VII Congreso de la Asociación Latinoamericana de Población, Foz do Iguaçu, 2016
Fertility postponement and regional patterns of
dispersion in age at first birth
[Preliminary version, August 2016]
Mathias Nathan
mathias.nathan@cienciassociales.edu.uy
Ignacio Pardo
ignacio.pardo@cienciassociales.edu.uy
Programa de Población, Facultad de Ciencias Sociales Universidad de la República
Uruguay
ABSTRACT
This study describes the evolution of dispersion in age at first birth in several countries going through the postponement transition. We examine the evolution of mean age at first birth and its associated standard deviation since 1975, as well as changes in age-specific fertility rates, using data from Human Fertility Database and Human Fertility Collection for 21 countries in 8 regions. Our results show that the postponement transition comes along with an overall increase of heterogeneity in age at first birth, with marked regional differences in terms of level and pace of increase. We also identify four patterns of dispersion that might be linked to specific socio-historic contexts and demographic features in each region. In comparison to countries from Europe, North America and East Asia, Chile and Uruguay exhibit the highest levels of dispersion from the onset of fertility postponement to the present, due to the persistence of significant early childbearing rates.
1. Background
The postponement of childbearing (i.e., a rising mean age at first birth) has been the central focus of tempo studies in fertility research (Balbo, Billari & Mills 2013). A large body of research has concentrated on understanding its driving forces and consequences (Beets 2010; Billari, Liefbroer & Philipov 2006; Kohler, Billari & Ortega 2002; Mills, Rindfuss, McDonald & Te Velde 2011; Ní Bhrolcháin & Beaujouan 2012; Sobotka 2010; 2004). However, little attention has been given to the evolution of heterogeneity of age at first birth across countries and regions.
Kohler et al. (2002) introduced the term “postponement transition” to describe the widespread and pervasive shift towards a late-fertility regime in European countries since 1970. They also predicted the “rectangularisation” of fertility patterns, meaning the concentration of childbearing into an increasingly narrow age interval, once the increase of mean age at first birth (MAB1) approaches its limits. In contrast, Sobotka (2004; 2010) found evidence supporting the hypothesis of a rising heterogeneity in first birth timing. Further research carried out in developed countries showed increasing differences in age at first birth between social groups, particularly in U.K. and U.S. populations (McLanahan 2004; Ravanera & Rajulton 2006; Rendall et al. 2010; Sobotka 2010). Philipov (2015) also showed that heterogeneity increased after the onset of fertility postponement and remains at high levels towards the end of the postponement transition, in line with an increasing diversification of the life course.
According to Rosero-Bixby, Castro Martín & Martín-García (2009) and Esteve, García-Roman, Lesthaeghe & López-Gay (2012), some Latin American countries have experienced the onset of the postponement transition since 2000. Among them, Argentina, Brazil, Chile and Uruguay -all belonging to the Southern Cone- seem to be at
the forefront of this change1. Due to the persistence of high teenage fertility rates in Latin
American countries (CEPAL 2012; Rodriguez & Cavenaghi 2014), the partial shift towards late childbearing may produce higher heterogeneity in the age-schedule of first births. Chile and Uruguay, for instance, were shown to go through the first stages of fertility postponement with a lower MAB1 and a higher standard deviation than
1 Esteve et al. (2012) state that several countries in Latin America have already begun the postponement
phase of the Second Demographic Transition (Lesthaeghe 2014) by examining the evolution of women aged 25-29 still childless, between 1970 and 2011. We included this figure in Annex-Table 1 with updated information for Argentina, Bolivia, Chile and Venezuela.
developed countries (Nathan, Pardo and Cabella 2016). This pattern is also reflected in the emergence of bi-modal curves of hazard rates of first birth by age (Lima et al. 2015; Nathan 2015; Nathan et al 2016).
Whether a rising MAB1 comes along with an increasing dispersion of first birth rates by age is still an open question. While previous studies focused on specific countries or regions, mostly in Europe, there is lack of recent studies comparing the evolution of heterogeneity of age at first birth in different low-fertility settings. Uncovering different regional patterns may contribute with a more thorough description of how the postponement transition unfolds in each case
The aim of this study is to describe the evolution of dispersion in age at first birth in several countries going through the postponement transition. We seek to identify regional trends and patterns, paying special attention to the onset of postponement in South America (Chile and Uruguay) and its specific features.
2. Data and methods
We examine the joint evolution of the period mean age at first birth (MAB1) and its standard deviation (sdMAB1), using data from the Human Fertility Database (HFD) and
the Human Fertility Collection (HFC)2 between 1975 and 2014. We select 21 countries
from 8 regions, comprising Northern, Western and Southern Europe, Central & Eastern Countries, Post-Soviet countries, North America, East Asia and South America (see Annex, Table 2). Given that countries started the postponement transition at different points in time, we plot the evolution of selected indicators without a calendar marker. We
also analyze the coefficient of variation (CV1)3, the share of first birth rates before age
20 and after age 29, and the shape of age-specific fertility rates in selected years.
2 Max Planck Institute for Demographic Research (Germany) and Vienna Institute of Demography
(Austria). Data available at www.humanfertility.org and www.fertilitydata.org.
3. Results
Our results confirm that increasing dispersion of first birth rates by age is a pervasive trend throughout the postponement transition, as stressed by Sobotka (2004) and Philipov (2015) (Figure 1). This might be the result of several driving forces, such as different rates of changes in reproductive preferences, the often-cited de-standardization of the life course and compositional changes in the population. In some countries, though, the sdMAB1 seems to stabilize during the later stages of postponement (for instance, Austria, Japan, Portugal, Sweden, Taiwan and The Netherlands).
However, there are marked differences among countries and regions, both in the level and pace of change in MAB1 and sdMAB1. For instance, countries in Northern Europe show a very similar pattern of steady increase in both MAB1 and sdMAB1, while post-Soviet countries show a much modest increase in MAB1 and do not reach high levels of dispersion. In East Asia, Japan and Taiwan increased its sdMAB1 notoriously, having started from very low levels. On the other hand, England and Wales and the United States, exhibit the highest sdMAB1 within the set of developed countries. In both countries dispersion increased considerably while exhibiting a slow pace of postponement. Interestingly enough, Chile and Uruguay, the only two South American countries with currently available data for first birth in HFD and HFC, seem to be starting the postponement transition at the end of the 90’s with higher levels of dispersion in comparison to the rest of the countries.
Figure 1. Evolution of Mean Age at First Birth (MAB1) and its standard deviation (sdMAB1) by region and country
Which processes lie behind an increase in sdMAB1? Different patterns of change in fertility timing can have an effect on heterogeneity, increasing or decreasing it. Descriptions tend to focus on two specific changes: the decline of adolescent fertility and the increase of fertility at age 30 and more. When those two trends do not evolve at the same rate, they might foster increasing heterogeneity in period measures (see Annex, Fig. 1). 3 4 5 6 7 sd MAB1 20 22 24 26 28 30 32 MAB1 CHL URY South America 3 4 5 6 7 sd MAB1 20 22 24 26 28 30 32 MAB1 AUT GBRTENW NLD Western Europe 3 4 5 6 7 sd MAB1 20 22 24 26 28 30 32 MAB1
FIN NOR SWE Northern Europe 3 4 5 6 7 sd MAB1 20 22 24 26 28 30 32 MAB1 JPN TWN East Asia 3 4 5 6 7 sd MAB1 20 22 24 26 28 30 32 MAB1 BLR RUS UKR Post-Soviet countries 3 4 5 6 7 sd MAB1 20 22 24 26 28 30 32 MAB1 ESP GRC PRT Southern Europe 3 4 5 6 7 sd MAB1 20 22 24 26 28 30 32 MAB1 BGR CZE HUN Central & Eastern Europe
3 4 5 6 7 sd MAB1 20 22 24 26 28 30 32 MAB1 CAN USA North America
¿Standard deviation or coefficient of variation?
A simple and available measure such as sdMAB1 seems to be the best option in order to capture dispersion of first births, but when the mean changes it might be appropriate to measure dispersion in terms relative to the mean, using the coefficient of variation.
Two cases might help us illustrate this debate on which one of both measures better capture dispersion (Figure 2). In the case of Sweden, sdMAB1 increased through the last 25 years, while the CV1 remained quite stable (17%-18%). The reason is pretty straightforward: while sdMAB1 was increasing, MAB1 was too, and almost at the same pace. Considering the different evolutions of sdMAB1 and CV1 (the former increased, while the latter did not), how can we interpret the evolution of dispersion in the timing of first births in Sweden?
In the case of the USA, we need to look at the very last years. The sdMAB1 has stalled since 2003. Meanwhile, CV1 started to decrease. Statistically, the explanation is also straightforward: MAB1 went on increasing, while sdMAB1 did not. The substantial change beneath those aggregate numbers is the decreasing share of first birth rates at younger ages (not shown), which produce an increase in MAB1 but do not change sdMAB1. This is a more peculiar trend but raises similar questions. Is it accurate to stick to CV1 and state that dispersion decreased in the USA during the last years?
Four different patterns of age-schedule for first birth rates are noticeable. First, there is a set of countries (Northern Europe, The Netherlands) which exhibit low dispersion levels both at the onset and at advanced phases of fertility postponement (Fig. 4a). Second, countries with very low dispersion at the onset and high increase of sdMAB1 (Central Europe and, to a lesser extent, East Asia) (Fig. 4b). Third, there is a group of countries which show high dispersion levels at the onset and low increase of sdMAB1 (countries from Southern Europe) (Fig. 4c). Finally, there is a fourth group of countries that reached the highest dispersion levels without having reached an advanced stage of the postponement transition (U.K., U.S. and South American countries) (Fig. 4d).
Figure 4a. Share of ASFR1 in Sweden and The Netherlands (selected years)
Figure 4c. Share of ASFR1 in Greece and Portugal (selected years)
Figure 4d. Share of ASFR1 in Great Britain, USA and Uruguay (most recent data points)
This last group tends to depart from the first three. In this case, %ASFR1 do not show a
normal-distribution type of curve. Instead, it appear to reflect a polarized pattern4, which
is consistent with some of the recent literature for these three countries (Chandola et al. 2002; Nathan et al. 2016; Sullivan 2005). Given that most research tend to focus on the most usual normal-shaped pattern of dispersion, these polarized %ASFR1 shapes are an appealing object for further research.
The emergence of this pattern in Uruguay and Chile might be more visible by comparing %ASFR1 in both South American countries with selected countries at the same stage of
4 Literature on income distribution works on this distinction intensely (Duclos et al 2004). In fertility
research, despite efforts made by Sullivan (2005), a mainstream index to standardize and measure polarization of first birth fertility is still lacking.
the postponement transition –i.e., same MAB1-. When MAB1=23.7, Uruguay and Chile show a slightly more polarized pattern (figure 5), while at the end of the Uruguayan time-span (2011), with MAB1=24.7, this pattern becomes much more pronounced and Uruguay gets very similar to the USA -where MAB1 reached 24.7 in 1996 (figure 6) and differs from the rest of the countries. Something very similar can be observed with Chile at its highest levels of MAB1 (not shown).
Fig. 5. Percentage distribution of age-specific fertility rates for first birth, sdMAB1 and modal age for selected countries with MAB1=23.7
Fig. 6. Percentage distribution of age-specific fertility rates for first birth, sdMAB1 and modal age for selected countries with MAB1=24.7
0% 2% 4% 6% 8% 10% 12% 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 % A SF R 1 Age
Uruguay 1996 Chile 2003 United States 1982 England&Wales 1975 Norway 1977 Taiwan 1978
MAB1= 23.7 sdMAB1 Mode Uruguay 1996 5.8 19 Chile 2003 5.7 19 United States 1982 5.0 20 England&Wales 1975 4.7 22-24 Norway 1977 4.3 21 Taiwan 1978 3.9 23 0% 2% 4% 6% 8% 10% 12% 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 % A S F R 1 Age
Uruguay 2011 United States 1996 England&Wales 1986 Norway 1984 Taiwan 1985 MAB1= 24.7 sdMAB1 Mode Uruguay 2011 6.5 18-19 United States 1996 5.8 19 England&Wales 1986 5.1 24-25 Norway 1984 4.4 24 Taiwan 1985 4.0 23-24
4. Conclusions
Our results tend to support the increasing heterogeneity hypothesis of age at first birth in several countries and regions: increasing dispersion of first birth rates by age is generally observed as MAB1 increases. However, distinctive regional patterns arise, as they are linked to specific socio-historic contexts and demographics features, including the duration and pace of the postponement transition. In particular, South American countries show the highest level of dispersion at the onset of fertility postponement and even so they experience an important increase in sdMAB1 in a short time span.
Different distributions of age-specific fertility rates of first birth arise during the postponement transition. In this regard, it is possible to identify four basic patterns of dispersion, embedded in two groups: the more usual normal-shaped pattern and the more polarized one. The latter is much less frequent. It was observed in the U.S. during the ‘90s and it is especially noticeable in Uruguay and Chile nowadays. Two underlying processes explain the emergence of this pattern: the persistence of high fertility rates at younger ages (see Annex, Fig.1), often observed in women from lower social strata, and the postponement of first births pioneered by women in middle and upper socio-economic strata.
The first of these patterns can be interpreted within the general framework of the de-standardization of the life-course, while the second is frequently observed through the lens of social status polarization. In any case, it is necessary to gain a better understanding of the driving forces beneath an increasing dispersion of first births.
Finally, it is not possible yet to determine how dispersion of first births might evolve as postponement comes to an end, although a decrease in the pace of postponement seems to go along with a stalling sdMAB1. Yet, how should be interpreted a rising sdMAB1 when CV1 does not change? Moreover, should we discard sdMAB1 as a dispersion measure when dealing with non-Gaussian shaped curves? Due to the fact that heterogeneity of age at first birth remains at high levels towards the end of the postponement transition, what is to be expected in terms of dispersion when MAB1 reaches its “limits”? Will countries and regions converge in terms of age-schedule?
References
Balbo, N., Billari, F. C., & Mills, M. (2013). Fertility in advanced societies: A review.
European Journal of Population, 29, 1–38
Beets, G. (2010). The Demography of the Age at First Birth: The Close Relationship between Having Children and Postponement. In Beets, G., Schippers, J. & Le Velde, E. (Eds), The Future of Motherhood in Western Societies. New York: Springer.
Billari, F. C., Liefbroer, A. C. & Philipov, D. (2006). The postponement of childbearing in Europe: driving forces and implications, Vienna Yearbook of Population
Research, 4: 117.
Burkimsher, M. (2015) Europe-wide fertility trends since the 1990s: Turning the corner from declining first birth rates, Demographic Research, 32(21): 621-656.
Chandola, T., Coleman, D.A & Hiorns, R.W. (2002). Distinctive features of age-specific fertility profiles in the English-speaking world: Common patterns in Australia, Canada, New Zealand and the United States, 1970-98. Population Studies, 56(2): 181-200.
Duclos, J., Esteban, J. & Ray, D. (2004). Polarization: concepts, measurements, estimation. Econometrica, 72: 1737-1772.
Esteve, A., J. García-Roman, R. Lesthaeghe & A. López-Gay (2012). The “Second Demographic Transition” Features in Latin America: the 2010 Update, Barcelona: Centre d’Estudis Demografics, Universitat Autonoma de Barcelona, Working paper.
Kohler, H.P., Billari, F.C. & Ortega, J.A. (2002). The emergence of lowest-low fertility in Europe during the 1990s. Population Development Review, 28(4): 641-680. Lesthaeghe, R. (2014). The second demographic transition: A concise overview of its
development. Proceedings of the National Academy of Sciences, 111(51): 18112-18115.
Lima, E., Zeman, K., Castro, R., Nathan, M. & Sobotka, T. (2015). Bimodal Age-Specific Fertility Profiles in Latin America: The Case of Chile and Uruguay. Population Association of America, 2015 Annual Meeting.
McLanahan, S. (2004). Diverging destinies: how children are faring under the Second Demographic Transition. Demography, 41(4): 607-627.
Mills, M., Rindfuss, R., McDonald, P. & Te Velde, E. (2011). Why do people postpone parenthood? Reasons and social policy incentives. Human Reproduction Update, 17(6): 848–860.
Nathan, M. (2015). La lenta transición hacia un régimen de fecundidad tardía en Uruguay: los cambios en la edad al primer hijo entre 1978 y 2011. RELAP, Año 9, No. 17, pp. 37-60.
Nathan, M., Pardo, I. & Cabella, W. (2016). Diverging patterns of fertility decline in Uruguay. Demographic Research, 34(20): 563-586.
Ní Bhrolcháin M. & Beaujouan É. (2012). Fertility postponement is largely due to rising education enrolment, Population Studies, 66(3): 311-327
Philipov, D. (2015). Dispersion of first births over age: more on fertility postponement in Europe. Presented at VID Fertility and Family Discussion Group, Vienna, 1 October 2015 (unpublished manuscript).
Ravanera, Z. & Rajulton, F. (2006). Social Status Polarization in the Timing and Trajectories to Motherhood. Canadian Studies in Population, 33(2): 179-207. Rendall, M., Aracil, E., Bagavos, C., Couet, C., De Rose, A., Di Giulio, P., Lappegard,
T., Robert-Bobée, Is. Rønsen, M., Smallwood, S., & Verropoulou, G. (2010). Increasingly heterogeneous ages at first birth by education in Southern-European and Anglo-American family-policy regimes: A seven-country comparison.
Population Studies, 64(3): 209–227.
Rodriguez, J. & Cavenaghi, S. (2014). Adolescent and youth fertility and social inequality in Latin America and the Caribbean: what role has education played? Genus, 70(1): 1-25.
Rosero-Bixby, L., Castro-Martín, T. & Martín-García, T. (2009). Is Latin America starting to retreat from early and universal childbearing? Demographic Research, 20(9): 169-194.
Sobotka, T. (2004). Postponement of childbearing and low fertility in Europe. Amsterdam: Dutch University Press.
Sobotka, T. (2010). Shifting parenthood to advanced reproductive ages: Trends, causes and consequences”. In.: J.C. Tremmel (ed.) A Young Generation Under Pressure? BerlinHeidelberg: Springer-Verlag: 129-154
Sullivan, R. (2005). The age pattern of first-birth rates among U.S. women: the bimodal 1990s. Demography 42(2): 259−273.
Annex
Table 1: Percent women 25-29 still childless in Latin American countries, 1970-2011
1970-77 1978-85 1990-97 1998-2005 2006-2011 Argentina 30.4 32.9 36.9 39.1 Bolivia 19.1 19.3 22.9 20.2 Brazil 29.5 28.3 29.2 30.8 39.9 Chile 15.6 26.1 27.2 31.4 39.7 Colombia 27.1 27.2 29.1 29.4 Costa Rica 22.1 22.2 25.9 36.1 Ecuador 18.8 20.6 23.9 23.4 24.8 El Salvador 25.6 26.4 Mexico 23.2 24.1 27.6 30.2 Nicaragua 15.5 14.7 17.5 Panama 17.7 21 24.5 26.1 28.3 Peru 26.3 33.3 Puerto Rico 23.5 25.2 33 Uruguay 32.8 32.1 34.4 43.7 Venezuela 26.9 27.2 28.2 33.4
Source: Esteve et al. (2012), computed from IPUMS data files. Data for Argentina, Bolivia, Chile and Venezuela in 2006-2011 (blue numbers) was not available in the original table and therefore introduced by the authors of this paper using census tabulations from national statistical offices (Argentina 2010; Bolivia 2012; Venezuela 2011) and microdata from Encuesta de Caracterización Socioeconómica Nacional of Chile (CASEN 2011).
Table 2. Countries by region, data source and time-span
Region & Country Source Time-span
Western Europe
Austria AUT HFD 1984-2014
The Netherlands NLD HFD 1975-2012
United Kingdom (England and Wales) GBRTENW HFC 1975-2007
Northern Europe Finland FIN HFD 1982-2009 Norway NOR HFD 1975-2012 Sweden SWE HFD 1975-2011 Southern Europe Greece GRC HFC 1975-2008 Portugal PRT HFD 1975-2012 Spain ESP HFC 1975-2008
Central & Eastern Europe
Bulgaria BGR HFD 1975-2009
Czech Republic CZE HFD 1975-2014
Hungary HUN HFD 1975-2009 Post-soviet countries Belarus BLR HFD 1975-2012 Russia RUS HFD 1975-2010 Ukraine UKR HFD 1975-2013 North America Canada CAN HFD 1975-2009
Unites States of America USA HFD 1975-2013
South America Chile CHL HFD 1992-2005 Uruguay URY HFC 1978-2011 East Asia Japan JPN HFD 1975-2012 Taiwan TWN HFD 1976-2010
Figure 1. Share of first birth rates (FBR) before age 20 and after age 29 by region and country 0 10 20 30 40 50 % F B R bef o re age 2 0 0 10 20 30 40 50 % FRB after age 29 BGR CZE HUN Central & Eastern Europe
0 10 20 30 40 50 % F B R bef o re age 2 0 0 10 20 30 40 50 % FRB after age 29 JPN RUS East Asia 0 10 20 30 40 50 % F B R bef o re age 2 0 0 10 20 30 40 50 % FRB after age 29 CAN USA North America 0 10 20 30 40 50 % F B R bef o re age 2 0 0 10 20 30 40 50 % FRB after age 29
FIN NOR SWE Northern Europe 0 10 20 30 40 50 % F B R bef o re age 2 0 0 10 20 30 40 50 % FRB after age 29 BLR RUS UKR Post-Soviet countries 0 10 20 30 40 50 % F B R bef o re age 2 0 0 10 20 30 40 50 % FRB after age 29 CHL URY South America 0 10 20 30 40 50 % F B R bef o re age 2 0 0 10 20 30 40 50 % FRB after age 29 ESP GRC PRT Southern Europe 0 10 20 30 40 50 % F B R bef o re age 2 0 0 10 20 30 40 50 % FRB after age 29 AUT GBRTENW NLD Western Europe