children. The decline was much greater than in Japan, 1.5 children per woman during the thirty year period. Indonesian women experienced a modest rise in family size between 1960 and 1970. At the peak, their average was well below the average found in South Korea. Significant decline has been confined to the last fifteen years during which the average number of children dropped by 0.6 children. Note 14
To summarize and to extrapolate to the experience of the region as a whole, working age women in South Korea, and probably Taiwan and Singapore, experienced a substantial decline in childrearing responsibilities during the high growth era, 1960-90. Change in Indonesia has been more recent. Thailand's experience no doubt falls somewhere between that of the South Korea and Indonesia. In Japan, the childrearing responsibilities of women were already substantially diminished by 1960 and have continued to decline.
CHILDREN
Investment in “human capital” is widely regarded to have been an essential element of Asian economic success. At the outset of the high growth era, the countries in our study had relatively healthy and well-educated populations. Moreover, health status and rates of educational attainment improved rapidly during the high growth era. The linkages between population and human capital formation are examined in some detail in this volume by Jensen and Ahlburg (chapter x), Ahlburg and Jensen (chapter y), Huang (chapter z), and elsewhere by Lloyd and Montgomery, 1996 among others.
In this section, we identify changes in demographic characteristics of the family that may bear on human capital or the quality of children. To do so requires a shift in perspective from that of the mother to that of the child. In the process we consider several key questions: Has there been a significant decline in the average size of the families in which children are being raised? How recently did that decline begin? Was it preceded by a period during which the average family size increased? Are there other important aspects of the family in East Asia that might have a bearing on their economic success?
Answering these questions is more difficult than one might at first suppose, for several reasons. First, if we follow a birth cohort through their childhood, their average family size changes as siblings are born and die. For some purposes, such as assessing competition among siblings for mother's time, family size at young ages is of greater interest. For other purposes, such as assessing competition among siblings for financial resources, family size at older, more costly ages, is of greater relevance. Because of the limited availability of data, we rely on measures of family size for older children whose mothers have completed their childbearing. Data available for the US suggests, however, that family size at older ages is highly correlated with family size at younger ages. Note 15
The second complication is that examining demographic variables from the point of reference of children involves different calculations and different results. In determining the average size of families from a mother's perspective, each family, large or small, carries equal weight. When children are the unit of analysis, however, each child carries equal weight. Because there are more children in larger families, the average size of the sibset, as demographers refer to it, is greater. In a population with high variance in childbearing, average sibset size will substantially exceed the average number of children. Note 16
Average sibset size can be calculated for South Korea and Indonesia using published data reporting the number of ever-married women by their age and the number of surviving children. (For similar data for Latin America and sub-Saharan Africa, see Lloyd and Desai, 1992.) Values for South Korea and Indonesia are plotted in Figure 11.
Sibset size substantially exceeds the average number of surviving children reported above, but the trend is quite consistent with the changes in the number of surviving children. South Korea experienced an extended period during which successive cohorts of children were raised in larger families. But during the last two decades completed sibset size has declined by about 1.5 members—those entering the labor force in the early 1970s had 5.5 children in their families and those entering in the early 1990s had 4 children in their families. In the absence of a resumption of high fertility, further decline in completed sibset size to under three children is imminent.
The Indonesian experience differs from the South Korean experience in two important respects. First, a significant decline in sibset size has not yet begun. The value for those entering the labor force in the early 1990s is only slightly below the peak of the early 1980s. Second, the rise in average sibset size prior to 1980 was much more gradual in Indonesia than South Korea's earlier decline. This may reflect differences in the pace of mortality change, but it may also reflect the greater variance in Indonesian family size as compared with South Korea.
The Indonesian and South Korean results taken together show that the decline in sibset size is so recent that any significant impact on the human capital or quality of the labor force prior to 1990 is unlikely except in Japan. The average sibset size of entering workers in Indonesia is much greater than the sibset size of departing workers, those who entered in the 1950s. Hence, the size of the family of origin of the working age population is rising, not declining. Moreover, it appears likely that this has been the case for some time. The rise could not have been particularly great, but there is no evidence that current workers were raised in smaller families than in the past, thus enhancing their human capital formation.
South Korea is further along in the transition to smaller families. But even there, the average quality of the labor force has not been influenced in any important way, to this point, by the decline in sibset size. The average working age adult in 1990 had 5.0 siblings in his or her family, as compared with 5.2 siblings in 1980 and 4.8 siblings in 1970. Note 17
Perhaps in Japan, the overall quality of the labor force has been influenced in important ways by changes in sibset size, but probably nowhere else in Asia. There may of course be other channels by which demographic changes influence human capital formation. Changes in age structure may have affected educational budgets, for example. Delayed or reduced childbearing may have influenced educational decision making by women.
Demographic features of the family, other than the number of children present, also may have an important bearing on human resource development. Age at marriage and first birth are relatively high and increasing so that few children are born to young women. Births outside of marriage are rare and rates of divorce are low; hence, children are more likely to be members of intact and, possibly, more economically viable family units. Many children are raised in extended households so that they are less subject to swings in available resources that otherwise naturally arise over the lifecycle.
Of these factors, the adverse health impact, for mother and child, of either early or late childbearing has been most extensively investigated (Lloyd, 1995; Montgomery and Lloyd, 1996). Much less is known about the direct human capital impact of the other demographic features. There are clearly identifiable ways, however, in which the resources available to children are enhanced when they are raised in intact and/or extended households. Because these conditions may have contributed to the human resource advantage held by many countries in East Asia, a brief review is warranted.
As discussed above, women have delayed marriage and childbearing so that relatively few children are born to young women. By 1995 in Taiwan, for example, only 5% of all births were to women under the age of 20 and only one-quarter were to women aged 20-24. Nearly two-thirds were to women aged 25-34 (Taiwan, 1996). In India, nearly one-quarter of all births were to teenagers and nearly 40% to women in their early twenties (IIPS, 1995). Of the study countries, only young women in Indonesia are engaged in childbearing to any significant extent. About 15 percent of all births were to teenagers in 1988-91 as compared with 18 percent in 1967-70.
Several demographic forces insure that a high percentage of children live with both parents. Few children are born to unmarried women, parents are unlikely to divorce, and, given high rates of life expectancy, both parents usually survive until their children are grown. In recent years, the percentages of children under the age of 15 living in a household headed by a lone parent ranged between 5 and 6% for females in Japan, Korea, Singapore, Thailand, and Indonesia and between 5 and 8% for males in those same countries. A somewhat higher percentage of boys (12%) and girls (10%) lived with one parent in Taiwan in 1990. In contrast, 28% of American children under the age of 18 did not live with both parents (including step-parents) and 49% did not live with both birth parents (Hogan and Lichter, 1995).
A final issue is how changes in the extended family system may have influenced the resources, time and financial, available to East Asian families engaged in childrearing. The extended, multi-generation family potentially serves an important role in human capital formation. Childbearing and human resource investment is heavily concentrated within the life cycle and, in a nuclear family system, falls most heavily on young parents whose income and wealth may be low compared to what they can expect later in life. In the absence of constraints on indebtedness, parents can accumulate debt in order to invest in their children. However, the existence of capital market imperfections may lead parents to under invest in their children in the absence of other institutional arrangements (Becker and Tomes, 1976). The multi-generation, extended family has historically been the institution that makes resources available to children without relying on capital markets (Mason and Miller, 1996).
The extent to which children are living in three generation rather than nuclear households has been relatively stable. In Japan, the proportion of children under the age of 15 living in nuclear households increased very modestly from 64% in 1970 to 69% in 1990. In South Korea, children are also somewhat more likely to live in nuclear households than previously (Feeney and Mason, 1997).
In Taiwan, by contrast, the proportion of children under the age of 15 living with grandparents increased from 24% in 1976 to 28% in 1993 (author’s calculations based on Family Income and Expenditure Surveys). In other respects, the Asian household is undergoing substantial change, but when it comes to childrearing it is the resilience of the extended family rather than its deterioration that is most noteworthy. Low mortality, by increasing survivorship of grandparents, and low fertility, by reducing the extent to which siblings must share grandparents, may be contributing to the continuation of the extended family by influencing the extent to which populations can accommodate three generation stem families.
THE ELDERLY
The rapid growth of the elderly population is one of the key features of the demographic transition emphasized in our description of macrodemographic change in East Asia. The demographic transition has important implications for the life circumstances of the elderly that bear on their economic welfare, income inequality (Oshima, chapter 5), saving rates (Williams and Higgins, chapter 6; Lee et al., chapter 7; Toh, chapter 8), and rates of economic growth.
The decline in mortality over the demographic transition insures that a much larger portion of the population survives to old age and that those who do will live longer. The survival curves for 1906 Taiwan and 1996 Japan, charted in Figure 2, show the extremes in the chances of surviving to old age. In a high mortality population fewer than one in five survive to age 60; in a low mortality population more than nine in ten reach old-age.
Until the later stages of the transition, the increases in the expected years of life for those who survive to old age are more modest than may be appreciated. Most of the early gains in life-expectancy lead to more years lived at young ages not old ages. In Taiwan, for example, life-expectancy at birth for females increased from 65.4 years in 1956 to 77.8 years in 1994, an increase of 12.4 years. Only a little more than a third of that increase, 4.5 years, came from a rise in the number of years lived at old age. In 1956, women who survived to age 60 could expect to live an additional 16.9 years as compared with 21.4 years in 1994. As life-expectancy reaches higher levels, the gains are much more heavily concentrated at the older ages. For example, eighty percent of the increase in female life-expectancy between 1990 and 1994 in Taiwan was concentrated at ages 60 and older (DGBAS, various).
With population aging the institutional arrangements by which the elderly gain access to resources become increasingly important to their well-being. Family support systems are dominant in traditional societies, but in modern settings new institutions evolve that supplement or, to some extent, supplant the family. The demographic forces influence the family support system in very direct ways. In a traditional setting, elderly often have no surviving spouse and must rely on their few surviving children or grandchildren. In a modern setting, elderly are much more likely to have a surviving spouse and as many surviving children as in a pre-transition setting. During the transition, the elderly are more likely to have a surviving spouse and they have more surviving adult children.
The elderly in East Asia were less likely to be widowed and more likely to be married in 1990 than earlier. Singapore had the smallest gain in the percentage currently married for women, 3 points, and Japan and Taiwan had the largest gains, with 9 and 20 percentage points, respectively ( Table 9). The especially large increase in Taiwan is probably a consequence of the unusual sex ratio discussed above. Even with these gains, a minority of elderly women are married.
In 1970, elderly men were two to three times more likely than elderly women to have a surviving spouse. Except in Taiwan, between seventy and eighty percent of men 65 and older were married. In Singapore the percentage of elderly men married declined slightly, but elsewhere the percentage married increased between 1970 and 1990. The gains were somewhat smaller for men than for women except in South Korea.
Adult children play an important role in any viable family support system for the elderly and for elderly women in particular. Confucian traditions heavily influence family obligations in Japan, South Korea, Taiwan, Singapore, and, to some extent, Thailand. Filial piety, the obligation of son to father, is an important organizing principle that guides many family relationships including living arrangements, lines of authority, and inheritance and inter vivos transfers. A woman marries into the family of her husband and her position within the family is defined by the position of her husband, even after his death (Cho and Moto, 1994; Fricke et al., 1994).
The elderly are experiencing an enormous rise and fall in the number of surviving adult children as documented for South Korea, Japan, and Indonesia in Figure 10. In South Korea, demographic change from 1960 to 1990 led to a large increase in the number of children. In 1960, women who were 65-69 had 3 surviving children (the value plotted for women 35-39 in 1930 in Figure 10). By 1990, women aged 65-69 averaged 4.5 surviving children, the largest number in South Korea's history. In the coming years, however, the number of surviving children will be declining and by 2020, women in their early sixties will have about 2 surviving children. Note 18
Elderly women, those 65-69, in Japan experienced an increase in the number of surviving children until the 1960s and early 1970s. Since then the number of surviving children has declined. Women reaching their 65th birthday between 1995 and 2000 are the first cohort in this century to average only two surviving children. Indonesia represents the other extreme among the study countries. The trend in the number of surviving children is upward for the elderly and will continue to be until early in the 21st century.
It is difficult to assess the extent to which demographic changes and other social and economic forces are affecting the family support system in Asia. Data on living arrangements, however, suggests that the family continues to play an important role but one that is slowly eroding. The great majority of the elderly live in a family household. Roughly speaking, at least 9 our of 10 elderly men and at least 8 out of 10 elderly women are living with family members. The sole exception to this generalization is Taiwan with its large immigrant elderly male population. Rates of institutionalization are relatively low. Those living outside of family households typically live alone. Women in Indonesia, South Korea, and Japan were most likely to be living by themselves, but much less so than in the US where 37 percent of elderly women lived alone in 1990 (US, 1992). Note 19
In some Asian countries a shift towards non-family living arrangements has emerged during the last few decades. In South Korea and Japan, the percentage of elderly living in family households has declined. The changes for men have been relatively modest; for women, they have been more substantial. In both countries, a rise in the percentage living alone and, to a lesser extent, in institutions are responsible for the decline in family living arrangements.
The elderly in Japan and Taiwan are also moving away from multi-generation living arrangements. In Japan, the percentage of elderly males living with children declined from 72 percent in 1970 to 54 percent in 1990. The percentage of elderly women living with their children declined from 70 percent to 52 percent during the same period. In Taiwan, the percentage of elderly men living with children or any family member from a younger generation declined from 80 percent to 62 percent between 1976 and 1993. For elderly women, the percentage declined from 87 percent to 72 percent.
South Korean elderly are also shifting away from multi-generation living arrangements. The proportion of elderly men and women who were the parent of a household head declined while the percentage who were the head or spouse of the head increased between 1970 and 1990. Moreover, the average number of children (0-19) and prime age adults (20-64) living in households headed by the elderly declined precipitously in South Korea between 1970 and 1990.
In neither Thailand nor Indonesia do we find evidence of a shift away from multi-generation living arrangements. In Thailand, the proportion of elderly who lived in a household headed by one of their children increased between 1970 and 1990; the proportion who headed their own households declined. The average number of prime age adults living in households headed by the elderly increased between 1970 and 1990. In Indonesia, the proportion of elderly heading households increased modestly while the proportion who lived in a household headed by a child declined by about five percentage points during the 1980s. The number of prime age adults in households headed by the elderly did not change ( Table 10).
To this point, the extended family remains an important source of support for the elderly in Asia. Many elderly continue to live with their children. Many live near their children and often receive financial and personal support. However, shifts away from the extended family are clearly underway in the more developed East Asian countries. The changes are most evident in Japan and South Korea. In Taiwan changes are coming at a somewhat slower pace. In the least developed of the high-performing economies, Thailand and Indonesia, there is little to indicate a shift away from a family-based support system.
CONCLUSION
The demographic experience of the countries of East Asia is of special interest. They were the first countries to make the transition from high fertility to replacement fertility during the post-World War II era. That experience is much more thoroughly documented by censuses, surveys, and vital statistics than the same transitions that occurred much earlier in the West. Thus, East Asia provides the first real opportunity for detailed study of the first two phases of the demographic transition.
The broad outlines of demographic transition are similar across countries, but there are great differences in the timing, the speed, and the magnitude of change. Demographic events in the developing world are very different than earlier changes experienced in the West. In the developing world, mortality rates dropped much more rapidly, population growth rates reached much higher levels, the swings in age structure have been much more dramatic, and in many countries childbearing dropped much more precipitously. Post-World War II baby booms were modest and short-lived as compared with the experience of the US and many other countries in the West. Because of these differences, the impact of population in the West may be a poor guide to what might be expected elsewhere.
Differences between East Asia and other countries of the developing world are also considerable. East Asia held an “advantage” in the 1950s. Life-expectancy was already relatively high and rates of childbearing were relatively low. Moreover, East Asia proceeded relatively rapidly to the high levels of life expectancy currently prevailing. Rapid growth with equity, declining poverty, increased education, and other aspects of development, clearly contributed greatly to the rapid improvements in mortality. The shift to low fertility was even more dramatic in East Asia. Development no doubt played an important role, but fertility levels are much more loosely connected to development than are mortality levels. The countries of East Asia that experienced rapid fertility decline also had strong policies and well-funded programs aimed at reducing childbearing (see Tsui, ).
In addition to explaining the ways in which trends in the demography of East Asia are distinctive, this chapter describes the most important ways in which the demographic underpinnings of the economic have been influenced. Six changes stand out.
First, the countries of East Asia experienced an enormous increase in their populations. Between 1950 and 1995, every country but Japan saw their population double or triple in size. Japan, South Korea, Singapore, and Taiwan and Java, home to most of Indonesia's population, are among the most densely populated countries in the world.
Second, population growth is slowing. Japan's population will cease growing and begin to decline within the next decade. The other countries in the region will experience a substantial amount of additional population growth, but growth rates are declining and the end of the high population growth era is clearly within sight.
Third, the region has experienced large swings in child dependency, evident at both the aggregate and the family level. With gains in child survival, the percentage of the population who were children increased substantially. Likewise, women experienced a rise in the number of surviving children to unprecedented levels. The rise was followed by an even more precipitous decline in child dependency as reduced rates of childbearing overwhelmed further gains in child survival.
Fourth, the productive potential of the populations of the region have increased enormously during the last three decades. In 1960 about 40-45 percent of the population fell in the working ages. By 1990, about 60 percent were concentrated into the 20-64 age span in Japan, South Korea, Taiwan, and Singapore. In these countries, the number of working age persons increased by a third because of changes in age structure.
Fifth, changes in marriage and family are increasingly apparent. Age at marriage has increased to unusually high levels and many observers express concern that many young adults in Japan, South Korea, Taiwan, and Singapore will never marry. The elderly are increasingly likely to live independently of their children. However, in many respects, the traditional family has proven to be a resilient institution. Except in Japan, young adults rarely establish separate households until after they marry. Few children are born to single mothers. Rates of divorce are relatively low. Many elderly continue to live with their children.
The sixth change is population aging. A basic feature of human life, the length of life, has changed. In the past, reaching old age was the exception. Now it is rule. In the past, a small portion of the population included the elderly. Now, the elderly are the fastest growing segment of the population. In Japan, the number 65 and older exceeds the number under 15. The other countries of East Asia are following the same path as Japan.
The demographic transition in East Asia is far from complete. The population is younger than is consistent with the end of the transition. During the first half of the 21st century, the countries of East Asia other than Japan will experience substantial increases in their total populations. This growth will be concentrated at older ages leading to a major shift toward an elderly population. This is very similar to what is being experienced in the West, but aging in East Asia is occurring more rapidly.
NOTES
Note 1 Population data in this chapter not otherwise referenced are from the digital version of United Nations (1995). There are occasional insignificant differences, probably due to rounding errors, between the digital data set and the published volume.
Note 2 Not all traditional societies are demographically the same, of course, and all experience fluctuations in levels of fertility and mortality. The same is true for modern societies. The differences between traditional and modern demographic regimes are far larger than the variations within them, however, and the dichotomy stands easily in the face of empirical variability. The level of fertility in traditional societies almost never falls below 4 births per woman, for example, and the level of fertility in modern societies almost never rises above 3 births per woman. In traditional demographic regimes 20 percent or more of all children born typically die before reaching their first birthday and this figure would rarely fall below 15 percent. In modern demographic regimes the same figure would almost never be as high as 5 percent.
Note 3 “Stabilization” is a considerable, if justifiable, simplification. In the centuries long perspective of the demographic transition, levels of fertility and mortality in developed countries have “stabilized.” Changes that appear small in this context, however, may loom very large when focusing on the current situation in developed countries. Mortality risks continue to decline, and there is the possibility that the length of life may be considerably extended. Fertility levels in developed countries fluctuate substantially, if at much lower levels than in the more distant past. The level of fertility in many developed countries falls short of population replacement, giving rise to concerns of population decline and excessive population aging.
Note 4 Stylized depictions of the demographic transition appear prominently in the demographic literature, but the increasing death rates that occur during roughly the last third of the transition have generally been ignored or stylized out of the picture. See for example Coale (1974:23), Cleland (1994:232), Keyfitz (1977:24), McNamara (citing Keyfitz 1977, 1982:146), Nam and Philliber (1984:43), Stockwell and Groat (1984:37), Thomlinson (1976:23), and Weeks (1989:74).
Note 5 More generally, population momentum refers to population change that results from an imbalance between current age distribution and current levels of fertility and mortality. See Preston and Guillot 1998.
Note 6 Unless otherwise noted to the contrary, income data comes from the Penn World Tables, Version Mark 5.6a (Summers and Heston, 1991 updated).
Note 7 Several with more than 2 million population in 1960, e.g., Cuba, Lebanon and North Korea, are excluded from this analysis because no income data are available for them in the Penn World Tables. The dividing line between high and low fertility countries, rather clearly indicated in Figure 5, is 4.5 children per woman. The figure of 6.7 births per woman is the median TFR for the 81 countries with TFRs greater than this level. The figure of 2.8 births per woman is the median TFR for countries with TFRs lower than this level.
Note 8 Were income data available for Cuba, it might be included on the list of high fertility countries that had achieved replacement fertility by the early 1990s. Whether Cuba should be included as a high fertility country is ambiguous. In 1960-64, Cuba's TFR was above the 4.5 births per woman dividing high and low fertility countries in 1960-64. In the 1950s, however, was well below this value.
Note 9 Using the estimates prepared by the United Nations Population Division, total fertility rates for the quinquennial periods 1950-54 through 1990-94 were plotted for all 113 countries. Taiwan, counted in the UN estimates as part of China, is counted here as a separate country using data from the Taiwan Demographic Fact Books. Forty-two countries experienced pronounced fertility decline for at least the three five year periods prior to 1990, which we took as the minimum sufficient to compute a reliable rate of decline. The cutoff was set at 1990, rather than at 1995, because limited data available for the 1990-94 period at the time the estimates were produced means that the 1990-94 estimates are in significant part extrapolations from earlier data.
For each of these 42 countries the period of fertility decline was determined judgmentally from the plots and a straight line fitted to the relevant points by least squares. The rate of decline is minus the slope of this line, expressed in units of children per woman per decade. The initial level is taken to be the average of the points from 1950-54 through the beginning of the decline, or the 1950-54 point if the decline began in this period or before. The time at which fertility decline began is estimated as the time at which the fitted decline line was at the initial level.
The median rate of decline for these 42 countries is 1.04 children per woman per decade. Twenty-three countries had a rate of decline greater than one and a 1990-94 TFR of less than 4 children per woman. Data for these countries are shown in Table 3. The table also shows data for Malaysia, which had a decline of slightly less than 1 (0.94) child per woman per decade.
Note 10 The year in which the fertility transition is completed is calculated using UN estimates and projections of the total fertility rate (United Nations, 1995). The exact year is estimated by linear interpolation based on rates reported for five-year intervals. The duration of the fertility transition is then calculated as the difference between the year replacement fertility is achieved and the year in which fertility decline began.
Estimates of the duration of the fertility transition may prove to be inaccurate for countries that have not yet achieved replacement fertility. Recently available data (Population Reference Bureau, 1998) indicates that Mexico may complete the transition more quickly than projected. In contrast, Jamaica will not achieve replacement fertility as rapidly as projected. Despite the uncertainties, fertility transitions accomplished in under three decades are clearly beyond the reach of any of the countries outside of East Asia.
Note 11 Systematic comparison with western experience is problematic on account of limited data and scattered sources. Much of the work in historical demography of the west, including the well known Princeton Fertility Study, provides little or no data on total fertility rates. The compilation by Mitchell (1975) includes crude birth and death rates and infant mortality rates, but not total fertility rates. We have used Keyfitz and Flieger (1968) for France and Sweden, Brunborg (1988) for Norway, Coale and Zelnick (1963) for the United States, and Lutz (1987) for Finland.
Note 12 Note that each point of the 1970 distribution is moved to the right by 20 years, because everyone who survives gets 20 years older, and down, because some people do not survive. When mortality is low, however, nearly all persons survive until reaching the older age groups, so most movement is over rather than down.
It is of course possible that demographic developments in Thailand over the next 40 years will not unfold so neatly as these projections suggest. Fertility may fall to and remain at levels substantially below two children per women, as has happened in many developed countries over the past half century. In this case there will be fewer young persons than are shown in Figure 8. Increases in fertility are less likely, but would have the opposite effect.
The illustration serves better than one based entirely on historical change, however, because particularistic circumstances (notably war) create distortions which, while they do not affect the fundamental character of the change, do make it more difficult to discern.
Note 13 Age at marriage is measured using the “singulate” mean age at marriage (Hajnal, 1953).
Note 14 Some details of the nature of source data and method of plotting in Figure 10 may be useful. Many population censuses, including those taken in Indonesia and South Korea, include a question on number of surviving children. These data are typically used to produce tables of women by age and number of surviving children, from which average numbers of surviving children to women in each age group may be calculated.
For Korea, censuses are available at five year intervals from 1960 through 1995, yielding a time series of average numbers of surviving children to women aged 35-39.
For earlier years we use older women in 1960. The average number of surviving children to women aged 35-39 in 1955, for example, is estimated as the average number to the same cohort in 1960, i.e., the average number of surviving children to women aged 40-44 in 1960, and similarly for earlier years.
The values estimated from older women are of course biased low, since some children have died between the reference time and the subsequent census. Comparison of estimates from successive censuses shows, however, that the impact of mortality is modest at these ages and does not effect our conclusions.
The situation for Indonesia is very similar, for although censuses have been taken at 10 year intervals, large nationally representative surveys are available from which similar information can be obtained.
Data on surviving children is not available for Japan, for which the indirect methods described in Mason (1987) have been used.
Note 14aThe calculations in Table 8 include estimates for women who were younger than 35-39 in 1990. We assume that these women bear no additional children if they have reached or exceeded 2 children in South Korea or 2.5 children in Indonesia. Younger women are assumed to stop at 2 children in Japan and South Korea and 2.5 children in Indonesia. Assuming that women of childbearing age stop at 3 children in Indonesia increases the family size reported in Table 8 by 0.2 children in 1990, by smaller amounts in 1980 and 1985, and not at all in 1975 and earlier.
Note 15 US estimates are available for the 1930s through the 1960s, a period during which family size was comparable to East Asia's. For each cohort of children, the average family size (size of the sibset to be more precise) increased by about 20% between the preschool years and the teen years. Calculation based on Hernandez, 1993, Table 2.4 p. 43).
Note 16 A simple example illustrates this. Consider a population consisting of three families with 0, 2, and 4 children. The average family has 2 children, but children are being raised in families with an average of (2x2 + 4x4)/6 = 3 1/3 children. However, in a population in which all women bear 2 children, sibset size will also be 2.
Note 17 These values are calculated using the age distribution of the adult population and the values reported in Figure 11 to calculate a weighted average.
Note 18 This discussion is based on estimates of the number of surviving children for women who are middle aged or older. No account is taken of the impact of adult mortality on the number of surviving children. See discussion with regard to Figure 10 in the text.
Note 19 Estimates for Asian economies presented in this section are based on published census reports and special tabulations prepared by the second author. Details are available on request.
REFERENCES
G.W. Barclay, Colonial Development and Population in Taiwan Princeton, New Jersey: Princeton University Press, 1954.
Becker, Gary S. and Kevin M. Murphy. 1988. The family and the state, Journal of Law and Economics XXXI:1-18.
Becker, Gary S. and Nigel Tomes. 1976. Child endowments and the quantity and quality of children. Journal of Political Economy 84:S143-162.
Bongaarts, John. 1987. The projection of family composition over the life course with family status life tables. Pages 189-212 in John Bongaarts, Thomas Burch, and Kenneth Wachter, Family Demography. Oxford: Clarendon Press.
Brunborg, Helge. 1988. Cohort and period fertility for Norway: 1845-1985. Rappporter 88/4 Central Bureau of Statistics of Norway.
Cleland, John. 1994. Different pathways to demographic transition. In Francis Graham-Smith, ed., Population—the complex reality: A report of the Population summit of the world's Scientific academies. Cambridge: University Press.
Coale, Ansley J. 1974. The history of world population. In The Human PopulationA Scientific American book. San Francisco: W. H. Freeman and Company.
Coale, Ansley J., and Melvin Zelnick. 1963. New Estimates of Fertility and Population in the United States. Princeton, New Jersey: Princeton University Press.
Lee-Jay Cho and Moto Yada, eds. 1994. Tradition and Change in the Asian Family. Honolulu: East-West Center.
Demeny, Paul. 1968. Early fertility decline in Austria-Hungary: A lesson in demographic transition. Daedalus 97(2):502-522.
Feeney, Griffith, and Kiyoshi Hamano. 1990. Rice price fluctuations and Fertility in late Tokugawa Japan. Journal of Japanese Studies 16:1-29.
Feeney, Griffith, and Andrew Mason. 1997. Population in East Asia. Paper prepared for the Conference on Population and the Asian Economic Miracle, held at the Program on Population, East-West Center, Honolulu, Hawaii, January 7-10, 1997.
T. Fricke, J.S. Chang, and L.S. Yang. 1994. Historical and ethnographic perspectives on the Chinese family. In Arland Thornton and Hui-Sheng Lin, eds. Social Change and the Family in Taiwan. Chicago: University of Chicago Press.
Hajnal, John, 1953. Age at marriage and proportions marrying. Population Studies 7(2):111-136.
Donald J. Hernandez, 1993. America's Children. New York: Russell Sage Foundation.
Dennis P. Hogan and Daniel T. Lichter, 1995. Children and youth: Living arrangements and welfare. Pages 93-141 in Reynolds Farley, ed., State of the Union: America in the 1990s. New York: Russell Sage Foundation.
IIPS, 1995. National Family Health Survey (MCH and Family Planning), India, 1992-93. Bombay: International Institute for Population Sciences.
Institute of Population Problems. 1996. The 49th Abridged Life Tables (April 1, 1995 - March 31, 1996). Tokyo: Ministry of Health and Welfare. [In Japanese.]
Japan Ministry of Health and Welfare. No date. The 15-th Life Tables. Statistics and Information Department, Minister's Secretariat, Ministry of Health and Welfare. [In Japanese with some English headings.]
Japan, Statistics Bureau. 1996. Japan Statistical Yearbook 1996 (CD-ROM) Tokyo: Statistics Bureau.
Jones, Gavin. 1997. "The Demise of Universal Marriage in East and South-East Asia," in G Jones, ed., The Continuing Demographic Transition Oxford: Clarendon Press: 51-79.
Keyfitz, Nathan. 1977. Applied Mathematical Demography. New York: John Wiley & Sons, Inc.
Keyfitz, Nathan, and Wilhelm Flieger. 1968. World Population: An Analysis of Vital Data. Chicago: The University of Chicago Press.
Lloyd, Cynthia B. and Sonalde Desai. 1992. Children's Living Arrangements in Developing Countries. Population Research and Policy Review 11: 193-216.
Lutz, Wolfgang. 1987. Finnish Fertility Since 1722. Helsinki: The Population Research Institute, Finnish Population and Family Welfare Federation.
McNamara, Regina. 1982. Demographic transition theory. In John A. Ross, ed., International Encyclopedia of Population, Volume One. New York: The Free Press.
Mason, Andrew and Tim Miller. Forthcoming. Dynasties, Intergenerational Transfers, and Lifecycle Income. In A. Mason and G. Tapinos, eds., Sharing the Wealth: Demographic Change and Economic Transfers between Generations. Oxford: Oxford University Press.
Mitchell, B. R. 1975. European Historical Statistics, 1750-1970. New York: Columbia University Press.
Nam, Charles B., and Susan G. Philliber. 1984. Population: A Basic Orientation. Second Edition. Englewood Cliffs, New Jersey: Prentice-Hall, Inc.
Preston, Samuel H. and Michel Guillot. 1997. Population dynamics in an age of declining fertility. Genus, Vol. 53, No. 3-4.
Stockwell, Edward G., and H. Theodore Groat. 1984. World Population: An Introduction to Demography. New York: Franklin Watts.
Summers, Robert and Alan Heston, 1991. The Penn World Table (Mark 5): An Expanded Set of International Comparisons, 1950-1988. Quarterly Journal of Economics 105(2):327-366. (Mark 5.6a update which includes data to 1991.)
Taiwan DGBAS, Statistical Yearbook of the Republic of China, various years.
Thomlinson, Ralph. 1976. Population Dynamics: Causes and Consequences of World Demographic Change. New York: Random House.
United Nations. 1995. World Population Prospects: The 1994 Revision. New York: United Nations.
US Bureau of the Census. 1992. 1990 Census of Population Washington, DC: US Government Printing Office.
Weeks, John R. 1989. Population: An Introduction to Concepts and Issues. Belmont, California: Wadsworth Publishing Company.
World Bank. 1993. The East Asian Miracle. Oxford: Oxford University Press.
Griffith Feeney is a Senior Fellow at the East-West Center Program on Population, Honolulu, Hawaii, email gfeeney@hawaii.edu.
Andrew Mason is Professor of Economics at the University of Hawaii and a Senior Fellow at the East-West Center Program on Population, Honolulu, Hawaii, email amason@hawaii.edu.
