Peter McDonald¹

In a century that has seen phenomenal growth in the number of people inhabiting the planet and massive efforts to bring the growth of world population under control, it is ironic that the emerging demographic issue at the end of the century is concern about low fertility in the advanced countries.

For several reasons, there is a degree of reluctance to raise the issue of the effects of low fertility on the age structure of the population. First, world population continues to grow at a massive rate; the United Nations medium projection shows world population increasing from 5.7 billion in 1995 to 9.4 billion in 2050 — and this projection is based on the assumption that fertility will continue to fall in developing countries during this period. There is a fear, therefore, that discussion of low fertility in advanced countries may jeopardise efforts to bring about the still-essential fertility declines in developing countries.

Second, opponents of immigration in countries such as Australia fear that recognition of low fertility may lead governments to pursue vigorous migration programs in an attempt to compensate for the effects of very low fertility. Lesthaeghe, Page and Surkyn have demonstrated that if governments were to take this approach, they would be wrong.² Immigration is a very inefficient approach to slowing down ageing; ageing is only significantly slowed by massive immigration. Also, in the context of low fertility, to keep the total population at a constant level, Lesthaeghe et al. show that there would need to be huge swings in the level of migration in relatively short time sequences. Finally, and most tellingly, they indicate that, again in the context of low fertility, large-scale migration leads in about fifty years to the migrant population dominating the total population, with the original inhabitants being numerically overwhelmed. This would be politically unsustainable in all advanced countries, including Australia. It is also worth mentioning that, contrary to popular wisdom, in Australia, the fertility rate of the Australia-born population is actually higher (by a small amount) than the fertility of the non-Australia born.³

Third, feminists fear that discussion of low fertility will provide weight to the arguments of traditionalists to send women back into the house. For example, in 1997, the Austrian Defence Minister, Werner Fasslabend, was reported to have said that women had a patriotic duty to have at least two children to increase the number of future earners. In reaction to the objections of Austrian women about the gender inequity implied by this statement, his colleague, the Austrian Family Minister, Martin Bartenstein, added that social policy for men and women could be equalised ‘when it becomes a reality that the raising of children is done by men and women’.⁴ In other words, the male breadwinner model of the family is still shaping Austrian policy and, under this model, women’s place is to look after children — and they should be at home looking after more of them.

Fourth, when fertility fell to around replacement level in the 1930s in several countries, pronatalist movements were associated with fascists and eugenicists. Hence, pronatalism has a very murky past. These associations of pronatalism are not far distant today. Fascism in Australia at present targets non-white immigration, but promotion of white Australian babies would be a logical progression. Furthermore, there are already modern-day eugenicists who express fears of a fall in the mean IQ of Australians because higher-educated women have fewer children that lesser-educated women.⁵ The well-known book, The Fear Of Population Decline, by Teitelbaum and Winter, makes the point that there is a tendency for societies in fear of population decline to adopt pronatalist policies ‘verging on the draconian’.⁶

As a person who believes in the need to bring world population growth under control, who believes that massive immigration is not a sensible policy direction for Australia, who believes in the continued advance of feminism and is opposed to fascism and eugenicism, I am reluctant to pursue an issue which may jeopardise all of these causes. On the other hand, low fertility in advanced countries is a reality which cannot be ignored. Indeed, failure to address the issue of low fertility in a sensible and non-draconian way will undoubtedly contribute to the arguments of the opponents of these causes.

Table 1 shows the level of the Total Fertility Rate in several countries in 1990 and in 1995/1996. The Total Fertility Rate is a measure of the average number of children that women would have in their lifetime if they experienced the fertility rates applying at each age in the given year. By any known historical standard, fertility today is exceptionally low. It is lowest in countries in southern and eastern European (with the exceptions of Poland and Yugoslavia) and in Germanic countries. It is also very low in Japan. Among the advanced countries, fertility is relatively high (above 1.6 children per woman) in English-speaking countries, Scandinavian countries and in France.

It must be emphasised that fear of population decline is very well founded when fertility is sustained at levels as low as those now prevailing in many of the countries shown in Table 1. The number of births in the next 40 years or so is influenced by the current age structure as well as by the fertility rate. For countries such as Australia and the United States with relatively young age structures, a relatively youthful age structure means that the number of births will exceed the number of deaths for many years to come. However, for several countries in Table 1 with older age structures, the number of deaths already exceeds the number of births. In Russia, the deficit of births compared to deaths is approaching one million per year, and is increasing.

The mathematics are simple. If women, on average, have just one child (and some of the countries in the table are close to this level), then the size of the generation will halve in one generation which, in demographic terms, is about 28 years. In 56 years, the generation size will only be a quarter of what it was two generations beforehand. In a population with fertility at the current Italian level, after the impact on the crude birth rate of the current age structure has been wiped out (about 40 years), the population in the subsequent 100 year period will fall to just 14 per cent of its initial level. Just as a young age structure provides a momentum of population increase, an old age structure provides a momentum of population decline. To stop population decline in this situation, fertility would have to return, for a time, to the highest levels ever achieved by European countries.

Many would argue that a fall in the number of people living in many of the countries shown in Table 1 would not be a bad thing. But sustained very low fertility is not the way to achieve the goal of population decline because of the disruptions that are caused to society. It is interesting that we are quick to make this argument about China’s explicit one-child policy, but not about the implicit one- child policies that apply in European countries. The latter are seen as having freedom of choice, but little cognisance is given to the institutional constraints upon choice.

Sustained very low fertility will change cultures. In particular, the place of children in the culture will be minimised and social institutions will adapt to the relative absence of children. A subsequent reversal of the trend would be difficult and slow. If the long-term aim is to reduce a population’s size, sustained fertility at the current Australian level will achieve this result in good time without major social disruption. Those who take a very long-term view of population growth point out that, in the very long run, world population growth must be close to zero.⁷ In most instances, it would be preferable if this result was characterised by gentle swings rather than by violent fluctuations.

The Total Fertility Rate can be misleading if fertility is temporarily low or high. In particular, the rate may be temporarily high if successive groups of women have their children at younger and younger ages (as was the case in Australia in the 1950s and 1960s) and temporarily low if successive groups of women have their children at increasingly older ages (as has been the case in Australia since the mid 1970s). Thus, an issue for debate at the present time is whether or not fertility is only temporarily low because of the delay of childbearing to older ages that has continued over the past two decades. It is inevitable, however, that a proportion of births delayed will never occur, primarily because life circumstances change and a birth does not suit new circumstances, but also because the biological time clock eventually rings midnight.

A very interesting recent paper by van de Kaa shows that women in most European countries have high expectations about the number of children that they will have when they are young but that, as women get older, the number of children that they eventually have falls well short of their early expectations.⁸ The extent of reduction tends to be higher for those who have more to lose by reducing their attachment to the paid labour force, that is, those for whom the opportunity cost of withdrawal from the paid labour force is highest. Similar data are now available for Australia from the Negotiating the Life Course Survey conducted by the Australian National University (ANU) in early 1997. This survey also indicates that young Australian women expect to have an average of more than two children. For example, in this survey, Australian women aged 20-24 expect to have an average of 2.33 children. Evidence presented later, however, suggests that they will fall well short of this expectation.

The ANU survey also shows that the expectations of Australian women who have completed secondary school and have a post-school qualification (the more highly educated) drop sharply from age group 20-24 (2.55 children expected on average) to age group 30-34 (1.81 children expected). By age group 30-34, this more highly educated group had actually had only 1.17 children on average and so their chance of meeting their expected number of 1.81 children would have to be doubted. At the other end of the education spectrum, those who did not complete secondary school and had no post- school qualification had no change in their expected family size from age group 20-24 through age group 30-34 and their achieved fertility at age group 30-34 (1.96 children on average) was relatively close to their expected number (2.40 children). Some commentators expect that fertility in low fertility countries will rise automatically in the future because desired family sizes of young women are above the actual current fertility levels. My view is that achieved fertility will remain below early expected fertility as expectations are modified by the reality of institutional constraints. That is, if women are to achieve the number of children they would prefer to have, there must be changes in social institutions; it will not happen of its own accord.

In another paper, I have argued that the countries with very low fertility are countries in which gender equity has advanced in individual-oriented social institutions such as education and market employment, but has been very slow to advance in family-oriented social institutions such as the tax-transfer system, the system of industrial relations, the availability and affordability of family-support services, and most particularly, the family itself.⁹ On the other hand, countries with relatively high fertility have done better in advancing gender equity in family- related social institutions. Policy directions which are premised upon the male-breadwinner model of the family, usually more implicitly than explicitly, will lead to lower fertility. In other words, if a woman is offered major rewards through education and market employment but those rewards are severely curtailed the more children that she has, then many women will have fewer children than they set out to have. Evidence for the point that I am making is provided by the fact that, among the advanced countries, those which have higher labour-force participation rates for women, that is, those that make it easier for women to combine work and childbearing, have higher levels of fertility.

In the context of the preceding sections, not unexpectedly, my view is that we should be aiming in Australia to sustain fertility at about its recent level of 1.8 to 1.9 children per woman. And indeed, there has been very little movement in the Total Fertility Rate in Australia over the past 20 years. However, there is emerging evidence of a downward trend in the 1990s (see Table 1).¹⁰ Furthermore, the Total Fertility Rate is a summary measure which can conceal important sub-trends which may be more indicative of future fertility levels than the TFR. In a recent paper, S. K. Jain and I have pointed out that, while the Total Fertility Rate has dropped by only a small amount in the last 20 years, there have been important sub-trends in Australian fertility.¹¹

For successive age cohorts of Australian women, the mean age at first birth is increasing, the average length of the intervals between births is widening, more women are having no children and the mean age at last birth is not increasing. These components determine the level of completed fertility of a cohort, and these trends indicate that completed fertility is falling sharply for each successive age cohort. There has also been a considerable increase in the proportion of births which are born to women who are not currently married (ex-nuptial births). Parity progression rates (the proportion of women with a given number of children who go on to have another child) have fallen over the past 20 years, but only very slowly, with the exception of the rapid drop in progression from having no births to having one birth.

The purpose of this paper is to examine what further light can be shed upon the future of fertility in Australia using the results of the 1996 Census. The restoration to the 1996 Census questionnaire of a question which asks women how many children they have ever had provides a new opportunity to investigate aspects of fertility in Australia. In particular, it allows us to investigate differences in fertility among women with differing characteristics. Such information is generally not available from birth registration statistics and requires very large sample surveys to collect reliable estimates. We trust that this fundamental piece of information about Australian society will not be dropped from any subsequent census.

Statistics on children ever born to women (issue) have two problems. First, there is a tendency for women who have never had a child to not answer the question. Attempts are made on the questionnaire to remind women that they should complete this question even if they have never had a child. In general, the levels of ‘issue not stated’ are comparatively low in the 1996 Census, especially in the main child-bearing ages. In this paper, I shall be relying mainly on the statements of women aged 25-29 and 35-39 years. At these ages, respectively, the levels of issue not stated were 4.4 per cent and 3.8 per cent, compared to 7.6 per cent and 6.4 per cent at the 1986 Census. The levels of issue not stated in 1996 are only a little above levels not stated for questions such as labour force status. Therefore, for the purposes of this paper, it has been assumed that ‘issue not stated’ is no more likely to mean zero issue than any other issue. That is, women with ‘issue not stated’ can be excluded from the calculations of proportions and means. In terms of the proportion of women who are childless, this assumption may lead to estimates slightly below the actual levels.

The second difficulty with statistics of children ever born is that, for women at younger ages, especially those under 35 years, these statistics represent incomplete fertility. That is, we do not know how many more children the women will have in the future. Hence a fall over time in the average number of children for women aged 25-29 years, for example, could reflect either a delay of childbearing, with the same number of children being born to these women ultimately, or a fall in the ultimate number of children that they have. More particularly for this paper, one group of women may have a higher number of children at age 25-29 years than another group simply because they have their children at an earlier age. Both groups could end up with the same number of children by the time they have completed their childbearing. This point needs to be kept in mind in the interpretation of the results.¹²

The average number of children has fallen at all ages (except 15-19 years) between the 1986 and 1996 Censuses (Table 2). The most substantial drops are for age groups 25-29 and 30-34. This partly reflects the postponement of fertility to later ages, but they also point to an overall decline in fertility with each successive age cohort. The distributions of the number of children ever born to women aged 35-39 in 1986 and to women aged 45-49 in 1996 (the same age cohort) are almost the same (Table 3), indicating that the fertility of women aged 35-39 can be considered to be very close to the cohort’s completed fertility. Thus women aged 35-39 in 1996 can be expected to end up with slightly more than 2.01 children per woman. This compares with a completed fertility for women aged 40-44 in 1986 of 2.48 children per woman. This reiterates the Jain and McDonald finding that completed fertility is declining sharply for successive age cohorts. Nevertheless, the completed fertility of women aged 35-39 in 1996 will be close to replacement level (2.06 children per woman).

However, the question remains as to how much the falls in average fertility at ages less than 35-39 will lead to future declines in the completed fertility of age cohorts. Women aged 25-29 in 1986 added an average of 0.91 children as they aged from 25-29 to 35-39. If this same average number were to be added by the cohort aged 25-29 in 1996, then their fertility at age group 35-39 will be 1.70, indicating a completed fertility well below the long-term replacement level. Applying this same approach to those aged 30-34 in 1996, their completed fertility would be 1.85. While these cohorts could add a higher average number of children in the next ten years because their low fertility may reflect postponement of fertility, rather than truncation, it seems inevitable that the 25-29 year-old cohort in 1996 will have a completed fertility of no more than 1.8 children per woman. This means that the 1996 cross-sectional Total Fertility Rate is indicative of what the completed fertility of women aged 25-29 years in 1996 will eventually be. Consequently, a rise in the Total Fertility Rate in the short term is very unlikely as the current level approximates the levels of completed fertility of central cohorts. It should also be noted that an argument can be mounted that the 1996 cohorts aged 25-29 and 30-34 will in fact add less to their fertility in the next decade than those who were at the same ages in 1986, if conditions for childbearing in the next ten years are unfavourable. Indeed, the following analysis shows that there were sharp reductions in the proportions of women with three or more children in the 1986-1996 decade.

While it is the average number of children per woman which determines population growth, individual women do not have the average number. They have a given number of children and these numbers have a distribution which tells us a great deal about fertility behaviour. Taking the distribution of children born to women aged 35-39 as indicative of their completed fertility, around 16 per cent will have no children, 13 per cent will have one child, 37 per cent will have two, 23 per cent will have three and 11 per cent will have four or more. The commonly accepted norm is two children and indeed the average number of children for this group of women is 2.01, but these data indicate a very broad spread of outcomes with only 37 per cent actually having two children. One in three women in this age cohort has three or more children.

This is the main difference between fertility in Australia and fertility in those European countries which have very low fertility. Very few women in countries with very low fertility have more than two children. Giorgi has shown, for example, that falls in the progression from the first to the second child and more particularly from the second to the third child account for most of the drop in Italian fertility over the past 20 years.¹³ The difference in the fertility rates of Italy and Australia, therefore, is not due to differences in the proportions of women who have no children. It is mainly due to differences in the proportions having more than two children.

Table 3 shows sharp increases between 1986 and 1996 in the proportions of Australian women in each age group who have no children and fairly sharp falls in the proportions who have three or more children. The distribution is clearly shifting downwards. Examining these trends, we might predict that those currently aged around age 30 will have something like the following distribution of completed fertility:

No.	%
0	22
1	16
2	35
3	20
4	5
5+	2

This distribution would yield average completed fertility of about 1.77 children per woman, close to the current Total Fertility Rate. The important point to note from this distribution, however, is that more than 50 per cent of the total fertility is contributed by the 27 per cent of women who have three or more children. The other 73 per cent of women, with 0-2 children, contribute less than 50 per cent. This indicates the extent to which the current level of fertility in Australia is dependent upon around one-quarter of all women having three or more children However, Table 3 shows that the proportion with three or more children is trending downwards. If all the women in the distribution above who now have three or more children had two children instead, the average fertility in Australia would fall to 1.4 children per woman. If the proportions having no children and one child were to increase simultaneously then Australian fertility would fall to the levels of Spain and Italy. There are movements in Australia which call for no Australian woman to have more than two children. This is not the right approach to a sustainable population.

By examining the differences in fertility among sub-groups in the society, we can obtain further insight into the future directions of fertility. For example, if married women have the highest fertility and there is a trend away from women being married, then we could expect fertility to fall. Table 4 shows trends in fertility for women based on three characteristics, marital status, living arrangement and labour-force status. Table 5 shows trends over the same time period in the proportions of women in each category of these characteristics. The mean number of children ever born declined between 1986 and 1996 in both age groups for every category in Table 4, except for women who have never married. Never married women aged 35-39 years had markedly higher average fertility in 1996 than in 1986, but their fertility level was still well below that of married women.

Considering marital status, by far the largest group of women at each census were those who were currently married. This means that changes for this group will have a larger impact on overall fertility than changes for other groups. Tables 4 and 5 show that the average fertility level of married women dropped markedly between 1986 and 1996 and, at the same time, the proportion of women who were currently married also fell. Thus, these trends have a combined downward effect upon the overall fertility level. This is not countered by women having children in de facto relationships, a group whose proportion of the population increased between 1986 and 1996, because the average fertility of women in de facto relationships also fell between 1986 and 1996.

The pattern of decline is similar if we consider labour-force status. Fertility declined between 1986 and 1996 for all three labour-force categories of women. However, the decline is greater for those in the labour force. At the same time, the proportion of women in the labour force increased sharply between 1986 and 1996. Thus, again, both the changes in fertility and the changes in population composition are in the direction of fertility decline.

Furthermore, data not shown in the tables indicates that, between 1986 and 1996, fertility fell by a larger amount for women of low education than for women of high education. For women aged 35-39, the average number of children ever born for those with university degrees was 1.62 in 1986 and 1.55 in 1996. For those with no post-school qualifications, the fall in the decade was from 2.34 to 2.15.

In the light of the above findings, it seems most unlikely that fertility will rise in future because the trends in the composition of the population and in the fertility of individual categories of women all point to further decline. Sustaining fertility at about current levels will not be achieved by reversing the directions of changes in population composition. A return to early and universal marriage and a return to a substantial majority of women being outside the paid labour force is not a possibility. Indeed, I have argued that to attempt to reverse these trends would force fertility lower.¹⁴ Countries with very low fertility are all countries in which levels of gender equity are low.

Table 6 shows the average number of children ever born to women aged 25-29 and 35-39 at the 1996 Census according to a selection of characteristics. If we look at the data by qualification obtained, fertility in both age groups falls as education rises, but especially in the younger age group. If fertility at age group 35-39 is taken as near-to-completed fertility, then women with a bachelor degree or higher have about 1.55 children on average compared to those with no qualifications who have 2.15 children. Other education categories lie between these two extremes. The differences in fertility by education are much larger at age group 25-29 years, probably mainly reflecting the delay of childbearing for the more educated women.

Somewhat similar patterns emerge for women in different occupation groups. It should be pointed out that, to have an occupation in the census, a woman must be employed, so the occupation data shown do not include women who were not employed. Fertility in age group 35-39 is lowest for professionals (1.61) and highest for women working in the trades or as production process workers or labourers (2.01), a result which might have been expected on the basis of the differences by education level. The next block of data in Table 6 shows, however, that when we restrict the analysis to wives in registered marriages, there is a much smaller difference between the fertility of professionals on one hand (2.02) and the fertility of those in trades and labouring occupations (2.23). The immediate thought from this result is that the lower fertility of all professionals might be due to a lower proportion of these women being wives in a registered marriage. That is the lower fertility is due to a lower marriage rate. This appears not to be the case as 66 per cent of professionals and 69 per cent of women in trades and labouring were wives in registered marriages; that is, there was almost no difference in their marriage rates. Hence, the lower fertility of all professionals must have been due, in large measure, to the lower fertility of professionals who were not wives. This is confirmed by other data in Table 6. Professionals living in de facto relationships had much lower fertility than those in trades and labouring occupations who were living in de facto relationships. The fertility differences by occupation were even wider among women who were neither married nor in a de facto relationship (for age group, 35-39, an average of 0.76 for professionals and 1.40 for trades and labouring).

Finally, Table 6 shows differences in fertility by levels of family income. The comparisons are shown only for women who were living in couple relationships because family income is only comparable for couples. Table 6 shows that, for wives, the differences in fertility by income level for age group 35-39 are small, although there is a tendency for fertility to fall as income rises. Mirroring the results for occupation, however, the fertility differences by income level for women living in de facto relationships were much wider with fertility falling sharply as income rises.

Thus, fertility is highest in every socio-economic group for women who are wives in registered marriages and, for these women, the differences between groups are not large. Compared to 10 years ago, the proportions of women in all categories who are in registered marriages has declined and this has contributed substantially to the fall in fertility overall. By age group 35-39, however, socio-economic differences in fertility are mainly due to the higher fertility of women of lower socio-economic status who are not in a registered marriage. Overall, the conclusion is that the trend towards lower fertility is a trend for almost all types of women. That is, sustaining a higher level of fertility would not be achieved by targeting sub-groups of women. Instead, maintaining the current level of fertility requires programs directed at all women, and all men.

Data from the 1996 Census of Australia indicate that fertility in Australia is very unlikely to rise in the future from the current level of about 1.8 births per woman. On the other hand, future falls in fertility appear to be quite likely. For those categories of women where it was possible to examine the change in fertility over time, almost all had experienced a fall in fertility between 1986 and 1996. Furthermore, trends in the composition of the population are in the direction of groups that have lower fertility. In particular, women are achieving increasingly higher qualifications and are becoming much more likely to be engaged in the paid labour force. Higher-educated women have fewer children than lower-educated women and fertility falls as involvement in the paid labour force rises. Thus, on present trends, Australia can expect its fertility rate to fall to lower levels. I have argued here and in an earlier paper that a further fall in Australia’s fertility level is not in Australia’s long- term interest or, for that matter, in the long-term interest of almost any developed country.

The solution to this situation does not lie in restricting women’s education and labour force participation. This is clearly unacceptable to most women and, nowadays, to most men. The solution also does not lie in increasing the migrant intake. In demographic and social terms, migrants are not a substitute for children. The solution does not lie in draconian approaches to increasing the birth rate such as banning contraception or abortion or encouraging baby farming. There is only one viable approach and that is to institute conditions under which women and men can participate in paid employment and have the number of children that they say they want to have.

The new liberal economic agenda which dominates present policy formulation does not provide much reassurance that these conditions will be achieved. There is a very strong direction in present policy to cut services for families with children. Most notably and most importantly, formal child care services are becoming increasingly unaffordable. The National Commission of Audit Report urges that support from government to families with children should be limited to only ‘poor’ families.¹⁵ Support for middle-class families with children is negatively labelled as ‘middle-class welfare’. The tag of ‘middle-class welfare’ is even being extended to such basic supports for children as primary school education. Of course, it is more efficient in the short term for governments to cut their support of children. It is the longer-term consequences that are the problem. The National Commission of Audit argues that we need to be making these efficiency changes so that we can deal with the future burden of an ageing population, but it is obvious that when people have fewer children as a result of these policy directions, the ageing problem will be considerably worse than the Commission envisaged. There seems to be a blind faith among policy makers that Australians will continue having children despite the institutional hurdles that they will have to jump in order to do so. Many will do so, but many others will reduce the number of children that they intended to have.

Lack of support for families with children is not just a direction of government policy. It is also the direction of employment policy. The family-friendly employment policies of the late 1980s are being reversed in the interests of lower costs. And, indeed, it is cheaper for an employer not to support the family needs of an employee. The workplace, if anything, is reverting to the male-breadwinner model under which the employee is expected to put the workplace first, to be available out of hours and not to have children who get sick or have school holidays. Insecurity of employment is also a major determinant of fertility decision- making. If you feel that your job may be cut or that you may have to move around the country in order to stay in work, then you will avoid long-term decisions — and children are for the long-term. The labour market rewards the flexible and the mobile, but family life is not flexible or mobile.

Low fertility is not a product of people rejecting the family way of life. As the fertility preference surveys cited earlier show, most young people still want to have a partner and to have children. However, they change their minds in the face of reality. Low fertility is the result of conflict between a liberal economic agenda and the persistence of social institutions which are premised upon the male-breadwinner model of the family. It is this combination which is fatal to contemporary childbearing. The liberal economic agenda needs to find space for the fulfilment of longer-term social and economic goals. It needs to find space for children. This can only be achieved through programs which emphasise gender equity and programs which recognise that governments must be involved in providing very substantial transfers through services or money to those who have children. In the 19th century when policies in relation to children were first introduced by governments, there was a recognition that children and their education were valuable to the whole society, not just to their parents. In sharp contrast, policy today says to parents: you had them, you pay for them. We shall all pay for this approach in the long term.

References