Recent Trends in Fertility and Household Formation in the Industrialized World

This article analyses the recent period changes in fertility in industrialized countries from the point of view of the varying degrees of fertility postponement at younger ages and subsequent partial recuperation at later ages. It does so through both period and cohort indicators. The outcome is that current period total fertility rates (PTFRs) are largely diverging because of differential recuperation in the various western countries and Japan, and because of the strong reaction to the economic and social overhaul since 1989 in Eastern Europe. A simple end of postponement would not bring the PTFRs back to replacement level fertility in the vast majority of countries, unless this is being accompanied by much larger rises in fertility past age 30 than witnessed so far. The article concludes with a pattern description and updating of trends concerning the destandardization of household formation. The data from the 1990s round of Fertility and Family Surveys provide the update. Common and more ideosyncratic causes of the shifting forms of household formation are being discussed. Also here the conclusion is that this destandardization is likely to progress further, particularly in Eastern Europe and Japan. 1. The Second Demographic Transition: Characteristics and Diversity Patterns of household formation in the industrialized nations of the world have undergone marked changes during the second half of the twentieth century. Compared to the 1960s there have been sustained trend reversals in ages at first marriage and ages at first parenthood, with mean ages often returning to levels that existed before World War II. Fertility levels have declined almost without exception, and countries that still have period total fertility rates (PTFRs) at or close to the replacement level of 2.08 children are exceptional. By contrast, many national PTFRs have dipped below 1.5 children and have remained there for more [ 121 ] This is the paper presented at the Fourth Welfare Policy Seminar, Families in the New Century, National Institute of Population and Social Security Research, Tokyo, March 14, 2000. The authors would like to thank D. Coleman, K. Kiernan, H. Kojima, P. McDonald, D. Philipov, and S. Sherbov for their help in locating and forwarding missing pieces of demographic information. Special thanks are also due to M. Corijn at the UNECE (United Nations Economic Commission for Europe) for making available additional FFS country reports and to H. Page for critical reading. * Interface Demography, Vrije Universiteit Brussel, Pleinlaan 2, B-1050 Brussels. R. LESTHAEGHE and G. MOORS 122 than a decade. Despite increased education, mean ages at first sexual experience have continued to decline. Hence, the number of years between first sexual intercourse and parenthood has increased dramatically. During this “interim phase” a plethora of patterns of household formation developed. First, there is a pattern that is still characterized by early marriage and childbearing in many Eastern European countries, but present trends show an increasing deviation from these old historical features. In a second group of countries, especially residence in the parental household has been prolonged and entry into partnership is still predominantly passing through a marriage, but often with postponed parenthood. In a third set of countries, home leaving and residential independence come earlier, but periods of single living, of living with age mates, and of cohabitation have filled this “interim phase.” Moreover, procreation outside wedlock, a historical custom that had been pushed by 1900 to the far corners of the European fringe,1 has again become more widespread. Finally, divorce rates of married couples and separation rates of cohabitants have reached high levels, and postmarital cohabitation has replaced remarriage. A variety of explanations have been offered to account for these changes. None of these are mutually exclusive (Lesthaeghe 1998). First, according to neoclassic microeconomic reasoning, increased female education led to more female economic autonomy, higher costs of entry into a union, higher opportunity costs associated with childbearing and childrearing, and to greater assertiveness in favor of more symmetrical gender roles. This explanation focuses predominantly on the better-educated and career-oriented population segments. Second, Easterlin’s (1976) relative deprivation theory accounts for similar responses among the other population strata by pointing to the increased need for extra household income, to be provided by female participation in the labor force, required to meet increased consumption aspirations. Third, ideational theory adds factors of cultural change to the economic explanations, such as reduced legitimacy of normative regulation and authority, increased secularism and individual ethical autonomy, and above all growing respect for individual choices and hence increased tolerance for alternative life cycle structuring.2 1 In 1900–1910 high illegitimacy (indicated by values of Coale’s index Ih above 0.100) was found in such diverse European provinces as Corinthia (0.219) and Salzburg (0.141), Austria; Oberbayern (0.126) and Mittelfranken (0.116), Germany; Krasso Szorzeny (0.211), Temers (0.171), and Torontal (0.176), Hungary; Lazio (0.135), Italy; Braganca (0.113) and Setubal (0.221), Portugal; Muntenia (0.151), Moldovia (0.167), and Oltenia (0.148), Romania; and Murcia (0.200), Spain. National levels of Ih above 0.050 for this period are found for Austria, Denmark, Germany, Hungary, Portugal, Romania, and Sweden (cf. Coale and Treadway 1986). 2 The ideational factors are not strictly endogenous since (1) several manifested themselves already during the “first demographic transition” (e.g., secularization), and (2) they are also responsible for increased female education to start with (e.g., antidiscrimination ideology). Obviously, various countries experienced the impact of these factors to varying degrees and at different times, but there is still a high degree of heterogeneity that is not readily accounted for by these economic and ideational factors (see, e.g., Lesthaeghe 1995). This heterogeneity is particularly striking in the patterning of home leaving and household formation. Hence, the paradigms cited above need to be further complemented by mechanisms such as: 1. Different patterns of diffusion across social strata within countries 2. Different policies and policy responses producing temporal period effects as well as longer lasting patterning 3. Different reactions to periods of economic hardship, particularly in Eastern Europe 4. Country or region specific cultural traits that produce leads and lags or that are directing responses along more idiosyncratic lines. We shall have ample opportunity to refer to such supplementary mechanisms in what follows. In presenting this update of demographic trends, we first direct attention to fertility. We argue that in this field a dominant patterning has been emerging, characterized by systematic postponement of parenthood, and that countries can be ordered according to a fairly coherent stepwise progression. In the second part of the essay, we report on the unfolding of patterns of home leaving and couple formation. Here we encounter a considerably greater diversity: the “second demographic transition” takes different paths despite a common body of factors that should steer it in a more uniform direction. 2. Fertility Trends and the Second Demographic Transition The historical fertility transition, i.e., the “first transition,” was characterized by increased fertility control that predominantly manifested itself by fertility reductions at higher ages. The degree of control typically followed a learning curve with contraceptive efficiency increasing monotonically with age (Coale and Trussell 1974), with marriage duration (Page 1977) and parity (Henry 1953). This reduction of fertility at older ages led to declines in the mean ages at childbearing (MACs), a trend reinforced in western countries by declining ages at marriage as well. The most recent element contributing to this historical pattern was the adoption of efficient forms of contraception (pill, IUDs), which, especially in the late 1960s, eliminated most unplanned pregnancies at older ages and further reduced fertility beyond age 30. In other countries, predominantly in Eastern Europe, access to legalized abortion fulfilled a similar role. The second demographic transition, by contrast, is characterized by the adoption of efficient contraception at early ages and by the overall postponement Recent Trends in Fertility and Household Formation in the Industrialized World 123 R. LESTHAEGHE and G. MOORS 124 of parenthood. The contraceptive learning curve now has a steep rise at young ages (typically before age 20) and becomes markedly less dependent on union duration and parity. Together with the postponement of marriage and the adoption of new living arrangements, fertility now declines prior to age 30. This general postponement of parenthood is the hallmark of the second demographic transition as far as the fertility pattern is concerned. During this phase PTFRs decline below the replacement level and record low levels are being reached. As is well known (e.g., Ryder 1980), a tempo shift in fertility to older ages is a strong factor leading to the rapid fall of period overall fertility. Once this trend is set in motion, two new questions emerge: 1. To what extent and for how long will such a tempo shift be maintained? 2. To what degree will successive cohorts recuperate after age 30 for the fertility foregone prior to that age? An end to the tempo shift definitely has the potential to raise the PTFR again (Bongaarts and Feeney 1998), but the magnitude of this “end to postponement” effect depends strongly on the degree of fertility recuperation past age 30 (Lesthaeghe and Willems 1999).3 Consequently, we need to inspect recent cohort patterns for signs of such fertility recuperation rather than to rely on period measures of parity-specific fertility (TFR1, TFR2, etc.) and period measures of tempo (such as parity-specific mean ages at childbearing—i.e., MAC1, MAC2, etc.). During the second demographic transition the age at first sexual intercourse has declined for both sexes. This was obviously a part of the “sexual revolution” and of the general normative and ethical change occurring since the 1960s. But the learning curves of contraceptive use-effectiveness do not exhibit the same steepness at young ages in all places. In several countries distinct subpopulations with slower learning have emerged. These subpopulations exhibit high teenage pregnancy rates and often high teenage fertility rates as well. Already in the late 1960s a rise in prenuptial conceptions and precipitated marriages occurred in many countries (sex was learned faster than efficient contraception), but by the mid-1970s nothing of this bulge was left. But in other countries this pattern has been maintained for much longer or has been on the rise. In several instances it has lead to a high incidence of teenage single motherhood and is associated with increased child poverty (e.g., in the United States and the United Kingdom; cf. 3 The adjusted PTFR proposed by Bongaarts and Feeney is only the level to which the PTFR would rise again in the absence of further postponement if the parity-specific PTFRs are constant. Tested against the Belgian cohort experience, this strong assumption proved to be the equivalent of a 100% recuperation of fertility after age 30 for fertility foregone prior to this age by the cohort born in 1965 compared to the cohort born in 1960. As shown in section 2.3, the actual recuperation falls short of this level in many countries, and the Bongaarts-Feeneyadjusted PTFR may, therefore, give a too optimistic estimate of any prospective fertility level. Recent Trends in Fertility and Household Formation in the Industrialized World 125 Bradbury and Jäntti 1999). The presence of such subpopulations is readily detectable from a bump prior to age 25 in period schedules of age-specific fertility (cf. Chandola et al. 1999), from the presence of young single mothers living on their own or in their own parental household (three generations), and from the proportions of children currently being raised in single-parent households headed by women younger than 25. These fertility features of the second demographic transition are contingent on two other demographic variables: (1) the nuptiality pattern as it existed and evolved prior to the 1960s, and (2) the path followed during the phase of contraceptive modernization. With respect to the first variable, the old cleavage along the Hajnal-line dividing Europe in a western and an eastern half is of significance again. In the West, the mean ages at first marriage (MAFMs) rose after 1965, whereas they remained low in the East. Communist policies reserving housing for married couples, stimulating female labor force participation and eliminating unemployment in general, undoubtedly contributed to the maintenance of the historically earlier marriage pattern. Now the issue is whether the features of the second demographic transition are currently spreading to Eastern Europe as well: are ages at marriage increasing as a result of the development of alternative and often childless living arrangements? The second variable—the modernization of contraception—equally produces an East-West divide, with the eastern area relying much more on abortion and on traditional nonsupply methods. “Roller coaster” policies with waves of liberalizations and restrictions concerning access to abortion (e.g., Stloukal 1998) combined with the lack of support for hormonal contraception have left the East with significantly lower contraceptive effectiveness. Hence, during the 1970s and 1980s Eastern European countries still faced the problem of unplanned pregnancies for women at older ages and still had ample room for fertility declines at ages above 30. The question again is whether the eastern countries are currently following the West in reducing fertility at younger ages and in producing the typically western tempo shift to older ages. 2.1. Mapping Current Fertility Patterns, 1995–1997 The present situation can be sketched by plotting the national fertility levels (PTFRs) against the fertility tempo indicator (MAC1 or the mean age at first childbearing). This plot is given in Figure 1, and the data are listed in Table 1. At present, only 3 of the 35 countries considered are at replacement-level fertility or close to it: Iceland, the United States, and New Zealand. Of the 12 Eastern European populations, 9 have an early reproductive pattern with MAC1 below 24, and all of them still have mean ages at first childbearing lower than 26. R. LESTHAEGHE and G. MOORS 126 However, this early start of reproduction has not prevented them from having steep declines in PTFRs during the 1990s. Only 2 countries—Yugoslavia (which contains Kosovo in this data set) and Croatia—have PTFRs close to 1.7 or just above it. Eight Eastern countries have PTFRs lower than 1.5, and 5 have dipped below 1.3 children. In this set, we are not counting former East Germany (GDR), which had a PTFR value of barely 0.95 in 1996. The western countries have developed much later ages at first childbearing. There are only 3 countries for which MAC1 is below 26: Iceland, the United States, and Portugal. The majority are located within the 26–28-year range, and 4 countries have MAC1 values above 28: Switzerland, France, Germany (former Federal Republic of Germany [FRG]), and especially the Netherlands with the latest start of fertility of all. The fertility levels, however, vary widely and are comprised between a PTFR of 1.1 and replacement fertility. On the whole, most Scandinavian countries and non-European countries (United States, Canada, Australia, New Zealand) have the highest levels (above 1.6), whereas the Mediterranean ones all have much lower levels (between 1.1 and 1.5). The general lesson to be drawn from Figure 1 is that the earlier starters have by no means higher fertility during the 1990s. The causes at work will appear more sharply in the next section when a decomposition of changes in PTFRs is made for two successive periods. Before turning to details, we also wish to point out several other features indicated in Figure 1. The different symbols used reveal the countries that have P TF R 2.1