Analyzing Contemporary Fertility [1st ed.] 9783030485184, 9783030485191

Citation preview

The Springer Series on Demographic Methods and Population Analysis 51

Robert Schoen  Editor

Analyzing Contemporary Fertility

The Springer Series on Demographic Methods and Population Analysis Volume 51

Series Editor Kenneth C. Land, Department of Sociology, SINET, Duke University, Durham, NC, USA

In recent decades, there has been a rapid development of demographic models and methods and an explosive growth in the range of applications of population analysis. This series seeks to provide a publication outlet both for high-quality textual and expository books on modern techniques of demographic analysis and for works that present exemplary applications of such techniques to various aspects of population analysis. Topics appropriate for the series include: General demographic methods, Techniques of standardization, Life table models and methods, Multistate and multiregional life tables, analyses, and projections, Demographic aspects of biostatistics and epidemiology, Stable population theory and its extensions, Methods of indirect estimation, Stochastic population models, Event history analysis, duration analysis, and hazard regression models, Demographic projection methods and population forecasts, Techniques of applied demographic analysis, regional and local population estimates and projections, Methods of estimation and projection for business and health care applications, Methods and estimates for unique populations such as schools and students. Volumes in the series are of interest to researchers, professionals, and students in demography, sociology, economics, statistics, geography and regional science, public health and health care management, epidemiology, biostatistics, actuarial science, business, and related fields. Ideas and proposals for additional contributions to the series should be sent either to Kenneth C. Land, Series Editor, Department of Sociology and Center for Demographic Studies, Duke University, Durham, NC 27708-0088, USA E-mail: [email protected] or to Evelien Bakker, Publishing Editor, Social Sciences Unit, Springer, Van Godewijckstraat 30, P.O. Box 17,3300 AA Dordrecht, Netherlands, E-mail: evelien. [email protected] More information about this series at http://www.springer.com/series/6449

Robert Schoen Editor

Analyzing Contemporary Fertility

Editor Robert Schoen Population Research Institute Pennsylvania State University University Park, PA, USA

ISSN 1389-6784     ISSN 2215-1990 (electronic) The Springer Series on Demographic Methods and Population Analysis ISBN 978-3-030-48518-4    ISBN 978-3-030-48519-1 (eBook) https://doi.org/10.1007/978-3-030-48519-1 © Springer Nature Switzerland AG 2020 Chapter 3 is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/). For further details see licence information in the chapter. This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG. The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Preface

The contemporary fertility scene is unique in world history. While a few less developed populations have yet to complete the first Demographic Transition from high to low rates of birth and death, most developed countries are experiencing a Second Demographic Transition, with many having sustained fertility far below replacement. The current volume surveys this heterogeneous landscape. It offers new perspectives, new measures, re-assessments of the nature of fertility intentions, and a multi-faceted examination of the phenomenon called multipartner fertility. The primary audience is social demographers and students of human reproduction, but the volume is also relevant to all those interested in families and children, and to many economists, sociologists, and statisticians. As editor, let me express my appreciation to Springer and its editorial staff, especially Evelien Bakker. I also want to thank Lowell Hargens, Kenneth C. Land and Delores C.  Schoen for their support. My greatest thanks are to the authors who contributed to this volume, and who give it its value. San Francisco, CA, USA March 2020

Robert Schoen

v

Contents

1 Introduction����������������������������������������������������������������������������������������������    1 Robert Schoen Part I Contemporary Perspectives on Fertility 2 Social Capital, Gender Competition, and the Resurgence of Childlessness����������������������������������������������������������������������������������������    9 Robert Schoen and Lowell Hargens 3 Uncertainty and Narratives of the Future: A Theoretical Framework for Contemporary Fertility������������������������   25 Daniele Vignoli, Giacomo Bazzani, Raffaele Guetto, Alessandra Minello, and Elena Pirani 4 Social Contagion Effects in Fertility: Theory and Analytical Strategy��������������������������������������������������������������   49 Nicoletta Balbo and Nicola Barban 5 Context of Interracial Childbearing in the United States��������������������   65 Zhenchao Qian and Yifan Shen Part II Fertility Intentions 6 Do Reproductive Attitudes and Knowledge Explain Race-Ethnic-Nativity Differences in Unintended Fertility?����������������   91 Karen Benjamin Guzzo, Sarah R. Hayford, and Vanessa Wanner Lang 7 Regional Fertility Differences in India��������������������������������������������������  133 Esha Chatterjee and Sonalde Desai

vii

viii

Contents

Part III The Demography of Multipartner Fertility 8 Multi-Partner Fertility in Europe and the United States ��������������������  173 Elizabeth Thomson, Edith Gray, and Marcia J. Carlson 9 Welfare Regimes and Fertility in Second Unions����������������������������������  199 Mariana Fernández Soto, Ana Fostik, and Benoît Laplante 10 Trends in Years Spent as Mothers of Young Children: The Role of Completed Fertility, Birth Spacing, and Multiple Partner Fertility����������������������������������������������������������������  237 Christine R. Schwartz, Catherine Doren, and Anita Li 11 Where’s Daddy? Challenges in the Measurement of Men’s Fertility��������������������������������������������������������������������������������������  257 Lindsay M. Monte and Jason M. Fields Part IV Issues of Measurement 12 Measuring the Prevalence of Multipartner Fertility Independent of Fertility Level����������������������������������������������������������������  287 Robert Schoen 13 Cross-Sectional Average Length of Life by Parity: Illustration of US Cohorts of Reproductive Age in 2015����������������������  293 Ryohei Mogi and Vladimir Canudas-Romo

Chapter 1

Introduction Robert Schoen

Since the year 1900, there has been phenomenal growth in the size of the world’s population (see Table 1.1). While the growth rate has declined since the year 2000, it remains quite high by historical standards. Some 60 years ago, “runaway” population growth was seen as an existential threat. In 1958, nuclear scientist Harrison Brown was quoted in the New York Times as saying that rapid population growth “is the most urgent and most critical problem confronting the world.” Paul Ehrlich’s best-selling 1968 book, The Population Bomb, sounded an urgent alarm about population growth and the imminent dangers of overpopulation. Today, those concerns seem archaic. The last 60 years have seen the demographic transition from high to low birth and death rates spread to nearly every part of the world. Yet, because of population momentum, population growth continues at a substantial rate and world population size today is some 4.7 times what it was in 1900. The absolute size of a population, whether of the world or of any political subdivision, does matter. In the long term, there are limits to the carrying capacity of any region. In the short term, as Jared Diamond (1997) described, larger numbers of persons in a polity means a more complex and hierarchical social, economic, and political system. Now most developed countries are experiencing a Second Demographic Transition (SDT), which has shifted the demographic focus from the family to the individual. The SDT is associated with unprecedented fertility declines, high levels of nonmarital fertility that often involve multiple partners, later and less marriage, and high rates of divorce. Importantly, should they choose to do so, most people have the capability to limit their fertility. The emergence of “lowest-low” fertility in a number of countries, where women on average have only 1.2 or 1.3 children, was an unexpected development. In R. Schoen (*) Population Research Institute, Pennsylvania State University, University Park, PA, USA e-mail: [email protected] © Springer Nature Switzerland AG 2020 R. Schoen (ed.), Analyzing Contemporary Fertility, The Springer Series on Demographic Methods and Population Analysis 51, https://doi.org/10.1007/978-3-030-48519-1_1

1

2

R. Schoen

Table 1.1  Size and growth of the world’s population Year 0 1700 1800 1900 1960 1999 2019

Number of persons in the world (approximate, in millions) 190 660 990 1650 3000 6000 7700

Annual population growth Rate (%) – 0.07 0.41 0.51 1.00 1.78 1.25

Source: Roser et al. (2020) Note: Growth rate (r) calculated from P(t) = P(0) ert, where P(t) is the population at time t

hindsight, it is less surprising. Shifts in the economy and in the societal division of labor, especially with regard to gender, have played a large role. The value of children to parents has markedly declined, while the demands of parenthood have increased. People have adapted to those new circumstances, and have chosen a better life over a larger family. In the terms of Kingsley Davis’ (1963) Theory of Change and Response, they acted to resist the relative loss of status that would have resulted from having more children. Pronatalism has atrophied, and no “invisible hand” stopped the fall in fertility at replacement level (roughly 2.05 to 2.10 children). Presently, nearly all developed countries have below replacement fertility, as do many other nations. The contemporary scene thus shows great heterogeneity in demographic behavior, uncertainty in family norms and gender relations, and a future that is far from clear. Here, we focus on advancing current fertility research in four broad areas: Contemporary Perspectives on Fertility, Fertility Intentions, the Demography of Multipartner Fertility, and Issues of Measurement. Part I, Contemporary Perspectives on Fertility, begins with Chap. 2 by Robert Schoen and Lowell Hargens. We view the rise in childlessness, the complete absence of childbearing, from a social capital perspective. If children provide fewer resources to their parents, it is reasonable to expect that more parents will have no children. We first examine contemporary family building patterns, which reveal that childlessness is a distinct dimension of fertility. Then, using measures from parity status life tables, we find that many developed countries have substantial proportions of women remaining childless, proportions that have been underestimated by a focus on childlessness in cohorts at the end of their reproductive years. Chapter 3, by Daniele Vignoli, Giacomo Bazzani, Raffaele Guetto, Alessandra Minello, and Elena Pirani, analyzes the role of uncertainty in fertility decision making. Arguing that economic constraints cannot explain contemporary European fertility, they see the rise in uncertainty as the missing element. Under uncertainty, fertility decisions are seen as being made based on expectations, imaginaries, and narratives. Understanding individuals’ “narratives of the future” can make an important contribution to explaining fertility behavior.

1 Introduction

3

Chapter 4, by Nicoletta Balbo and Nicola Barban, considers “social contagion” effects in fertility. A sizeable body of research has described how social networks and diffusion processes influence how populations adopt new demographic behavior. In their review, Balbo and Barban survey the recent empirical evidence that has emerged at the micro level, linking micro relationships to macro effects on fertility. They discuss the major mechanisms of “contagion”, describe solutions to the difficult problem of disentangling interaction effects from selection and contextual effects, and provide an overview of the data sets available to analyze peer effects on fertility. In Chap. 5, Zhenchao Qian and Yifan Shen examine interracial fertility in the United States. While past theories of acculturation and assimilation have focused on intermarriage, the diminished salience of marriage and the rise of cohabitation and nonmarital fertility point to a need for closer study of interracial parentage. In 2015, one in seven American births was interracial or interethnic. Qian and Shen separately analyze ten types of interracial/ethnic fertility, showing both the importance and the complex effects of mother’s educational attainment and marital status. Part II, Fertility Intentions, begins with Chap. 6 by Karen Benjamin Guzzo, Sarah R. Hayford, and Vanessa Wanner Lang. Race-ethnic-nativity differentials in unintended fertility have been well documented, but inadequately explained. Guzzo, Hayford, and Lang first examine whether different measures of reproductive attitudes and knowledge explain race-ethnic-nativity differences. They find few significant predictors, which leave significant intergroup differences unexplained. They then perform a factor analysis, which demonstrates that the same reproductive attitudinal and knowledge factors do not arise across race-ethnic-nativity groups, supporting their conclusion that attitudinal and knowledge factors are of limited value in explaining observed group differences in unintended fertility. Chapter 7, by Esha Chatterjee and Sonalde Desai, come to a somewhat similar conclusion in quite a different way. Chatterjee and Desai analyze regional fertility differentials in India, a nation characterized by extreme diversity. Using the India Human Development Survey, they strive to identify factors that affect fertility preferences and those that affect the ability to implement those preferences. They find that socioeconomic characteristics account for a considerable proportion of regional differences in fertility preferences but, in a longitudinal analysis, find that those characteristics are unable to account for any substantial fraction of the differentials in unintended births. Part III, The Demography of Multipartner Fertility, begins with Chap. 8 by Elizabeth Thomson, Edith Gray, and Marcia J. Carlson. They provide the first estimates of multipartner fertility (MPF) across 15 countries and over time, using the Harmonized Histories that were largely generated by the Generations and Gender Programme. Most of the more developed countries surveyed show modest levels of MPF which have been fairly stable over time. The United States is an outlier, with MPF accounting for some 22% of all births. In Chap. 9, Mariana Fernández Soto, Ana Fostik, and Benoît Laplante treat MPF as part of the dependent variable. Using data for 6 representative Western countries, they analyze the influence of welfare regimes on the likelihood of a birth in a second

4

R. Schoen

union. Decomposition results show that the contribution of second union births is directly related to welfare benefits in 5 of the 6 countries, the United States being the single exception. There are significant interactions with women’s educational level, with the value of the benefits paid increasing the likelihood of a second union birth among low educated women, regardless of parity. Chapter 10, by Christine R. Schwartz, Catherine Doren, and Anita Li, explores the time women spend as mothers of young children. In the United States between 1967–69 and 2010–17, the length of time spent with children under the age of 6  years fell for women with a high school or more education, but increased for women with less than a high school education. The increased prevalence of MPF played a significant role, as it added between 0.75 and 1.23 years to the average time women without a college degree spent as mothers of young children. In Chap. 11, Lindsay M. Monte and Jason M. Fields approach MPF by asking “Where’s Daddy?” The 2014  U.S.  Survey of Income and Program Participation (SIPP) collected data from both men and women, and directly inquired about MPF. Despite those advantages, the SIPP data have internal inconsistencies. Monte and Fields, intimately familiar with the SIPP data, work to reconcile those discrepancies in order to provide estimates of MPF along with more detailed characteristics of male fertility. The findings support their argument that demographers should pay more attention to patterns of male fertility, because it differs in significant ways from that of female fertility. Part IV, Issues of Measurement, begins with Chap. 12, where I examine the structural relationship between MPF and fertility level. Since a person must have at least 2 children to be at risk of MPF, the level of fertility is inextricably connected to the prevalence of MPF. I assume that a woman’s fertility depends only on her age and not her parity, i.e. her number of children ever born, and then calculate models where parameter s, the likelihood of a birth by a different partner, is the same for all parities 2 and above. As expected, the prevalence of MPF in a cohort of women, and of maternal half siblings among their children, varies greatly with both s and the level of fertility. Those calculations also provide a basis for estimating parameter s from a population’s fertility level and the prevalence of MPF, allowing populations with different levels of fertility to be directly compared. Chapter 13, by Ryohei Mogi and Vladimir Canudas-Romo, introduces a new measure, CALP, the cross-sectional average length of life by parity. CALP, a period measure that uses the cohort fertility of all women of reproductive age, gives the average length of time women spend at each parity during their reproductive years. Using data from the Human Fertility Database, Mogi and Canudas-Romo calculate CALP for the United States in 2015. The results show that women aged 12 to 50 spend an average of 18 years with no children. Many questions about contemporary fertility remain to be answered, with many more not yet asked. This volume seeks to take a step toward finding some of those answers, in the hope that a better understanding of today will lead to a better understanding of tomorrow.

1 Introduction

5

References Brown, H. (1958, November 20) As quoted in the New York Times, p. 35. Davis, K. (1963). The theory of change and response in modern demographic history. Population Index, 29, 344–366. Diamond, J. (1997). Guns, germs, and steel. New York: Norton. Ehrlich, P. R. (1968). The population bomb. Sierra Club/Ballantine Books. Roser, M., Ritchie, H., & Ortiz-Ospina, E. (2020). World population growth. Published online at OurWorldInData.org. Accessed 23 Feb 2020.

Part I

Contemporary Perspectives on Fertility

Chapter 2

Social Capital, Gender Competition, and the Resurgence of Childlessness Robert Schoen and Lowell Hargens

2.1  Introduction Contemporary fertility in most of the more developed world is near or substantially below replacement level. In 2018, South Korea broke new ground with a Total Fertility Rate (TFR) of 0.98. The proportion childless is a component of overall fertility, but one with special significance. A first birth represents a major change in life circumstances, and thus has an impact that is qualitatively different from that of subsequent births. Here we examine contemporary patterns of childlessness in more developed nations from a comparative perspective, and employ synthetic cohorts to reflect recent behavior at all ages. We begin by looking at the major theoretical perspectives offered to explain fertility behavior, and at past empirical work on patterns of childlessness.

2.2  Theoretical Perspectives on Fertility Dynamics The earliest and most prominent framework for explaining recent developments in demographic behavior argues that they constitute a “second demographic transition” (SDT) that is being driven by a cultural shift toward value orientations that

R. Schoen (*) Population Research Institute, Pennsylvania State University, University Park, PA, USA e-mail: [email protected] L. Hargens Separtment of Sociology, University of Washington, University Park, PA, USA e-mail: [email protected] © Springer Nature Switzerland AG 2020 R. Schoen (ed.), Analyzing Contemporary Fertility, The Springer Series on Demographic Methods and Population Analysis 51, https://doi.org/10.1007/978-3-030-48519-1_2

9

10

R. Schoen and L. Hargens

emphasize individual autonomy and self actualization (Lesthaeghe 1995, 2010; Van de Kaa 1987). According to the SDT, populations undergoing this shift increasingly emphasize secular views and the attainment of material aspirations, which in turn weaken traditional religious, community and family ties. The SDT is associated with changes in many features of family formation, including later ages at marriage, lower marriage propensities, higher divorce rates, and higher levels of non-marital cohabitation and non-marital fertility. Fertility in populations that have passed through the SDT is low, usually below replacement, and levels of childlessness are high. Women are increasingly active in the labor force, and a belief in gender equality gains traction. Despite its prominence, the SDT framework is only loosely descriptive of recent changes in the demographic features of developed societies, and because it offers no theory about how value orientations will change in the future, it yields no predictions about future demographic developments (Zaidi and Morgan 2017). Responding to some of the inadequacies of the SDT approach, several analysts have developed alternatives that emphasize changing gender relationships as playing a key role in producing the demographic changes that the SDT seeks to explain. The most developed of these is the “gender revolution” (GR) framework proposed by Goldscheider et al. (2015). According to the GR approach, women’s increased labor force participation in developed countries at first produced stresses on family life that resulted in the trends (marital instability, lower fertility, etc.) that the SDT was designed to explain. However, the GR framework holds that these changes in women’s roles will also eventually bring about changes in men’s roles wherein re-­ negotiated spousal relationships feature increased involvement of men in the home and family. The GR framework thus holds that in the long run the high levels of family marital disorganization present in the first state of the gender revolution will be reduced and that fertility will rebound. The GR approach is structurally, not ideologically, based, and offers a good analysis of the retreat from the family and the inherent strengths of stable unions. It readily accommodates the fertility effects of the restructuring of the economy (Seltzer 2019) and the increasing uncertainties of the labor market (Mills and Blossfeld 2013; Vignoli et al. 2012). On the other hand, the GR framework simply assumes that couples desire children and does not allow for the possibility that males’ and females’ interests in childbearing may compete rather than being in harmony, or that male adaptations to the first stage of the gender revolution may be to reassert the importance of gender differences rather than to embrace more egalitarian relationships. Finally, although there are signs that the sex role re-­negotiations postulated by the GR framework are incipient in a few populations, substantial empirical support for the idea that this will be a general transformation is lacking. A third framework for explaining current childbearing trends in developed countries is the “children as social capital” (CSC) approach (Schoen et al. 1997). This framework considers the forces that sustain fertility as well as those that constrain it, and identifies the “social capital” of children, not their economic or psychological value, as the principal motivator of fertility. Children’s social capital consists of the resources accruing to parents as a result of the social connections created by

2  Social Capital, Gender Competition, and the Resurgence of Childlessness

11

having children, and include valuable ties to kin and community. Each parent gains from the social capital provided by a child, but the gains are often parent-specific because each parent has their individual interests and ties. Finally, given that the social capital created by childbearing is the chief motivation for parents to have children, factors that change the level of social capital created by children will produce changes in fertility behavior. The CSC perspective identifies three drivers of low fertility and high levels of childlessness. (a) The social capital typically associated with children has declined. Weakened kin ties, governmental social safety net programs, and the greater participation of both parents in the labor force have reduced the social capital benefits conferred on parents by children. (b) Gender competition has undermined family stability and weakened the connection between men and their children (Schoen 2010). Men increasingly avoid the legal constraints of marriage and opt for cohabitation, which brings no long term commitments to a partner. As cohabitation gives greater power to the person with more resources, it is increasingly attractive to some women as well. Men and women also have a differential stake in children. At union dissolution children generally go with their mothers, and because unions have become fragile, men are discouraged from investing resources in their children. In fact, recent decades have seen men shift much of the economic burden of childrearing to women. Mothers are often forced to support both their children and themselves, perhaps with some assistance from the state. The strength of the push back from men against the Women’s Movement has not been adequately appreciated, and underscores the sharp decline in the social capital value of children to their fathers. (c) The negative impact of children on parental relationships has been seriously underestimated. Research on marital quality has consistently found that relationship quality declines with the first birth (Amato et al. 2003). Gendered differences and the division of labor by sex within couples widen, and great time, energy, and economic demands are put on parents, undermining relationship quality and union stability. While a first birth creates new ties between the parents, it interferes with other important relationships, including labor force attachments, and thus can cause a net loss of social capital. The combination of those three factors can be seen as producing a multiphasic response (cf. Kingsley Davis 1963), pushing fertility to sub-replacement levels in many countries. There is no biological imperative to have children, hence traditional societies always needed strong pronatalist pressures (Blake 1972). When those pressures disappear, fertility can plummet, as it has in ostensibly “familistic” East Asian societies. In general then, the CSC framework suggests that populations in the developed world will tend to exhibit low fertility accompanied by high and increasing levels of childlessness. To set the stage, we next look at the literature on the demography of childlessness.

12

R. Schoen and L. Hargens

2.3  Contemporary Childlessness Levels and Trends Aside from suggestive evidence for preindustrial England (Weir 1984; Schofield 1984), which shows variation between 4% and slightly over 20%, little is known about the prevalence of childlessness among populations of adult women earlier than the mid 1800s. Rowland (2007) presented data on the proportion of women who were childless at age 45 for cohorts born as early as the 1840s and up into the 1950s for seven developed countries, and found that the incidence of childlessness exhibited long-term cycles that corresponded to overall fertility. Specifically, after rising to a peak for cohorts born between 1895 and 1909, the proportion childless fell to low levels among the cohorts born between 1930 and 1945. Peak levels of childlessness among the first group of cohorts include 32% in Australia, 28% in the Federal Republic of Germany, and 25% in the United States, while the lowest levels for the second cohort group include 9% for Australia, 10% for the Federal Republic of Germany, and 9% for the United States. Childlessness then started to rebound among subsequent birth cohorts in these nations. The Childlessness in Europe volume, edited by Kreyenfeld and Konietzka (2017), examines more recent cohorts in the United States and a number of European countries. In some Western European countries, including Germany, the United Kingdom, Netherlands, and Switzerland, about 20% of all women born around 1965 have remained childless. In Southern and Eastern Europe, childlessness has risen to 15% or higher, and Sobotka (2017) sees childlessness across Europe converging at a high level. For the United States, childlessness reached 18% for cohorts born in the 1950s (Frejka 2017). While high compared to recent decades, those proportions childless have not exceeded the levels experienced by the cohorts born in the late 1800s and early 1900s. Childlessness can be seen as the outcome of a series of decisions to not have a child. Pronatalist pressures from family, society, and religion have waned over the last half century, while the costs and risks of parenthood have risen (Hobcraft and Kiernan 1995). Cultural perspectives on childbearing have changed in tandem with recent declines in fertility. Childless couples have gone from being considered “deviant” to being socially acceptable (Basten 2009), and the term “childless” itself has been challenged by the term “childfree”, which emphasizes the positive aspects of not having children.

2.4  Data and Methods 2.4.1  Multidimensional Scaling of Parity Progression One way to gauge the importance of childlessness in the structure of contemporary fertility patterns is to examine the family building process that underlies the parity distributions of a heterogeneous group of contemporary national populations. We collected data on the parity progression ratios for women through ages 45 to 49 in

2  Social Capital, Gender Competition, and the Resurgence of Childlessness

13

countries represented in either the UNdata website of the United Nations Statistics Division (United Nations, Dept of Economic and Social Affairs 2013) or the Human Fertility Database (HFD) provided jointly by the Max Planck Institute for Demographic Research and the Vienna Institute of Demography. These two sources employ different open ended upper parity categories for different countries, but provide data that allow the construction of parity progression ratios, through the progression from parity four to parity five, for 80 countries (see Appendix A). We began our analysis by measuring the Euclidean (i.e. straight line) distances between the sets of parity progression ratios for each pair of the 80 countries in our data. We then used multidimensional scaling (MDS) analysis to summarize and construct a spatial representation of the 3160 {i.e. (802–80)/2} distances.

2.4.2  Measuring Childlessness The existing literature largely measures childlessness from a cohort perspective by examining the proportion of women who are childless at the end of their childbearing years, generally taken as 45–50 years of age. The birth cohort approach has the advantage of reflecting the actual behavior of a real group of people. However, at each data year, the cohort measure reflects a single cohort born some 45 years earlier. Younger cohorts of that year, who have not yet completed their childbearing, are excluded from the analysis. That is a serious shortcoming, as the fertility patterns of the younger cohorts are informative, and discarding their behavior forfeits valuable data. The alternative is to employ life table methods. A life table whose single decrement is to first birth can show the proportion childless at age 50 in a hypothetical cohort of persons who, at every age, experience the first birth rates of a given period. A parity status life table (PSLT) goes further, and shows the implications of age-­ parity-­specific birth rates for parity composition as the cohort passes through its reproductive years. We follow that approach here. Parity status life tables for 24 nations are available in the Human Fertility Database (Jasilioniene et  al. 2015). Those countries are identified in Table 2.1. The HFD data also provide the period TFR, the cohort proportion childless at the end of the reproductive period (denoted 45–49 P0), and the period mean age of mother at first birth (TMAB1).

2.5  Results 2.5.1  M  ultidimensional Scaling over 80 Contemporary Populations The results of the multidimensional scaling are shown in Fig.  2.1. Appendix Table  2.1 lists the countries graphed and their abbreviations. We found that two dimensions provide a good representation of the parity progression ratios.

14

R. Schoen and L. Hargens

Table 2.1  Fertility levels, measures of childlessness and age at first birth for 24 countries, by region and time period Period TFR Eastern and Central Europe Belarus 2000–04 1.26 2005–09 1.40 2010–14 1.60 2015–16 1.73 Croatia 2010–14 1.48 2015–17 1.43 Czechia 2000–04 1.17 2005–09 1.41 2010–14 1.48 2015–17 1.63 Estonia 2000–04 1.37 2005–09 1.64 2010–14 1.59 2015–17 1.59 Hungary 2000–04 1.30 2005–09 1.33 2010–14 1.32 2015–17 1.48 Lithuania 2000–04 1.29 2005–09 1.39 2010–14 1.57 2015–17 1.64 Poland 2000–04 1.26 2005–09 1.31 2010–14 1.30 2015–16 1.32 Russia 2000–04 1.28 2005–09 1.41 2010–14 1.66 Slovakia 2000–04 1.23 2005–09 1.31

PSLT P0

45–49 P0

ΔP0

TMAB1

.140 .154 .147 .153

.045 .041 .068 .054

.095 .113 .079 .099

24.6 25.0 25.4 25.9

.214 .243

.114 .131

.100 .112

28.4 29.3

.225 .180 .189 .171

.060 .063 .067 .077

.165 .117 .122 .094

27.1 28.0 28.6 28.4

.177 .161 .209 .204

.046 .027 .072 .092

.131 .134 .137 .112

25.9 26.1 27.2 28.2

.240 .242 .280 .270

.082 .073 .083 .103

.158 .169 .197 .167

27.1 28.2 28.6 28.3

.232 .218 .182 .151

.041 .057 .139 .132

.191 .161 .043 .019

25.3 25.9 27.0 27.5

.295 .275 .295 .307

.101 .107 .134 .147

.194 .168 .161 .160

26.0 26.6 27.1 27.4

.150 .158 .149

.079 .066 .079

.071 .092 .070

24.5 24.9 25.4

.291 .260

.099 .098

.192 .162

26.1 27.3 (continued)

2  Social Capital, Gender Competition, and the Resurgence of Childlessness

15

Table 2.1 (continued) Period 2010–14 Slovenia 2000–04 2005–09 2010–14 2015–17 Ukraine 2000–04 2005–09 2010–13 Western Europe Austria 2000–04 2005–09 2010–14 2015–17 Denmark 2000–04 2005–09 2010–14 2015–16 Finland 2000–04 2005–09 2010–14 2015 Netherlands 2000–04 2005–09 2010–14 2015–16 Norway 2000–04 2005–09 2010–14 Sweden 2000–04 2005–09 2010–14 2015–17 Switzerland 2000–04

ΔP0 .099

TFR 1.46

PSLT P0 .203

45–49 P0 .104

TMAB1 28.0

1.22 1.40 1.57 1.59

.214 .204 .186 .195

.088 .087 .096 .116

.126 .117 .090 .079

28.0 28.7 28.8 29.0

1.14 1.35 1.48

.180 .182 .198

.062 .045 .059

.118 .137 .139

24.3 24.6 25.0

1.38 1.40 1.44 1.51

.266 .264 .247 .229

.152 .170 .182 .206

.114 .094 .065 .023

27.3 28.1 28.9 29.4

1.75 1.84 1.74 1.75

.176 .154 .163 .163

.139 .141 .138 .137

.037 .013 .025 .026

28.2 28.6 29.2 29.4

1.74 1.84 1.79 1.65

.236 .217 .229 .252

.159 .173 .191 .202

.077 .044 .038 .050

27.9 28.2 28.6 29.1

1.73 1.74 1.73 1.66

.186 .183 .172 .186

.167 .176 .181 .179

.019 .007 −.009 .007

28.7 29.0 29.4 29.9

1.80 1.92 1.84

.160 .132 .137

.118 .121 .118

.042 .011 .019

27.6 27.9 28.6

1.66 1.88 1.91 1.83

.178 .139 .139 .166

.130 .134 .132 .125

.048 .005 .007 .041

28.7 28.8 29.2 29.6

1.41

.256

.206

.050

29.0 (continued)

16

R. Schoen and L. Hargens

Table 2.1 (continued) Period 2005–09 2010–14 2015–16 Southern Europe Portugal 2000–04 2005–09 2010–14 2015 Spain 2000–04 2005–09 2010–14 2015–16 Asia Japan 2000–04 2005–09 2010–14 2015–17 Taiwan 2006–09 2010–14 North America Canada 2000–04 2005–09 2010–14 2015–16 United States 2000–04 2005–09 2010–14 2015–17

ΔP0 .016 −.013 −.018

TFR 1.46 1.52 1.54

PSLT P0 .228 .203 .192

45–49 P0 .212 .216 .210

TMAB1 29.7 30.4 30.8

1.46 1.38 1.29 1.31

.118 .125 .140 .145

.095 .097 .086 .077

.023 .028 .054 .068

27.3 28.3 29.5 30.2

1.26 1.38 1.32 1.34

.236 .228 .251 .256

.129 .140 .156 .176

.107 .088 .095 .080

29.9 28.8 30.5 30.9

1.32 1.32 1.40 1.43

.307 .320 .300 .284

.125 .161 .218 .259

.182 .159 .082 .025

28.7 29.1 29.7 30.1

1.07 1.08

.295 .308

.094 .119

.201 .189

29.4 30.4

1.53 1.63 1.61 1.55

.242 .222 .237 .255

.163 .178 .192 .191

.079 .044 .045 .064

27.9 28.3 29.0 29.6

2.04 2.07 1.88 1.81

.142 .136 .151 .173

.162 .155 .141 .127

−.020 −.019 .010 .046

25.5 25.9 27.0 27.9

The horizontal dimension in Fig. 2.1 has by far the largest eigenvalue in the distance matrix, comprising 75% of the sum of the five eigenvalues. At the right end of Dimension 1 are high fertility countries, including Burundi (Buru), Palestine (Pale), and Burkina Faso (B-F). At the left end are low fertility countries, including Luxembourg (Lux), Spain (Sp), and Japan (Jpn). The correlation between countries’ positions on the horizontal dimension and their TFRs equals 0.95, indicating that this dimension substantially captures the variation in nations’ overall fertility.

2  Social Capital, Gender Competition, and the Resurgence of Childlessness

17

Fig. 2.1  Multidimensional scaling analysis of parity progression ratios for 80 contemporary populations

Variation on the vertical dimension of Fig. 2.1 is present chiefly on the left hand side of the figure. There Dimension 2 distinguishes between countries with low levels of childlessness, such as South Korea (SKor), Portugal (Por), and Belarus (Bela), which have proportions childless (P0) of 0.06 or lower, from countries with high P0 values, such as Barbados (Barb), Finland (Fin), and Ireland (Ire), all with proportions of 0.18 or above. Although less pronounced, the childlessness gradient is present across the rest of the horizontal domain. For example, toward the right hand side of the graph, Peru and Kyrgyzstan (Kyr), with proportions childless of 0.05 and 0.04 respectively, are below The Philippines (Phil) and Bahrain (Bahr), which have proportions of 0.12 and 0.09 respectively. For the entire set of 80 countries, the correlation between Dimension 2 and the proportion childless among women 45 to 49 equals 0.82. Although the eigenvalue associated with Dimension 2 is considerably smaller than that associated with Dimension 1, together they comprise 90% of the sum of the eigenvalues of the distance matrix. After the sizeable variation in overall fertility, the variation in childlessness constitutes the most important component of contemporary national fertility patterns.

18

R. Schoen and L. Hargens

2.5.2  M  easures of Childlessness in 24 Contemporary Populations Most analysts see low values of the TFR as indicating a population’s status in the first or second demographic transitions. From this perspective, the results in Fig. 2.1 suggest that there is no reason to expect that high levels of childlessness will be a feature of future fertility patterns in developed countries. We therefore turn to additional types of evidence to make the case that high levels of childlessness are likely to be a key component of fertility in those countries. Table 2.1 presents several fertility measures for each of the 24 nations for available periods beginning in the year 2000 (see Appendix B for details). The period TFRs for each population are shown in the second column of the table, confirming that we are examining low fertility countries. Only 7 of the 24 ever show a TFR as high as 1.7 (USA, Belarus, and 5 Northern/Western European countries). The United States, with a TFR of 2.074 in 2005–09, is the only country to approximate replacement level fertility. The lowest TFR, 1.068, is in Taiwan 2006–09. The two measures of childlessness are shown in columns 3 and 4. The parity status life table value for childlessness, PSLT P0, shows the average proportion childless at the end of the reproductive years from parity status life tables calculated from the observed age-parity-specific birth rates of each year in the interval. The highest life table proportion childless is 0.320 for Japan 2005–09; the lowest is 0.118 for Portugal 2000–04. Of the 24 countries, 15 have at least one interval with a PSLT P0 greater than 0.20. The PSLT P0 values show no clear trend over the 2000–17 period. For 4 countries (Croatia, Portugal, Taiwan, and Ukraine), the PSLT P0 is increasing; for 5 countries (Austria, Czechia, Lithuania, Slovakia, and Switzerland), the PSLT P0 is decreasing; and for the other 15 countries there is no consistent trend. The 45–49 P0 column shows the more common measure of childlessness, the average proportion at parity zero for women at ages 45–49. The highest cohort proportion childless is 0.259 for Japan 2015–17; the lowest is 0.027 for Estonia 2005–09. In contrast to the PSLT P0 measure, only 4 of the 24 countries (Austria, Finland, Japan, and Switzerland) have an interval where the cohort proportion childless is as high as 0.20. Also unlike the PSLT P0, the 45–49 P0 values show a temporal trend, as 13 of the 24 countries show increases in cohort proportions childless. Only 2 countries (Portugal and USA) show a downward trend, while 9 (Belarus, Denmark, Norway, Russia, Slovakia, Sweden, Switzerland, Ukraine, and Netherlands) show no consistent trend. Thus, an upward time trend emerges when completed cohort fertility is considered. This trend is masked in the PSLT, which reflects the fertility of multiple cohorts and therefore serves to anticipate future cohort levels of childlessness. A striking feature of the corresponding values in columns 3 and 4 of Table 2.1 is that the PSLT P0 is almost always larger than the cohort 45–49 P0. The difference is shown in the fifth column of Table 2.1, which is labeled ΔP0. In only five of the 88 observations does the 45–49 P0 value exceed the PSLT P0 value, and in none of

2  Social Capital, Gender Competition, and the Resurgence of Childlessness

19

these cases is the difference larger than .02. In contrast, many of the other 83 differences are of substantial magnitude, especially for nations in the lower left hand region of Fig.  2.1. These include Belarus, Czechia, Estonia, Hungary, Poland, Slovenia, and the Ukraine. Until the early 1990s these former members of the Soviet bloc had pronatalist policies that encouraged fertility among young women but also made abortion widely available so that relatively few of them bore more than two children. Subsequent governmental policy changes in these countries have led to lower period fertility measures (Sobotka et  al. 2008). In these countries, cohorts who were young adults in the 1980s have recently completed their childbearing years, and they exhibit both the low rates of childlessness and low overall fertility levels shown in Fig. 2.1. Their PSLT P0 values, which reflect the parity progression rates present in recent years, imply substantially higher levels of childlessness in the years to come. In general, then, the fact that the PSLT P0 values almost always exceed their corresponding 45–49 P0 values suggests that the future will see increases in most countries’ levels of childlessness. That these increases can occur relatively quickly when there are large differences between these measures is shown by Japan, where the level of childlessness for cohorts completing their childbearing years doubled to over 25 percent from the early 2000s to 2015–2017. Given its currently high ΔP0 values, Taiwan may well provide another instance of a surge in cohort childlessness. A further indication of trends in childlessness may be provided by trends in the mean age of mothers at first birth, the values of which are shown in Column 6 of Table 2.1. Those TMAB1 values reveal a strikingly consistent upward trend. Only in Czechia, Hungary, and Spain is there the slightest deviation from strictly monotonic increases. In the 24 study countries, nearly all mean ages at first birth are between 25 and 30 years. The lowest TMAB1 values are in Ukraine (24.3 and 24.6), Russia (24.5 and 24.9), and Belarus (24.6). The highest are in Switzerland (30.4 and 30.8), Taiwan (30.4), Portugal (30.2) and Japan (30.1). First births are occurring relatively late in the mother’s life, and are becoming even later. Since postponing is often a prelude to foregoing, the TMAB1 trend also suggests that proportions childless may continue to rise.

2.6  An Overview and the Likely Future We have examined childlessness in low fertility countries from two perspectives. The first used multidimensional scaling on parity progression ratios for 80 contemporary populations. The results showed that family building was largely comprised of two factors: the overall fertility level and the proportion childless. Next we examined period and cohort childlessness since the year 2000 in the 24 countries for which data are available in the Human Fertility Database. The results from period parity status life tables show that childlessness is approaching historically high values, levels not revealed by previous analyses based on the cohort

20

R. Schoen and L. Hargens

proportion childless at ages 45–49. That cohort measure, while indicating lower levels, shows a steady upward trend in childlessness in 13 of the 24 countries studied. Prospects for increasing childlessness are reinforced by the steady rise in the mean age at first birth in all 24 populations. There is good reason to think that the demography of the future will not repeat that of the past. Around the year 1900, the economic system of developed countries could not support universal marriage and childbearing. Now it is social factors that constrain fertility in those countries. The post-industrial world, where women are an essential part of the labor force, is far removed from the strictly gendered world of the first half of the twentieth century. Accordingly, to try and predict future trends, we look at the present results in the light of our theoretical perspective. Globalization, two-earner families, reductions in manufacturing, and an uncertain post-industrial economy appear likely to continue (Lutz et al. 2006). Pronatalist pressures are weak and likely to weaken further. There is no sign that the social capital value of children will rise, and children now bring many fewer valuable social or economic benefits to their parents. The perceived stronger bond between mother and child, in a context of relationship instability, discourages men from parenthood, and from supporting their children. The risk of harming the relationship may further deter both partners from becoming parents, suggesting that large families will remain infrequent and that childlessness will rise. Indeed, high proportions childless may well be a permanent feature of more developed populations, with many nations already having over a fifth of all women permanently childless. To echo the analysis in Preston’s PAA Presidential Address (Preston 1984), a high proportion childless is likely to reinforce patterns of less societal support for children and more resources devoted to the elderly, whose proportion of the total population will continue to rise. That scenario is likely to have many social implications. For example, long term care facilities will likely expand, while school bond issues and child welfare are likely to face increased obstacles. It is important to consider relationship stabilization and parenthood separately. Relationships are a complex combination of cooperation and competition (Stoetzel 1946), but stable relationships can offer partners substantial advantages. Those include mutual support, sexual gratification, economies of scale, and a long term planning horizon. In the Judeo-Christian tradition, as expressed by Genesis 2:18, “It is not good for the man to be alone”. That applies to the woman as well. Children are a different matter, however. The absence of children, and the stresses, costs, and division of labor issues they bring, can strengthen union stability. The same considerations apply to same-sex unions, which are now widely recognized legally. In the near term, their fertility is likely to remain demographically negligible, as it depends on adoption or advanced reproductive technologies that are still in their infancy. Partnerships are flexible, and can accommodate gender equality and dual careers. A re-equilibration that increases the equality between partners is quite likely, as the ideology of gender equality becomes more accepted and the earnings of men and women become more equal. Nonetheless, equal power is best sustained by equal contributions. Marriage is basically an economic institution, hence contributions are best measured in economic terms. Re-equilibration is thus more a matter of women

2  Social Capital, Gender Competition, and the Resurgence of Childlessness

21

increasing their economic role than men increasing their domestic role. Absent institutional support, childrearing is likely to undermine the career of at least one partner, and to aggravate gender competition within the relationship by introducing new and difficult demands on the couple’s resources and priorities. Those considerations suggest a future characterized by a two-earner couple norm, continued downward pressure on fertility, and high proportions childless.

Appendices Appendix A: Multidimensional Scaling We began by collecting recent data on completed parity distributions for countries included either in the UNdata website (72 countries) or in the Human Fertility Database (8 countries: Canada, Japan, the Netherlands, Norway, Portugal, Sweden, Taiwan and the United States). A list of all countries and their abbreviations in Fig. 2.1 is given below. Although the countries represented in the UNdata website give data for varying numbers of parity categories, we were able to gather data for all 80 countries in our analysis for parities zero through parity 5 and above. From these data we constructed five parity progression ratios for each of the 80 countries. We analyzed the 80 by 5 matrix of parity progression ratios using Stata’s mds command (see https://www.stata.com/manuals13/mvmds.pdf), and began by calculating standardized Euclidean distances between the set of parity progression ratios for each pair of countries. The mds command then carried out a principal coordinates analysis (“classical metric scaling”) of the 3160 Euclidean distances calculated in the first step. Plotting the eigenvalues associated with the five dimensions produced by this analysis showed that there was little decline after the third dimension, suggesting that a two dimensional representation of the distances is appropriate (Davison 1983, p. 69). Figure 2.1 presents the 80 countries’ positions on the first two dimensions yielded by our analysis. Abbreviations for Countries in Fig. 2.1  Albania Australia Austria Azerbaijan Bahamas Bahrain Barbados Belarus Bhutan Bolivia

Alb Aus Aust Azer Bah Bahr Barb Bela Bhu Bol

Luxembourg Macedonia Malawi Malta Mauritius Mexico Moldova Montenegro Morocco Mozambique

Lux Mac Mala Malt Maur Mex Mol Mont Mor Moz (continued)

22 Botswana Brazil Bulgaria Burkina Faso Burundi Canada Chili Columbia Costa Rica Croatia Czechia Dominican Rep. Ecuador Estonia Finland France Georgia Ghana Greece Hungary India Indonesia Ireland Israel Jamaica Japan Kyrgyzstan Latvia Lesotho Lithuania

R. Schoen and L. Hargens Bots Braz Bul B-F Bur Can Chl Col C-R Cro Cz D-R Ecu Est Fin Fr Geo Gha Gr Hun In Indo Ire Is Jam Jpn Kyr Lat Les Lit

Nepal Netherlands New Zealand Norway Palestine Paraguay Peru Philippines Poland Portugal Romania Russian Federation Rwanda South Korea Serbia Slovakia Slovenia Spain Sri Lanka Suriname Sweden Taiwan Tajikistan Thailand Trinidad & Tobago Turkey Ukraine Uruguay United States Venezuela

Nep Neth NZ Nor Pale Para Peru Phil Pol Por Rom RFed Rwa SKor Serb Slvk Slvn Sp SL Suri Swe Tai Taj Thai T&T Tur Ukr Urug US Ven

Appendix B: Fertility Data TFR values in Table 2.1 are based on yearly TFRs reported in the Human Fertility Database (HFD). We report the mean of the period TFR values for each group of years shown. Similarly, PSLT P0 values in Table 2.1 are means of proportions of women age 49 at parity zero for each group of years reported in the HFD period fertility life tables. Values for mean age at first birth are also means of annual values for each group of years reported for each country in the “table mean ages at birth” section of the HFD period fertility tables.

2  Social Capital, Gender Competition, and the Resurgence of Childlessness

23

We estimated the cohort childlessness proportions in Table 2.1 using HFD data on exposure by year, age and parity. For each country and group of years, we summed the exposure values for ages 45 through 49 and divided that sum by the sum of the exposure values for all parities for those ages. In a few cases exposure values were not available for all five ages. In these cases we averaged the ages reported by the HFD.

References Amato, P. R., Johnson, D. R., Booth, A., & Rogers, S. J. (2003). Continuity and change in marital quality between 1980 and 2002. Journal of Marriage and Family, 65, 1–22. Basten, S. (2009). Voluntary childlessness and being childfree. St. John’s College, Oxford University and Vienna Institute of Demography, Future of Human Reproduction Working Paper #5. Blake, J. (1972). Coercive pronatalism and American population policy. University of California, Berkeley International Population and Urban Research Preliminary Paper No. 2. Davis, K. (1963). The theory of change and response in modern demographic history. Population Index, 29, 345–366. Davison, M. L. (1983). Multidimensional scaling. New York: Wiley. Frejka, T. (2017). Childlessness in the United States. In M. Kreyenfeld & D. Konietzka (Eds.), Childlessness in Europe: Contexts, causes, and consequences (pp. 159–179). Cham: Springer. Goldscheider, F. K., Bernhardt, E., & Lappegard, T. (2015). The gender revolution: A framework for understanding changing family and demographic behavior. Population and Development Review, 41, 207–239. Hobcraft, J., & Kiernan, K. (1995). Becoming a parent in Europe. Paper presented at the European Population Conference, Milano, 4–8 September. Jasilioniene, A., Jdanov, D.  A., Sobotka, T., Andreev, E.  M., Zeman, K., & Shkolnikov, V.  M. (2015). Methods Protocol for the Human Fertility Database. Downloaded 7/25/2019 from the Human Fertility Database website. Kreyenfeld, M., & Konietzka, D. (2017). Analyzing childlessness. In M.  Kreyenfeld & D. Konietzka (Eds.), Childlessness in Europe: Contexts, causes, and consequences (pp. 3–15). Cham: Springer. Lesthaeghe, R. (1995). The second demographic transition in Western countries: An interpretation. In K. O. Mason & A.-M. Jensen (Eds.), Gender and family change in industrialized countries (pp. 17–62). Oxford: Clarendon Press. Lesthaeghe, R. (2010). The unfolding story of the second demographic transition. Population and Development Review, 36, 211–251. Lutz, W., Skirbekk, V., & Testa, M. R. (2006). The low fertility trap hypothesis: Forces that may lead to further postponement and fewer births in Europe. Vienna yearbook of population research, 4, 167–192. Mills, M., & Blossfeld, H.-P. (2013). The second demographic transition meets globalization: A comprehensive theory to understand changes in family formation in an era of rising uncertainty. Chapter 2. In A.  Evans & J.  Baxter (Eds.), Negotiating the life course: Stability and change in life pathways. Dordrecht: Springer. Preston, S.  H. (1984). Children and the elderly: Divergent paths for America's dependents. Demography, 21, 435–457. Rowland, D. T. (2007). Historical trends in childlessness. Journal of Family Issues, 28, 1311–1337. Schoen, R. (2010). Gender competition and family change. Genus, 66, 95–119.

24

R. Schoen and L. Hargens

Schoen, R., Kim, Y. J., Nathanson, C. A., Fields, J., & Astone, N. M. (1997). Why do Americans want children? Population and Development Review, 23, 333–358. Schofield, R. (1984). English marriage patterns revisited. Journal of Family History, 10, 2–20. Seltzer, N. (2019). Beyond the Great Recession: Labor market polarization and ongoing fertility decline in the United States. Demography, 56, 1463–1493. Sobotka, T. (2017). Childlessness in Europe: Reconstructing long-term trends among women born 1900–1972. In M. Kreyenfeld & D. Konietzka (Eds.), Childlessness in Europe: Contexts, causes, and consequences (pp. 17–53). Cham: Springer. Sobotka, T., St’astna, A., Zeman, K., Hamplova, D., & Kantorova, D. (2008). Czech Republic: A rapid transformation of fertility and family behaviour after the collapse of state socialism. Demographic Research, 19(14), 403–454. Stoetzel, J. (1946). Sociologie et demographie. Population, 1, 79–89. United Nations, Department of Economic and Social Affairs, Population Division. (2013). World Fertility Report 2012. New York: United Nations. Van de Kaa, D. (1987). Europe’s second demographic transition (Population Bulletin 42). Washington DC: Population Reference Bureau. Vignoli, D., Drefahl, S., & DeSantis, G. (2012). Whose job instability affects the likelihood of becoming a parent in Italy? A tale of two partners. Demographic Research, 26(2), 41–62. Weir, D. R. (1984). Better never than late: Celibacy and age at marriage in English cohort fertility, 1541–1871. Journal of Family History, 9, 340–354. Zaidi, B., & Morgan, S.  P. (2017). The second demographic transition theory: A review and appraisal. Annual Review of Sociology, 43, 473–492.

Chapter 3

Uncertainty and Narratives of the Future: A Theoretical Framework for Contemporary Fertility Daniele Vignoli, Giacomo Bazzani, Raffaele Guetto, Alessandra Minello, and Elena Pirani

Imaginative forecast of the future is this forerunning quality of behavior rendered available for guidance in the present. John Dewey

3.1  Introduction Understanding the relationship between economic and fertility trends is a challenge for demographic research. Karaman Örsal and Goldstein (2018) took a large group of middle–high income countries and looked at the relevant data from the post–war period onwards. They showed that, since 1970, good economic conditions have led to higher fertility, while bad economic conditions mean lower fertility, suggesting a pro–cyclical trend (see also Myrskylä et  al. 2009). However, a close look at the demographic trends at the beginning of the twenty-first century casts doubts on this kind of interpretation. Economic indicators suggest that European countries are currently moving out of the Great Recession, whereas fertility trends are not so positive. For instance, in 2009 Northern European economies resumed economic growth, but their total fertility started to decrease substantially. In Norway, total fertility dropped from 1.98 in 2009 to 1.6 in 2018, the lowest ever in peacetime; similar changes, and even lower fertility levels, have been observed in Denmark, Finland, Iceland and Sweden (Comolli et al. 2019). On the other side of Europe, Mediterranean countries such as Italy, Greece and Spain, after a period of fertility rebound, re–entered, in the same period, a regime of lowest–low fertility, with total fertility around 1.3.

D. Vignoli (*) · G. Bazzani · R. Guetto · A. Minello · E. Pirani University of Florence, Florence, Italy e-mail: [email protected]; [email protected]; [email protected]; [email protected]; [email protected] © The Author(s) 2020 R. Schoen (ed.), Analyzing Contemporary Fertility, The Springer Series on Demographic Methods and Population Analysis 51, https://doi.org/10.1007/978-3-030-48519-1_3

25

26

D. Vignoli et al.

Explanations for fertility decisions based on structural constraints—such as labor, housing condition, or income—may account for a substantial share of cross– country differences in fertility, nevertheless important questions remain unanswered, posing major challenges to contemporary demographic theories. Given the disjuncture between economic and fertility trends, what are the drivers of the low fertility in contemporary European societies? The central explanation we put forward for this new state–of–affairs is the rise of uncertainty. The future is inevitably uncertain, and uncertainty is a structural factor implied in any long–term decision– making process, such as the fertility choice. In addition to this fundamental uncertainty, though, we would argue that recent economic developments in Europe—namely, the increasing speed and volatility of outcomes of globalization, and the new wave of technological changes—have amplified uncertainty in people’s life, adding a contingent component of economic uncertainty. Economic uncertainty makes it increasingly difficult for individuals to imagine their future, choose between alternatives, and form strategies. Notwithstanding its theoretical importance, uncertainty is rarely considered in traditional explanations of fertility. We suggest that the interpretation of recent fertility trends needs a clear action theory where uncertainty has a central role. We argue that fundamental uncertainty needs to be conceptualized and operationalized taking into account that people use works of imagination, producing their own narrative of the future—i.e. imagined futures embedded in social elements and their interactions. The medium and long–term future cannot be predicted with any degree of certainty, but people can sketch out their personal narratives of the future, and, on the basis of these, take decisions. In the “life course cube”, a re-conceptualization of the life course as a set of interdependences between time, life domains, and levels of analysis, this is referred to as the “shadows of the future” (Bernardi et al. 2019: 4). The narratives of the future become potent driving forces for fertility intentions: people might plan a child according to or despite uncertainty, irrespective of structural constraints and their subjective perceptions. Fertility intentions follow the desire for childbearing and anticipate concrete behavior by reflecting the combined effect of desired fertility and situational constraints (Thomson and Brandreth 1995). Fertility intentions have been generally regarded as a fairly suitable predictor of behavior at the individual level (Westoff and Ryder 1977; Schoen et al. 1999), provided that a time frame for the realization of the intention is set (Régnier-Loilier and Vignoli 2011). We continue by offering a view on the main conceptualizations of uncertainty and defining the theoretical perspective we adopt. Then, a brief review of existing theories on reproductive behavior, from past to more recent theoretical approaches, is presented. Overall, each one of the previous approaches added various and novel tesserae to the mosaic of low fertility; nevertheless, none of them explicitly addressed the role of (rising) uncertainty. We continue by proposing a novel framework (the Narrative Framework) for the study of fertility decisions under uncertain conditions based on expectations, imaginaries and narratives: we argue that narratives of the future might help to disentangle the nature and the role of the elements

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

27

involved in the fertility decision–making process. Then, we address the causal validity of the narrative framework for contemporary fertility. We conclude by highlighting the advantages of taking into account narratives of the future in fertility research.

3.2  The Notion of Uncertainty In the literature, there is an ongoing debate over the definition of uncertainty and its relevance for social dynamics. A valuable starting point is the classical work of Knight ([1921] 2006), who distinguished between risk and uncertainty. In a risky condition, the probability distribution of future events is completely known, and outcomes can be calculated or, at least, estimated by classifying on the basis of a known probability; in an uncertain condition, outcomes are not homogeneous enough to be estimated through probability calculus, or they are purely unknown. More recent definitions of uncertainty do not pose a particular challenge to the Knightian distinction between risk and uncertainty, but they highlight specific, or recurrent, features in the proposed notions of uncertainty. In Table 3.1, we offer a synthetic schema of the most important definitions of uncertainty given in the recent economic and sociological literature, classifying various types of uncertainty by their source. All the identified conditions of uncertainty imply that future outcomes cannot be measured probabilistically. A common source in creating personal uncertainty for a given actor, that we find in some but not all authors’ conceptualizations, are social interactions and the roles of other actors. Human behavior can never be totally predicted. Thus, personal decisions and plans based on others’ expected (rational) actions may easily fail. Individuals may learn from past experiences, reflect on their cognitive processes, and act strategically using their imagination and creativity: these capacities allow them to shift away from the expected course of action. A second conceptualization of uncertainty focuses on the quality and the quantity of the available information needed for the selection and evaluation of the alternative courses of actions and their consequences. Uncertainty may arise because information is missing or because there is no feasible access to information due to ignorance or because of the limited time availability for collecting it. Finally, the last column of Table 3.1 shows the more radical condition of uncertainty, or fundamental uncertainty, that can be an effect of the other two, but may also arise independently. Under fundamental uncertainty, the effects of the present action cannot be successfully forecast or estimated; the list of possible future outcomes is not complete, nor can the elements involved in the course of action and their roles be known with precision. In the context of reproductive behavior, a fertility decision is always taken in a condition of fundamental uncertainty: despite the level of uncertainty experienced by individuals and couples, a fertility decision has to be taken in the present. Through this decision, people plant a seed that will germinate and grow in their

28

D. Vignoli et al.

Table 3.1  Types of uncertainty classified by their sources Social interaction Davidson (1996)

Dequech (2000)

Lane and Maxfield (2005)

Elster (2009)

Beckert (2016)

Semantic uncertainty: actors are uncertain about what a proposition of other actors means Strategic uncertainty: strategic action of other actors is a spiral source of uncertainty that cannot be defined Social interaction uncertainty: third parties’ actions cannot be accurately predicted, even in game theory models

Information Epistemological uncertainty: a complex situation involving too much information that cannot be computed Ambiguity: relevant information is missing or cannot be accessed

Fundamental Ontological uncertainty: creativity and innovation cannot be predetermined but only observed retrospectively Fundamental uncertainty: the creativity implied in the future cannot be deducted from present information Ontological uncertainty: refers to the entities that inhabit the world, their modes of interaction, and the results of their interaction

Truth uncertainty: actors are uncertain about whether well– defined propositions of future consequences are true or not Information gathering Brute uncertainty: no uniform distribution of cases can be invoked uncertainty: the gathering of information cannot be rationally stopped

Complexity uncertainty: a complex situation does not allow utility maximization

Tuckett and Nikolic (2017)

Fundamental uncertainty: innovation is unpredictable and cannot be estimated in present calculus

Radical uncertainty: equivocal situations in which uncertainty about the outcomes of actions is so profound that it is both difficult to set up the problem structure to choose between alternatives and impossible to represent the future in terms of a knowable and exhaustive list of outcomes to which to attach probabilities

Own elaboration based on Beckert (2016)

short– and long–term future. Parenthood is a unique and irreversible experience, where what can be learned from the past does not often apply to the present and to the future, not least because children change like their parents through the years, and because children differ from one another. Fundamental uncertainty has always accompanied fertility decisions; this is not a novelty, then. Nevertheless, recent societal changes experienced by post–industrial societies—i.e. changes classified under the umbrella of globalization, and the

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

29

neo–liberal policies that accompanied it—added new contingent elements of uncertainty to life plans. The impact of economic uncertainty on demographic behavior has been at the core of demographic research since its earliest years. However, it has been recognized that a “harsh new world of economic insecurity” (Hacker 2019: xvi) only appeared as of the 1980s. This was a world characterized not only by increasing instability in individuals’ employment, but also by rising inequalities (OECD 2011). Although inequality and uncertainty in the economic context are not the same thing, they are strongly interwoven (Hacker 2019). In the era of globalization, economic uncertainty is also amplified by the intensification of worldwide social relations through the information and technology revolution. Social interactions are more and more numerous and complex, and the media often multiply uncertainty as growing information does not imply its intelligibility and utility. For a majority of citizens, the media are an essential source of information on complex economic issues (Boomgaarden et al. 2011), which also evaluate, filter, and simplify information. The perception of economic uncertainty is thus strongly anchored in public images produced by the media and other leading opinion formers, like politicians. During the years of the Great Recession, media news contributed to the emergence of a European public sphere with a pessimistic view of a stagnant, underperforming continent (Davis Cross and Ma 2013). The Great Recession was popularized by a tsunami of news that focused on the crisis as the evil of contemporary European societies (Cepernich 2012), even in countries that did not experience a real economic recession. This is a novelty compared to previous recessions. In the Great Depression of the 1930s, when a rapid surge in unemployment was followed by a drastic drop in fertility, economic information was not as amplified and diffused as it is in the era of globalization. Classical theoretical perspectives on fertility did not deal with the issue of fundamental uncertainty, nor with the increased salience of uncertainty due to globalization dynamics. In the Narrative Framework, we acknowledge the concept of uncertainty and study how a fertility decision can be taken according to or despite it.

3.3  Classical Perspectives on Low Fertility In the second half of the twentieth century, the two most influential perspectives on fertility have been the New Home Economics (NHE) (Becker 1964) and the Second Demographic Transition (SDT) (van de Kaa 1987; Lesthaeghe 1995). Becker (1981), following a strict microeconomic approach, considers fertility behavior as an individual action oriented to utility maximization. The concept of utility remains largely undetermined (Strandbakken 2017), but thanks to this indeterminacy the microeconomic approach has been applied to family decisions as well as to almost all domains of social life. Utility maximization implies that the availability of higher economic resources—driven, for example, by the increasing contribution of women to household income—may have ambiguous effects on fertility. On the one hand, women’s employment fuels permanent (household) income and may foster fertility

30

D. Vignoli et al.

(income effect). On the other hand, women’s employment reduces the time for childrearing; therefore, working and having children may become competing tasks (substitution effect). Sociological approaches refer to the latter mechanism as evidence of role incompatibility—or the inability to combine mother and worker roles in a modern economy where home and workplace are separated. Despite the existence of empirical evidence on this kind of substitution effect (Matysiak and Vignoli 2008, 2013), the application of a strict economic approach to fertility behavior may create a stylized and unrealistic type of family agency,1 in which partners calculate the costs and benefits of a child, discounting the actual cost in the light of future utility (Caldwell 1982). Usually, human actions are a mix of different ideal types of agency (Weber 1978 [1922]), and fertility decisions, in particular, are complex decisions where interests, values, opportunities, and social ties interact. For example, a different explanation for declining fertility in spite of increasing levels of household income is provided by Schoen et al. (1997). They opposed simplifying the fertility decision to the economic cost of children, and claimed for their value as social resources: As kin ties weaken and children bring comparatively fewer social or economic resources to their parents, fertility is unlikely to recover in low–fertility populations. The other classical perspective on low fertility, the Second Demographic Transition (SDT; van de Kaa 1987; Lesthaeghe 1995), builds, instead, on the sociological foundations of value change and individualization. The idea is that, in post– modern societies, individuals, in particular women, reprioritize career and self–actualization over family and childbearing. Even if not stated explicitly in the original formulation of the theory, this argument has often been used to argue that the increase in women’s education and employment anticipated fertility decline in Western countries during the second half of the twentieth century (Guetto et al. 2015). The SDT does not contemplate economic uncertainty, and from the NHE one can only indirectly conclude that it may matter. Income level is important because people make decisions subject to budget constraints and if the budget is uncertain, they cannot make such decisions. Generally speaking, in these two frameworks the demand for fertility is conceived as being determined by permanent (household) income, the opportunity cost of children, tastes, and self–realization needs—all factors assumed to be relatively stable over individuals’ life courses, and subject to a very slow pace of change at the societal level. In recent years, however, the role of (increasingly pervasive) uncertainty cannot be disregarded in the fertility decision– making process. After all, on this uncertainty will depend important fluctuations in income, wealth, and preferences (Hacker 2019). More recent developments in demographic theory stress the importance of gender equity within couples and at the societal level in understanding trends and cross–country differences in fertility rates (McDonald 2000). According to the

1  The concept of agency refers to the human capacity to act independently or strategically irrespective of the influence of the social structure (Emirbayer and Mische 1998: 988).

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

31

theory of multiple equilibria (Esping-Andersen and Billari 2015) or gender revolution (Goldscheider et al. 2015), (very) low fertility rates would be a temporary phenomenon following the rise in female employment. However, fertility rates would, these theories proffer, tend to return to replacement level as societies adapt to new women’s roles. Beyond being challenged by the recent evidence of a fertility fall in gender–egalitarian Nordic countries, this approach does not assign any relevant role to rising uncertainty. The change in gender role attitudes in recent decades have inspired a further interpretative line. Specifically, the diverging perspectives between the male and the female member of the couple, namely any competition between them, have been argued to be an underappreciated element in the literature. This competition, it is suggested, would potentially foster a retreat from marriage and contribute to low fertility as men increasingly withhold economic support from their children (Schoen 2010). Once again, even if economic aspects are part of the framework, especially in terms of power in the relationship, uncertainty is not accounted for.

3.4  Objective or Perceived Economy? In demographic research, economic uncertainty has so far been viewed as an individual risk factor, mainly related to the labor market (e.g., unemployment, short– term contract jobs, underemployment, or a combination of these; Mills and Blossfeld 2013; Vignoli et al. 2012, 2019; Dantis and Rizzi 2020). A persistent experience of employment uncertainty may lead to the perpetual postponement of family formation and, as a result, to a smaller family or even to no family at all (Busetta et al. 2019). Several studies tried to assess the effect of objective employment indicators on fertility (see Alderotti et al. 2019 for a meta–analysis of micro–level research findings for Europe). What is generally not considered in these studies is that objective indicators alone are perhaps not good proxies of the perceived economic condition, because individuals differ in the extent to which they feel, tolerate and react to the same objective condition. In the light of these limitations, a different stream of research focused on the effects of perceived economic uncertainty on fertility. Ranjan (1999) claimed that perceived economic uncertainty played an important role in fertility trends in post– communist countries after the communist regimes had collapsed there. In the last years, information regarding perceived economic conditions is more frequently included in surveys. For instance, the availability of individuals’ perception of economic  conditions in the German Socio Economic Panel (GSOEP), enabled Kreyenfeld (2009, 2016), Bhaumik and Nugent (2011) and Hofmann and Hohmeyer (2013) to explore its relationship with fertility. Even if these studies recognized that fertility choices are not only influenced by the objective side of uncertainty, they fail to recognize the future-oriented nature of uncertainty. An increasingly popular approach used in demography to identify the fertility decision–making process is the socio–psychological framework of the Theory of Planned Behaviour (TPB) (Ajzen 1991), which stems from the Theory of Reasoned

32

D. Vignoli et al.

Action (Fishbein and Ajzen 1975). In this conceptual model, an action is the result of actors’ attitudes toward the behavior, subjective norms (dependent on the relevant others’ perceptions of the behavior) and perceived behavioral control (self– efficacy) (Ajzen and Klobas 2013). These constructs are operationalized within the TPB in a hypothetical situation in the next three years. The empirical validation of the TPB is highly problematic and much debated (Schoen et al. 1999), especially in terms of the role of background factors and structural constraints (Mencarini et al. 2015). Nonetheless, the TPB is one of the few forward–looking approaches developed for the study of a fertility decision–making process. The TPB tries to predict fertility behavior with a set of elements that still rely on a deterministic approach, disregarding an individual’s capacity to deviate from the expected course of action. From our perspective, hence, the TPB misses one crucial element in its forward– looking approach, namely the imaginative capacity of human agency. In the Narrative Framework, the TPB elements are part of the structural constraints that shape the course of action, but individuals may also deviate from an expected course of action thanks to their imaginative capacity.

3.5  The Narrative Framework We propose a conceptual framework—the Narrative Framework—to investigate the fertility decision–making process in a state of fundamental uncertainty. The study of the future is a growing field in many branches of economics, sociology, psychology, psychoanalysis and anthropology, but it is still not considered while analyzing contemporary fertility. We build upon these theoretical bases, complementing our proposal with a conceptual distinction between expectations, imaginaries and narratives. In real life, the separation between expectations, imaginaries and narratives is often blurred, not least because each one influences the other. For analytical purposes, a conceptual distinction is necessary, however. People form their own expectations on the basis of structural and contingent constraints. Although expectations come from structural constraints and past experiences, their influence is not deterministic, and the knowledge of the past does not include all future possibilities: the future is not merely a “statistical shadow of the past” (Davidson 2010: 17; Beckert and Bronk 2018). Expectations are thus the foundation for the imaginaries of the future, although they often do not coincide. Imaginaries draw on expectations, but they may also deviate from the expected future, thanks to the imaginative capacity of humans. Structural constraints, expectations and imaginaries find their proper place in narratives of the future, the less abstract level of the imaginative capacity, able to sort them in an intelligible and actionable manner. The aforementioned elements are included in the narrative of the future and, at this level, they influence fertility intentions. We propose a graphical representation to better exemplify the decision–making process (Fig.  3.1). Each of the four levels—narratives of the

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

33

Fig. 3.1 Fertility decision–making process under conditions of uncertainty: a stylized representation of the Narrative Framework

future, imaginaries, expectations and structural constraints—stems from the previous, but it is not fully determined by it; in fact, each element can exceed what can be expected from the previous elements. Fertility intentions, of course, will also have effects on the structural constraints and the past experiences of the next courses of action. The fertility decision–making process is a clear example of a situation of fundamental uncertainty where expectations, imaginaries and narratives of the future matter together with any structural constraints and past experiences. Typical questions arising during the decision–making process are related to structural constraints: they might be of a micro nature (e.g., housing or labor circumstances), of a meso nature (e.g., the role of familial or friendship networks), or of a macro nature (e.g., the context for balancing paid work and family life). But these objective conditions cannot alone predict fertility intentions: Facing the same structural constraints, people do not necessarily make the same choices. For example, an uncertain labor condition may not be an obstacle to having a child if strong economic growth is expected, but it may inhibit fertility when coupled with the expectation of economic decline (expectations). However, neither expectations nor structural constraints alone can predict a fertility decision: human beings still have agency—i.e. an imaginative capacity, something which allows them to deviate from the expected course of action. For example, a wishful future involving numerous descendants or a strong belief in the sacredness of family (imaginaries) may encourage childbearing notwithstanding adverse economic expectations or a condition of income hardship. Expectations and structural constraints together with (family) imaginaries contribute to the definition of a narrative of the future driving the fertility decision– making process. Here positive fertility intentions may be formulated despite the uncertainty of the future or fertility may be avoided according to the condition of uncertainty.

34

D. Vignoli et al.

The following subsections present the different elements of the narrative theoretical framework.

3.5.1  Structural Constraints The contingency of action under uncertain conditions is limited by structural constraints, such as cultural frames, conventions, rules, and institutional settings (Beckert and Bronk 2018). All these elements reduce the potentially limitless possibilities of action by imposing a limitation (Offe 1998). This limitation, however, does not imply the entire predictability of the future. Indeed, the limitations deriving from structural constraints can be encouraged, moderated or thwarted by the agency of the elements involved (i.e., individuals, organizations, social groups, technology, and so forth) and by their contingent interpretations. The fertility decision-making process  is  clearly influenced by the presence of structural constraints. In Western countries, the fall in fertility rate from the first to the second generations of migrants from high–fertility countries shows how the social context influences fertility decisions (Kulu 2005). Different generations of immigrants are, indeed, influenced by different structural constraints: local cultural values, as opposed to country of origin cultural values, seem to affect fertility decisions differently for the first and the second generations of immigrants. While the fertility decisions of first–generation migrants are likely to be strongly influenced by the predominant cultural values in the origin countries, second–generation migrants, being exposed to cultural values of the destination country, are more likely to distance themselves from the values of their parents (Guetto and Panichella 2013). In this case, the influence of the social context in fertility decision varies in relation to the age and the period of exposure to a different culture and institutional setting over the life course. The influence of the structural constraints on decision and action is not always mechanical, however. Bourdieu ([1980] 1990), for example, referred to the concept of habitus as, also, a resource for action. In his original framework, the habitus, shaped by past experiences, is incorporated into the body with a pre–conscious set of expectations about the future. Nonetheless, structural constraints, despite being naturalized and taken for granted, can also be “strategically mobilized in accordance with the contingencies of particular empirical situations” (Emirbayer and Mische 1998: 978). Structural constraints often shape the life course with a silent reduction in possible actions. But they can also be strategically used by the actors, for example thorough their inclusion or exclusion in an expected future. Within the second generations of migrants it may, for example, happen that the traditional values of the country of origin are deliberately used to build a personal identity different from that of native peers. In this case, the number of children may increase despite the expected influence of local cultural values, thanks to the strategic use of structural constraints.

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

35

3.5.2  Expectations Beliefs about the future are the compass guiding the decision–making process: “When making decisions, actors associate certain future results with the course of action they are contemplating, connecting numerous outcomes with different possible decisions. These perceptions are known as expectations” (Beckert 2016: 35). Expectations represent the whole complex system of beliefs about the future that an individual takes into consideration in the decision–making process. Past experiences act in a fundamental way here, but the past cannot determine the outcome of the process per se: expectations of the future play a key role in selecting actionable decisions from the available set. In the social sciences, expectations have taken on different roles and levels of importance in action theory, across the years and across different disciplines. Sociological approaches have traditionally been more interested in understanding the role of the past in explaining present actions than in the role of expected futures. Even when expectations are considered as an essential part of a given course of action, they seem to reflect something that has already happened in the past. Schütz (1962), for instance, devotes particular attention to the role of expectations in his theory of action: expected typicality of the events informs the course of action. However, typicality remains a concept anchored in the past and cannot account for the role of expectations and individual agency in social reality.2 On the other side, expectations of future gain, or utility, play a key role in economists’ accounts of economic dynamics. In mainstream economics, the possibility of an expected long– term equilibrium in the markets is connected to the capacity of individuals to forecast successfully and to make investments and consumptions in the light of this predicted future. In our Narrative Framework, let us imagine a hypothetical young woman in her late twenties, one who has completed her education. Supported by her social and cultural environment, which promotes a two–children norm, she might expect a future family with children. This family expectation, however, clashes with her economic expectations: she thinks she will never get a permanent contract, at least not in the near future; moreover, she is concerned about possible difficulties in reconciling work and family. This example suggests how expectations are embodied in the currently available set of actions and influence the decision–making process, until a fertility intention is formulated. Regardless of their truthfulness, rationality or plausibility, expectations are real, and able to guide the fertility decision–making process. Expectations shape an expected future, but human agency may deviate from the expected course of action toward an imagined future.

2  The role of future expectations has also been considered in sociological approaches relying on some forms of Rational Action Theory, such as in the field of social stratification and inequality (Goldthorpe 2007), as well as in the stream of research focusing on the educational aspirations of immigrants and their descendants (Kao and Tienda 1995).

36

D. Vignoli et al.

3.5.3  Imaginaries Imagination is the capacity to place oneself in one or more imagined situations, hypothesizing their effects. But also, and more radically, it means the capacity to imagine a wishful future that cannot be deducted from the present. Imaginaries are imagined futures that join together elements of the present with some normative value orientations; they can be related to collective outcomes (e.g., an egalitarian or a carbon–free society) but also to individual goals (e.g., a family with many children or living in a house with a pool). Fertility decisions are often connected with a family imaginary that may be seen either as being wishful (e.g., because of the parenthood experience) or as being frightening (e.g., because of the reduction in free time). Importantly, imaginaries can themselves be a cause of uncertain futures because they make people deviate from the expected future. They also represent an important tool to cope with uncertainty: imaginaries allow actors to move beyond inherited thought–patterns and categories; to invent entirely novel ideas; to spot emerging patterns; to choose between visualized but counterfactual options (Bronk 2009). Imaginaries help to de–routinize the course of action by replacing routines with action models characterized by a higher level of consciousness and reflexivity. Continuing with our previous example, in the imagined future our hypothetical young woman sees herself married with two children in an owner occupied home, close to her parents who may help her. Possibly she has a permanent job contract.

3.5.4  Narratives The gap between imaginaries and the present course of action is filled by narratives.3 When imaginaries are associated with a hypothetical course of action, they constitute a narrative of the future. The interplay between structural constraints and agency can be disentangled through the study of narratives that, especially for long– term decisions, provide a goal and show how it can be reached with specific elements, actions, and limitations. In this sense, a narrative of the future subsumes structural constraints, expectations, and imaginaries. Narratives of the future perform four functions: (i) they select the key elements of the story and avoid what is considered irrelevant for the events at stake (selection); (ii) they interpret their value and meaning (interpretation); (iii) they connect the elements in temporal order identifying the causes and effects of the action (causal modelling); and iv) they support the action rationally and emotionally (action support) (Fig. 3.2). The selection process refers to a basic cognitive function. During social action, actors can only focus attention on a small number of elements. For example, what 3  There are different meanings of narratives across different disciplines, and a review of the use of the term is far beyond the scope of this chapter (for an introduction, see Emirbayer and Mische 1998).

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

37

Fig. 3.2  From imaginaries to fertility intentions: the four functions performed by a narrative of the future

matters most in the conscious deliberation of a fertility decision between personal and partner’s labor contracts, household income and savings, housing condition, public services or peers’ and parents’ opinions is not obvious. Relevant information for the decision process is selected—consciously or unconsciously—from an almost infinite possible set. In any given moment, social action can consider only a limited set of elements and information, both because of cognitive limitations and because of the risk of inaction due to excess information. The zone of attention contains both preconscious elements, given by the habitus, as well as innovative elements, given by the capacity of imagination. The interpretation process of the selected elements consists of two main phases: typification and classification. The first step of an actor in interacting with a new element of the context is the recognition of the new element’s analogies with things already experienced, namely typification (Schütz 1967). The second step is to classify the selected elements. For example, after the selection of the stability of labor contract as a key element in the fertility decision–making process, the next step would be to assess to what extent the current contract can be considered as being stable. The classification process often follows a matrix of binary oppositions (Lévi-­ Strauss 1963) (e.g., stable/precarious, enough/not enough, short–term/long–term), but it may also involve a more complex system of relationships (e.g., economic sectors). Lines of separation may be nuanced and the classification of elements may not be easy, especially in long–term decisions where there is fundamental uncertainty. The classification process always happens in the light of specific imaginaries. While imaginaries may remain in the background in a decision–making process with clear alternative outcomes, in the case of a decision involving fundamental uncertainty, a narrative of the future tries to align the selected elements in the direction of the imagined future. For example, while in many cases unemployment is considered as

38

D. Vignoli et al.

a deterrent to parenthood, its expected negative influence on plans for parenthood may be overlooked when there is an imagined wished–for future involving a child. The definition of the necessary conditions for reaching the goal (e.g., a family with children) implies a causal path involving the indispensable elements to achieve the imaginary (e.g., a stable job is necessary to save money and to rent a house with n bedrooms and thus to plan childbearing). The process of setting the necessary conditions and elements for reaching the goal is the causal modelling function of the narratives of the future. Thanks to the capacity of the imagination, people may consider alternative combinations of elements or means–end sequences. Narratives reflect the interplay between individual agency and structural constraints in a given context. Narratives help people to coordinate their social action in a condition of uncertainty, and allow for the construction of everyday meanings and their implicit causal mechanisms (Bruner 1990). The connection between the social elements of the past, present and future through causal mechanisms also sustains the emotional commitment of individuals to act despite or according to the uncertainty they face (the action support function of narratives). All in all, narratives provide reasons for action. Irrespective of the extent to which these narratives may be false or the actions questionable, they have the power to reduce world complexity (selection process). Narratives make a given environment more intelligible and actionable (thanks to interpretation and causal modeling) and support the ongoing efforts of dealing with uncertainty (action support). Importantly, the more the decision to be taken has important, long–term effects, as in the case of fertility, the more a conscious narrative of the future is needed to help with selection, interpretation, causal modeling and to support the action. To conclude the example introduced in the previous Sections, our young woman could assign a prevailing importance to the pre–conditions considered as necessary to start a family and under which children should be raised (e.g., property ownership and a good income). In a first phase this leads our young woman to postpone childbearing in order to achieve economic stability. Then, while keeping faith with her two–children fertility ideal, approaching her mid–thirties she might consider having only one child as a way to adhere at least partially to her imagined future. Personal narratives of the future are not only a matter of psychological attitudes or individual intentionality, but they are also the place where the social context (structural constraints) takes an intelligible form and provide elements and reasons for action. They are the hinge that keeps the link between individual and society, favoring their mutual influence and interdependence and, at the same time, allowing for a separate accounting of both sides. Personal narratives will never be totally socialized, nor can they ever be totally independent of context. Hence, personal narratives of the future are anchored in existing cultural and institutional frames, as well as public images produced by the media and other powerful opinion formers. Based on socially–constructed perceptions, people build their personal narratives of the future to act according to or in spite of uncertainty, irrespective of structural constraints and their subjective perceptions. The building blocks of personal narratives are thus shared narratives produced by several agents of socialization, such as parents, peers and the media (Vignoli et  al. 2020a). Through the analysis of

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

39

narratives and their building blocks the researcher can find hints to put together the causal explanatory chain of fertility, which is under construction.

3.6  The Causal Power of Narratives We posit that in an era of amplified and pervasive uncertainty the role of narratives of the future gain importance in facilitating or inhibiting fertility decisions irrespective of—objective and subjective—structural constraints. The Narrative Framework, however, shares together with other explanatory models of social action the two main problems of nomological explanations described by Davidson: namely specific causality and generic causality problems (Davidson 1980 [1976]). First, the narrative explanation is not backed up by a general law able to explain all similar actions under the same law (specific causality problem): another mental status may cause the same action. Second, the narrative explanation cannot state what is the final cause of the phenomenon observed (the generic causality problem): another level in the causal chain may represent a deeper level in causal explanation, for example, hormones or neurological connections. Narratives may actually be invoked as a way of dealing with the specific causality problem, contributing to “choosing among various plausible interpretations of an action in terms of possible reasons” (Stueber 2008: 42). Notwithstanding the fact that different (pre or post) rationalizations of fertility decision will always be possible, narratives attribute reason(s) to the action. Using the words of Uebel (2012: 43–44), narratives “deepen the causal claim by spelling out the context of the attributed reason, embedded in a personal and social history”. In this sense, a personal narrative may be able to explain the intention to have a child, although this does not necessarily impede alternative narratives in the mind of another person with the same constraints. For example, the question of whether the increase in objective economic uncertainty may inhibit fertility intentions (Busetta et al. 2019), or rather facilitates them (consistently with the uncertainty reduction framework; Friedman et al. 1994) cannot be solved under a generic law. Only in the light of a narrative of the future, do the objective indicators of economic uncertainty find their proper role (selection) and value (interpretation) in relation to fertility intentions (causal modeling), providing individuals with sufficient levels of commitment to allow for action (action support). As stated, the causal chain may always have one more deep level, internal or external to individuals, to be used as the final causal reason to act (Freese 2009). The concept of mechanism helps in framing the limits of the causal chain in the social sciences, and, particularly, in the Narrative Framework. An explanation based on social mechanisms does not aspire to set up the final cause over the observed event; instead, it seeks to explain how the action is brought about and under which conditions it can take place (Hedström 2005). Whereas this aim is shared by numerous explanatory models, the added value of the mechanisms approach is to describe the

40

D. Vignoli et al.

elements of a continuous and contiguous causal chain (Hedström and Swedberg 1998; Hedström and Bearman 2009) leading to fertility decisions. The claim for the continuity and contiguity of causes of the decision–making process allows for a shift in the focus from the final cause to the plausibility of this cause in the light of the whole action chain under consideration. In this sense, the economic uncertainty that permeates contemporary globalized societies may be seen only with difficulty as the final cause of a fertility decision. However, it may affect the action chain. A concrete example may, once again, prove useful. Given unsatisfactory housing conditions for a couple, the degree of expected uncertainty in a partners’ future labor condition and income may play a crucial role in the formulation of the intention to have or to not have a child. After all, these labor uncertainties sustain a narrative of the future in which it seems impossible to improve housing conditions. In this case, economic uncertainty may not be the final cause (the generic causality problem remains unsolved). But it leads to the decision because it highlights that the negative housing condition cannot be changed (causal chain continuity) and this defines the couple’s narrative of the future (the specific causality problem is addressed). Not all the actions are taken under conscious deliberation and through the evocation of narratives. In daily life, routine prevails and imitation may play a central role in guiding actions without the need for narratives, or intentions or motivations. In these cases, narratives can be seen as an ex–post justification of a previous course of action without any specific claim for causality and any reason for the action. Nonetheless, the fertility decision is life–changing and is intimately matched with a narrative of future parenthood, irrespective of how simplistic or infeasible this future may appear to an external observer (Todd et al. 2013). On the one hand, narratives help individuals in selecting the relevant information from a given context, its interpretation and causal modeling. On the other hand, narratives provide crucial information to the researcher on what elements need to be considered to be relevant in a fertility choice. The researcher is, thus, helped in embedding the proposed explanation of the fertility intention in the social reality experienced by the actors.

3.7  Research Examples The proposed Narrative Framework might offer a powerful approach to frame and operationalize the role of uncertainty in the fertility decision–making process. This is a novel approach in fertility research, making it hard to cite studies that successfully applied said framework. Nonetheless, the literature offers evidence that, on top of the actual economic outlook or objective insecurity, the perception  of one’s own economic situation, or anticipation of future downturns inhibits childbearing. Some studies use direct questions to respondents about how insecure they feel their own economic situation (Bhaumik and Nugent 2011; Kreyenfeld 2009) or how insecure they feel their jobs to be (Bernardi et al. 2008; Hanappi et al. 2017). The level of subjective well-being has been employed to capture unobserved amenities

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

41

of the job, such as prestige, infrastructure, or welfare provisions in the study of fertility intentions (Vignoli et  al. 2020b).  Other  studies use survey questions about respondents’ knowledge about possible future events (Trinitapoli and Yeatman 2011). Liefbroer (2005), employing a five–wave panel survey among young Dutch adults spanning 13 years, found that the timing of entry into motherhood is affected by anticipated costs to one’s career and to one’s level of individual autonomy, and by an anticipated increase in one’s sense of security. He also illustrated that anticipated costs to one’s career and spending power, and anticipated rewards in terms of one’s sense of security and quality of the partner relationship affect the timing of entry into fatherhood. In the following, we present two more studies where the importance of uncertain futures is explicitly considered. The understanding of the complex interplay between the agency capacity and the structural constraints is at the core of recent developments in family research (Johnson-Hanks et  al. 2011; Huinink et  al. 2015). For instance, the qualitative research of Bernardi et al. (2008) studied the different roles played by an insecure job on fertility choices in what were once East and Western Germany after reunification. The authors sustain that parents brought up in different cultural contexts (communist and capitalist) were socialized to different values and cultural frameworks. This circumstance has consequences for the role that labor insecurity may play in fertility decision. Their conclusion was that socialization and cultural values play a crucial role in shaping imaginaries about the right job for having a child: in the West, fertility is usually postponed until after reaching career goals, whereas in the East childbearing and professional life represent parallel paths. While this explanation of the emergence of a life course narrative may account for the specific case under study, different contexts may see different generative mechanisms of narratives and elements at work (i.e., shared narratives of media and peers). In addition, cultural values and cognitive frameworks may remain valid during a long period and influence expectations, imaginaries and narratives. But, equally, they may sharply change and support a new life course narrative. Alternatively, a recent article by Gatta et al. (2019) proposed an operationalization of two dimensions of perceived employment uncertainty—stability and resilience—and tested their relevance for predicting fertility intentions, net of socio–economic structural constraints. To this end, they relied on a unique survey that covers an array of variables measuring employment uncertainty vis á vis respondents’ fertility intentions, the Trustlab survey for Italy (Aassve et al. 2018). Perceived resilience to job loss seems of particular relevance in fertility planning, outperforming objective indicators of employment status and characteristics. The observed significance and strength of association between perceived employment resilience and fertility intentions remains strong after the introduction of person–specific controls for individuals’ risk attitudes in the model equation. In addition, the effect of perceived resilience to job loss did not vary significantly in regions with a higher share of fixed–term contracts or higher unemployment rates. In sum, the study by Gatta et al. (2019) advanced the importance of considering how different expectations of the future influence fertility intentions, net of the actual individual–level

42

D. Vignoli et al.

employment circumstances, person–specific heterogeneity in risk attitudes, and taking into account the moderating role of the macroeconomic context. These research examples highlight how expectations, imagination, and the ability to devise different scenarios may play a major role in planning the future.

3.8  Conclusions We argued that narratives of the future constitute a crucial element in the fertility decision–making process. In a condition of fundamental uncertainty, the future is not ergodic or merely the statistical shadow of the past (Davidson 2010: 17) and, as such, subject only to random changes (Beckert and Bronk 2018). A major shortcoming in the application of the traditional fertility frameworks is connected to backwards reasoning (Johnson-Hanks et al. 2011): indicators and statistical models consider what already happened in life, without taking into account the sources of uncertainty of the expected future. At the micro–level, actors are always in a present condition where the past already took place—often independently of the actors’ wishes—and the future is yet to come in a specific form. Hence, planning a present action means aligning elements of the past in the light of an expected or imagined future. There is no doubt that understanding historical trajectories is indispensable to understanding the social phenomena of the present. However, events in the social world cannot be explained by the past alone. Actors’ decisions are determined by more than existing structures and past experiences—they are shaped in equal measure by perceptions of the future (Beckert 2016: 35).

The action under uncertain conditions requires narratives of the future capable to reduce uncertainty and sustain commitment because outcomes are not necessarily “implied in the present” (Buchanan and Vanberg 1991: 170). Individuals who are uncertain about their future income or earning opportunities may shy away from long–term commitments and, thus, postpone leaving the parental home, setting up their own household, and having children. The fundamental uncertainty in fertility decisions, reinforced during globalization, makes narratives of the future pivotal in generating a level of commitment sufficient enough to act: Narratives create experienced rather than just abstract ‘knowledge’: they provide support for action founded on an emotionally coloured and subjective feeling of ‘knowing’ what will happen (Tuckett 2018: 74).

We believe that the Narrative Framework will help to understand contemporary fertility dynamics. We do not advocate that perceived economic uncertainty is the only factor responsible for the fertility decline observed in recent years across Europe and the US. However, analyses that simultaneously include numerous objective indicators such as the unemployment rate, the cost of public debt and consumer confidence index, do not entirely explain the decline in birth rates in Europe and the US, 2008–2013 (Comolli 2017; Matysiak et al. 2020). Moreover, value change can

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

43

of course be an additional factor contributing to fertility decline, but this is most likely to be a long–term trend, which is unlikely to be so concentrated in the aftermath of the Great Recession. Here, we argue that part of this unexplained fertility decline can be clarified by the rise of uncertainty, a condition in which the future cannot be deduced from present information. Narratives of the future are contingent tools to cope with the uncertainty that people face: they are not just a mirror of the socialization period, but they are always shaped by contingent external forces that may accomplish or thwart the established and expected life course. Indeed, narratives allow people to act according to the uncertainty they face (e.g. avoiding having children) or despite uncertainty (e.g. trying to have children). We conclude that the study of fertility decisions cannot disregard the condition of uncertainty in which they are taken and, especially, its future–oriented nature. The increasing uncertainty of a given prospect does not imply more unintelligible or chaotic behavior. Rather, the role of uncertainty in the fertility decision–making process can be assessed. Subjective reasoning and decision–making processes often rely on what people expect will happen, or what they are trying to achieve, and this became more important in the era of globalization–induced uncertainty. The Narratives Framework contributes to the study of the decision–making process in a condition of uncertainty, allowing the researcher to assess whether, to what extent, and what elements of an uncertain context influence the fertility decision. We wish that future research will operationalize the Narrative Framework, testing its components in connection with fertility, while disentangling the effects of uncertainty from other factors. Acknowledgments  The authors acknowledge the financial support provided by the European Union’s Horizon 2020 research and innovation programme/ERC Consolidator Grant Agreement No 725961 (EU–FER project Economic Uncertainty and Fertility in Europe, PI: Daniele Vignoli).

References Aassve, A., Mencarini, L, Chiocchio, F., Gandolfi, F., Gatta, A., & Mattioli, F. (2018). Trustlab Italy: A new dataset dataset for the study of trust, family demography and personality. Dondena Working Papers 115, Bocconi University. Ajzen, I. (1991). The theory of planned behaviour. Organizational Behavior and Human Decision Processes, 50(2), 179–211. Ajzen, I., & Klobas, J. (2013). Fertility intentions: An approach based on the theory of planned behaviour. Demographic Research, 29(8), 203–232. Alderotti, G., Vignoli, D., Baccini, M., & Matysiak, A. (2019). Employment uncertainty and fertility: A network meta–analysis of European research findings (DiSIA Working Papers 2019/06), University of Florence. Becker, G. S. (1964). Human capital. New York: Columbia University Press. Becker, G. S. (1981). A treatise on the family. Cambridge, MA: Harvard University Press. Beckert, J. (2016). Imagined futures. Fictional expectations and capitalist dynamics. Cambridge: Harvard University Press.

44

D. Vignoli et al.

Beckert, J., & Bronk, R. (2018). An introduction to uncertain futures. In J. Beckert & R. Bronk (Eds.), Uncertain futures: Imaginaries, narratives, and calculation in the economy (pp. 1–36). Oxford: Oxford University Press. Bernardi, L., Klärner, A., & von der Lippe, H. (2008). Job insecurity and the timing of parenthood: A comparison between eastern and Western Germany. European Journal of Population, 24(3), 287–313. Bernardi, L., Hiunink, J., & Settersten, R. A. (2019). The life course cube: A tool for studying lives. Advances in Life Course Research, 41, 100258. Bhaumik, S. K., & Nugent, J. B. (2011). Real options and demographic decisions: Empirical evidence from east and West Germany. Applied Economics, 43(21), 2739–2749. Boomgaarden, H. G., van Spanje, J., Vliegenthart, R., & de Vreese, C. H. (2011). Covering the crisis: Media coverage of the economic crisis and citizens’ economic expectations. Acta Politica, 46, 353–379. Bourdieu, P. (1990) [1980]. The logic of practice. Stanford: Stanford University Press. Bronk, R. (2009). The romantic economist: Imagination in economics. Cambridge: Cambridge University Press. Bruner, J. (1990). Acts of meaning. Cambridge, MA: Harvard University Press. Buchanan, J.  M., & Vanberg, V.  J. (1991). The market as a creative process. Economics & Philosophy, 7, 167–186. Busetta, A., Mendola, D., & Vignoli, D. (2019). Persistent joblessness and fertility intentions. Demographic Research, 40(8), 185–218. Caldwell, J. (1982). Theory of fertility decline. In London. New York: Academic. Cepernich, C. (2012). Storie di subprime, downgrading, spread e default. La narrazione della grande crisi tra informazione e popolarizzazione. Comunicazione Politica, 3, 409–440. Comolli, C. L. (2017). The fertility response to the great recession in Europe and the United States: Structural economic conditions and perceived economic uncertainty. Demographic Research, 36(51), 1549–1600. Comolli, C. L., Neyer, G., Andersson, G., Dommermuth, L., Fallesen, P., Jalovaara, M., Jónsson, A., Kolk, M., & Lappegård, T. (2019). Beyond the economic gaze: Childbearing during and after recessions in the Nordic countries. Stockholm Research Reports in Demography, 16. Dantis, C., & Rizzi, E. L. (2020). Transition to first birth during the Great Recession: the case of Greece. Genus, 76(1), https://doi.org/10.1186/s41118-019-0070-1. Davidson, D. (1980). [1976]. Hempel on explaining action. In D.  Davidson (Ed.), Essays on actions and events (pp. 261–276). Oxford: Clarendon Press. Davidson, P. (1996). Reality and economic theory. Journal of Post Keynesian Economics, 18(4), 479–508. Davidson, P. (2010). Risk and uncertainty. In R. Skidelsky & C. W. Wigström (Eds.), The economic crisis and the state of economics (pp. 13–29). New York: Palgrave Macmillan. Davis Cross, M. K., & Ma, X. (2013). EU crises and the international media. In ARENA working papers 3/2013, Center for European Studies. Oslo: University of. Dequech, D. (2000). Fundamental uncertainty and ambiguity. Eastern Economic Journal, 26(1), 41–60. Elster, J. (2009). Excessive ambitions. Capitalism and Society, 4(2), 1–30. Emirbayer, M., & Mische, A. (1998). What is agency? American Journal of Sociology, 103(4), 962–1023. Esping-Andersen, G., & Billari, F. C. (2015). Re-theorizing family demographics. Population and Development Review, 41(1), 1–31. Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behavior: An introduction to theory and research. Boston: Addison-Wesley. Freese, J. (2009). Preferences and the explanation of social behavior. In P. Bearman & P. Hedström (Eds.), Oxford handbook of analytic sociology (pp. 94–114). Oxford: Oxford University Press. Friedman, D., Hechter, M., & Kanazawa, S. (1994). A theory of the value of children. Demography, 31(3), 375–401.

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

45

Gatta, A., Mattioli, F., Mencarini, L., & Vignoli, D. (2019). Employment uncertainty and fertility intentions: Stability or resilience? (DiSIA working papers 2019/12), University of Florence. Goldscheider, F., Bernhardt, E., & Lappegård, T. (2015). The gender revolution: A framework for understanding changing family and demographic behavior. Population and Development Review, 41(2), 207–239. Goldthorpe, J. H. (2007). On sociology, second edition. In Volume one: Critique and program. Stanford: Stanford University Press. Guetto, R., & Panichella, N. (2013). Geographical mobility and reproductive choices of Italian men. European Sociological Review, 29(2), 302–315. Guetto, R., Luijkx, R., & Scherer, S. (2015). Religiosity, gender attitudes and women’s labour market participation and fertility decisions in Europe. Acta Sociologica, 58(2), 155–172. Hacker, J.  S. (2019). The great risk shift: The new economic insecurity and the decline of the American dream. New York: Oxford University Press. Hanappi, D., Ryser, V. A., Bernardi, L., & Le Goff, J. M. (2017). Changes in employment uncertainty and the fertility intention–realization link: An analysis based on the Swiss household panel. European Journal of Population, 33(3), 381–407. Hedström, P. (2005). Dissecting the social. Cambridge: Cambridge University Press. Hedström, P., & Bearman, P. (Eds.). (2009). The Oxford handbook of analytical sociology. Oxford: Oxford University Press. Hedström, P., & Swedberg, R. (1998). Social mechanisms: An analytical approach to social theory. Cambridge: Cambridge University Press. Hofmann, B., & Hohmeyer, K. (2013). Perceived economic uncertainty and fertility: Evidence from a labor market reform. Journal of Marriage and Family, 75(2), 503–521. Huinink, J., Kohli, M., & Ehrhardt, J. (2015). Explaining fertility: The potential for integrative approaches: Introduction to the special collection ‘theoretical foundations of the analysis of fertility’. Demographic Research, 33(4), 93–112. Johnson-Hanks, J. A., Bachrach, C. A., Morgan, S. P., & Kohler, H.-P. (2011). Understanding family change and variation: Toward a theory of conjunctural action. New York: Springer. Kao, G., & Tienda, M. (1995). Optimism and achievement: The educational performance of immigrant youth. Social Science Quarterly, 76(1), 1–19. Karaman Örsal, D. D., & Goldstein, J. R. (2018). The changing relationship between unemployment and total fertility. Population Studies, 72(1), 109–121. Knight, F.H. (2006) [1921]. Risk, uncertainty and profit. New York: Cosimo Classics. Kreyenfeld, M. (2009). Uncertainties in female employment careers and the postponement of parenthood in Germany. European Sociological Review, 26(3), 351–366. Kreyenfeld, M. (2016). Economic uncertainty and fertility. In K. Hank & M. Kreyenfeld (Eds.), Social Demography Forschung an der Schnittstelle von Soziologie und Demografie (pp. 59–80). Wiesbaden: Springer VS. Kulu, H. (2005). Migration and fertility: Competing hypotheses re-examined. European Journal of Population, 21(1), 51–87. Lane, D., & Maxfield, R. (2005). Ontological uncertainty and innovation. Journal of Evolutionary Economics, 15(1), 3–50. Lesthaeghe, R. (1995). The second demographic transition in Western countries: An interpretation. Gender and family change in industrialized countries. In K. O. Mason & A. Jensen (Eds.), Gender and family change in industrialized countries (pp. 17–62). Oxford: Clarendon Press. Lévi-Strauss, C. (1963). Structural anthropology. New York: Basic Books. Liefbroer, A. C. (2005). The impact of perceived costs and rewards of childbearing on entry into parenthood: Evidence from a panel study. European Journal of Population, 21(4), 367–391. Matysiak, A., & Vignoli, D. (2008). Fertility and women’s employment: A meta-analysis. European Journal of Population, 24(4), 363–384. Matysiak, A., & Vignoli, D. (2013). Diverse effects of women’s employment on fertility: Insights from Italy and Poland. European Journal of Population, 29(3), 273–302.

46

D. Vignoli et al.

Matysiak, A., Sobotka, T., & Vignoli, D. (2020). The great recession and fertility in Europe: A sub-­ national analysis. European Journal of Population, online first. McDonald, P. (2000). Gender equity in theories of fertility transition. Population and Development Review, 26(3), 427–439. Mencarini, L., Vignoli, D., & Gottard, A. (2015). Fertility intentions and outcomes: Implementing the theory of planned behavior with graphical models. Advance in Life Course Research, 23, 14–28. Mills, M., & Blossfeld, H. P. (2013). The second demographic transition meets globalization: A comprehensive theory to understand changes in family formation in an era of rising uncertainty. In A. R. Evans & J. Baxter (Eds.), Negotiating the life course: stability and change in life pathways (pp. 9–33). London: Springer. Myrskylä, M., Kohler, H.-P., & Billari, F.  C. (2009). Advances in development reverse fertility declines. Nature, 460, 741–743. OECD. (2011). Divided we stand: Why inequality keeps rising. OECD Publishing. https://doi. org/10.1787/9789264119536-en.. Accessed 18 Feb 2020. Offe, C. (1998). Political economy: Sociological perspectives. In R. E. Goodin & H. D. Klingemann (Eds.), A new handbook of political science (pp. 675–690). Oxford: Oxford University Press. Ranjan, P. (1999). Fertility behaviour under income uncertainty. European Journal of Population, 15(1), 25–43. Régnier-Loilier, A., & Vignoli, D. (2011). Intentions de fécondité et obstacles à leur réalisation en France et en Italie. Population, 66(2), 401–431. Schoen, R. (2010). Gender competition and family change. Genus, 66(3), 95–120. Schoen, R., Kim, Y. J., Nathanson, C. A., Fields, J., & Astone, N. M. (1997). Why do Americans want children? Population and Development Review, 23(2), 333–358. Schoen, R., Astone, N. M., Kim, Y. J., Nathanson, C. A., & Fields, J. M. (1999). Do fertility intentions affect fertility behavior? Journal of Marriage & Family, 61(3), 790–799. Schütz, A. (1962). The problem of social reality. Collected papers I. The Hague: Martinus Nijhoff. Schütz, A. (1967). The phenomenology of the social world. Evanston: Northwestern University Press. Strandbakken, P. (2017). The study of consumption in sociology. Beyond utility theory. In H. Leiulfsrud & P. Sohlberg (Eds.), Concepts in action (pp. 136–150). Leiden: Studies in critical Social Science. Stueber, K. R. (2008). Reasons, generalizations, empathy, and narratives: The epistemic structure of action explanation. History and Theory, 47(1), 31–43. Thomson, E., & Brandreth, Y. (1995). Measuring fertility demand. Demography, 32(1), 81–96. Todd, P., Hills, T., & Hendrickson, A. (2013). Modeling reproductive decisions with simple heuristics. Demographic Research, 29(24), 641–662. Trinitapoli, J., & Yeatman, S. (2011). Uncertainty and fertility in a generalized AIDS epidemic. American Sociological Review, 76(6), 935–954. Tuckett, D. (2018). Conviction narrative theory and understanding decision-making in economics and finance. In J. Beckert & R. Bronk (Eds.), Uncertain futures: Imaginaries, narratives, and calculation in the economy (pp. 62–82). Oxford: Oxford University Press. Tuckett, D., & Nikolic, M. (2017). The role of conviction and narrative in decision-making under radical uncertainty. Theory & Psychology, 27(4), 501–523. Uebel, T. (2012). Narratives and action explanation. Philosophy of the Social Sciences, 42(1), 31–67. van de Kaa, D. J. (1987). Europe’s second demographic transition. Population Bulletin, 42(1), 1–59. Vignoli, D., Drefahl, S., & De Santis, G. (2012). Whose job instability affects the likelihood of becoming a parent in Italy? A tale of two partners. Demographic Research, 26(2), 41–62. Vignoli, D., Tocchioni, V., & Mattei, A. (2019). The impact of job uncertainty on first-birth postponement. Advances in Life Course Research, online first. Vignoli, D., Guetto, R., Bazzani, G., Minello, A., & Pirani, E. (2020a). Economic uncertainty and fertility: Narratives of the future. In DiSIA working papers 2020/1. Florence: University of. Vignoli, D., Mencarini, L., & Alderotti, G. (2020b). Is the effect of job uncertainty on fertility intentions channeled by subjective well-being? Advances in Life Course Research, 100343.

3  Uncertainty and Narratives of the Future: A Theoretical Framework…

47

Weber, M. (1978 [1922]). Economy and society: An outline of interpretive sociology. Berkeley: University of California Press. Westoff, C.  F., & Ryder, N.  B. (1977). The predictive validity of reproductive intentions. Demography, 14(4), 431–453. Open Access  This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made. The images or other third party material in this chapter are included in the chapter’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

Chapter 4

Social Contagion Effects in Fertility: Theory and Analytical Strategy Nicoletta Balbo and Nicola Barban

4.1  Introduction According to the linked lives principle of the life course approach (Elder et al. 2003; Elder Jr. 1985), lives are experienced interdependently and individuals develop in synchrony with significant others, who influence their life choices and trajectories (Elder et al. 2003). We therefore need to take into account that the decision to have a child is not made in a vacuum. It is instead shaped by the interaction and exchange of resources with relevant people surrounding the individual or, more often, the couple. Everyday life demonstrates the importance of forces working at an intermediate, meso-level, that pertains to the web of social relationships in which an individual is embedded (i.e., an individual’s social network). These network-related forces act in parallel to other individual/couple (e.g., education, age, socio-economic status) and contextual factors (e.g., welfare regime), which can enhance or buffer their effects. Diffusion and social interaction theories (e.g., Bongaarts and Watkins 1996; Montgomery and Casterline 1996) have already highlighted the importance of social relationships for fertility decisions (i.e., meso-level determinants). Since the 2000s, recognition of the relevance of social networks on childbearing decisions has increased, with scholars more often turning to social interaction effects (e.g., social multiplier) to explain fertility differentials across time and space at macro-­ level (e.g., Raab et al. 2014). At the micro-level, theoretical efforts to explain how an individual’s network of relatives, friends, coworkers, peers (i.e., the so-called relevant others) might influence an individual’s fertility decision-making can be N. Balbo (*) Bocconi University, Milan, Italy e-mail: [email protected] N. Barban University of Essex, Colchester, UK e-mail: [email protected] © Springer Nature Switzerland AG 2020 R. Schoen (ed.), Analyzing Contemporary Fertility, The Springer Series on Demographic Methods and Population Analysis 51, https://doi.org/10.1007/978-3-030-48519-1_4

49

50

N. Balbo and N. Barban

found in several studies (Kohler 2001; Bernardi 2003; Bühler and Philipov 2005). We can identify two, relatively disconnected, bodies of research adopting a meso-­ level explanatory approach. One focuses on the exchange of resources that can affect fertility decisions (social capital) with relevant others (e.g., Bühler and Philipov 2005; Philipov et al. 2006). The other examines whether relevant others’ opinions and behaviors shape reproductive choices (social interaction) and if so, how (e.g., Montgomery and Casterline 1996; Kohler 2001). Although both streams of research investigate the potential manner in which social networks influence fertility behavior, they seem to have developed in parallel. There is no acknowledgement that they focus on complementary dimensions of the same aspect and have not seemed to benefit from one another. In this chapter, we aim at reviewing both explanatory approaches, looking at the different sides and aspects of an individual’s social network. We look into the potentially differing roles that a social network might play on an individual’s fertility decision-making by considering it as a source of relevant resources, as well as a place where interactions between people mean that they are exposed to and influenced by others opinions and behaviors.

4.2  Social Capital The resources and support that an individual has access to as a result of personal relationships might influence their decision to have a child. These resources are part of an individual’s social capital. The concept of social capital has been studied extensively in several sociological theories. Sociologists have focused on different aspects of the concept, providing multiple definitions and operationalizations (e.g., Granovetter 1973; Bourdieu 1986; Coleman 1988; Lin et al. 2001; Flap and Völker 2004; Van der Gaag 2005). Fertility research has borrowed the concept, using it to define resources that individuals have access to through reciprocal and trust-based exchange between network members. Resources consist of goods, information, money, and the capacity to work, as well as influence, power, and active help (Bühler and Philipov 2005). This body of research has shown that network resources are often taken into account during fertility planning and that more supportive network relationships positively influence fertility intentions (e.g., Bühler and Philipov 2005; Philipov et al. 2006; Bühler and Fratczak 2007). There is evidence of the fact that the effect of fertility-relevant network supportive resources, namely childcare assistance, is institutionally filtered, that is the positive effects of informal childcare change under different welfare regimes (Balbo and Mills 2011a). If the effect of specific fertility-relevant resources (i.e., emotional support and informal childcare) is one potential channel via which social capital affects fertility, a second channel lies in the quality and strength of the social ties that generate these resources. The first channel has mainly been investigated in a body of sociological research on social capital (e.g., Snijders 1999), which stresses the importance of looking at resources that are instrumental in reaching a certain goal. Other scholars have instead emphasized the importance of social ties, especially within the family of

4  Social Contagion Effects in Fertility: Theory and Analytical Strategy

51

origin, from which individuals can potentially draw supportive resources (Astone et  al. 1999; Balbo and Mills 2011b). A reverse perspective has been adopted by Schoen et al. (1997), who have shown that children create social capital for parents, and this is part of the reason why individuals have children.

4.3  Social Interaction Building on previous studies (e.g., Bongaarts and Watkins 1996; Montgomery and Casterline 1996), we define social interaction as the general mechanism via which relevant others’ (e.g., relatives, friends, peers, colleagues) opinions and behaviors affect an individual’s choices. The importance of social interaction on an individual’s fertility decision-making has been increasingly acknowledged (e.g., Kohler 2001; Bernardi 2003). At the macro-level, researchers have often turned to diffusion and social interaction effects to explain the persistent diversity of fertility behavior between geographical areas or over periods of time (Kohler et al. 2002, 2006). As an example, these effects are assumed to amplify the behavioral impact of certain socio-economic and institutional changes (i.e., the so-called social multiplier, Billari 2004). At the micro-level, different disciplines (e.g., sociology, demography and economics) have focused on different channels through which social interaction might work. By bringing together the contributions from these different disciplines, we can identify five main social interaction channels: social influence; social learning; social pressure; cost-sharing dynamics; and network externalities. The first two mechanisms have been proposed in sociological and demographic research (e.g., Montgomery and Casterline 1996; Kohler 2001). Social influence references to consensus in peer groups that constrain attitudes and behaviors, whereas social learning relates to how individuals gain knowledge from others. An additional channel that has been highlighted by more qualitative demographic research (Bernardi 2003) is social pressure, defined as the individual’s perception of what relevant others approve or disapprove. Economic research provides the last two possible complementary channels through which social interaction might work: cost-sharing dynamics and network externalities (Kuziemko 2006; Balbo and Barban 2014). Cost-sharing dynamics refers to the opportunity for people consuming similar goods or experiences to share the costs and uncertainty associated with it. Network externalities are the increase in benefit, or surplus, that an individual derives from an experience when the number of other people consuming it increases (Katz and Shapiro 1985). These two mechanisms emphasize two different aspects of the same sharing process: the former focuses on the cost and the latter stresses the benefit. A certain experience not only generates a particular value in itself, but can also produce additional value when ‘consumers’ of such an experience interact with one with another. This is called the synchronization value and is the essence of a sharing process (Liebowitz and Margolis 1995). Moreover, literature has provided empirical evidence of the fact that different types of “relevant others” can influence an individual’s choice to have a child:

52

N. Balbo and N. Barban

siblings (Kuziemko 2006; Lyngstad and Prskawetz 2010), colleagues (Asphjell et al. 2013; Pink et al. 2014), and friends (Balbo and Barban 2014). The majority of these studies have highlighted a positive effect that is, however, limited in time.

4.4  A  Simple Model of Social Interactions in Fertility Behavior In order to understand the contribution of social effects to fertility change, it is essential to formalize how the social process develops with a model. A simple algebraic representation from Casterline (2001) can be used to identify different components of social influence.

Yit = Xit b + a å Yj, t -1Wij + e it



(4.1)

Where Y is a measure of fertility for individual i at time t; X are sets of conventional determinants of fertility such as age, educational attainment or marital status; W are weights for the salience of the jth person for the ith person. The parameter α can be interpreted as the causal effect of significant others’ fertility on determining individual fertility. The model, in this simple formulation, does not take into account that individual fertility can be influenced by other persons’ characteristics or attitudes, beliefs and behavior that might affect fertility decision-making. Moreover, it restricts the radius of social influence within individuals of reproductive age. A related important aspect of this formulation is that Xit are not correlated with the same characteristics from individual j. This implies, for instance, that the timing of marital status or educational choices are not directly influenced by others. This is certainly a limitation, as the channels of influence can be multiple. Raab et  al. (2014), for example, show that there is pronounced similarity in the early socioeconomic trajectories of siblings. Equation (4.1) moreover assumes that the effect of multiple others is additive. However, there might be nonlinear effects, or threshold effects, meaning that the social effects become relevant after a relative number of significant others experience childbearing. This specification also considers a very short time dependence, as Yit is influenced by events experienced by individual j at time t-1. Effects may be cumulative over time or have a different time dependency. Last, the choice of the value for the weights W indicating salience of relationship between i and j, is something that can be difficult to establish a priori. A possible solution is to investigate different types of social ties separately, for instance siblings, friends and colleagues, and set the weights equal to 1, providing the same salience to all. Various extensions of this model are possible. In particular, it is recommended to study fertility in a time-to-event framework because of data censoring. An example of a semi-parametric model is depicted in Eq. (4.2). This class of models study the hazard of experiencing childbearing that depends on time t. The implicit assumption

4  Social Contagion Effects in Fertility: Theory and Analytical Strategy

53

is that the social effect on fertility affects the risk of an individual’s childbearing in a proportional way and does not change the shape of the hazard function described by h0(t).

hi ( t ) = h0 ( t ) exp { Xit b + a å Yj, t -1Wij + e it }



(4.2)

Equation (4.2) can be approximated by discrete-time models that model the probability of having a child using regression for binary data such as logit, probit or cloglog models. Equation (4.3) shows a discrete-time model using logit regression. Discrete-time models can easily accommodate both time-variant and time-invariant covariates that affect the “risk” of childbearing. Typical time-invariant covariates are background characteristics, such as parental education or region of birth. The fertility of peers can be included in the model as a time-variant variable that modifies the probability of having a child by the respondent.



æ h (t ) ö = Xit b + a å Yj, t -1Wij + e it log ç i ç 1 - h ( t ) ÷÷ i è ø

(4.3)

4.5  Challenges in Measuring Social Interaction Effects The estimation of the causal effect of social influence in fertility behavior can be troublesome. Once a causal model of influence has been specified, such as the simple model described in the previous paragraph, the estimation of a direct social influence effect is threatened by different sources of bias that affects the precision and validity of the estimates. We identify three sources of confounders: the common confounders problem, the reflection problem, and the endogenous network problem. Common confounders, or the correlated unobservable problem, arises when individuals belonging to the same social groups are influenced by the same factors that cannot be observed. Siblings share the same parents and extended family, classmates live in the same geographical area, and friends share the same teachers in schools. If these factors influence life course decisions, individuals belonging to the same social groups will experience higher similarity in their behavior, even if there is no direct influence from one to the other. For example, coworkers may experience apparent social influence in fertility behavior, because their workplace has adopted a family-friendly policy allowing flexible working or facilitating childcare. As workers belonging to the same company can benefit from the same policies, this will result in an apparent spread of fertility behavior among colleagues. The second source of bias called the reflection problem, or simultaneity, refers to the basic question: who influences whom? The reflection problem has been introduced in the econometric literature of peer effects by (Manski 1993, p. 532) as the following: (page 532).

54

N. Balbo and N. Barban The term reflection is appropriate because the problem is similar to that of interpreting the almost simultaneous movements of a person and his reflection in a mirror. Does the mirror image cause the person’s movements or reflect them? An observer who does not understand something of optics and human behaviour would not be able to tell.

In other words, this means that by observing outcomes of individuals from the same group it may hard to distinguish the direction of the social effect. In the case of fertility behavior, we are interested in identifying the causal effect of one’s childbearing on others’ fertility behavior. As childbirth is a discrete event, we might expect that there is a clear sequence of events. That is, childbirth of individual i at time t can be only influenced by events of individual j at time t-k. There are, however, some factors to consider. Childbirth is the outcome of a pregnancy, and therefore the decision process related to fertility started before the observed outcome. We also need to consider that individuals may be affected by the pregnancy of relatives, friends and colleagues, and not by the actual childbirth. If this is the case, the time lag to include in our estimation models can be different. At the same time, not all the pregnancies result in live births, making the decision about what is the event of interest more difficult. Moreover, the decision to have a child may happen well before conception, as the probability of conception differs substantially across individuals. This may result in couples of friends who decide to have a child at the same time, but experience childbearing in different years. Kuziemko (2006) calculated that the probability of experiencing childbirth within 6 months among individuals who decided to conceive at the same moment is around 14%. Another important aspect affecting simultaneity is parity. Demographic literature has stressed the importance of considering parity separately, as the decision processes related to having a first child might be different form higher parity. It is important to understand if higher parity births have an effect on first children, for instance. Cost-­ sharing dynamics might be higher among individuals experiencing the same event. At the same time, individuals who have already experienced childbearing will not have the same degree of social learning from other individuals. The last source of bias is the endogenous group selection problem. Individuals choose their friends and their peers based on multiple characteristics. These characteristics may affect the probability of childbearing. Preferences, attitudes and expectations about the timing of parenthood or family formation may be important in the formation of relationships and group belonging. In other words, network formation may be endogenous to the outcome (childbirth) of interests. If this is the case, we would over-estimate the social contagion effect. This problem is limited to chosen relationships, such as friendships. Siblings, on the contrary, are exogenous as you do not choose them. Understanding the nature of other forms of relationships may be complicated. Generally speaking, individuals do not choose their coworkers, but the decision of the employer and the type of company might be correlated with fertility attitudes. For example, one might decide to work for the public sector rather than the private for more flexible time, or work for companies who have better parental benefits.

4  Social Contagion Effects in Fertility: Theory and Analytical Strategy

55

4.6  Statistical and Econometric Approaches What strategies can be adopted to correctly estimate direct causal social effects in fertility? To solve the common confounder problem, it is necessary to understand what the common unobservable bias is. For instance, it could be that individuals from the same school are exposed to the same set of confounding variables. A common solution, in this case, is the use of fixed effects that explicitly account for some part of the omitted variable and selection bias. For instance, including a school fixed effect in a model for peer effect among school friends can remove the common source of confounders, leading to unbiased estimates. Asphjell et al. (2013) include fixed effects for year, month, region and industry when studying the effect of coworkers’ fertility on the probability of having a child. Kuziemko (2006) includes individual fixed effects in studying the social effects of siblings’ fertility, implicitly assuming that there are no differences in the baseline hazard by parity. Individual fixed effects are in fact possible only when looking at multiple spells to allow for individual-specific heterogeneity. However, this limits the analysis when looking at non-repeated events such as first childbirth. Other possibilities include controlling for factors that might be affected by common confounders but cannot directly affect individuals’ decision. In our previous work (Balbo and Barban 2014) we control for peers’ fertility, that is former classmates who were never friends with respondents. In this way, we argue that we control for common confounders due to the same social environment. Our results show non-negligible effects of peers, suggesting common confounder effects that would otherwise inflate the social effect of interest. Another possible solution is the use of instrumental variables. Possible instrumental variables, in this context, are friends of friends or siblings of friends that have no direct effects on the respondents. This “exclusion restriction”, however, is difficult to completely justify as we cannot completely assume that the effects of, for instance friends’ siblings, are not exposed to the same confounders, for example if they live in the same neighborhood. To control for endogenous network formation, the best solution would be to study a random assignment of peers and their social effects, making the network exogenous de facto. This is obviously not possible in most cases. Some pseudo-experimental designs, however, have been used to understand peer effects. For instance, the Dartmouth College roommates’ research design, pioneered by Sacerdote (2001), exploits the fact that Dartmouth College matches freshman roommates randomly. This random assignment is used to study the social effects in education, and to the best of our knowledge, has not been used to study fertility. In principle, similar studies that exploit random allocation of coworkers, for instance, could be used to study social effects in fertility behavior. Last, a possible way to control the presence of bias in the analysis is the use of falsification tests, or placebo tests, that look at the social effect of random individuals to understand if the analysis is robust. In our previous study mentioned above (Balbo and Barban 2014), we assigned random individuals as friends and peers to understand if the effect was driven by different sizes of social networks. Asphjell et al. (2013) used a similar strategy by running the analysis on three placebo peer groups: (1)

56

N. Balbo and N. Barban

workers of the same firm and industry, but different workplace, (2) future coworkers; (3) siblings of coworkers. They found no effects on these random assignments of coworkers. Last, it is possible to test the robustness of the identification strategy, by looking at outcomes that cannot be transmitted. Cohen-Cole and Fletcher (2008) tested previous methodology to assess peer effects by looking at characteristics for which social effects are implausible. In particular, they look at the spread of height, acne and headaches in high schools using Add Health. They found that previous methodology is affected by common confounder biases.

4.7  The “Second Wave” of Social Interactions in Fertility Traditional models of social influence in fertility behavior aimed at describing macro-dynamics in fertility try to understand the emergence of differentials in fertility across time and space. These studies (e.g., Bongaarts and Watkins 1996; Montgomery and Casterline 1996) mainly refer to micro-level social interaction effects to explain macro-level trends, such as persistent diversity of fertility behaviors across countries or over a period of time (i.e., multiple equilibria and path dependence), or amplification of the behavioral impact of institutional changes (i.e., social multiplier effects). However, such bodies of research had not directly modelled, and thereby tested, social interaction effects at the individual-level. More recently, starting approximately from the 2010s a “second wave” of studies focusing on social influence and fertility behavior has emerged. Thanks to the availability of new longitudinal or retrospective data on linked individuals, as well as advancement in the methodology for dyadic and network data, several studies that estimate social contagion effects in fertility have been published. In this section, we aim to review a sample of papers that examine the social influence of siblings, friends and coworkers on an individual’s fertility behavior. These studies, that are summarized in Table 4.1, provide robust evidence of the significant effect that different “relevant others” have in shaping an individual’s fertility behavior, at least in the short-run. The mechanisms highlighted to explain fertility interaction effects are those described above in Sect. 4.3, although sometimes the label of the same mechanism varies from paper to paper (e.g., social influence is sometimes called emotional contagion).

4.7.1  Siblings’ Influence on Fertility Studies focusing on cross-sibling effects on fertility (Lyngstad and Prskawetz 2010; Hart and Cools 2019; Raab et  al. 2014) mainly use register data from Northern European countries, such as Norway and Finland. Raab et  al. (2014) provide descriptive evidence of the fact that family-formation trajectories among dyad of siblings are significantly more similar than any unrelated dyad. They engage in

4  Social Contagion Effects in Fertility: Theory and Analytical Strategy

57

Table 4.1  Summary of results from selected papers on social interactions in fertility Type of social influence Authors Lyngstad and Siblings Prskawetz (2010)

Country and period Norway

Data Register data

Main results Methodology Positive cross-­sibling Continuous-time effects on first birth hazard models for women’s first and second births. Hart and Siblings Norway Register No cross-sibling Instrumental Cools (2019) data effect on third birth variable regression Raab et al. Siblings Finland Register Siblings show similar Sequence analysis (2014) data family-­formation trajectories Dyadic discrete Friends United Add health A friend’s Balbo and time event States childbearing Barban increases a woman’s history models (2014) risk of becoming a mother United Add health Teenage pregnancies Instrumental Fletcher and Friends variable States are influenced by Yakusheva regressions peers (2016) Discrete time Pink et al. Coworkers Germany LIAB data, Colleagues’ event history childbearing (2014) that increases the risk for models combines survey and other colleagues to register data have a(nother) child Instrumental Asphjell Coworkers Sweden IFAU Coworkers’ births variable discrete et al. (2013) register data influence an individual’s timing of time hazard models childbearing Instrumental Register Both a colleague’s Büyükkeçeci Coworkers The Netherlands data sibling and a sibling’s variable discrete et al. (2020) and time hazard colleague influence siblings models an individual’s transition to parenthood

sequence analysis and show that sibling similarity seems to be stratified by gender and socio-economic status, whereas other shared parental characteristics do not seem to play a major role in explaining such similarity. Next to this study, adopting a more holistic life-course approach because they look at the entire life trajectories, there are two other papers (Lyngstad and Prskawetz 2010; Hart and Cools 2019) focusing on cross-sibling effects on specific fertility transitions. Lyngstad and Prskawetz (2010), using continuous-time hazard models on Norwegian register data and focusing on women only, show that a sibling’s childbearing increases the risk for the other sibling to have a first child, whereas such effect is negligible for second births. Along this line, Hart and Cools (2019), adopting a fully causal approach that

58

N. Balbo and N. Barban

uses two instrumental variables again on Norwegian register data, show that cross-­ sibling effects are not visible on the third birth too. Therefore, there seems to be evidence of sibling influence on an individual’s likelihood of becoming a parent that might also explain overall similarity in family-formation patterns between siblings, whereas no cross-sibling effects are found for higher-order fertility transitions.

4.7.2  Friends’ Influence on Fertility Very few studies focus on friends’ influence on fertility, and this is mainly due to the lack of suitable data able to measure both an individual’s friendship network as well as friends’ fertility history. To our knowledge, the two studies (Balbo and Barban 2014; Fletcher and Yakusheva 2016) that look at friends’ effects on reproductive behavior use the National Longitudinal Study of Adolescent to Adult Health, that collects very rich information on a representative U.S. sample of individuals who were followed from adolescence into adulthood and for whom we have detailed information on their fertility history and their high-school mates and friends, who are in turn respondents in the survey. In our study published in 2014, using a series of discrete time event history models we show that dyads of female friends influence each other in their reproductive choices. Specifically, we found that a friend’s childbearing significantly increases the risk for the other woman to become a mother, and that is true even after the high-school period. Such cross-friend effects are short-term and curvilinear: it increases after the friend’s childbearing, reaching a peak around 2 years later, and then decreases, becoming negligible. In estimating friends’ influence, we take into account contextual confounders by including in the model previous high-school mates, who were not friends, and we control for selection bias by jointly estimating the risk of childbearing and the likelihood of a friendship. Instead, Fletcher and Yakusheva (2016) uses two instrumental variables, that are, the share of grademates whose mothers had a teenage pregnancy and grademate average age at menarche, to show that the likelihood of a teenage girl experiencing a pregnancy is influenced by the teen pregnancy rate among high-school female friends. Both studies therefore provide evidence of a significant cross-friend effect on first birth among women, both during adolescence as well as later in life.

4.7.3  Coworkers’ Influence on Fertility An interesting study that combines cross-sibling effects together with coworkers’ influence is the very recent work of Büyükkeçeci et al. (2020). They show, using Dutch register data, positive effects of colleagues’ fertility, as well as siblings’ fertility on transition to parenthood. Such kinds of social influences are observed after the second year following these events. By making use of instrumental variables within a discrete time event history model, the study finds that the instruments first

4  Social Contagion Effects in Fertility: Theory and Analytical Strategy

59

influence the individual’s network partners and then the individual through the network partners. These findings provide evidence for social spillover effects across an individual’s different network domains, showing how fertility spreads from the workplace to the family, as well as the other way around. Such work integrates previous studies showing cross-sibling effects on fertility, as the ones described in the previous subsection, with research uncovering coworkers’ positive influence on an individual’s fertility. The latter body of research consists of studies (Pink et al. 2014; Asphjell et al. 2013) that find a relevant influence of coworkers on an individual’s likelihood of having a child. Pink et  al. (2014) use a combination of survey and register data in Germany to show that colleagues’ childbearing increases the risk for other colleagues to have a(nother) child, and such effects appear to be short-termed but particularly strong in the first year after coworkers’ childbearing, and then decreasing thereafter. Asphjell and colleagues (2013) use dynamic discrete choice optimization problem with a finite horizon to model how social influence in the workplace affects an individual’s decision to have a child. They use IFAU data, a dataset that contains a diverse administrative register covering the entire Swedish population, and find a significant effect of coworkers on an individual’s timing of childbearing. Moreover, they show that the influence of a coworker’s childbearing is weaker under higher uncertainty in terms of work-related costs of childbearing, while it increases when women are at the end of their fertile ages.

4.8  Data Availability and Potential of Big Data As highlighted in the previous section, innovation in the collection of micro-data from multiple actors has been pivotal for the “second wave” of social effects in fertility. Let us now describe in more detail the main characteristics, advantages and disadvantages of each data source.

4.8.1  Administrative Data Administrative data – also called register data- from which it is possible to extract information on the life courses of the entire population, have been used to study social effects in fertility among siblings and coworkers. Siblings can be identified in the administrative data by linking multiple generations of individuals, that is by linking parents and children. The length of the timespan covered by these data in the Nordic countries is essential to study long processes such as fertility. Administrative datasets that cover main demographic events were created decades ago in the Nordic countries, making it possible to follow the fertility trajectories of multiple generations. These data often combine different administrative registers created for different scopes. Data are updated every time there is a change in family composition, providing a complete snapshot of the entire population. In some cases, detailed

60

N. Balbo and N. Barban

geographical information is available, making it possible to link individuals who live in the same neighborhood. Lyngstad and Prskawetz (2010) and Hart and Cools (2019) both use rich Norwegian register data to estimate the social effect of siblings’ fertility. Similarly, Buyukkececi and colleagues (Büyükkeçeci et  al. 2020) make use of Dutch administrative information to uncover the spiller contagion effect across multiple network domains, such as workplace and family. Alternatively, Raab et al. (2014) use data from Finland to address the issue of siblings’ similarity in life course. At the same time, the linkage of demographic registers with data on companies is sometimes possible and exploited, as in the work undertaken by Asphjell and colleagues (2013) who investigate the spread of fertility behavior among coworkers. Similarly, Pink et  al. (2014) use a combination of survey and register German data to show that colleagues’ childbearing increases the risk for other colleagues to have a(nother) child, at least in the short-term.

4.8.2  Household and Multi-actor Surveys Longitudinal surveys of household or extended family networks are increasingly available. These data, when they cover a sufficient time-span, can be used to study the social effects of fertility among family members. Household panel studies, such as UKHLS (Understanding Society) in the United Kingdom; PSID in the United States; SOEP or PAIRFAM in Germany; or the Netherlands Kinship Study, provide a source of demographic events among family members that can be used to study social interaction effects. Although the possibility of linking different respondents is very promising, these data sources rarely provide enough information to provide good estimates of social effects in fertility. Time-span needs to be large enough to observe fertility events of multiple respondents. Despite the abundance of information on social contexts, background characteristics, attitudes and expectations, sample size is often not sufficient to obtain robust estimates. Moreover, the issue of attrition makes inference from these data sources fragile.

4.8.3  Add Health There is one survey that dedicated incredible attention and resource on network analysis. This is the National Study on Adolescent to Adult Health (Add Health). Add Health is a nationally representative cohort drawn from a probability sample of 80 U.S. high schools and 52 U.S. middle schools, representative of U.S. schools in 1994–1995 with respect to region, urban setting, school size, school type, and race or ethnic background. Add Health includes a network component in Wave I and II where the entire social network of friends in selected schools has been collected. These data have been studied extensively in relation to health behaviors. A very important feature of Add Health is that respondents were asked to provide names of

4  Social Contagion Effects in Fertility: Theory and Analytical Strategy

61

up to ten friends in Wave I. These friends were then included in the sample and followed over time. In this way, we have a snapshot of the friendship network during high school that can be used to study long-term demographic events over time. In addition, in Wave III, respondents are asked to provide information on the relationship with the previously reported friends. A computer name generator also provides names of possible friends from the same school, based on common characteristics. Respondents can say that they are still friends, they are no longer friends, only acquaintances, or that they were never friends. This retrospective information on friendship has been used to separate the social influence effect of friendship on fertility from common confounders (Balbo and Barban 2014). There are several limitations in this design. First, the social network is static at high school, as there are no new friends entering the survey. There is no information, for instance, of friends from college or the workplace. In addition, we have information only on dyads and we cannot reconstruct more complex friendship networks. Nevertheless, this study pioneered the analysis of long-term friendship networks and represents one of the best data sources for the study of social interaction on life demographic behavior.

4.8.4  Other Social Networks With the explosion of internet social network websites, lives are connected to a degree that is unprecedented in human history. In 2019 Facebook reached in 2.37 billion users, with continuous growth in all continents. People continuously post pictures and update their biographies, including information on births. There is now an increasing interest in using data from internet social network websites to estimate demographic variables. Individual information on networks, however, are restricted to researchers and require the authorization of users. We are not aware of studies that use these data for investigating social effects in fertility, at the moment. We do not exclude, however, that specific surveys who target a sample of individuals and tackle the delicate issues of privacy could be designed in the near future. Such data, complemented with more traditional data sources could give insight on the dynamics of social influence among friends and relatives. There are, however, other types of “social network data” that have much fewer privacy issues. These is increasingly available data on historical family networks, from internet genealogy websites and historical micro-censuses. Internet genealogy websites provide the platform for millions of genealogy enthusiasts to complete information on their ancestors by creating their family trees. The most interesting example is FamiLinx (Kaplanis et al. 2018). FamiLinx is based the public information available on Geni. com, a genealogy-driven social network that is operated by MyHeritage. Users can decide whether they want their profiles to be public or private. New or modified family tree profiles are constantly compared to all existing profiles, and if there is high similarity to existing ones, the website offers the users the option to merge the profiles and connect the trees. FamiLinx is a database created for scientific purposes that contains public profiles of individuals from Geni.com. Substantive data quality

62

N. Balbo and N. Barban

issues have been corrected by removing incorrect family relationships and by resolving differences on individual records reported by multiple genealogists. The public FamiLinx database contains basic demographic information on 43 million individuals, derived by 12  million “founders”, that are individuals who have no parents in the database. For 4 million individuals, information on year of death and year of birth are available. By combining this information and reconstructing family networks, we suppose it is possible to study diffusion processes in fertility, for example by looking at age at birth among individuals belonging to the same extended family network. Similarly, micro-census data, available publicly from IPUMS, can be used to create historical “registers” and understand social influence in fertility, especially in periods of fertility transition.

4.9  Discussion In the present chapter we aimed to review the existing literature on social interaction effects on fertility. We described the mechanisms that have been identified as the processes through which fertility behavior may become contagious among different types of “relevant others”, such as friends, kin, and coworkers. In doing so, we highlighted the main sociological theories explaining the diffusion phenomenon at the micro-level, as well as the analytical strategies used by existing studies. Such strategies aim to overcome the challenges deriving from the difficulty of disentangling “pure” social interaction effects from other confounding forces, such as selection and contextual effects. The relevance of this identification problem is evident in the active and large debate on possible empirical strategies to measure diffusion and contagion dynamics. In this piece of work, we also provided a useful overview of the existent data suitable to analyze peer effect on fertility, also discussing the potential use of new data sources, such as digital or social network data, as well as historical genealogies. The availability of new longitudinal or retrospective data on individuals’ networks, as well as advancement in the methodology to identify social interaction effects on an individual’s behaviors, enabled the rise of a recent “second wave” of studies focusing on social influence and fertility. These studies provide robust evidence of a significant effect that different “relevant others” have in shaping an individual’s fertility behavior, at least in the short-term. We believe that this body of research could be fruitfully extended in two key ways. First, building on the recent work of Buyukkececi and colleagues (Büyükkeçeci et al. 2020), we could explore more complex network structures that allow to (i) take into account indirect social contagion effects, such as friends of friends influence; (ii) social spillover effects from one network to the other, such as from family to workplace to friendship circle and vice versa. Second, future studies could extend our knowledge on peer effects by looking at their impact on other life-course transitions and family-formation behaviors, such as leaving home, marriage, divorce or residential mobility. Research on these outcomes is still very scarce if not virtually absent.

4  Social Contagion Effects in Fertility: Theory and Analytical Strategy

63

Finally, we believe that the advancements in the empirical strategies used to properly identify social contagion effects might allow future research to better uncover the relevant role that peer effects have for policies. On the one hand, social interaction effects should be more carefully taken into account as amplifier forces on interventions to support childbearing in low fertility settings. On the other hand, a thorough understanding of social contagion mechanisms is crucial to increase the effectiveness of policies that, for example, aim at reducing teenage pregnancies in advanced societies or adopting contraception in developing countries.

References Asphjell, M. K., Hensvik, L., & Nilsson, P. (2013). Businesses, buddies, and babies: Fertility and social interactions at work (Working Paper No 2013:8). Uppsala: Uppsala University, Center for Labor Studies. Astone, N. M., Nathanson, C. A., Schoen, R., & Kim, Y. J. (1999). Family demography, social theory, and investment in social capital. Population and Development Review, 25(1), 1–31. Balbo, N., & Barban, N. (2014). Does fertility behavior spread among friends? American Sociological Review, 79(3). Balbo, N., & Mills, M. (2011a). Social capital and pressure in fertility decision-making: Second and third births in France, Germany and Bulgaria. Population Studies, 65(3), 335–351. Balbo, N., & Mills, M. (2011b). The influence of the family network on the realization of fertility intentions. Vienna Yearbook of Population Research, 2011(9), 179–206. Bernardi, L. (2003). Channels of social influence on reproduction. Population Research and Policy Review, 22(5–6), 427–555. Billari, F. (2004). Becoming an adult in Europe: A macro(/micro)-demographic perspective. Demographic Research, Special 3, 15–44. Bongaarts, J., & Watkins, S. (1996). Social interactions and contemporary fertility transitions. Population and Development Review, 22(4), 639–682. Bourdieu, P. (1986). The forms of capital. In J.  G. Richardson (Ed.), Handbook of theory and research for the sociology of education (pp. 241–258). New York: Greenwood. Bühler, C., & Fratczak, E. (2007). Learning from others and receiving support: The impact of personal networks on fertility intentions in Poland. European Societies, 9(3), 359–382. Bühler, C., & Philipov, D. (2005). Social capital related to fertility: Theoretical foundations and empirical evidence from Bulgaria. Vienna Yearbook of Population Research, 3, 53–81. Büyükkeçeci, Z., Leopold, T., van Gaalen, R., & Engelhardt, H. (2020). Family, Firms, and Fertility A Study of Social Interaction Effects. Demography (forthcoming)., 57(1), 243–266. Casterline, J. B. (2001). Diffusion processes and fertility transition: Introduction. In J. B. Casterline (Ed.), Diffusion processes and fertility transition: Selected perspectives (pp. 1–38). Washington, DC: National Research Council, National Academy Press. Cohen-Cole, E., & Fletcher, J. M. (2008). Is obesity contagious? Social networks vs Environmental factors in the obesity epidemic. Journal of Health Economics, 27(5), 1382–1387. Coleman, J.  S. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94(Suppl), S95–S120. Elder, G.  H., Jr. (Ed.). (1985). Life course dynamics: Trajectories and transitions, 1960–1980. Ithaca: Cornell University Press. Elder, G. H., Johnson, M. K., & Crosnoe, R. (2003). The emergence and development of life course theory. In J. T. Mortimer & M. J. Shanahan (Eds.), Handbook of the life course (pp. 3–19). Cham: Springer US.

64

N. Balbo and N. Barban

Flap, H., & Völker, B. (2004). Creation and returns of social capital: A new research program. London: Routledge. Fletcher, J.  M., & Yakusheva, O. (2016). Peer effects on teenage fertility: Social transmission mechanisms and policy recommendations. American Journal of Health Economics, 2(3), 300–317. Granovetter, M.  S. (1973). The strength of weak ties. American Journal of Sociology, 78(6), 1360–1380. Hart, R., & Cools, S. (2019). Identifying interaction effects using random fertility shocks. Demographic Research, 40(10), 261–278. Kaplanis, J., Gordon, A., Shor, T., Weissbrod, O., Geiger, D., Wahl, M., Gershovits, M., Markus, B., Sheikh, M., Gymrek, M., Bhatia, G., MacArthur, D. G., Price, A. L., & Erlich, Y. (2018). Quantitative analysis of population-scale family trees with millions of relatives. Science, 360(6385), 171–175. Katz, M., & Shapiro, C. (1985). Network externalities, competition and compatibility. American Economic Review, 75(3), 424–440. Kohler, H.-P. (2001). Fertility and social interaction: An economic perspective. Oxford: Oxford University Press. Kohler, H. P., Billari, F. C., & Ortega, J. A. (2002). The emergence of lowest-low fertility in Europe during the 1990s. Population and Development Review, 28(4), 641–681. Kohler, H.-P., Billari, F. C., & Ortega, J. A. (2006). Low fertility in Europe: Causes, implications and policy options. In F. R. Harris (Ed.), The baby bust: Who will do the work? Who will pay the taxes? (pp. 48–109). Lanham: Rowman & Littlefield Publishers. Kuziemko, I. (2006). Is having babies contagious? Estimating fertility peer effects between. Siblings: Princeton University Manuscript. Liebowitz, S. J., & Margolis, S. E. (1995). Are network externalities a new source of market failure? Research In Law And Economics, 17(1), 1–22. Lin, N., Cook, K. S., & Burt, R. S. (2001). Social capital: Theory and research. New Brunswick: Aldine Transaction. Lyngstad, T.  H., & Prskawetz, A. (2010). Do siblings’ fertility decisions influence each other? Demography, 47, 923–934. Manski, C.  F. (1993). Identification of endogenous social effects: The reflection problem. The Review of Economic Studies, 60(3), 531–542. Montgomery, M. R., & Casterline, J. B. (1996). Social learning, social influence, and new models of fertility. Population and Development Review, 22, 151–175. Philipov, D., Spéder, Z., & Billari, F. C. (2006). Soon, later, or ever? The impact of anomie and social capital on fertility intentions in Bulgaria (2002) and Hungary (2001). Population Studies, 60(3), 289–308. Pink, S., Leopold, T., & Engelhardt, H. (2014). Fertility and social interaction at the workplace: Does childbearing spread among colleagues? Advances in Life Course Research, 21, 113–122. Raab, M., Fasang, A. E., Karhula, A., & Erola, J. (2014). Sibling similarity in family formation. Demography, 51(6), 2127–2154. Sacerdote, B. (2001). Peer effects with random assignment: Results for Dartmouth roommates. The Quarterly Journal of Economics, 116(2), 681–704. Schoen, R., Kim, Y., Nathanson, C., Fields, J., & Astone, N. M. (1997). Why do Americans want children? Population and Development Review, 23(2), 333–358. Snijders, T. A. B. (1999). Prologue to the measurement of social capital. The Tocqueville Review, 20(1), 27–44. Van der Gaag, M. (2005). Measurement of individual social capital. PhD dissertation, The Netherlands: University of Groningen, Sociology Department.

Chapter 5

Context of Interracial Childbearing in the United States Zhenchao Qian and Yifan Shen

Much of the literature on integration of individuals of various racial/ethnic backgrounds focus on interracial marriage (Choi 2020; Qian and Lichter 2007). Interracial marriage is an important indicator of racial/ethnic relations in the United States (Alba and Nee 2003). It breaks down racial/ethnic barriers and connects together diverse families, friends, and other social networks. In recent decades, intermarriage among African Americans, Hispanics, Asian Americans, American Indians, and whites has increased. Educational attainment is one important factor. More minority individuals with high levels of educational attainment have increased rates of interracial marriage between whites and minorities (Qian and Lichter 2011). Education creates contact opportunities in schools and workplaces, reduces social distance, increases interactions, builds friend and/or romantic relationships, and facilitates intermarriage. Yet, interracial marriage may represent just one form of racial/ethnic integration. Dating, casual sexual relationships, cohabitation, and marriage undergo a winnowing process and each transition selects couples who share high levels of education and have same racial/ethnic backgrounds (Blackwell and Lichter 2004). This selection process means that cohabiting and other casual relationships tend to involve couples with lower levels of education and/or different racial/ethnic backgrounds compared to marital relationships (Joyner and Kao 2005). Casual sexual relationships, by nature, are unstable and short term because stable relationships would transition to cohabitations or even marriage (Sassler et  al. 2018). However, the impact of such relationships may have a long-lasting impact when children are involved. In 2007, nearly 40% of the births were born outside of marriage (Cherlin 2010). Births born to unmarried mothers tend to have low education and socioeconomic status (Cherlin 2010; Lichter et al. 2006). Higher rates of Z. Qian (*) · Y. Shen Department of Sociology, Brown University, Providence, RI, USA e-mail: [email protected]; [email protected] © Springer Nature Switzerland AG 2020 R. Schoen (ed.), Analyzing Contemporary Fertility, The Springer Series on Demographic Methods and Population Analysis 51, https://doi.org/10.1007/978-3-030-48519-1_5

65

66

Z. Qian and Y. Shen

interracial relationships among casual than among marital relationships also suggest that disproportionately more children born to unmarried mothers are multiracial. Clearly, a focus on intermarried couples and their fertility behavior does not present a comprehensive picture of racial/ethnic integration, assimilation, and boundary blurring. Unmarried couples and their childbearing may also spur the growth of America’s multiracial population and contribute to strengthening or weakening of racial boundaries across racial/ethnic groups (Alba 2009; Lichter and Qian 2018a). In this paper, we study patterns and trends of all births born to parents of various racial/ethnic pairings in select years between 1980 and 2016. We further distinguish births by mothers’ marital status (married versus unmarried). Data come from the National Vital Statistics System (NVSS), National Center for Health Statistics, which provides annual demographic and health data for all the births and includes specific information on mothers’ marital status, mothers’ and fathers’ educational attainment and race/ethnicity. The information allows us to track increases in births born to parents of different racial/ethnic backgrounds and explore how mothers’ marital status and educational attainment play a role in interracial fertility.

5.1  Literature Review Interracial marriage, as an engine of social change, breaks down social barriers, builds as a family two individuals of different racial/ethnic backgrounds, and connects together, to a varying degree, two sets of relatives, friends, and social networks. Interracial fertility often follows and biracial or multiracial children born to intermarried couples promote population diversity in ways different from influx of racial/ethnic minority immigrants and natural growth of racial/ethnic minority populations (Lichter and Qian 2018a). Multiracial individuals may identify or be identified one race or two or more races. Their contributions to racial diversity are not straightforward and depend on their race and socioeconomic status (Bratter 2018). Historically, black-white individuals identified black due to African Americans’ unique experiences—the rule that one drop of black blood made one black solidified the black/white boundary (Davis 1991). This rule did not apply to those of American Indian-white descent and many identified white (Eschbach 1995). Clearly, multiracial individuals differ in racial classification and could either reinforce or blur racial group boundaries (Fu 2008). Multiracial individuals in prior censuses could only identify one single race yet have been able to choose more than one race since the 2000 census. However, not all multiracial individuals choose more than one race. For example, children born to interracially married black-white couples are less likely to identify as biracial (white and black) than those born to interracially married American Indian-­ white (as white and American Indian) or Asian American-white couples (as white and Asian American) (Bratter 2007; Lichter and Qian 2018a). Differences in racial classification among multiracial individuals reflect the salience of race and varying

5  Context of Interracial Childbearing in the United States

67

degrees of discrimination and prejudice against racial minorities in American society (Okamoto 2014). Racial classification of multiracial individuals reflects challenges of multiracial individuals’ living experiences in American society. Such challenges were especially difficult in the past when interracial marriage was rare and multiracial children were few. After all, the Supreme Court did not rule that laws forbidding people of different races to marry were unconstitutional until 1967. Intermarried couples overcame barriers to marry, but low acceptance of intermarriage in the American public, their lack of social support to bear and raise children, and their concern for their multiracial children to grow up in such an environment might discourage fertility, thus had fewer children than their racially endogamous counterparts (Childs 2005). Over time when interracial marriage became more widely accepted as a result of improved racial relations and reduced social barriers across racial groups, intermarried couples may find it easier to receive social support and more likely to have number of children similar to their endogamous counterparts. This leads to our first hypothesis: The rates of births born to interracially married parents increase over time. Racial differences in fertility are strong in the United States. Racial minority women, except for Asian Americans, have more children on average than whites, which have slowed down the decline in total fertility rates in the United States compared to other more developed countries. In 1990, TFR ranged from 2.0 to 3.0 among five racial/ethnic groups (3.0, 2.5, 2.2, 2.0, and 2.0, respectively, among Hispanics, blacks, American Indians, whites, and Asian Americans) (Hamilton et al. 2003). Classical assimilation theory underscores intermarriage to be an important stage of assimilation for racial minorities and immigrants (Gordon 1964; Park and Burgess 1969). Minority or immigrant spouses in interracial marriage assimilated into mainstream society and adopted the host population’s cultural patterns. If true, we may expect fertility rates of interracial couples be more similar to those of white endogamous couples than to those of their minority counterparts. Choi and Goldberg (2018) find evidence that pregnancy rates among interracial couples are generally closer to those of endogamous whites in the 2000s, with the exception of white wife-black husband couples. In contrast, Alba and Nee’s (2003) reformulated theory challenges the classical assimilation. They argue that contemporary immigrants and minorities are racially distinctive, subject to prejudice and discrimination, and may not intentionally seek to assimilate. However, the quest for better lives through schooling and work creates opportunities for social interactions and potentially leads to intimate relationships with whites. Consequently, intermarriage between minorities and whites may not be a simple integration of minority individuals into the majority but can be a two-way integration process involving majority and minority populations (Alba and Nee 2003). Thus, interracial fertility between whites and each minority group is likely to fall somewhere between the two groups rather than at the level of whites (Fu 2008). Over the years, racial/ethnic differences in fertility have become much smaller. In 2018 TFRs converged among all racial/ethnic groups ranging from 1.5 for Asian

68

Z. Qian and Y. Shen

Americans to 2.0 for Hispanics (1.8 for blacks, 1.7 for American Indians, and 1.6 for whites) (Martin et al. 2019). As a result, racial differences in fertility are likely to play a small role in explaining differences in interracial fertility. Yet, as interracial marriage varies by race/ethnicity, social barriers across racial/ethnic groups remain to be important. Lower social barriers between two racial groups promote interracial marriage and affect other aspects of social life, including fertility behavior (Fu 2008). Hispanics, a pan-ethnic group, include people of different racial backgrounds and Hispanic whites are more likely to marry non-Hispanic whites than Hispanic nonwhites (Qian and Cobas 2004). On the other hand, African Americans remain least likely to marry whites among all minority groups (Qian and Lichter 2011). Such differences indicate that race of the minority spouse matters and influences rates of fertility among intermarried couples. It is expected that fertility is more likely among Hispanic-white couples than among black-white couples, despite racial convergence in fertility. Fertility differences by educational attainment are well documented (Yang and Morgan 2003). Highly educated individuals invest more time in education and career, and have children later in life than their less educated counterparts (Brand and Davis 2011). Fertility rates are low among college-educated women due to delay in childbearing and high opportunity costs to childbearing (Blossfeld 1991; Ellwood and Jencks 2004). However, highly educated women are much more likely to have marital childbearing compared to their less educated counterparts. They marry at later ages but tend to stay married and center on family life around marriage. Asian Americans, including those who marry whites, are more likely to marry and less likely to cohabit than individuals of other racial groups, and the likelihood of marriage between Asian Americans and whites increases by educational attainment (Qian and Lichter 2007). Thus, we hypothesize that the likelihood of fertility among Asian American-white parents has a strong educational gradient. Marriage as a social institution has declined in recent decades, especially among those with less education (Kalmijn 2013; Sassler and Lichter 2020). Instead of marriage, more people with less education cohabit or stay single. Those who do marry experience a higher level of marital disruption (Cherlin 2020). They delay or forgo marriage but do not necessarily postpone having children (Cherlin 2010; Edin and Kefalas 2005; Guzzo and Hayford 2020). Only 4% of the births were born outside marriage in 1950, but this figure increased to 28% in 1990 and 41% in 2012 (Cherlin 2010; Martin et al. 2013). According to data from the National Survey of Family Growth between 2006 and 2010, 26% of the births born to college educated mothers was unmarried, but the figures were 49%, 41%, and 39%, respectively, for those with less than high school, high school, and some college (Sweeney and Raley 2014). The likelihood of intermarriage between African Americans and whites was low and did not vary strongly by educational attainment (Qian and Lichter 2007). However, given higher levels of nonmarital fertility among less educated women, we may expect that rates of interracial fertility between African Americans and whites decline by educational attainment and increase over time. Intermarriage between Hispanics and whites is also strongly influenced by educational attainment yet nonmarital childbearing was also common among less educated Hispanics (Qian

5  Context of Interracial Childbearing in the United States

69

and Lichter 2007). The combination of these two factors leads to the hypothesis that interracial fertility between Hispanics and whites is high and increases over time, but does not vary strongly by educational attainment. American Indians are likely to follow the patterns of Hispanics, because of strong differences in educational attainment and marital formation between American Indians living in cities and living on reservations (Eschbach 1995; Snipp 2002). Furthermore, racial/ethnic differences in fertility by context of childbearing remain strong. African Americans, in particular, have moved away from marriage to a large extent and exhibit strong disconnections between marriage and childbearing (Cherlin 2020). Among the births born to African Americans in 2012, 72% were to unmarried women, while the corresponding figures were 29% for unmarried non-­ Hispanic white women and 54% for unmarried Hispanic women, respectively (Martin et al. 2013). These figures were 17%, 37%, and 67% respectively, for white, Hispanic and black mothers in 1990 (Sweeney and Raley 2014). In addition, among women giving births outside marriage, black mothers were more likely to be unpartnered and less likely to be cohabiting than were white mothers (Cherlin 2020; Manning et al. 2014). Unmarried births to Asian Americans were not available but likely much lower than those in other racial groups. We hypothesize that the likelihood of unmarried fertility was much greater than the likelihood of marital fertility between blacks and whites and much lower between Asian Americans and whites. The differences would be smaller among Hispanic-white parents and American Indian-white parents because interracial marriage was common among Hispanics and American Indians with high levels of educational attainment and nonmarital fertility was relatively high among Hispanics and American Indians with less education. Previous literature has identified two important interconnected findings about interracial relationships: (1) individuals go through various relationships over the life course and each relationship transition is associated with a lower likelihood of moving into an interracial relationship and (2) interracial relationships are more common among dating and cohabiting relationships than among married relationships (Joyner and Kao 2005). The explanation for these findings is straightforward: individuals may explore interracial relationships early in the life course but when they decide to settle, they tend to find someone just like themselves (Kao et  al. 2019). These patterns are consistent with the winnowing process that interracial relationships become less likely when young men and women grow older and people in interracial relationships may not transition into marriage due to racial/ethnic barriers (Blackwell and Lichter 2004). Consequently, a focus on intermarried couples’ fertility has missed a significant number of births born to unmarried mothers who may be cohabiting or in singlehood. Census and American Community Survey data rely on individuals’ relationships with the householder to identify children and their parents. Cohabiting partners’ information is available but the births born to cohabiting couples may only be linked to one parent if cohabitation dissolves (Lichter et al. 2016; Lichter and Qian 2018b; Raley 2001). Without knowing dissolved partners’ race information, births born to unmarried interracial parents are underestimated. Vital statistics data are based on

70

Z. Qian and Y. Shen

births, which include not only marital but also nonmarital births and provide information about both the mother and father, regardless of their relationship status, which make it possible to estimate the number of births born to unmarried parents of different racial backgrounds. We can explore whether births born to unmarried mothers are more likely to be interracial than those born to married mothers and whether racial pairing or mothers’ educational attainment of interracial parents are associated with levels of interracial fertility. Previous studies on multiracial children are limited to those born to interracially married couples, and miss those born to unmarried mothers. Our study fills this void.

5.2  Data and Methods The National Vital Statistics System (NVSS) of National Center for Health Statistics has publicly available demographic and health data for births occurring every year since 1968. Collecting data from birth certificates filed in vital statistics offices of each state and District of Columbia, the NVSS natality microdata are a complete census of new births for the entire country. For example, the total number of births was 3,617,981 in 1980, 4,118,907 in 2004, and 3,956,112 in 2016. For our purpose, these data provide trends and patterns of births born to parents of different racial backgrounds. Each birth certificate includes information on parents’ race/ethnicity, education, marital status, nativity status but mothers’ information has much higher completion (non-missing) rates than that of fathers. All states collect in their birth certificates data on race of the mother and father, but only 22 states (Arizona, Arkansas, California, Colorado, Florida, Georgia, Hawaii, Illinois, Indiana, Kansas, Maine, Mississippi, Nebraska, Nevada, New Jersey, New Mexico, New York, North Dakota, Ohio, Texas, Utah, and Wyoming) asked whether father and mother, respectively, are of Hispanic origin prior to 1993. Thus, in order to include Hispanics in earlier years, we relied on data from the 22 states for the period between 1980 and 1993 and from all the country between 1993 and 2016. Race/ethnicity for mothers and fathers is classified, respectively, as non-Hispanic white, non-Hispanic black, non-Hispanic American Indian, non-Hispanic Asian, and Hispanic. Each year less than .55% of mothers report their race as “unknown or not stated” or “other races” and are dropped from our analyses. The percent of fathers who belong to “other races” is also negligible, but every year a small minority of fathers whose race or Hispanic origin is “unknown or not stated.” We exclude the unknown or not stated cases in our analysis. Fortunately, the share is small and does not influence the findings. Hispanics is a pan-ethnic group inclusive of those from various racial backgrounds, but are treated here as one racial/ ethnic group because of its unique position in the United States (Perez and Hirschman 2009). In addition to race and ethnicity, birth certificates include information on mothers’ age (continuous variable), marital status (married or not married), education (coded as less than high school, high school, some college, and college or more),

5  Context of Interracial Childbearing in the United States

71

nativity (native born or foreign born), and birth order (1, 2, 3, or 4+). Fathers’ education and nativity are only available for a few years. As a result, our analysis relies mostly on mothers’ information including age, marital status, education, and nativity. Our first goal is to examine whether percent of births born to interracial parents has increased over time. We use two time periods: the first period from 1980 to 2004 is limited to the 22 states in which the Hispanic origin question was asked. The second time period from 1993 to 2016 covers all the states in the United States. We explore how trends differ by mothers’ marital status and parents’ racial pairing (white-white, black-black, Hispanic-Hispanic, American Indian-American Indian, Asian American-Asian American, black-white, Hispanic-white, American Indian-­ white, Asian American-white, and minority-minority). We also examine further how fertility by racial pairing varies by mothers’ age, marital status, educational attainment, and nativity status over time. We then conduct multivariate analysis to examine trends in births by parents’ racial pairing. Log-linear models, which have long been used to study intermarriage patterns across racial/ethnic boundaries, are employed. One main advantage of these models is their ability to study patterns and trends of births of parents’ racial pairing by controlling for changes in marginal distributions by parents’ racial/ethnic distributions over time. We can answer the extent of the increase in interracial fertility after controlling for growing population group sizes of racial/ethnic minorities. In addition, we can explore how temporal trends in interracial fertility vary by mothers’ educational attainment and marital status. The basic log-linear model takes the following form:

log Fijmt = b 0 + b iMR + b jFR + b mME + b tT + b itMRT + b jtFRT



(5.1)

where Fijmt is the expected number of births born to mothers in race/ethnicity i and fathers in race/ethnicity j, and mothers’ education m at time t. b 0 is the constant, b iMR b jFR is mothers’ (fathers’) race/ethnicity (i, j  =  white, b ME African American, American Indian, Asian American, and Hispanic), and m is mothers’ educational attainment (m = less than high school, high school, some college, and college or more). b tT denote year (t = 1980 or 1993 for data from 22 states in the first series and t = 1993 or 2016 for data from all the states in the second series). We also account for the two-way interactions between mothers’ race/ethnicity and time ( b itMRT ) and the two-way interactions between father’s race/ethnicity

(

)

and time ( b jtFRT ).

log Fijnt = b 0 + b iMR + b jFR + b nM + b tT + b itMRT + b jtFRT



(5.2)

Model 2 is the same as Model 1, except that it replaces mothers’ educational attainment with mothers’ marital status b nM .

( )

72

Z. Qian and Y. Shen

5.3  Results

trends among 22 states that collected Hispanic origin data from parents trends among all states imputed for all states in 1980 and 1990

Fig. 5.1  Trends in share of interracial births among all births

18

16

Year

20

20

10 20

04 20

00 20

93 19

90 19

19

80

0

% Among all births 5 10

15

Our first goal is to document trends in interracial fertility over the period when interracial marriage was on the rise. Figure 5.1 reveals a significant increase in the number of births born to parents of different racial backgrounds. The percent of multiracial births increased from 6% in 1980, to 9% in 1993, and to 12% in 2004. This trend is based on data from the 22 states in which the Hispanic origin question was included in the birth certificates. The share of multiracial births was about one percentage point lower in 1993, 2000, and 2004, respectively, among all the states in the U.S. For ease of comparisons, we combine the two data series by imputing the percentages of multiracial births for the entire country in 1980 and 1990, by calculating the ratios of multiracial births for all the states to those for the 22 states in 1993, 2000, and 2004 and multiplying the corresponding percentages for the 22 states in 1980 and 1990 by the median of the three ratios (1993, 2000, and 2004). The dotted line connecting 1980, 1990, and then 1993 is the imputed data. The percent of multiracial births for the entire country increased from 5.4% in 1980 to 14.3% in 2016, nearly three times as high as in 1980. In other words, one out of seven births were born to parents of different racial backgrounds, which indicates a strong increase in interracial relationships. Such a strong increase could be due to two factors: one is that interracial fertility increases as a result of wider acceptance of interracial marriage and other relationships; and the other could be rises in number of interracial relationships as a result of increases in racial minority populations.

5  Context of Interracial Childbearing in the United States

73

15

% among all births

12 9 6 3 0

1980

1990

1993

2000

2004

2010

2016

Year black-white American Indian-white minority-minority

Asian American-white

Hispanic-white

Fig. 5.2  Percent of births born to minority-white and minority-minority parents by relationship type and year

Figure 5.2 shows in detail how each type of interracial relationships contributes to shares of multiracial births in select years. Interracial relationships are classified as black-white, American Indian-white, Asian American-white, Hispanic-white, and minority-minority (black-Hispanic, for example). As in Fig. 5.1, Fig. 5.2 uses the imputed figures in 1980 and 1990 for the entire country (rather than for the 22 states). Percent of births born to interracial parents increased for each relationship type between 1980 and 2016, by 2.7 times for African American-white parents, .4 times for American Indian-white parents, 1.7 times for Asian American-white parents, 1.2 times for Hispanic-white parents, and 4.2 times for parents of two different minority groups. Among interracial parents, about half were Hispanic-white parents, with a declining share from 55% in 1980 to 46% in 2016, in large part due to a growing share of Hispanic immigrants who are less likely to be in interracial relationships than their US born counterparts. Notably, the share of births born to African American-white parents increased from 14% in 1980 to 19% in 2016 and the share of births born to parents of two different minority backgrounds increased from 9% to 18% over the same period. The increases were most rapid among these two groups, a pattern not witnessed in interracial marriage, suggesting more of these births were born to unmarried parents. In contrast, the share of births born to American Indian-white parents declined from 8% in 1980 to 4% in 2016 and the percent of births born to Asian American and white parents remained unchanged at about 14% over the period. We now examine how births born to interracial parents by mothers’ race/ethnicity and marital status. For 1980 and 1990, Table 5.1 uses imputed percentages as in Fig. 5.1, but keeps the population size of mothers from the original 22 states for

Unmarried 14.09 (67378) 13.99 (140540) 14.90 (267300) 18.25 (341432) 19.62 (372606) 20.93 (439023) 21.55 (443919)

African American Married Unmarried 1.87 1.80 (133469) (84729) 3.86 3.03 (125867) (113770) 4.55 3.49 (195290) (173946) 6.65 4.87 (184652) (198740) 8.26 6.10 (173334) (195311) 11.14 7.89 (157390) (223523) 12.57 9.62 (172152) (231969)

American Indian Married Unmarried 34.42 35.50 (8114) (2520) 42.36 37.60 (8433) (4608) 50.48 38.88 (15897) (9179) 55.82 41.68 (15651) (12464) 57.76 38.11 (14513) (14998) 61.89 40.68 (13068) (17433) 62.73 43.46 (11727) (15612)

Asian American Married Unmarried 23.51 33.30 (50010) (2420) 19.83 34.70 (93436) (8168) 20.32 32.66 (125237) (12434) 20.55 35.44 (166870) (18694) 21.59 34.93 (186905) (21967) 22.36 37.44 (197316) (27655) 20.33 37.39 (236631) (30882)

Hispanic Married 13.88 (236327) 14.95 (348025) 15.07 (393906) 15.04 (462598) 14.97 (504501) 18.26 (435948) 21.00 (430032)

Note: births whose mother or father has unknown or other races OR marital status is unknown/missing are dropped from the sample Total number of mothers by race/ethnicity and marital status are given in parentheses (the figures for 1980 and 1990 refer to the 22 states)

2016

2010

2004

2000

1993

1990

1980

White Married 2.92 (1228641) 4.09 (1208079) 4.38 (2006073) 5.57 (1849979) 6.44 (1726648) 7.72 (1518759) 8.66 (1476144)

Table 5.1  Percent of births born to parents of different racial/ethnic groups by mothers’ race/ethnicity, marital status, and year Unmarried 12.89 (48527) 10.56 (133794) 10.18 (176959) 11.86 (241866) 10.88 (320875) 12.89 (385189) 15.85 (381241)

74 Z. Qian and Y. Shen

5  Context of Interracial Childbearing in the United States

75

each racial/ethnic group (in parentheses). Percent of births born to minority-father-­ white-mother parents is relatively small due to large size of white endogamous relationships. Among married white mothers, percent of multiracial births increased from 2.9% in 1980 to 8.7% in 2016. Among unmarried white mothers (cohabiting or other unmarried relationships), the figures increased from 14.1% to 21.6% over the same time period. The greater percent of births born to interracial parents among unmarried white mothers than among married white mothers suggests that there are more interracial relationships among unmarried white mothers. Percent of having a birth with fathers of another racial/ethnic group was about the same in 1980 between married and unmarried African American mothers, but increased faster for the married (12.6% in 2016) than for the unmarried (9.6% in 2016). American Indians, the smallest racial group, had the highest percent of multiracial births. The percent of births (with husbands of another racial/ethnic group) increased from 34% in 1980 to 63% in 2016 among married American Indian mothers and increased from 36% to 43% over the same period among unmarried American Indian mothers. The patterns between married and unmarried mothers for African Americans and American Indians are opposite to those for white mothers: shares of interracial relationships are greater among married African American and American Indian mothers than among their unmarried counterparts. This suggests a much stronger likelihood of non-marital births to same-race African American or American Indian parents. Among married Asian American mothers, shares of births with non-Asian American husbands declined slightly from 24% in 1980 to 20% in 2016, reflective of the decline in interracial marriage among Asian Americans due to a rapid increase in the Asian American population over the years. Yet, among unmarried Asian American mothers, the percent of multiracial births was relatively high and increased over time. Proportionately, unmarried Asian American mothers were more likely to be in interracial relationships than their married counterparts. Among Hispanic mothers, the percent of multiracial births increased among the married, from 14% in 1980 to 21% in 2016, and increased from 13% to 16% over the same period among the unmarried. Overall, these results demonstrate the importance of changes in population size of various racial/ethnic groups in percent of births born to interracial parents by context of childbearing (married or unmarried). Our second goal is to explore similarities and differences in basic characteristics by parents’ racial/ethnic pairing. Table 5.2 presents the characteristics such as mothers’ age, educational attainment, nativity, and marital status in 1993 and 2016 (1993 was the first year in which birth certificates of all states in the United States include a question on whether the mother or father is of Hispanic origin). For white-­minority parents, we distinguish further whether the mother is white or a member of a given racial minority group. Notable changes over the period between 1993 and 2016 include an approximate two-year rise in mothers’ age at birth, tremendous improvement in mothers’ educational attainment, declined shares of foreign-born Hispanics and Asians, and increased shares of foreign-born whites and blacks over the period. Another significant increase is share of the births born to unmarried mothers for every racial/ethnic group except for Asian Americans. For example, the percent of births born to unmarried mothers almost doubled from 10.6% to 20.5% among

1993 Mothers’ age Mean 28.0 26.0 SD 5.5 6.0 Mothers’ education (%) 20.1 Less than 10.8 high school High 35.9 41.8 school Some 25.1 25.7 college BA or 28.3 12.5 higher Mothers’ nativity (%) Native 95.4 90.2 born Foreign 4.6 9.8 born Mothers’ marital status (%) Married 89.4 52.8 Unmarried 10.6 47.2 Mothers’ current birth order (%) 1 32.0 26.4

Both White

25.9 6.0 30.9

43.0 19.9 6.2

93.4 6.6

58.4 41.6 22.6

25.5 5.8

57.3

27.8

10.4

4.4

30.6

69.4

67.8 32.2

31.8

36.2

92.3 7.7

90.7

9.3

36.0

19.8

26.7

17.5

29.3 5.4

33.2

57.0 43.0

29.6

70.4

10.6

24.1

40.6

24.7

25.2 5.9

31.9

67.8 32.2

13.1

86.9

19.5

31.4

35.4

13.7

26.8 6.0

Black mother Both Both Both Minority white American Asian African Both American Minority father American Hispanic Indian

Table 5.2  Descriptive statistics by parents’ racial/ethnic pairing in 1993 and 2016

33.8

51.3 48.7

5.7

94.3

11.8

22.7

42.5

23.0

25.0 6.0

Black father white mother

34.1

83.4 16.6

32.2

67.8

19.8

28.3

37.0

14.9

27.6 5.7

Hispanic mother white father

33.8

73.9 26.1

5.1

94.9

15.4

23.1

39.8

21.7

26.0 5.9

Hispanic father white mother

28.5

76.5 23.5

2.5

97.5

10.3

24.6

41.7

23.5

25.7 5.7

Indian mother white father

32.7

74.8 25.2

2.4

97.6

11.5

22.7

43.8

22.0

25.4 5.8

Indian father white mother

38.4

91.5 8.5

73.3

26.7

39.9

26.0

26.8

7.3

29.8 5.5

Asian mother white father

35.4

83.5 16.5

7.6

92.4

34.2

26.2

29.8

9.8

28.2 6.0

Asian father white mother

76 Z. Qian and Y. Shen

2 3 4+ Total 2016 Mothers’ age Mean 29.5 28.2 SD 5.3 5.9 Mothers’ education (%) 5.7 11.1 Less than high school High 18.6 31.2 school Some 29.7 35.7 college BA or 46.0 21.9 higher Mothers’ nativity (%) Native 92.9 79.2 born Foreign 7.1 20.8 born Mothers’ marital status (%) Married 79.5 41.7 Unmarried 20.5 58.3

Both African Both American White 31.7 26.9 19.3 20.3 17.0 26.4 2,122,323 347,559

31.2 4.8

26.8 5.8 20.8

36.5 35.2 7.5

98.9 1.1

33.2 66.8

28.0 6.1

32.0

32.9

24.2

10.9

43.8

56.2

51.5 48.5

90.7 9.3

86.0

14.0

62.8

15.9

12.9

8.4

Both Asian American 31.5 16.3 16.0 104,232

Both American Both Hispanic Indian 27.7 22.4 19.1 18.8 21.4 36.1 483,937 13,241

44.5 55.5

21.7

78.3

22.3

37.5

28.8

11.4

27.9 6.0

55.7 44.3

13.7

86.3

29.9

38.4

24.2

7.5

28.3 6.0

Black mother Minority white Minority father 26.9 28.8 18.4 18.8 21.5 20.5 35,021 7874

41.0 59.0

4.5

95.5

22.8

35.8

30.3

11.2

27.9 5.8

Black father white mother 27.3 17.8 21.1 33,348

68.8 31.2

25.4

74.6

34.7

35.9

22.2

7.2

29.2 5.8

Hispanic mother white father 31.0 18.6 16.4 59,104

61.4 38.6

6.1

93.9

31.0

34.8

25.1

9.1

28.5 5.7

Hispanic father white mother 29.1 18.4 18.8 70,817

60.4 39.6

1.2

98.8

20.7

37.7

30.3

11.3

27.6 5.6

Indian mother white father 28.1 19.6 23.8 7960

58.1 41.9

2.1

97.9

22.5

36.4

30.6

10.4

27.5 5.6

Indian father white mother 28.5 18.8 20.0 9560

88.3 11.7

60.6

39.4

67.5

21.5

9.0

2.0

32.8 5.1

82.3 17.7

12.8

87.2

59.3

25.8

12.1

2.9

31.3 5.3

Asian father white mother 30.1 18.2 16.4 12,383

(continued)

Asian mother white father 31.9 16.8 12.9 21,660 5  Context of Interracial Childbearing in the United States 77

32.4 26.4 17.7 23.5 87,524

34.8 27.9 17.4 19.9 24,739

29.3 26.6 18.6 25.5 67,650

37.1 34.2 16.2 12.5 203,607

26.5 26.9 21.3 25.2 650,625

22.6 23.1 18.9 35.3 13,085

Black father white mother

Black mother Both Both Minority white American Asian Both American Minority father Hispanic Indian

Note: Missing data on mothers’ educational attainment, nativity, and race are excluded

Both African Both American White Mothers’ current birth order (%) 1 32.3 26.8 2 30.2 25.6 3 18.5 19.4 4+ 19.0 28.2 Total 1,685,484 355,912

Table 5.2 (continued)

34.7 29.7 17.7 17.8 103,505

Hispanic mother white father 32.2 28.3 18.5 21.0 118,460

Hispanic father white mother 29.5 28.5 18.8 23.2 8676

Indian mother white father 30.5 28.6 18.5 22.3 11,698

Indian father white mother 39.5 32.3 16.1 12.1 41,449

Asian mother white father

35.5 30.5 16.8 17.2 23,833

Asian father white mother

78 Z. Qian and Y. Shen

5  Context of Interracial Childbearing in the United States

79

white-white parents, from 47.2% to 58.3% among African American-African American parents, from 41.6% to 66.8% among American Indian-American Indian parents, and from 32.2% to 48.5% among Hispanic-Hispanic parents. The trend coincided with the rapid rises in cohabitation and female-headed single families (Qian 2014). In 2016, mothers’ age at birth was highest among Asian American-Asian American parents (31.2), despite 37.1% of the births being the first born, followed by white-white parents (29.5). Mothers’ age at birth was about 1.5 years younger among African American-African American and Hispanic-Hispanic parents and nearly 3 years younger among American Indian-American Indian parents (despite 35% of the births being fourth- or later-order births). Among minority-white parents, mothers’ age at birth was higher among Asian American-white parents (32.8 for Asian mother-white father parents and 27.5 for Asian father-white mother parents) but lower among African American-white, Hispanic-white, and American Indian-white parents than among white-white parents. Although we don’t know whether mothers’ earlier-order births were with the same partner, mothers’ age at birth suggests that interracial childbearing between whites and African Americans, Hispanics, and American Indians, respectively, started at younger ages than between whites and Asian Americans. The percent of mothers with completed college was highest when at least one parent is Asian American (67.5% among Asian American mother-white father parents, 59.3% among Asian American father-white mother parents, and 62.8% among Asian American-Asian American parents), followed by white-white parents (46.0%). Hispanic-white parents are also educationally selective—34.7% of mothers among Hispanic mother-white father parents and 31% of mothers among Hispanic father-white mother parents have completed college. These percentages are much higher than among Hispanic-Hispanic parents (10.9%), indicating that highly educated Hispanic mothers are more likely to be in relationships with whites than their less educated counterparts. In contrast, despite more mothers with college education among whites than among blacks, white mothers among white mother-­ black father parents have lower education than black mothers among black mother-­ white father parents. Among black-white relationships, 41.5% of white mothers but 31.7% of black mothers had high school education or less while 22.8% of white mothers but 29.9% of black mothers had completed college education. Although 56% of the mothers among Hispanic-Hispanic parents and 86% of the mothers among Asian American-Asian American parents were foreign born, only a quarter of the mothers among Hispanic mother-white father parents but 60.6% of the mothers among Asian American mother-white father parents were foreign born. The findings show that proportionately more Hispanic women in interracial relationships are US born than their Asian women counterparts. In 2016, two thirds of the births were born to unmarried mothers among American Indian parents, followed by 58% among African American parents. The figure was lowest (9%) among Asian American parents. Percent born to unmarried mothers also varied strongly among parents of different racial/ethnic backgrounds, ranging from 59% among white mother-black father parents and 44% among black mother-white father

80

Z. Qian and Y. Shen

parents to 12% among Asian American mother-white father parents and 18% among Asian American father-white mother parents. These descriptive analyses reveal that births born to African American-white parents were most likely to involve parents who started the relationships at younger ages, had less education, and were unmarried but the births born to Asian American-white parents, on the other hand, tend to involve parents who were older, had more education, and were married. Hispanic-­ white parents were closer to Asian American-white parents while American Indian-­ white parents resembled their African American-white counterparts. We turn to results from log-linear models in Table 5.3, which control for changes in marginal distributions of mothers’ and fathers’ race/ethnicity as well as mothers’ educational attainment for 1980, 1993, and 2016. Table 5.3 includes two models: the first model presents the odds of interracial fertility based on the 22 states in which the Hispanic origin question was included in 1980 and 1996 and the second model shows the same models for all the states in 1993 and 2016. In 1980, the odds of births born to African American-white parents were 1 relative to 100 births born to racially endogamous parents (either white or African American parents). Odds Table 5.3  Predicted odds of births born to parents of different racial/ethnic groups relative to those born to same racial/ethnic groups by mothers’ educational attainment and year 1980