The second demographic transition in Japan: a review of the evidence

Despite a half century of below-replacement fertility, Japan is typically not included in discussions and evaluations of the second demographic transition (SDT), a widely referenced framework for understanding family changes and attitudinal shifts associated with very low fertility. I address this limitation by drawing upon a range of published research and data sources to provide an empirical basis for thinking about how the Japanese experience does or does not conform to the general patterns of behavioral and attitudinal change associated with the SDT in the West. From this evidence, it seems clear that the prototypical pattern of family change in Northern and Western Europe has only partially emerged in Japan. The same is true of attitudes, particularly those related to gender. Consistent with depictions of the SDT in Europe, Japan has experienced substantial delays in marriage and childbearing along with notable increases in non-marital cohabitation and divorce. However, non-marital childbearing has remained at negligibly low levels and cohabiting unions have not emerged as an alternative to marriage. Attitudinal data show that endorsement of conventional family patterns and gender roles has declined, but remains at higher levels than in most SDT countries. Taken as a whole, these data describe a distinctive path to very low fertility in which universal forces of social and family change interact with strong normative expectations of two-parent families characterized by a clear gender division of labor.

Japan is not typically included in discussions and evaluations of the second demographic transition (SDT) despite a half century of below-replacement fertility, including 29 years with a total fertility rate (TFR) below 1.50. Japan's low profile in the SDT literature likely reflects two distinctive features of its demographic landscape -one behavioral and the other attitudinal. The former is the rarity of non-marital childbearing and the latter is limited evidence of growth in either individuation or female empowerment.
The decoupling of marriage and fertility is a defining behavioral feature of the SDT. This fundamental reshaping of family formation patterns has been widely documented in Europe and English-speaking countries, but has yet to occur in Japan or other low-fertility East Asian societies. Nonmarital births remain rare in Japan despite rapid growth in many other features of the SDT, including pronounced delays in marriage and childbearing, more lifelong singlehood, increases in cohabitation and premarital pregnancy, and relatively high divorce rates. The strong relationship between marriage and childbearing is a distinctive feature of the Japanese context that scholars have long recognized as inconsistent with conventional articulations of the SDT (e.g., Atoh 2002;Rindfuss et al., 2004). It is thus important to carefully consider how we should interpret the SDT and the insights it offers for understanding low fertility and social change in a society where marriage remains the only acceptable setting for childbearing.
Features of the social landscape considered fundamentally important in facilitating the SDT are increasing emphasis on individual (rather than family or group) goals or pursuits and growth in women's education, emancipation, and autonomy (Lesthaeghe, 2020). These attitudinal components of the SDT are less visible in the Japanese context. Women have experienced dramatic increases in educational attainment and female labor force participation rates are relatively high, but Japan remains one of the most gender-inegalitarian low-fertility countries (Brinton et al., 2018). The malefemale wage gap is among the largest in the OECD, the majority of women exit the labor force at least temporarily around the birth of their first child, and over half of female employees work in non-standard jobs. Furthermore, the division of household labor (domestic work and childrearing) is among the most unequal in the wealthy world. Along with the strong link between marriage and childbearing, limited evidence of individuation, descriptions of Japan's gender essentialist social organization (Brinton & Lee, 2016;Brinton & Oh, 2019) and growing economic uncertainty in the context of limited public support for families (Inaba, 2011;Tachibanaki, 2005), highlight the importance of considering the wide variety of social and economic pathways into and through the SDT (e.g., Lesthaeghe 2020; Lesthaeghe & Moors, 2000).
Earlier studies have stressed the importance of evaluating the generality of the SDT's combination of low fertility, family change, and attitudinal shifts beyond the Northwestern European countries where the framework originated (Lesthaeghe, 2010). This work has shown a great deal of heterogeneity among European countries in the pace and nature of change associated with the SDT (Lesthaeghe & Moors, 2000) and has evaluated the relevance of the SDT in societies such as the U.S. that differ markedly from those in Northwestern Europe (Lesthaeghe & Neidert, 2006). With this in mind, I provide a summary of research and recent data on trends in fertility, family behaviors, and attitudes associated with the SDT in the Japanese context.
The goals of this effort are two-fold. The first is to provide an empirical assessment of how the Japanese experience does and does not conform to the general patterns of behavioral and attitudinal change associated with the SDT in the West. The second is to provide a descriptive overview of demographic and attitudinal change for scholars interested in better understanding the social context of low fertility in Japan. 1 Throughout this paper, I assume readers' familiarity with the central tenets of the SDT framework and with related empirical research on low-fertility Western populations. 2 My descriptive overview of behavioral change draws heavily on recently published data from the Census and Vital Statistics as well as the summary report of the 15th National Fertility Survey (NFS) conducted in 2015. Data on attitudes and values come from the NFS, the Japan General Social Survey, and other recent surveys. I integrate these data with insights from demographic research published over the past two decades. I draw primarily on work published in English, but stress that there is a substantial body of excellent research on family change and attitudinal trends published in Japanese that may not be accessible to most readers.

Fertility
Japan's total fertility rate has been below replacement level longer than any other country. The TFR fell below 2.10 for the first time in 1957 and has remained below replacement level since 1974, reaching a historical low of 1.25 in 2005. Sobotka's (2008) indicator of the onset of SDT (i.e., the year in which mean age at first birth was two years higher than the post-WWII minimum) puts the beginning of the SDT in Japan in 1988, only a few years after Sweden, Finland, and the Netherlands. In 2020, Japan's TFR stood at 1.33, a level that is higher than in countries such as Korea (0.84), Spain (1.23), and Italy (1.27), but generally similar to the group of lowfertility countries typically included in discussions of the SDT. Japan's long decline in fertility and recent stabilization at very low levels reflect a combination of later age at marriage and childbearing, negligible non-marital childbearing, and growing proportions of women who never marry and thus never have children. 3 Women's mean age at first birth increased from 25.7 to 1975 to 30.3 in 2020 and trends in age-specific fertility rates show little evidence of recuperation at older 1 It is important to note that these goals are not novel. Similar questions have been addressed in previous research on low fertility and the SDT in Japan, including Atoh ( , 2002Atoh ( , 2008, Atoh et al., (2004), Brinton (2020), andOchiai (2014). This paper can thus be seen as an update and extension of these earlier studies. 2 Lesthaeghe (2010Lesthaeghe ( , 2020 offers comprehensive overviews of both theoretical and empirical research on the SDT. I also assume reader's familiarity with the historical demographic transition and efforts to understand the SDT in the context of earlier fertility decline (see e.g., Lesthaeghe & Neels 2002;van de Kaa 2004). Excellent treatment of the demographic transition in Japan can be found in Mosk (1977) and Taeuber (1960). 3 I focus primarily on trends in women's behavior for the sake of simplicity and convention. Corresponding data for men are, of course, similar. ages (National Institute for Population and Social Security Research, IPSS hereafter, 2022). As shown in Fig.1, cumulative fertility has declined across birth cohorts, with a particularly pronounced drop between the cohort of women age 15 in 1975 and their counterparts born 10 years later. There is no discernable evidence of fertility recuperation at older ages through the 1985 cohort. There is some evidence of recuperation for the 1995 cohort, but it is unlikely that completed fertility for these women will meaningfully differ from that of the 1985 cohort. Japan clearly belongs to the group of countries in which very low levels of fertility appear not to reflect period effects followed by some recuperation at older ages and thus a reversal in TFR decline (Lesthaeghe, 2010).
While trends in TFR arguably place Japan at the forefront of the SDT, extremely low levels of non-marital childbearing are at odds with the SDT's central focus on the decoupling of marriage and childbearing. In contrast to most European countries and English-speaking countries, where over 30% of children are now born to unmarried mothers, the corresponding figure in Japan has been stable at 1% or 2% since the mid-1950s. The various explanations offered for this distinctive feature of family formation in Japan include a history of legal discrimination against "illegitimate" children (especially with respect to inheritance), social stigmatization of unmarried mothers and bullying of their children, limited public income support for single-parent families, and the difficulty of balancing full-time employment and childrearing (Hertog, 2009;Hertog & Iwasawa, 2011).
Another relevant factor is widespread normative endorsement of children's need for both a mother and a father (Brinton & Oh, 2019;Hertog, 2008Hertog, , 2009. Like other East Asian countries, but in contrast to most low-fertility countries, a strong belief in the complementarity, but not the substitutability, of distinct maternal and paternal roles is arguably key to understanding why the link between marriage and childbearing remains so strong in Japan despite profound changes in other aspects of family behavior. Widespread, and firmly held, belief in the fundamental importance that children not be deprived of two parents is arguably part of the reason for the growing prevalence of marriages preceded by pregnancy with no accompanying growth in childbearing outside of marriage. While many premarital pregnancies (especially those of younger women) are aborted in Japan (Hertog & Iwasawa, 2011), pregnancy  is an increasingly important step in the marriage process (Fukuda, 2020;Raymo & Iwasawa, 2008). Recent research has shown that the percent of first marriages preceded by pregnancy doubled from 10% to 1980 to 19% in 2010, is over 30% for marriages under age 25 (Iwasawa & Kamata, 2014), and nearly 30% for women with a high school education or less (Mogi et al., 2020). Perceived need to marry in response to (often unplanned) pregnancy is interesting and somewhat puzzling in light of Japan's relatively high prevalence of single-parent families formed via divorce (discussed below).
Because marriage and childbearing remain closely linked, fertility decline is far less pronounced if we shift our focus from all women to currently married women. As shown in Fig.2, completed fertility was 2.22 for married women born in 1937-42 and has declined slowly, but steadily, to 1.86 for women born in 1965-70 (roughly equivalent to cumulative fertility for the 1985 cohort of 15-year-old women in Fig.1). The pace of decline in completed fertility picked up for women born after 1960, with decline across successive cohorts roughly 7% for the last two cohorts in Fig.2. Importantly, this recent decline in marital fertility is only partially explained by a rise in childless marriages. The proportion of women having their first child within the first four years of marriage has remained stable at around 80% over the past four decades while the prevalence of childless women among those married for 15-19 years increased from 3% to 1982 to 6% in 2015. Much more important for the decline in marital fertility is decline in the prevalence of women who have three or more children -from 32% to 1982 to 21% in 2015 (IPSS 2017a). It is also clear that declin-  ing fertility does not reflect substantial reductions in desired number of children. For example, among 18-34-year-old never-married female respondents to the National Fertility Surveys, the desired number of children declined from 2.29 to 1982 to 2.02 in 2015, figures that are almost identical to the expected number of children reported by married women in these surveys (IPSS 2017a). In the 2015 survey, only 7% of never-married women reported a desire to remain childless and only 8% wanted one child. The two-child norm is firmly entrenched in contemporary Japan.
These trends in marital fertility are important for thinking about the nature of the SDT in Japan. As noted above, the patterns described in Figs.1 and 2 are not consistent with a pattern of delayed marriage resulting in delayed fertility that is followed by recuperation at older ages. Rather, they are consistent with later childbearing and lower completed fertility reflecting delayed marriage, limited evidence of recuperation at older ages, a small decline in marital fertility, and especially with a decline in the proportion of women (and men) who ever marry. Taken as a whole, these data are consistent with a continued emphasis on having children, albeit in somewhat smaller numbers, conditional on marriage. Decomposition analyses have repeatedly demonstrated that changes in the age-specific proportion of women married accounts for the bulk of the observed decline in TFR (Atoh et al., 2004;Tsuya & Mason, 1995). In this context, it is essential to understand how and why patterns of marriage have changed.

Marriage
Three marriage trends are key to understanding the role of marriage in the SDT in Japan. The first is marked increase in the age of first marriage. The second is dramatic growth in the proportion who never marry. The third is the fact that long-term, marriage-like cohabiting unions have yet to emerge as an alternative to marriage. As discussed below, the prevalence of cohabitation has increased significantly (especially in recent years), but these unions are typically short-lived preludes to marriage and their growth should not be viewed as the emergence of long-term unions that serve as an alternative to marriage. Later and less marriage in Japan thus translates into later and less union formation to a much greater degree than it does in other lowfertility countries. This distinction is of critical importance in light of negligibly low levels of non-marital fertility because later and less marriage also translates into later and less childbearing.
Figure3 shows the age-specific proportion of women never married and women's mean age at first marriage, by year. Both sets of numbers clearly demonstrate the pronounced delay in first marriage. As late as 1980, over 90% of 30-34-year-old women had married and the mean age at first marriage was 25.2. In 2020, only 61% of 30-34-year-old women had married and the mean at first marriage was 29.4. In addition to the delay in first marriage and increasing variation in age at first marriage, it is clear that increasing proportions of women will never marry. In 1980, marriage was universal for women, with only 4.5% of 45-49-year-old women never married, but this figure increased to 19.2% in 2020 indicating that roughly one in five women in the 1970-74 birth cohort will never marry. The figures for men are even more extreme, with the percent never married at age 45-49 in 2020 reaching 30%. Data from the National Fertility Surveys show that small proportions of unmarried men and women are currently cohabiting or report having a partner (IPSS 2017a). For example, among never-married 35-39-year-old men and women in the 2015 survey, 71% reported not having a romantic partner. This evidence of decline not only in marriage, but also in union formation, clearly differs from conventional depictions of changing patterns of union formation associated with the SDT.
The close relationship between marriage and childbearing also means that an increase in lifelong singlehood will be accompanied by an increase in childlessness. Indeed, recent population projections (medium variant) are based on an assumption that the percent of women childless for cohorts born after 2000 will be 32% (IPSS 2017b). While never marrying and remaining childless could be seen as consistent with an SDT emphasis on individuation and the declining centrality of the family as an institution, empirical support for this pattern is relatively weak in Japan. While recent attitudinal data do offer some evidence of a decline in the attractiveness or salience of marriage as an institution (as discussed below), it is also clear that large majorities of young men and women wish to form families (via marriage).
Recent research describes a pattern of marriage postponement that can be characterized as a "drift into singlehood" and, by extension, into childlessness Yoshida 2016). In this process of drifting, economic uncertainty, work demands, and limited opportunities to meet potential partners appear far more relevant than a clear desire to avoid marriage and family formation. For example, among respondents to the 2015 National Fertility Survey who reported that they hoped to marry soon, insufficient financial resources were by far the most common perceived barrier to marriage for both men and women (IPSS 2017a: 18). Further, among 25-34-year-old unmarried men and women, roughly half reported that absence of a suitable partner was a primary reason for remaining single. Some evidence suggestive of decline in the attractiveness of marriage, or at least ambivalence toward marriage, can be found in relatively high proportions who report that they remain single because they don't yet feel the need for marriage or because they don't want to lose their freedom (IPSS 2017a: 19).
With respect to marriage (and, by extension, childbearing), trends in socioeconomic differentials are quite interesting for what they suggest about the nature of the SDT in Japan. Of particular importance is limited evidence that highly educated innovators have been at the forefront of a move away from marriage and family formation. Educational attainment has long been positively associated with marriage for men and recent studies have documented the relatively low marriage rates of low-educated men and those employed in non-standard work Piotrowski et al., 2015). It is true that highly educated women have long been less likely to marry (and thus more likely to remain childless), but it is not clear whether this has been a result of more resources and stronger desire to pursue career and other endeavors or greater difficulties in meeting and marrying similarly educated men (Raymo & Iwasawa, 2005. Importantly, the long-standing negative educational gradient in women's marriage has disappeared and perhaps reversed in recent years . These patterns presumably reflect stagnating wages, growing employment uncertainty (Raymo & Shibata, 2017), the high costs of childrearing, and limited public housing, income, and childcare support for young families. Taken as a whole, the evidence in Japan thus appears more generally consistent with marriage and family becoming a luxury (Park, 2022) than with an increasing avoidance of family in order to pursue individual goals.

Cohabitation
Many widely used sources of demographic data show that the prevalence of cohabiting unions has been much lower in Japan than in most low-fertility countries examined in research on the SDT. For example, recent data from the 15th National Fertility Survey (conducted in 2015) indicate that slightly less than 2% of unmarried men and women are currently in a cohabiting union while 7% of unmarried women report that they have ever cohabited. In conjunction with the low prevalence of romantic partners among unmarried men and women, this low cross-sectional prevalence of cohabitation suggests that, for many, later and less marriage means longer periods of singlehood (i.e., no romantic or sexual relationship).
Published research on cohabitation in Japan is quite limited, but the few papers available demonstrate that, not surprisingly, cohabitation experience is much higher among those who are married and has increased markedly in recent years. For example, Raymo et al., (2009) showed that about 20% of men and women born in the 1970s had ever cohabited, that most of these cohabiting unions were short-lived, and that about half resulted in marriage. This pattern is consistent with cohabitation as a prelude to marriage (Heuveline & Timberlake, 2004). Raymo et al., (2009) also showed that cohabitation is not a common setting for childbearing, but is a setting for premarital pregnancies that are "legitimated" via marriage.
Recent research demonstrates that cohabitation experience is increasingly common among recent marriage cohorts. Data from the 2010 and 2015 National Fertility Surveys indicate that as many as 39% of women who entered their first marriage in 2010-15 cohabited prior to marriage (Mogi et al., 2020). One innovation of Mogi et al.'s study is their expansion of the definition of cohabitation to include couples who lived together as married for some period of time prior to officially registering their marriage. One might argue that this is not actually cohabitation as conventionally defined or understood, but rather a period of unofficial or unregistered marriage. The counterargument is that, when cohabitation functions primarily as a prelude to marriage, a period of living together as married before official registration of marriage should be treated as pre-marital cohabitation. If this type of cohabitation is not included, the percent of couples who cohabited prior to marriage in the most recent marriage cohort is 33%, a figure that is still much higher than that observed in previous cohorts. A similar approach to measuring cohabitation finds that as many as 60% of women in recent marriage cohorts report ever cohabiting (Kojima, 2020). The much higher prevalence of cohabitation in this study may be due to selectivity of the internet monitor sample on which the analyses are based, but evidence that cohabiting unions tend to be short in duration is similar to the findings of Raymo et al., (2009) and Mogi et al., (2020). These data suggest that living together prior to marriage has lost much of its earlier stigma, but it is also clear that cohabitation remains closely linked to marriage and has not emerged as a relationship setting in which childbearing is common.
Another interesting finding from the study by Mogi and colleagues is evidence that the prevalence of cohabitation experience varies little by women's educational attainment, a pattern that resembles Raymo et al.'s, (2009) earlier findings, but differs from the strong negative educational gradient in cohabiting unions observed in many other countries (Lesthaeghe, 2020). However, the nature of cohabitation varies markedly in Japan. In particular, it is clear that cohabitation prior to or following pregnancy is increasingly more common among women with lower levels of educational attainment. Indeed, only 36% of the women with a high school degree who married in 2010-15 followed a conventional pathway to marriage, i.e., one without experience of either pregnancy or cohabitation (Mogi et al., 2020).

Divorce
Increases in divorce rates and stabilization at high levels are a hallmark feature of the SDT thought to reflect declining commitment to marriage as an institution. In Lesthaeghe's (2010: 215) words, "the onset of the rise in divorce was probably the very first manifestation of the accentuation of individual autonomy in opposing the moral order prescribed by Church and State." A similar claim might be made in Japan if "Church and State" is replaced with "widely shared and relatively rigid normative expectations regarding acceptable family behavior." With respect to divorce, Japan appears similar to other countries examined in the SDT literature. In the not so distant past, Japan was a universal marriage society (as shown in Fig.3) and divorce rates were low. Among ever-married 45-49-year-olds in the 1980 census, only 2.1% of men and 3.9% of women reported being currently divorced. These figures increased to 8.3% among men and 12.3% among women in the 2020 census. Because these figures do not include people who remarried after divorce and because roughly 20% of all marriages now involve at least one formerly married partner, it is important to also consider life-time probabilities of divorce. Earlier analyses of Vital Statistics data demonstrated that the percent of marriages ending in divorce within 12 years of marriage increased from 13% for the 1980 marriage cohort to 20% for the 1990 marriage cohort (Raymo et al., 2004). That same study showed that the synthetic cohort probability of marriage ending in divorce based on 2002 data was about one-third. Subsequent data indicate that the crude divorce rate peaked in 2002 and that lifetime dissolution probabilities are now closer to one-fourth (Iwasawa, 2019). Common explanations for divorce in Japan resemble those in the U.S. and other countries -communication problems and drifting apart, economic conflict, infidelity, and domestic violence (Alexy, 2020).
It is clear that divorce in Japan, as in the U.S. and many other countries, is strongly associated with educational attainment. Raymo et al., (2013) showed that the risk of divorce for university-educated women is about half that of women with a high school education. This differential is particularly important for our understanding of patterns of diverging destinies in the SDT (McLanahan, 2004) given the strong association between divorce and economic disadvantage. Divorce in Japan is distinctive not only for its ease (it only requires both spouses to sign a divorce registration), but also with respect to custody patterns. In 2020, 58% of all divorces involved at least one minor child and in 85% of those divorces, the mother received sole custody of the children (there is no joint custody in Japan). As a result, roughly one in ten families with children is now a single-mother family and the percent of single-parent families in poverty (48% in 2018) is among the highest among OECD countries (IPSS 2022;OECD 2022). Other recent work demonstrates that the negative educational gradient in divorce accounts for about one-fourth of the observed negative (maternal) educational gradient in economic disadvantage among families with at least one minor child (Raymo & Wang, 2021).

Maternal employment
The second demographic transition is closely intertwined with changing gender norms and attitudes (discussed in detail below) and increasing gender symmetry in family and social roles. Although not typically discussed together, theoretical articulations of the SDT and those of the "gender revolution" (England 2010;Goldscheider et al. 2015) have much in common. Of particular importance is waning endorsement of "separate spheres" and emphasis on both the intrinsic and strategic value of growing symmetry in men's and women's educational attainment and labor force participation. Importantly, the conflict between expanding opportunities for women's individual pursuits/self-actualization and limited change in gendered norms, expectations, and behavior within the family is central to theoretical explanations for very low fertility in familistic societies of Southern Europe and East Asia (McDonald, 2000(McDonald, , 2009. Although the female labor force participation rate in Japan is slightly higher than that in the U.S. (Shambaugh et al., 2017), this partially reflects the relatively high prevalence of unmarried (and thus unpartnered) women in Japan (Raymo and Fukuda 2016) and obscures the concentration of married women in insecure, low-wage, non-standard employment. Although 63% of married women below age 50 were in the labor force in 2015 (IPSS 2022), over half of these women are in non-standard employment (Raymo & Shibata, 2017) and married women's earnings account for about one-fourth of household income, on average (Nagase, 2018).
In contrast to many SDT countries characterized by a positive educational gradient in maternal employment, the relationship between mothers' educational attainment and employment is relatively weak in Japan (Raymo & Lim, 2011). One explanation for this pattern is the high prevalence of labor force exit around first birth which limits job opportunities for mothers seeking to reenter the workforce, regardless of their educational attainment or employment history. The percent of mothers leaving the labor force prior to the birth of their first child was 73% among those whose first birth was in 1985-89 and 58% for those whose first birth was in 2010-15 (IPSS 2017a: 52). Continuous employment is much more common among women who were employed in regular full-time jobs prior to childbirth, but the modal life course for mothers continues to involve at least some time out of the labor force, with implications for the quality of subsequent employment opportunities.
Several studies have linked this distinctive pattern of maternal labor force participation to Japan's highly unequal gender division of domestic labor (including childcare). Indeed, OECD data show that Japan has one of the most asymmetric divisions of domestic labor, a pattern that has been linked to both normative endorsement of intensive mothering and men's long work hours (e.g., Brinton & Oh 2019). Based on these trends and related attitudinal data (described below), it is debatable how far Japan has proceeded in the first half of the gender revolution (Esping-Andersen & Billari, 2015; Goldscheider et al. 2015), but it is clear that the second half (the emergence of egalitarian marriages) has yet to begin. Women's independence and empowerment central to the SDT may be emerging in Japan, but at a much slower pace than in most low-fertility societies.

Attitudes
Theorization of the SDT places fundamental emphasis on shifting attitudes and norms as a reflection of social and cultural change (Lesthaeghe, 2010(Lesthaeghe, , 2020van de Kaa 2001van de Kaa , 2004. In this section, I consider three different sets of attitudes central to the SDT framework. The first is attitudes toward family behaviors associated with the SDT. Theoretically, the increases in cohabitation, divorce, childlessness, and maternal employment documented above should be occurring in tandem with growing normative acceptance of those family behaviors (Rindfuss et al., 2004). That is, observed behavioral changes should be driven not only by economic forces or policy shifts, but also by attitudinal change (Liefbroer & Fokkema, 2008). The second is attitudes reflective of individuation, self-actualization, and declining institutional constraints on behavior. Attitudinal acceptance of SDT behaviors is thought to occur in a selfreinforcing interaction with increasing endorsement of the pursuit of individual life goals and the rejection of constraints associated with an institutionally structured life course. The third is attitudes related to gender roles and relationships. Increasing gender equality in higher education, the workplace, and the family are central to the social backdrop of the SDT, particularly as it is articulated with respect to lowfertility in Northern and Western Europe (e.g., Sobotka 2008). The trends in maternal employment described above provide some evidence consistent with growing gender symmetry in family roles, but other evidence suggests that these trends do not reflect fundamental change in the sharp gender division of labor that has long characterized Japanese families.
Two things should be clear from this section on attitudes in Japan. One is that recent trends are generally consistent with expectations of the SDT. The other is that attitudes consistent with expectations of the SDT are typically less prevalent in Japan than in many European countries (see Liefbroer & Fokkema 2008or Brzozowska 2021 for attitudinal data from other low-fertility countries). This may reflect a tension between forces of change and existing social structures and values posited in earlier research on family change in East Asia (e.g., Raymo et al., 2015). Because data availability necessitates focusing on change since the 1980s (or later), one thing that will not be clear from this description of recent attitudinal change in Japan is how dramatically different today's attitudes toward family, life, and gender are from those of the 1970s and earlier (see  for examination of attitudinal change over longer periods of time).

Family
Data on attitudes toward marriage and childbearing in Japan show both marked change and continued endorsement of conventional family organization and behavior. For example, earlier surveys demonstrated a notable increase in the percent of respondents who believe that men or women can lead fulfilling lives without marrying or without having children Choe et al., 2014). However, it is important to note that the percent strongly or somewhat accepting of these unconventional life paths was less than 40% in 2009 (Choe et al., 2014). It is also important to note that responses to simple attitudinal questions often obscure a good deal of contextual variation. One good example of this is responses to questions about marriage intentions/desires. Intentions to marry are nearly universal in response to a simple yesno question (IPSS 2017a), but there is significant variation in the strength of those intentions or desires, with large proportions desiring marriage conditional on timing, financial circumstances, and presence of a partner.  showed that, while about 40% of never married men and women report that they "definitely want to marry" and less than 3% "don't want to marry," the majority express more ambivalent views such as wanting to marry if possible, not caring if they marry, or not currently thinking about marriage. Evidence regarding perceptions of marriage as important or desirable thus depends not only on life circumstances (e.g., age), but also on how the question is asked.
Attitudes toward non-marital cohabitation are consistent with depictions of cohabiting unions in Japan as a precursor to marriage. For example, data from the National Fertility Surveys indicate strong support (and some evidence of increasing support) for the statement, "If a couple is going to live together, they should get married," with three-fourths of 18-34-year-old unmarried men and women agreeing strongly or somewhat in the 2015 survey. This figure suggests limited normative support for cohabitation as an alternative to marriage. Again, however, attitudes depend on how questions are asked, with a majority of respondents to a 1998 survey expressing neutral attitudes when presented with the negative statement "A young man and woman should not live together unless they are married" (Rindfuss et al., 2004).
Attitudes toward premarital sex have changed markedly in tandem with the trends toward later marriage and more sexual relationships prior to marriage. In 2015, only 11% of unmarried men and 15% of unmarried women disagreed with the statement "if the couple cares for each other, there is nothing wrong with premarital sex." Because the most effective types of contraception are not widely used in Japan (Sato & Iwasawa, 2006), these attitudes and associated behavioral change have contributed to growth in bridal pregnancies. This differs from patterns of change articulated in many depictions of the SDT that emphasize the role of effective contraception in decoupling sex and pregnancy from marriage.
Attitudes toward childbearing are consistent with the continued tight link between marriage and fertility. In 2015, two-thirds of 18-49-year-old married women endorsed the statement "if you marry, you should have children." Notably, this figure is much lower than in the 1992 survey when 88% strongly or somewhat agreed. Again, however, it is important to note that attitudes toward children depend upon how questions are worded. For example, the statement "a couple receives social recognition only after having a child" was endorsed by 25% of married women in 2018, down from 36% to 2008 (IPSS 2020). While far lower than support for the previous statement about the necessity of children within marriage, this also suggests relatively strong normative valuation of parenthood in contemporary Japan. Finally, it is clear that acceptance of non-marital childbearing is limited, with only 20% of respondents in 2000 and 2009 surveys accepting of this pathway to parenthood (Choe et al., 2014).
In an earlier study of change in attitudes between the early 1970s and early 1990s,  showed large declines in the percent opposed to divorce. However, changes in attitudes since the 1990s have actually been in the opposite direction.
In 2015, 70% of unmarried men and 60% of unmarried women agreed (strongly or somewhat) with the statement "couples should not divorce simply because they no longer get along," figures that are 8-10% points higher than in 1997. Taken as a whole, these attitudinal data suggest that endorsement of family changes associated with the SDT is weaker in Japan than in many other low-fertility societies. However, it is essential that we interpret these differences with an understanding that valuation of marriage and parenthood remains strong in all countries (Sobotka, 2008). Marriage is widely viewed as representing a higher level of commitment than cohabitation and is viewed as preferable setting for raising children.

Individuation
Often referred to as a collectivist society (Kitayama et al., 2009;Triandis, 2018), Japan seems an unlikely candidate for the growing emphasis on individual goals and self-actualization central to the SDT framework. Indeed, an earlier evaluation of attitudinal trends in the context of the SDT found limited evidence of growing individuation . In particular, Atoh found that endorsement of several statements about living for oneself remained stable at relatively low levels from the early 1970s to the early 1990s. Others have pointed to the seeming inconsistency between SDT theory and the apparent inverse relationship between individuation and very low-fertility, with fertility lowest in collectivist, familistic societies in Southern Europe and East Asia and higher in more individualistic Western-European and English-speaking countries (Brinton 2016).
As a society without the kind of historical connections between church and state emphasized by Lesthaeghe and other contributors to theoretical development of the SDT framework, Japan also seems an unlikely setting for a role of secularization in shaping family change. For example, data from the Japan General Social Survey show that only 10% of respondents said "yes" when asked "Is there a religion you believe in?" This figure has been stable since 2000. Among those who did respond "yes," only a small percentage responded that they were strong believers. 4 Although the wording of questions is different, these figures suggest even lower levels of religiosity than in the earlier data presented by . In this context, it is hard to measure secularization and even more difficult to speculate about the possible role of secularization in fertility decline and other family changes associated with the SDT. I therefore focus here on attitudes related to prioritization of individual goals/pursuits -what Lesthaeghe and others have called self-actualization.
To my knowledge, systematic analysis of attitudinal data salient to individuation is limited in Japan. One notable exception is a study by Lee & Fujita (2011) which provided some evidence of increasing individualism that appears to have slowed during the economic downturn that began in the early 1990s. Interestingly, however, their data show that only about 10-15% of respondents disagreed with the following statement: "When I do something, I want to respect harmony above my own opinion." In an examination of multiple measures from the World Values Survey, Hamamura (2012) concluded that collectivist attitudes remain strong in Japan. These findings suggest a meaningful role for social and normative forces limiting prioritization of individual goals, pursuits, and views.
Because previous research on individualistic attitudes is limited, I describe responses to several related questions in the JGSS in an effort to provide a general picture. In some rounds of the JGSS, respondents were asked how they felt about the following statement: "the family's well-being should be prioritized over individual well-being." In the 2017 and 2018 surveys, 50% of respondents agreed with the statement, about 40% chose the neutral option, and only 10% disagreed. Aother relevant question asked "It is safer to go along the with the opinion of the group even if your own opinion differs." In the 2017 and 2018 surveys, roughly one-third agreed, onethird chose the neutral response, and one-third disagreed.
Endorsement of the relative importance of the parent role (relative to one's role as an individual) is suggested by data showing that nearly half of married women agree that it is natural to expect that marriage and family require significant sacrifice of one's own lifestyle and individuality (this figure has actually increased from 35% to 1997) (IPSS 2017a). The 2018 National Family Survey provides further evidence that many Japanese couples perceive a tension between children and pursuing one's own interests, with 86% of respondents agreeing (strongly or somewhat) that couples should sacrifice their own interests while prioritizing their children (IPSS 2020). Taken as a whole, available data offer relatively weak evidence of growth in individuation.

Gender
As in other low-fertility countries, Japan has experienced large increases in women's educational attainment, employment, and life opportunities. It is important, however, to recognize that absolute and relative improvements in women's status and well-being have occurred in the context of marked gender inequality. Women's attendance at four-year universities is now approaching that of men, but Japan is one of the few countries in which women have yet to surpass men in university education. Women are also underrepresented at elite universities and in STEM fields (Uchikoshi, 2022). Similarly, the increase in women's employment largely reflects their growing representation in low-paying, part-time and other non-standard jobs (Nagamatsu, 2021). Women's low representation in managerial positions is one key reason why Japan ranks lower than any other wealthy country on the Global Gender Gap Index (World Economic Forum 2022). Gender inequality is also readily apparent in other measures like time spent on unpaid household labor and childrearing, with Japan having one of the largest gender gaps among OECD countries (OECD 2022). It is important that the trends in attitudes related to men's and women's roles in the family and society presented below be interpreted with this contextual backdrop in mind.
Attitudes regarding gender-based roles are clearly changing in Japan. For example, data from the National Fertility Surveys show a dramatic decline over time in agreement with the statement "husbands should work outside and wives should take care of the home." The percent of unmarried men agreeing (either strongly or somewhat) with this statement fell from over 60% in 199260% in to 31% in 201560% in (IPSS 2017a. For unmarried women, there has been less change over time, with about 30% agreeing with this statement over the past 20 years. Earlier research  also showed major declines in support for a clear gender division of labor between the early 1970s and early 2000s. While the direction and pace of attitudinal change is consistent with expectations of the SDT, the relatively high levels of continued support for "separate spheres" underlie descriptions of Japan as a "gender essentialist" society (Brinton et al., 2018;Brinton & Lee, 2016).
While support for a clear gender division of labor has waned, the extent to which this is true depends on the presence and age of children. Another question regularly asked in surveys is whether it is better for mothers to stay at home and not work for pay while children are young. Support for this view of mothers as the most appropriate caregivers for young children stood at 73% among married women in 2015 (down from 88% to 1992) (IPSS 2017a). In a comparative study of attitudinal data from the World Values Survey, Brinton & Lee (2016) showed that gender attitudes in Japan have become more egalitarian over time, but remain far more conservative than in most European countries. They describe gender attitudes in Japan as "prowork conservative," with clear attitudinal support for prioritization of men's employment when jobs are scarce and widespread belief that women want children and can derive as much fulfillment from being a housewife as from employment.
Responses of unmarried men and women to questions about ideal and expected life course trajectories also suggest changes in attitudes toward the gender division of labor among spouses. Among unmarried women, the percent responding that their ideal life course is full-time homemaker declined from 34% to 1987 to 18% in 2015 while the percent responding that they would ideally balance continuous employment with parenthood increased from 19 to 32% over the same period (IPSS 2017a). Interestingly, men's attitudes toward women's life course have changed even more rapidly. When asked about the life course that they hope their (future) wife will follow, the percent responding full-time homemaker declined from 38% to 1987 to 10% in 2015. It is, of course, possible that these data do not reflect fundamental change in gender attitudes as much as they do a growing recognition that breadwinner-homemaker marriages are no longer feasible for most couples.
Recently collected interview data reinforce this picture of tension between changes in some attitudes and continued normative adherence to a clear gender division of labor within marriage. For example, Nemoto and colleagues (2013) demonstrated that perceived social pressure to marry is not particularly strong among young men. They argued that low pressure to marry, in combination with limited change in perceptions regarding men's role as breadwinner, is particularly salient for increasing men's apprehension about marriage (or at least reducing their enthusiasm for marriage). Similarly, Brinton & Oh (2019) describe highly educated women's tacit acceptance of a highly unequal gender division of domestic work. Women may not like the highly gendered division of labor among spouses, but they appear to accept it as an unavoidable result of men's long work hours. It is not clear whether these findings should be viewed as inconsistent with expectations of the SDT or simply the result of workplace constraints on men's ability to participate more in domestic work.

Discussion
In this paper, I have sought to provide an empirical basis for assessing how the Japanese experience does or does not conform to the general patterns of behavioral and attitudinal change associated with the SDT in the West. Because the SDT is a powerful and purportedly universal framework for understanding the emergence (and persistence) of below-replacement fertility, there is great value in assessing its generality and understanding cross-national variation in the pace and nature of behavioral and attitudinal change. In Japan, evidence of behavioral change is mixed. Consistent with depictions of the SDT in Europe, Japan has experienced substantial delays in marriage and childbearing along with notable increases in non-marital cohabitation, divorce, and single-parent families formed via divorce. However, non-marital childbearing has remained at negligibly low levels and cohabiting unions have not emerged as an alternative to marriage. Attitudinal data present a similarly mixed picture, with support for conventional family patterns and gender roles declining, but remaining at higher levels than in most low-fertility Western countries typically considered in research on the SDT. Taken as a whole, these data describe a distinctive path to very low fertility in which universal forces of social and family change interact with strong normative expectations of two-parent families characterized by a pronounced gender division of labor.
It is impossible to discuss or understand low fertility and family change in Japan without attention to these two contextual features. Despite a wide array of family changes similar to those observed in other low-fertility societies, marriage remains essentially the only setting for childbearing. Relatively high levels of divorce in Japan suggest that the tight link between marriage and childbearing reflects ideals and aspirations more than rigid normative constraints regarding single parenthood. These ideals and aspirations appear to reflect fundamental beliefs about the distinctive roles and contributions of mothers and fathers and the importance of providing children with both (at least until the marriage is no longer sustainable). While maternal employment is increasing and desire for breadwinner-homemaker marriages has waned, the modal family continues to prioritize men's role as primary breadwinner and women's role as primary childcare provider and supplementary earner. Such families obviously exist in larger numbers in other low-fertility societies, but their prevalence is generally lower and normative endorsement is arguably weaker than in Japan.
This combination of gender essentialist views of the family and the close link between marriage and childbearing is not part of conventional articulations of the SDT. It is thus helpful to turn to the work of scholars who have noted and sought to understand the link between very low fertility and gender inegalitarian social structures and norms. Particularly notable is McDonald's (2000McDonald's ( , 2009 gender equity theory which posits the increasing salience of an "either-or" choice for women (and men) in societies where growth in women's opportunities outside of the family (i.e., education and employment) have outpaced changes in their normatively expected roles within the family. Recent extensions of these ideas have emphasized the role of gender essentialism in perpetuating the stickiness of gendered family roles and associated barriers to family formation in Japan (Brinton 2016;Brinton & Oh, 2019;Brinton et al., 2018). In this work, emphasis on the largely unquestioned social understanding of men's responsibility to be the primary breadwinner and women's responsibility for raising children provides a compelling basis for understanding distinctive features of low fertility in Japan: later and less marriage, the strong link between marriage and childbearing, and limited change in stated desires for marriage and children.
Scholars have long recognized that there are multiple pathways to very low fertility, and perhaps multiple versions of the SDT. These pathways include change driven primarily by economic uncertainty (Sobotka, 2008). This explanation is surely relevant for the changes in Japan described in this paper, especially to the extent that growing economic uncertainty and wage stagnation make it increasingly difficult to form and maintain families characterized by a highly gender-asymmetric division of labor. Alternative pathways also include demographic change in the context of deeply rooted social ideologies -Yu and Xie's emphasis in this special issue on the central importance of Confucianism for understanding family change in China is a good example. While these Confucian principles are also present to some degree in Japan, they are arguably of less importance than gender essentialist norms for understanding Japan's distinctive pattern of family change. In general, it is likely that features of an East Asian pattern of the SDT, including a strong link between marriage and childbearing and varying degrees of gender essentialist family organization reflect tension between rapid social and economic change and deep-rooted normative understandings of family structure, organization, and expectations (Raymo et al., 2015). It is essential to understand that, as in Northern and Western Europe, there is variation in the pace and nature of change associated with the SDT in East Asia reflecting variation in the historical, social, economic, and policy forces underlying observed change.
Ultimately, an evaluation of the SDT in Japan comes down to how one defines the SDT and the behavioral and attitudinal changes it includes. It is clear that the proto-typical pattern of family change in Northern and Western Europe has only partially emerged in Japan, with little evidence that the link between marriage and childbearing has weakened or that cohabitation has emerged as an alternative to marriage. The same is true of attitudes, with attitudinal shifts associated with the SDT occurring in Japan, but at a much slower pace than in most other low-fertility societies. On one hand, if the SDT is defined as the emergence of extended periods of below-replacement fertility in the context of growing heterogeneity in family forms, then Japan is unquestionably a forerunner. On the other hand, if the SDT is defined such that these family changes are driven primarily by attitudinal change -especially with respect to the meaning of marriage and gender egalitarian understanding of family roles -then it seems clear that Japan is a country in which only certain aspects of the SDT have emerged.

Funding and Competing Interests Not Applicable.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, 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 article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article'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. To view a copy of this licence, visit http://creativecommons.org/ licenses/by/4.0/.