Educational Assortative Mating, Bridal Pregnancy, and Childbearing in Contemporary Japan

Page created by Morris Fischer
 
CONTINUE READING
Educational Assortative Mating, Bridal Pregnancy, and Childbearing in Contemporary Japan

                                              Fumiya Uchikoshi
                                          (The University of Tokyo)

                                                 Ryohei Mogi
                    (Centre d'Estudis Demogràfics, Universitat Autònoma de Barcelona)

1. Introduction
This study examines the impact of educational assortative mating on fertility outcomes in contemporary
Japan. Previous studies argued that the relative improvement of women's better access to higher education
in developed nations accompanies an increase in the number of women having stable jobs and an increase
in homogamy and hypogamy marriage patterns. However, little is known about the consequence of
hypergamy (van Bavel 2012) on demographic outcomes. One exception is Nomes and van Bavel (2016),
which argued that an increase in hypogamy couples (wives have more education than husbands) is a main
determinant to explain the long-term decline in fertility in Belgium. These hypogamic couples are less
likely to have children because of relatively larger opportunity cost, differentials in marital timing, and
women’s preference to marry down.
           In the Japanese context, however, the impact of hypogamy on childbirth may have a different
mechanism. Uniqueness of Japan comparing other developed countries is the strong association between
marriage and fertility (Raymo, Musick, and Iwasawa 2015) and this feature may create the different
mechanism behind that relationship. Although a family structure and attitudes towards family values have
been transformed over the decades in many respects (e.g., a decrease in family size, increase in single person
household, or declining parent’s intervention towards marriage), the proportion of non-marital births is still a
quantitative minority (National Institute of Population and Social Security Research 2017). Hence,
childbirth mostly happened in marital union. However, recently the order of marriage and conception
becomes more flexible than in the past. The proportion of bridal pregnancy1 (Dekichatta kekkon), which is a
marriage preceded by pregnancy, among first births increased from 12.6% in 1980 to 27.9% in 2002, then it
has been stable (Ministry of Health, Labor, and Welfare 2000, 2010).
           Related with spouse pairing patterns, Raymo and Iwasawa (2008) argued bridal pregnancy
occurs among non-normative type of couples such as women’s educational hypogamy, since bridal
pregnancy is highly likely to be unintended (Raymo, Musick, and Iwasawa 2015). Moreover, as previous
studies suggested that marriage preceded by pregnancy is strongly associated with divorce risk and lower
quality of marital relationship (Teachman 2002), it is possible to assume that these unexpected and
non-normative marriage would not produce an additional childbirth. Therefore, according to those findings,

1
   We use bridal pregnancy as a term to describe a marriage preceded by pregnancy following Raymo and
Iwasawa (2008). While it sounds more common to use pre-marital pregnancy to describe it, pre-marital
pregnancy implies that it does not consider they get married and marriage timing. Therefore, using the
term of bridal pregnancy is more suitable in the Japanese context.
hypogamy in Japan have a positive association with the first childbirth, possibly mediated by its strong
linkage to the bridal pregnancy and a negative influence on an additional childbirth.
            Based on this research interest, this study examines the relationship between educational
assortative mating and fertility outcomes possibly mediated through marriage preceded by pregnancy in
Japan. In this study, Japanese General Social Survey 2009 Life Course Study (JGSS-2009LCS) is used.
One of the advantages of this survey is that it captures detailed information of respondent’s and spouse’s
demographic events and occupational careers in person-month level. Because of its exploratory nature of
the study, we attempt to estimate the hazards by parity (i.e. first, second, and third childbirth) using event
history model (Yamaguchi 1991). The total sample used for the analysis is 887 Japanese female aged 28 to
42 in 2008 who are currently married2.
            We explain the trend of educational structure and assortative mating in developed countries and
the unique trends in Japan in the following section. After defining our two hypotheses of the impact of
hypogamy onto childbirth by parity in the third section, data and method used are introduced. The results
are illustrated in the fifth section, then the conclusions, discussion, limitations and future research are
found in the final section.

2. Literature Review
2.1. Women’s better access to higher education and the end of hypergamy thesis
In most of developed countries TFR declined under the replacement level. It also has an aspect as a
process of reducing the differential fertility among the social classes. While in the past scholars have
examined the relationship between social mobility and differential fertility (Berent 1952; Duncan 1966),
the influence of women's educational attainment on fertility is paid an attention in recent years, possibly
because of the relative improvement of women's better access to higher education.
            However, since fertility consists of two partners, we can also focus on the influence of the male
partner’s social status (Corjin et al. 1996). For women, if marriage acts as an opportunity of social mobility,
it is decisive to consider the influence of not only female’s status but also male's educational attainment on
fertility (Huinink 1995). Moreover, besides of each partner’s socio-economic status, the combination
(assortative mating) can influence on the fertility (Dribe and Stanfors 2010). In the economic model of
division of labor, male candidates in the marriage market seek women's housekeeping role, and women
seek male partner’s earning potential (Becker 1991). Therefore, it is predicted that the fertility of a
women’s hypergamy couple (whose female status is lower than that of men) is the higher than the other
type of couples (Nomes and van Bavel 2016; Trimarchi and van Bavel 2017). Alternatively, as homogamy
couples are predicted to have similar values and are less likely to dissolve (Tzeng 1992), the fertility of
homogamous couple could be higher than heterogamy couples.

2  JGSS-2009LCS only asked their respondents about educational attainment of their current spouse. Since
this study examines the effect of educational assortative mating, we omitted divorced, widowed, and
remarried cases from the analysis.
Predicting a similar result through a different mechanism, scholars in demography have paid
more attention on increasing hypogamy couples in recent years. Due to women’s better access to higher
education over the decades, more women than men study at higher education in most of the developed
nations (DiPrete and Buchmann 2013; Esteve et al. 2016; van Bavel 2012). Esteve et al. (2016) presented
that the proportion of female college graduates age 25 to age 29 was higher than male in 139 countries in
2010. While the women’s hypergamy was the normative type of union in the past, this structural change in
higher education decrease the hypergamy couples (Esteve et al. 2016).
           Due to this changing in the assortative mating trend, hypogamy couples, which were
considered as non-normative couples, have decline relatively in its disadvantages (e.g. high risk of
divorce) in the United States (Schwartz and Han 2014). In European countries, on the other hand, the
number of female breadwinner has increased and hypogamy couples are predicted to be playing a new role
of division of labor between wives and husbands. Based on the increase in the number of hypogamy
couples in developed nations, the consequence of educational hypogamy has been intensively examined.
Among those literature, Nomes and van Bavel (2016) hypothesized that an increase in educational
hypogamy in Belgium contributed to the decline in fertility over the decades.
           They provided three causal mechanisms to hypothesize that hypogamic unions are negatively
associated with fertility. First, opportunity cost of childbearing could be considerably high in households
where women outearn their husbands. Secondly, there are differentials in marital timing between the lower
and higher educated. Therefore, women of hypogamic unions tend to postpone their fertility behaviors.
Third, women’s preference towards the number of children might be associated with expected spouse
pairings. Women who want to have more children might project to rely on husband’s earning potential,
while highly educated women who do not want to have children selectively choose partners whose status
are lower than themselves. Therefore, women of hypogamic unions are assumed to have fewer children
compared with wives of other types of assortative mating.

2.2. The Japanese context
2.2.1 Trends in educational assortative mating in Japan
Is the association of an increase in educational hypogamy and fertility applicable to other low fertility
countries in different contexts, such as East Asian nations? To broaden the perspective provided by Nomes
and van Bavel (2016), this study examines its applicability in the Japanese context.
           In Japan, the composition of educational assortative mating has changed differently from other
countries in which there has been an increase in educational homogamy (Blossfeld and Timm 2003; Mare
and Schwartz 2005). Japan has been characterized as a country where gender division both in public and
private spheres is one of the most rigid in the world. As a result of this gender asymmetric structure,
women are socially expected to quit the job after marriage or childbirth. Therefore, educational hypergamy
has been considered as normative (Raymo and Iwasawa, 2005).
           In terms of trend of educational assortative mating, there has been a decreasing trend in
educational homogamy (Fukuda, Yoda, and Mogi 2017; Miwa 2007). Raymo and Xie (2000), using the
10th National Fertility Survey in 1992, argued that the association between husband’s and wife’s education
decreased between the 1970-74 and 1988-92 marriage cohorts. Miwa (2007) also found that the level of
educational homogamy has decreased over the decades using different nationally representative datasets.
The latest study by Fukuda, Yoda, and Mogi (2017), using microdata from the Population Census, argued
a constant decline in educational homogamy in Japan from 1980 to 2010.
            In contrast to the interests in educational homogamy, the trend of educational hypogamy has
not been paid a scholar’s attention, but the proportion has gradually increased (Fukuda, Yoda, and Mogi
2017). A part of reason was its smaller share in educational assortative mating in Japan, where many
women are willing to marry up (hypergamy) even in the bad economic environment that has existed since
the 1990s (Raymo and Iwasawa 2005).

2.2.2 The relationship between educational assortative mating and fertility in Japan
Empirical study about the association educational assortative mating and fertility in Japan is limited. One
exception is Shirahase (2011), which argued that highly educated homogamic couples tend to have fewer
children to invest more in the child's education than lower educated couples, possibly because of its poor
public expenditure on education and thus private sectors (households) play a major role for education
expenditure among OECD countries.
            How does hypogamy influence on fertility in Japan? We propose two hypotheses drawn from
the same theoretical assumption: educational hypogamy couples more tend to be pregnant before marriage
than the other types of combination. Before introducing our two hypotheses, we explain this assumption.
Raymo and Iwasawa (2008), using four waves of National Fertility Survey in Japan, argued bridal
pregnancy occurs among couples of women’s educational hypogamy. This is because bridal pregnancy is
positively associated with unintendedness of childbearing (Raymo, Musick, and Iwasawa 2015). There are
two main reasons. Firstly, the effective contraception is not widespread (Konishi and Tamaki 2016). Use of
hormonal contraception (e.g. using pill) is authorized by the Ministry of Health, Labor and Welfare in Japan
in 1999 and most couples rely on using condom or withdrawal in sexual intercourse. Konishi and Tamaki
(2016), using a cross sectional survey among married and never-married women in Japan, argued that more
than half of never-married women do not use reliable contraceptive methods such as pill, even if they do not
have a current pregnancy intention. In addition, sexual activity at young age in Japan increased and has got
younger (Japanese Association for Sex Education 2013). Therefore, bridal pregnancy occurs unintendedly as
a result of sexual intercourse without an effective contraceptive use (Ohtani 1993).
            Secondly, the reason of marriage illustrates the feature of bridal pregnancy. Using the 14th
Japanese National Fertility Survey (2010), Iwasawa and Kamata (2014) compared reasons to get married
between bridal pregnant couples and the others. They presented 75.3 % of bridal pregnant couples decided to
get married because women conceived (Iwasawa and Kamata 2014), while only 4.0% of couples of other
first marriages responded to this category. This result indicates that it is strongly required to marry by the
time of first birth in the Japanese context.
While Nomes and van Bavel (2016) argued that educational hypogamy suppressed the level of
fertility, we predict that educational hypogamy is differently associated with the first and the second birth
due to the key feature of bridal pregnancy in Japan. As hypogamy couples are more likely to be bridal
pregnant, we hypothesize that educational hypogamy is positively associated with first childbirth.
           In contrast, we predict the relationship between hypogamy and the second childbirth would be
negative. There are two series of empirical studies that support this hypothesis. First, unintended pregnancy
decreases mother’s well-being (Institute of Medicine 1995), which is anticipated to lead to lower
childbearing desire. A theory of couple’s interdependence (Kelley 1979; Scanzoni 1979), which assumes
that couples are interconnected in terms of their behavioral and psychological exchange derived from
romantic relationship, predicts a pregnancy causes a conflict within a couple. If the pregnancy is unintended,
its negative impact on marital wellbeing will be strengthened, because the union is formed for protecting
their child from illegitimacy (Hertog 2009), rather than the relationship itself (Knab and Harknett 2006;
Surra, Chandler, and Asmussen 1987). Secondly, premarital pregnant couples tend to have higher
probability to divorce (Billy, Landale, and McLaughlin 1986; Knab and Harknett 2006; McCarthy and
Menken 1979; Surra, Chandler, and Asmussen 1987; Teachman 2002). Premarital pregnancy occurs among
non-normative type of couples such as women’s educational hypogamy (Raymo and Iwasawa 2008), which
is also positively associated with divorce (Tzeng 1992). Divorced women will face a considerable difficulty
to have an additional child unless they remarried a new partner in the Japanese context. Therefore, it is
highly likely that bridal pregnancy negatively influences on the second childbirth.

3. Hypothesis
On the one hand, if we follow the argument by Nomes and van Bavel (2016) that examine the impact of
educational assortative mating on childbirth in Belgium, we predict that, regardless of the order of
childbirth, hypogamy couples tend to have less likelihood to experience childbearing (hypothesis 1). On
the other hand, if we focus on the non-normative aspect of this spouse pairing, hypogamic unions have a
positive effect on fertility, but only for the first childbirth (hypothesis 2a). To add, educational hypogamy
is negatively associated with second childbirth and the relationship is partly explained by their propensity
to experience bridal pregnancy (hypothesis 2b).

4. Data and Methods
The data, JGSS2009-LCS we use is a retrospective data and contains marriage/divorce, childbirth, and
other life course history in person-month level. Hence, it fits our research interest because it allows us to
separate bridal pregnancy couples from other pregnancy couples. We created person-month data in which
the risk set limits the sample to female respondents and starts from (1) the time at their first marriage for
the first childbirth and (2) twelve months after the first childbirth for the second. Bridal pregnancy is
defined cases in which the first child was born within the first seven months from marriage. Respondent's
and spouse’s education were classified into junior high school, high school, vocational/junior colleges, and
university or more. Based on this classification, we created homogamy (husband’s education = wife’s
education), hypergamy (husband’s > wife’s), and hypogamy (husband’s < wife’s). Table 1 presents the
distribution of each variable at the beginning of observation. Duration longer than 180 months (=15 years)
from the beginning of observation was censored.

                                        Table 1. Descriptive statistics
                                                                           Second
                                                             First birth
                                                                            birth
                           Educational attainment
                                 Junior high school            1.13         1.15
                                 High school                   38.11        38.41
                                 Junior/two year colleges      42.39        43.48
                                 University and more           18.38        16.96
                           Spouse’s educational attainment
                                 Junior high school            2.93         3.34
                                 High school                   38.67        38.87
                                 Junior/two year colleges      18.15        17.88
                                 University and more           40.25        39.91
                           Age at marriage
                                 16-22                         14.54        14.42
                                 23-25                         31.34        31.83
                                 26-28                         33.26        32.76
                                 29-31                         14.09        14.42
                                 32-34                         4.51         4.27
                                 Over 35                       2.25         2.31
                           Leaving parental home               35.47        92.71
                           Educational assortative mating
                                 Homogamy (low)                22.77        22.84
                                 Hypergamy                     31.91        32.41
                                 Hypogamy                      20.29        20.42
                                 Homogamy (high)               25.03        24.34
                           Age at first birth
                                 16-22                                      9.11
                                 23-25                                      19.84
                                 26-28                                      32.06
                                 29-31                                      25.84
                                 32-34                                      8.77
                                 Over 35                                    4.38
                           Bridal pregnancy                                 26.87
                                          N                     887         867
                           Source: Authors’ calculation using JGSS2009-LCS data

5. Results
5.1 Descriptive results
We first estimated the descriptive survival rates of having first and second child depending on the types of
educational assortative mating. Since homogamy couples include both low and high educated, we separate
this type of pairing into low (junior high and high school) and high (vocational/junior colleges and
university) educated groups.
Figure 1 shows survival function of first birth among these four educational groups.
Homogamy couples of low educated are the most likely to have their first childbirth during the first half of
the survival time at month level and hypogamic couple follows. Homogamy of high educated are less
likely to have their child, but the gap in propensity to bear child gradually decreased and during the second
half of the survival months the propensity of high educated homogamy couples to bear child is almost the
same with other groups. Not surprisingly, these differentials in fertility timing occur because low educated
couples tend to marry earlier. Therefore, we need to examine whether educational hypogamic couples are
more likely to have their child compared with other groups by controlling demographic covariates.

                                                    Figure 1. Survival function of first birth
                      1.00

                      0.75
      Survival rate

                      0.50

                      0.25

                      0.00
                             0                 50                          100                          150                         200
                                                 Duration from time at marriage (month)

                                               Homogamy (high) 222                       Hypogamy 180
                                               Homogamy (low) 202                        Hypergamy 283

     Note: a log-rank test shows the difference of those survival functions between four groups is statistically significant (1% level).

                       Figure 2 shows survival functions of second childbirth among those groups. Compared with the
survival function of first birth in which more than 90% of the population experience their first birth, the
probability of couples to have their second child is slightly lower. It also demonstrates that the differentials
in birth risk between these four groups are not significant compared with the first birth sample although the
result of log rank test shows the difference of second birth risk between them is statistically significant at
10% level. In spite of these smaller difference, homogamy of low educated couples are more likely to have
their second child than other couples. However, a similar problem occurs as we face in the survival
function of first birth. In this case, it is highly likely that homogamy of low educated couples have their
first child earlier than other couples. Therefore, we need to control not only the timing of marriage but also
the timing of first child birth in estimating the propensity to have second child birth and its differentials
among four educational groups.
Figure 2. Survival function of second birth
                      1.00

                      0.75
      Survival rate

                      0.50

                      0.25

                      0.00
                             0                  50                          100                         150                         200
                                        Duration from 12 months after time at first birth (month)

                                                 Homogamy (high) 211)                    Hypogamy 177
                                                 Homogamy (low) 198                      Hypergamy 281

     Note: a log-rank test shows the difference of those survival functions between four groups is statistically significant (10% level).

5.2 The relationship between educational assortative mating and first childbirth
To capture the precise estimation of the effect of assortative mating on childbirth, Table 2 shows the
results of discrete time logit regressions which predict the propensity of each assortative mating to have
first childbirth. In Model 1, we examined the effect of educational attainment of each partner with other
covariates. Since in this study we pay a particular attention on educational hypogamy, the highest
attainment (i.e., university and more) is placed in a reference category. Compared with these university
graduates, other groups are more likely to have their first child. In particular, the positive association on
junior college graduates with first birth risk is statistically significant and high school graduates also
positively associated with first birth at marginal level (10%). As of spouse, high school graduates are more
likely to have their first child than university graduates and this is statistically significant at 5% level. The
relationship between junior college graduates and first birth risk is observed possibly because in Japan
male junior college graduates are a quantitative minority.
                       In Model 2, we added a series of educational coupling setting educational homogamy of high
educated as a reference category. In contrast to our hypothesis, the effect of educational hypogamy on first
birth risk is not significant, although the coefficient is positive. Other types of educational assortative
mating, homogamy of low educated and hypergamy couples, are neither associated with first birth risk at
statistically significant level.
Table 2. Results of the effect of educational assortative mating on first birth estimated by discrete time
                                                 logit model
                                                                         Model 1                 Model 2
                                                                      Coef.        S.E.       Coef.        S.E.
 Month at risk (ref: 1-6)
 7-12                                                               0.846***      (0.124)   0.846***      (0.124)
 13-24                                                              1.043***      (0.125)   1.043***      (0.125)
 25-36                                                              0.878***      (0.146)   0.879***      (0.146)
 37-60                                                              0.748***      (0.152)   0.750***      (0.152)
 61-96                                                               0.155        (0.213)    0.155        (0.214)
 over 97                                                            -0.538†       (0.315)    -0.541†      (0.315)
 Educational attainment (ref: university and more)
 Junior high school                                                  0.309        (0.406)    0.595        (0.585)
 High school                                                         0.203†       (0.113)    0.325        (0.255)
 Junior/two year colleges                                            0.244*       (0.103)    0.335†       (0.179)
 Spouse’s Educational attainment (ref: university and more)
 Junior high school                                                  0.140        (0.226)    -0.125       (0.466)
 High school                                                         0.207*       (0.089)    -0.034       (0.373)
 Junior/two year colleges                                            0.049        (0.103)    -0.029       (0.162)
 Age at marriage (ref: 29-31)
 16-22                                                               -0.017       (0.142)    -0.023       (0.142)
 23-25                                                               -0.024       (0.115)    -0.026       (0.116)
 26-28                                                               -0.018       (0.112)    -0.021       (0.112)
 32-34                                                               0.116        (0.194)    0.116        (0.194)
 Over 35                                                             0.174        (0.262)    0.180        (0.262)
 Leaving parental home                                               -0.015       (0.094)    -0.015       (0.095)
 Educational assortative mating (ref: homogamy (high))
 Homogamy (low)                                                                              0.126        (0.303)
 Hypergamy                                                                                   -0.101       (0.191)
 Hypogamy                                                                                    0.156        (0.283)
 Constant                                                          -4.121***      (0.154)   -4.120***     (0.156)
 Observations                                                             21580                   21580
 Log Likelihood                                                         -3482.951               -3482.718
 AIC                                                                    7003.901                7009.436
 Pseudo R square                                                          0.021                   0.021
 † p
second birth risk? Especially, are hypogamic couples, which have been regarded as non-normative, less
likely to have their second child?
             To answer this question, we examined the effect of educational attainment and its combination
on second birth risk in Table 3. In Model 1, the same variables shown in Table 2 except for the age at first
birth are used. In contrast to results of first birth, neither wife’s educational attainment nor spouse’s
educational attainment are not associated with childbirth. However, even if educational attainment as
baseline is not significantly related to childbirth, it is likely that its combination (assortative mating) might
affect the childbirth risk. In Model 2, we added three dummy variables to examine this point. The result
demonstrates that compared with homogamy of high educated, educational hypogamy is negatively
associated with second childbirth risk even after controlling covariates such as the age at marriage and the
age at first childbirth. The odds of hypogamic couples not having their second child are 2.33 times greater
than homogamy of high educated couples (1/exp(-0.824)=1/0.43=2.33).
             The negative association of educational hypogamy and second childbirth supports our
hypothesis. To elucidate a possible mechanism behind the relationship, we added a dummy variable that
will be 1 is a respondent experiences a bridal pregnancy and 0 otherwise. Although our hypothesis
assumes that the negative effect of educational hypogamy on second childbirth is explained by its
propensity to conceive her first child before marriage, the effect of bridal pregnancy is not significant and
positive. To add, the association between educational hypogamy and second childbirth is still marginally
significant and its negative effect remains.

 Table 3. Results of the effect of educational assortative mating on second childbirth estimated by discrete
                                                 time logit model
                                                   Model 1              Model 2                Model 3
                                                Coef.        S.E.    Coef.        S.E.      Coef.        S.E.
   Month at risk (ref: 1-6)
   7-12                                     1.386***     (0.193)    1.387***   (0.193)    1.387***    (0.193)
   13-24                                    1.672***     (0.183)    1.675***   (0.183)    1.675***    (0.183)
   25-36                                    1.541***     (0.193)    1.546***   (0.193)    1.546***    (0.193)
   37-60                                    1.158***     (0.200)    1.161***   (0.200)    1.161***    (0.200)
   61-96                                       0.112     (0.263)     0.123     (0.264)     0.124      (0.264)
   over 97                                     -1.051*   (0.443)    -1.015*    (0.444)    -1.012*     (0.444)
   Educational attainment
   (ref: university and more)
   Junior high school                          -0.172    (0.471)     -0.680    (0.641)     -0.677     (0.643)
   High school                                 -0.001    (0.138)     -0.189    (0.306)     -0.190     (0.306)
   Junior/two year colleges                    0.007     (0.125)     -0.118    (0.215)     -0.122     (0.215)
   Spouse's Educational attainment
    (ref: university and more
   Junior high school                          -0.202    (0.245)     0.733     (0.582)     0.725      (0.581)
   High school                                 0.084     (0.103)     0.946†    (0.496)     0.926†     (0.497)
   Junior/two year colleges                    -0.099    (0.125)     0.030     (0.185)     0.023      (0.185)
Age at marriage (ref: 29-31)
   16-22                                    -0.221       (0.215)    -0.233       (0.216)    -0.195       (0.225)
   23-25                                    -0.336*      (0.170)    -0.343*      (0.171)    -0.317†      (0.176)
   26-28                                    -0.135       (0.150)    -0.138       (0.150)    -0.124       (0.152)
   32-34                                    0.090        (0.304)    0.067        (0.304)    0.057        (0.305)
   Over 35                                  -0.158       (0.615)    -0.135       (0.613)    -0.158       (0.613)
   Leaving parental home                    0.211        (0.202)    0.247        (0.204)    0.244        (0.204)
   Age at first birth (ref: 16-22)
   23-25                                    -0.071       (0.184)    -0.051       (0.185)    -0.028       (0.189)
   26-28                                    -0.409*      (0.204)    -0.388†      (0.205)    -0.348       (0.215)
   29-31                                   -0.609**      (0.221)   -0.608**      (0.222)    -0.555*      (0.239)
   32-34                                   -0.815**      (0.273)   -0.816**      (0.274)    -0.750*      (0.295)
   Over 35                                 -1.514**      (0.508)   -1.522**      (0.507)   -1.447**      (0.521)
   Educational assortative matinge
    (ref: homogamy (high))
   Homogamy (low)                                                   -0.715       (0.439)    -0.702       (0.439)

   Hypergamy                                                        0.070        (0.224)    0.069        (0.224)

   Hypogamy                                                         -0.824*      (0.420)    -0.814†      (0.420)

   Bridal pregnancy                                                                         0.062        (0.103)

   Constant                               -4.561***      (0.348)   -4.554***     (0.349)   -4.618***     (0.365)

   Observations                                  29139                   29139                   29139

   Log Likelihood                              -2888.480               -2886.239               -2886.061

   AIC                                         5824.961                5826.478                5828.121

   Pseudo R square                               0.047                   0.048                   0.048

   † p
In Japan, while the increase in hypogamy couples are observed as well (Fukuda, Yoda, and
Mogi 2017), nuptiality and fertility change can be seen in the increase in bridal pregnancy. The aim of this
study was to investigate the effect of hypogamy onto fertility considering bridal pregnancy in Japan. Bridal
pregnancy is highly related to hypogamy (Raymo and Iwasawa 2008) and highly likely being unintended
(Iwasawa and Kamata 2014; Raymo, Musick, and Iwasawa 2015), which decreases the desire of childbirth
(Institute of Medicine 1995) and increases the divorce risk (Billy, Landale, and McLaughlin 1986; Knab and
Harknett 2006; McCarthy and Menken 1979; Surra, Chandler, and Asmussen 1987; Teachman 2002).
Therefore, our hypothesis is that educational hypogamy is positively associated with firstbirth, but
negatively associated with second childbirth.
            Although this is the tentative result, this study’s attempt to examine the relationship between
assortative mating and fertility outcome shows notable findings. First, hypogamic unions do not
necessarily have fewer children. In particular, no difference in first birth risk with statistical significance
was observed among four types of educational assortative mating. Second, however, educational
hypogamy is negatively associated with second childbirth compared with educational homogamy of high
educated. Third, in contrast to our prediction, bridal pregnancy is not associated with additional childbirth
and the effect of educational hypogamy on second birth remains even after controlling the effect of bridal
pregnancy. Therefore, these results partly support our hypothesis. Although the association between
educational hypogamy and firstbirth was not observed, we demonstrated the negative association of
educational hypogamy with secondbirth. However, bridal pregnancy does not explain this relationship.
Possibly, as we argue below, a part of educational hypogamy couples which were formed because of
unintended childbearing might have dissolved and was omitted from the sample of this study. If so, other
factors such as wife’s relatively higher earning potential than other types of couple affect lower probability
of childbearing. Alternatively, as Nomes and van Bavel (2016) argued, these non-normative couples are
deviated from other family norms, the two-child norm in this case.
            It should be noted that the abovementioned interpretation should be cautious due to a selection
problem. There are selection biases in bridal pregnancy and the second child birth. We only observe a
person who get pregnant before marriage and is married before childbirth, which means that we are not
able to observe a person who ends the pre-marital pregnant in abortion. Thus, the bridal pregnant sample in
this analysis is selected from the number of pre-marital pregnant. Secondly, we have to omit the divorced
sample due to the data limitation. JGSS-2009LCS did not ask the educational level of the first partner for
the divorced and widowed person. Therefore, it is not possible to include them as we are not able to create
assortative mating variable. As bridal pregnant couples have higher divorce risk (Billy, Landale, and
McLaughlin 1986; Knab and Harknett 2006; McCarthy and Menken 1979; Surra, Chandler, and Asmussen
1987; Teachman 2002), the results underestimate the negative impact of bridal pregnancy on second
childbirth at least. It is possible to anticipate that its effect could be larger than our results if we included
the information of divorced person. However, at the same time, this also indicates the robustness of our
results.
This study leaves an implication to disentangling the decline in TFR in Japan. While the main
factor to decline in TFR has been the changes of nuptiality behavior in Japan (Iwasawa 2002), recently the
decline in marital fertility started contributing to the decline in TFR (Hiroshima 2001; Kaneko 2004). Our
results present that hypogamy couples do not have statistically significant difference from the other type of
assortative mating couples, while they have lower probability to have second birth. This means that recent
decline in marital fertility may be caused by hypogamy couples. Similar to Belgium case (Nomes and van
Bavel 2016), the increase in hypogamy couples may contribute to the low fertility in Japan. In addition,
Japanese unique situation among developed countries (i.e., low non-marital childbirth rate and high bridal
pregnant proportion) may help to understand and generalize the new trend of assortative mating and its
consequences.

References
Becker, G. S. (1991). A Treatise on the Family. Massachusetts: Harvard University Press.
Berent, J. (1952). Fertility and social mobility. Population Studies 5(3): 244-260.
Billy, J. O. G., Landale, N. S., and McLaughlin, S. D. (1986). The effect of marital status at first birth on
     marital dissolution among adolescent mothers. Demography 23(3):329-349.
Blossfeld, H.-P. and A. Timm. (2012). Who Marries Whom? Springer Science & Business Media.
Corijn, M., Liefbroer, A. C., and de Jong Gierveld, J. (1966). It Takes Two to Tango, Doesn't It? The
     Influence of Couple Characteristics on the Timing of the Birth of the First Child. Journal of Marriage
     and Family 58(1): 117-126.
DiPrete, T. A. and Buchmann, C. (2013). The Rise of Women: The Growing Gender Gap in Education and
     What It Means for American Schools. New York: Russell Sage Foundation.
Dribe, M. and Stanfors, M. (2010). Family life in power couples.: Continued childbearing and union stability
     among the educational elite in Sweden, 1991–2005. Demographic Research 23(30): 847–878.
Duncan, O. D. (1966). Methodological Issues in the Analysis of Social Mobility. In: Smelser, N. J. and
     Lipset, S. M. (eds.). Social Structure and Mobility in Economic Development. Chicago: Aldine: 51-97.
Esteve, A, Schwartz, C. R., van Bavel, J., Permanyer, I., Klesment, M., and García-Román, J. (2016). The
     End of Hypergamy: Global Trends and Implications. Population and Development Review 42(4): 615–
     625.
Fukuda, S., Yoda, S., and Mogi, R. (2017). Three Decades of Educational Assortative Mating in Japan - A
     Micro-Data Analysis of Population Census 1980-2010. IPSS Working Paper Series (J) No.14. Tokyo:
     National Institute of Population and Social Security Research. (in Japanese)
Hertog, E. (2009). Tough choices: Bearing an illegitimate child in Japan. Stanford, CA: Stanford University
     Press.
Hiroshima, K. (2001). Shusshou ritu teika wo donoyou ni toraeruka? Nenreibetsu yuuhaiguu shussyouritu
     no monndaisei [How to Analyze Fertility Decline? : Drawback in Age-specific Marital Fertility Rate],
     Riron to Hoho [Theory and Methods], 16(2) 163-183. (in Japanese)
Huinink, J. (1995). Education, Work, and Family Patterns of Men: the Case of West Germany. In: Blossfeld,
     H. P. and Huinink, J. (eds.). The New Role of Women: Family Formation In Modern Societies. Boulder,
     CO: Westview Press: 247-262.
Institute of Medicine. (1995). The Best intentions: Unintended pregnancy and the well-being of children and
     families. Washington, DC: The National Academies Press.
Iwasawa, M. (2002). Kin'nen-no Kikan TFR hendo-ni okeru kekkonkodo oyobi fufu no shusshokodo no
     henka no kiyo-nitsuite [On the contribution of the changes in first marriage behaviors and couples:
     reproductive behaviors to the recent change in Japan’s total fertility rate]. Journal of Population
     Problems, 58(3) 14-44. (in Japanese)
Iwasawa, M. and Kamata, K. (2014). Marriage Preceded by Pregnancy and Women's Employment. Japan
    Labor Review 11(4):21–51.
Japanese Association for Sex Education. (2013). Sexual behavior of youth: The seventh survey of junior high
    school, high school, and university students in Japan. Tokyo: Japanese Association for Sex Education.
    (in Japanese)
Kaneko, R. (2004). Shoshika katei ni okeru fufu shusseiryoku teika to bankonka, kogakurekika, oyobi
      shussei kodo no sokutei [Decline in the fertility rate among married couples, the delay in marriage,
      the increased level of education, and the measurement of fertility behavior]." Journal of Population
      Problems 60: 37-54. (in Japanese)
Kelley, H. H. (1979). Personal relationships: Their structures and processes. Hillsdale, NJ: Erlbaum
    Associates.
Knab, J. and Harknett, K. (2006). ‘Shotgun’ marriages and relationship outcomes. Paper presented at the
      Annual Meeting of the American Sociological Association; Montreal, QC. Aug 10 2006.
Konishi, S. and Tamaki, E. (2016). Pregnancy intention and contraceptive use among married and unmarried
    women in Japan. Japanese Journal of Health and Human Ecology, 82(3), 110-124.
McCarthy, J. and Menken, J. (1979). Marriage, remarriage, marital disruption and age at first birth. Family
    Planning Perspectives, 11(1), 21-30.
Ministry of Health, Labor, and Welfare. (2000). Specified report of vital statistics 2000.
Ministry of Health, Labor, and Welfare. (2010). Specified report of vital statistics 2010.
Miwa, S. (2007). Long-Term Trends in Status Homogamy. In: Sato, Y. (ed.). Deciphering Stratification
    and Inequality. Trans Pacific Press: 140-160.
National Institute of Population and Social Security Research. (2017). Latest demographic statistics.
     Tokyo: National Institute of Population and Social Security Research. (in Japanese)
Nomes, E. and Van Bavel, J. (2016). Marital Fertility and Educational Assortative Mating Before, During,
    and After the Baby Boom in Belgium. Annales de démographie historique 132(2):139.
Otani, K. (1993). Gendai nihon shusshoryoku bunseki [Analysis of fertility in contemporary Japan]. Osaka:
    Kansaidaigaku Shuppansha. (in Japanese)
Raymo, J. M. and Iwasawa, M. (2005). Marriage Market Mismatches in Japan: an Alternative View of the
    Relationship Between Women's Education and Marriage. American Sociological Review 70(5):801–
    822.
Raymo, J. M. and Iwasawa, M. (2008). Bridal Pregnancy and Spouse Pairing Patterns in Japan. Journal of
    Marriage and Family 70(4):847–860.
Raymo, J. M., Musick, K., and Iwasawa, M. (2015). Gender equity, opportunity costs of parenthood, and
    educational differences in unintended first births: Insights from Japan. Population Research and
    Policy Review 34(2):179–199.
Raymo, J. M. and Xie, Y. (2000). Temporal and Regional Variation in the Strength of Educational
    Homogamy, American Sociological Review, 65(5), 773–781.
Scanzoni, J. (1979). Social exchange and behavioral interdependence. In: Burgess, R. L. and Huston, T. L.
     (eds.). Social exchange in developing relationships. New York: Academic Press: 61–98.
Schwartz, C. R. and Han, H. (2014). The Reversal of the Gender Gap in Education and Trends in Marital
    Dissolution.” American Sociological Review 79(4):605–629.
Schwartz, C. R. and R. D. Mare. (2005). “Trends in Educational Assortative Marriage From 1940 to 2003.”
    Demography 42(4):621–46.
Shirahase, S. (2011). Syoshika Syakai no Kaisou Kozou: Kaiso Ketsugo toshite no Kekkon ni Chakumoku
    shite (Stratification Structure in the Age of Declining Birth Rate: With a Focus on Marriage as
    Assortative Mating). In: Ishida, H., Kondo, H., and Nakao, K. (eds.). Gendai no Kaiso Syakai 2:
    Kaisou to Ido no Kouzou (Contemporary Stratified Society 2: Structure of Inequality and Mobility).
    Tokyo: University of Tokyo Press: 317-333. (in Japanese)
Surra, C. A., Chandler, M., and Asmussen, L. (1987). Effects of premarital pregnancy on the development of
    interdependence in relationships. Journal of Social and Clinical Psychology 5(1):123–139.
Teachman, J. D. (2002). Stability Across Cohorts in Divorce Risk Factors. Demography 39(2):331–351.
Trimarchi, A. and van Bavel, J. (2017). Educational Assortative Mating and Couples’ Fertility. Paper
    presented at the 2017 Annual Meeting of Population Association of America, April 27th, Chicago.
Tzeng, J. M. (1992). The Effects of socioeconomic heterogamy and changes on marital dissolution for first
    marriages. Journal of Marriage and Family 54(3):609–619.
Van Bavel, J. (2012). The Reversal of the Gender Gap in Education and Female Breadwinners in Europe.
    In: Lutz, W., James, K. S., Skirbekk, V., and Van Bavel, J (eds.). Vienna Yearbook of Population
    Research 10:127–154.
Yamaguchi, K. (1991). Event History Analysis. SAGE Publications.

Acknowledgement
The Japanese General Social Survey 2009 Life Course Study (JGSS-2009LCS) is designed and carried out
by the JGSS Research Center at Osaka University of Commerce (Joint Usage/Research Center for
Japanese General Social Surveys accredited by Ministry of Education, Culture, Sports, Science and
Technology).
You can also read