Declines in first and second birth rates and their effect on levels of fertility.

Kippen, Rebecca

This article analyses the fertility decline in Australia, focussing on the period from 1991 to 2000 when fertility data by women's age and parity status are available. It finds that much of the decline is due to increased childlessness and to the postponement of first births. Second birth rates for woman who are at parity one status have remained quite high. Higher parity births remain low but, expressed as a rate of those eligible to achieve them, have not declined significantly.

**********

An examination of fertility rates in Australia indicates a fall over time in both period and cohort fertility. Period total fertility has been below replacement level for twenty-five years, while cohorts of women completing their fertility in the first decade of the 21st century are likely to have around replacement-level fertility. Policy-makers and others in Australia are concerned about the potential for further fertility decline, given the impact this would have on the age structure of the population and its rate of growth.

This paper tests the hypothesis that the recent fall in fertility is due solely to changes in age-specific first-birth rates, resulting both from delayed parenthood and increased levels of childlessness. Age- and parity-specific fertility rates for the period 1991 to 2000 are constructed. These show that recent fertility declines are driven by changes in first-birth rates, and, to a lesser extent, by changes in second-birth rates. Rates for higher-order births have remained relatively constant. One implication of this pattern is an increasing prevalence of childlessness and one-child families in Australia.

The author then goes on to use age- and parity-specific fertility rates for the period 1991 to 2000 as the basis of four different projections of fertility and considers the likelihood of each of these projections given past trends in fertility.

FERTILITY IN AUSTRALIA--AN OVERVIEW

Period fertility

In 1961, at the height of the postwar baby boom, Australia's total fertility rate (TFR) stood at 3.56 births per woman, more than twice the 2002 level of 1.75. Most of the decline in fertility between 1961 and 2002 occurred in the first two decades from 1961. Fertility fell sharply in the early 1960s, plateaued briefly at around 2.9 births per woman from 1966 to 1971 and then plummeted again. The TFR stabilised at around 1.8-1.9 births per woman in the 1980s. However, in the 1990s, fertility again began to decline, this time much more slowly. Australia's TFR over the period 1950 to 2002 is shown in Figure 1.

The drop in the TFR in the 1960s and 70s was the result of falls in fertility at nearly all ages. Between 1961 and 1966, age-specific fertility rates declined by around 20 per cent for each age between 19 and 47 years. Between 1971 and 1979, fertility at each age declined between 20 and 60 per cent, with the greatest falls occurring for women aged 40 years and over. However, in the 1980s, fertility at older ages began to increase while declines at younger ages continued. The result was a stabilisation of the TFR over this period, with rises in fertility at ages 27 years and above compensating for continuing fertility declines for women in their teens and early twenties. In the 1990s falls in fertility at the younger ages continued apace while increases in fertility at the older ages slowed considerably, resulting in a decline in the TFR.

[FIGURE 1 OMITTED]

Figure 2 shows age-specific fertility rates for the years 1980, 1990 and 2002. A comparison of the age-specific fertility rates for 1980 and 1990 shows a substantial shift to older-age fertility, even though the TFRs for these years, 1.89 and 1.90 births per woman respectively, were virtually the same. Between 1990 and 2002, falls in fertility at the younger ages were no longer being fully compensated for by increases in fertility at the older ages.

The slight increase in the TFR from 2001 to 2002 (from 1.73 to 1.75) is unlikely to herald a reversal of the recent downward trend. Such minor deviations are common and were also seen in 1985, 1990 and 1992.

Cohort fertility

The total fertility rate and other period age-specific fertility rates are useful measures in that they can be used to quantify current fertility experience. However they do not relate to the experience of any actual individual or group; women bear children over time, not cross-sectionally. In addition, cross-sectional fertility measures can be misleading in that falls in period fertility may be the result of delayed childbearing (tempo) rather than reduced parity (quantum) for any particular group of women. In order to consider the experience of actual women, the fertility of birth cohorts can be traced through time.

[FIGURE 2 OMITTED]

The completed cohort fertility rate (CCFR) for Australia is also shown in Figure 1. In many ways, the story of Australian cohort fertility is similar to that of period fertility. The CCFR has been declining for all cohorts born since 1932, although no cohort with complete or near-complete fertility has yet experienced a CCFR below replacement level. The 1932 birth cohort bore, on average, 3.14 children per woman. This has declined to 2.15 children per woman for the 1960 birth cohort and is projected to decline further for later cohorts. The lower fertility of women born in 1960, compared to that of women born in 1932, was the result of falls in fertility of up to 50 per cent between the ages of 18 and 40 years. The most dramatic declines occurred for women in their twenties. This decline is continuing for later cohorts at younger ages for which data are available. The fall in cohort fertility is associated with delayed first birth. For example, in the 1986 Census the proportion of women aged 30 years (born 1955 to 1956) with zero parity was recorded as 26 per cent. For women aged 30 years in 1996 (born 1965 to 1966) this proportion had increased to 38 per cent.

Figure 3 shows age-specific fertility rates for women born in 1950, 1960 and 1970. The birth cohort of 1960 experienced much lower levels of fertility at the younger ages than did the cohort of 1950 and only slightly higher levels of fertility at the older ages. Women born in 1970 have lower fertility at every age up to 30 years of age than either women born in 1950 or 1960. Although fertility at ages 30 and over is likely to be higher for this cohort, it is unlikely that it will increase to the extent required for replacement-level fertility.

[FIGURE 3 OMITTED]

Fertility by age and parity

Ideally, Australia would have available a long series of annual births recorded by maternal age and previous parity in order for age- and parity-specific fertility rates (APFRs) to be calculated. This would allow an examination of both period and cohort rates by parity over time. The total previous number of children borne by the mother is recorded in all birth registers of the Australian states and territories (with the exception of Victoria) but, unfortunately, only Queensland, South Australia, Western Australia and Tasmania code this information. This means that the necessary parity data are available through the registration system for less than 40 per cent of Australian births.

However, in 1991, the Australian Institute of Health and Welfare's National Perinatal Statistics Unit (AIHW NPSU) established the Perinatal Data Collection. This database also collects information on birth order, for all states and territories. Using these data, APFRs can be calculated for each year from 1991 to 2000, the last year for which data are currently available. Age- and parity-specific fertility rates are calculated as the annual number of births to women of age a and parity i in year y, divided by the mid-year population of women at risk of such a birth; that is, the mid-year female population aged a of parity i. (1)

Figure 4 shows APFRs for women with zero parity, that is, rates of first birth. These fertility rates follow the trend for age-specific rates in general, in that fertility is falling under age 30 years and increasing over age 30 years, although not to the same extent. For example, the first-birth rate at age 24 years declined from 73 per 1000 to 48 per 1000 from 1991 to 2000, while the first-birth rate at age 34 years increased from 78 per 1000 to 100 per 1000 over the same period. The peak in first-birth fertility occurs around age 29 to 31 years in each year.

[FIGURE 4 OMITTED]

Rates of first birth are relatively low at all ages and did not exceed 125 births per thousand women over the period 1991 to 2000.

Age- and parity-specific fertility rates for women of parity one are shown in Figure 5. Again rates declined substantially over time for younger women and increased for older women during the 1990s. The striking feature of these fertility rates is how high they are for women in their twenties and early thirties. The rate of second birth for women with parity one exceeds 150 per 1000 for all ages between 20 and 34 years over the period in question. Women in their late 20s experienced second-birth rates of close to 250 per thousand in the early 1990s. This illustrates the strength of the two-child family norm in Australia, with women moving quickly from first to second birth. Fertility rates for women with parity one are higher at all ages than for women with zero parity or parities above one.

Fertility rates for women of parity two and three are shown in Figure 6. Rates for parity four plus are not shown but they follow the same pattern. Fertility rates for women with parity two and above are low and decline from the early twenties. These rates have remained relatively constant over the period 1991 to 2000. This underlines the finding of this paper that the decline in fertility over this period is due to delayed first and second births, rather than delayed higher-order births; delayed first births produce around 85 per cent of the decline in the TFR and delayed second births the remaining 15 per cent.

[FIGURE 5 OMITTED]

If first and second births are delayed in a context where the probability of a higher-order birth declines steadily with age, the result is a fall in the number of women achieving parities of three and above. In the recent past, women with three or more children have contributed considerably to completed fertility levels in Australia. McDonald (2) has argued that the relatively high levels of fertility in Australia, compared with levels experienced in some European countries, result from the higher proportion of women in Australia with three or more children.

First- and second-birth rates are still changing over time, with declines at the younger ages and increases at older ages. These changes will result in increasing proportions of childlessness and one-child families, decreasing proportions of women with higher-order births, and lower levels of total period and cohort fertility.

FUTURE FERTILITY

Concern about Australia's birth rate does not (or should not) centre around the current level of fertility. Projections show that if fertility remains where it is, or stabilises at a slightly lower level, then, combined with current levels of net migration, the population will stabilise within the next few decades with a sustainable age structure. (3) The concern is whether fertility will continue to fall, perhaps to the very low levels experienced in some countries of Europe. Although it is impossible to conjecture with any certainty about future levels of fertility, projections can be made based on past cross-sectional and cohort trends.

[FIGURE 6 OMITTED]

The real questions are whether fertility under the age of 30 will continue to decline and whether any such decline will be matched by increased fertility over the age of 30 years as cohorts move through to those ages. Over the recent past, fertility at older ages has been increasing for successive cohorts, but not to the extent required to compensate for falls in fertility at younger ages. Although no cohort with completed fertility has fewer than two births on average per woman so far, the trend shows a steady decline in completed fertility. The falls in fertility at younger ages are continuing. This suggests that falls in fertility at younger ages are not simply the result of delayed births that will occur at some point in the future, but rather indicate a shift to smaller completed family size.

PROJECTIONS OF FERTILITY IN AUSTRALIA

There have been relatively few projections of Australian fertility in recent years. Those that have been made are generally based on projections of period age-specific fertility rates.

Official population projections for Australia are produced by the Australian Bureau of Statistics (ABS). In their most recent publication, the ABS (4) use a high, medium and low fertility assumption. In the high assumption the TFR is assumed to increase slightly to 1.8. The ABS argues that this may occur given that fertility has 'fluctuated between 1.7 and 1.9 babies per woman since the late 1970s'. Under the medium and low assumptions, the TFR falls to 1.6 and 1.4 births per woman respectively before stabilising. Under all three scenarios, the mean age of fertility is projected to continue rising.

The Retirement Income Modelling Unit of the Federal Treasury (5) has projected age-specific fertility-rate trends using curve fitting, resulting in a gradual decline in the TFR to 1.65 in 2010 and 1.56 by 2050.

McDonald and Kippen (6) have projected fertility of 1.65 births per woman by 2008. They state that this 'seems justified' based on then-current levels of fertility in other industrialised countries, many of which have fertility rates lower than Australia's, and fertility rates in Australian capital cities, which averaged around 1.65 in 1998.

PROJECTING AGE- AND PARITY-SPECIFIC FERTILITY

In this paper I make four different assumptions about the future of APFRs for women with parity zero and parity one and assess what impact these have on Australia's age-specific fertility rates over the period to 2015. In all projections, rates for women with parities two and over are assumed to remain constant, in line with Australian experience over the period 1991 to 2000. (7)

The four assumptions are:

1. the postponement of fertility halts; that is, fertility stops declining at younger ages but continues increasing at older ages in line with the average annual linear change experienced over the period 1991 to 2000. Cohorts born from 1979 all experience that same projected fertility schedule;

2. age- and parity-specific fertility rates remain constant at 2000 levels;

3. age- and parity-specific rates continue to change at half the average annual linear change experienced over the period 1991 to 2000; and

4. age- and parity-specific fertility rates continue to change at the rate experienced over the period 1991 to 2000.

RESULTS

The TFRs and age-specific fertility rates resulting from each projection are shown in Figures 7 and 8 respectively. Under projection 1, fertility rates at younger ages stabilise while those at the older ages continue to increase. This results in a steady increase in the TFR from 1.75 in 2002 to 1.85 in 2015. The CCFR declines slowly from 2.15 births per woman for the 1960 birth cohort to 1.91 for women born in 1970 to 1971.

Under projection 2, with age- and parity-specific rates held constant, the TFR increases slightly from 1.75 in 2002 to 1.78 in 2015 and a slight shift to older age fertility results. This is because of changes in the proportions of the population by parity. Under this assumption, the cohort born 1970 to 1971 will have the following characteristics at the end of their childbearing years: 18 per cent will have zero parity, 16 per cent will have borne one child, 39 per cent will have two children and 18 per cent and 9 per cent will have three and four or more children respectively. This results in a CCFR of 1.87 births per woman.

Under projection 3, the TFR declines to 1.65 in 2015 and the shift to older age fertility continues, but at a slower rate than experienced over the 1990s. Under projection 4, the TFR declines to 1.52 in 2015. Because of continuing shifts in first- and second-birth rates, proportions of the population with parity zero and parity one continue to increase. For example, in 2000, 22 per cent of women aged 35 were childless and 17 per cent had one child. By the year 2015, under the assumptions of projection 3, these proportions increase to 27 per cent and 22 per cent respectively. Put another way, in 2000, 61 per cent of 35 year-old women had at least two children. By 2015 this proportion is projected to drop to 51 per cent.

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

CONCLUSION

Outcomes of the above four projections range from the TFR falling to 1.52 in 2015, to the TFR increasing to 1.85 in 2015. Age-specific fertility rates in Australia show no sign of stabilising and the fertility of recent cohorts shows no sign of convergence. Given these recent fertility trends both projections 1 and 2 are unlikely outcomes. If current shifts to lower and older-age fertility continue (projection 4), even if at a slower pace (projection 3), the result will be a continuing decline in both period and cohort fertility.

Population projections for Australia indicate that a long-term TFR below about 1.5 births per woman would be difficult to sustain demographically. (8) Therefore it is in the interests of Australian policy-makers to consider ways in which the decline in fertility may be halted. Given that the fall in fertility is the result of changes in first- and second-birth rates, research designed to identify ways of facilitating the transition to first birth, and the transition from first to second birth would seem to be prudent.

Acknowledgement

The author thanks Peter McDonald and the anonymous referees for their helpful suggestions. The author would also like to thank Elizabeth Sullivan, Director of the National Perinatal Statistics Unit of the Australian Institute of Health and Welfare, for providing unpublished births data from the National Perinatal Statistics Unit's Perinatal Data Collection.

References

(1) For detailed calculations, see R. Kippen, 'Trends in age- and parity-specific fertility in Australia', Working Papers in Demography, no. 91, The Australian National University, Canberra, 2003, Appendix A

(2) P. McDonald, 'Contemporary fertility patterns in Australia: first data from the 1996 Census', People and Place, vol. 6, no. 1, 1998, pp. 1-13

(3) P. McDonald and R. Kippen, 'Population futures for Australia: the policy alternatives'. Department of the Parliamentary Library, Research Paper no. 5, Canberra, 1999

(4) Australian Bureau of Statistics (ABS), Population Projections Australia 2002-2101, Cat. No. 3222.0, Canberra, 2003

(5) B. Bacon, Projecting the decline in fertility. Paper presented at Workshop on Declining Fertility in Australia: Policy and Research Issues, Australian Institute of Family Studies, Melbourne, 30 May, 2000

(6) McDonald and Kippen, 1999, op. cit.

(7) For detailed calculations, see R. Kippen, 2003, op. cit., Appendix B.

(8) McDonald and Kippen, 1999, op. cit.