Marriage Market Mismatches in Japan An Alternative View of the .pdf

Document Sample
Marriage Market Mismatches in Japan An Alternative View of the .pdf Powered By Docstoc
					 Marriage Market Mismatches in Japan:
 An Alternative View of the Relationship
 between Women’s Education and Marriage
 James M. Raymo                                  Miho Iwasawa
 University of Wisconsin-Madison                 National Institute of Population
                                                 and Social Security Research, Tokyo


 In Japan, unlike in most other industrialized societies, the decline in marriage rates has
 been most pronounced among highly educated women. Theoretical interpretations of this
 distinctive pattern of change have typically emphasized increasing economic
 independence for women and reductions in the gains to marriage. In this paper, the
 authors develop and evaluate an alternative explanation that emphasizes women’s
 continued dependence on men’s economic resources and decline in the relative supply of
 highly educated men. Using data from four rounds of the Japanese National Fertility
 Survey, the authors decompose the observed decline in marriage rates into changes in
 the propensity to marry and changes in the educational composition of the marriage
 market. Results indicate that change in the availability of potential spouses accounts for
 one-fourth of the decline in marriage among university-educated women and explains a
 substantial proportion of the growing educational differences in marriage. The
                                         decline in by Ingenta to :
 conclusion is that the relatively large Deliveredmarriage among highly educated
 Japanese women likely reflects both Mr Beppu (cid 35001789), KYUSHU SHIKA
Kyushu University Central Library;increasing economic independence and continuedDAIGAKU (cid
     2606), Kyushu University
 economic dependence on men. Central Library (2); Zasshi (cid 52024142), Kyushu University
    Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
                                Kyushu University Library (cid 43009622)
                                        Tue, 24 Jan 2006 03:55:30
     he trend toward later marriage in industri-       Institute for Population and Social Security
 T   alized countries (e.g., Kiernan 2000; Raley
 2000) is particularly pronounced in Japan
                                                       Research, NIPSSR hereafter, 2005). Not only is
                                                       first marriage taking place at later ages, the
 (Raymo 2003; Retherford, Ogawa, and                   proportions of Japanese men and women who
 Matsukura 2001). With mean age at first mar-          never marry are also projected to increase sub-
 riage reaching 27.6 for women and 29.4 for            stantially (NIPSSR 2002; Retherford et al.
 men in 2003, Japan is currently one of the lat-       2001).
 est marrying societies in the world (National            Consistent with sociological and economic
                                                       theories of marriage that emphasize concepts of
                                                       specialization and exchange (e.g., Becker 1991;
    Direct all correspondence to James M. Raymo,       Parsons 1949), the trend toward later and less
 Department of Sociology, University of Wisconsin,     marriage in Japan is most pronounced among
 1180 Observatory Dr., Madison, WI 53706 (jray-        highly educated women (Raymo 2003). The
 mo@ssc.wisc.edu). This work was supported by a        standard theoretical explanation for this pat-
 Health and Labor Science Research Grant to the        tern of change suggests that the gains to mar-
 National Institute of Population and Social Security  riage derived from spousal pooling of
 Research. Earlier versions of this paper were pre-    complementary specializations are lower for
 sented at the annual meetings of the Population       Japanese women whose economic prospects
 Association of America and the Population
 Association of Japan. We would like to thank Myra
                                                       most resemble those of their potential mates.
 Marx Ferree, Jerry Jacobs, Valerie Oppenheimer,       Interestingly, however, this “economic inde-
 Zhenchao Qian, Megan Sweeney, Yu Xie, and four        pendence hypothesis” has received little empir-
 anonymous referees for their helpful comments on      ical support in most other countries. Recent
 previous drafts.                                      studies of the United States and other industri-

                      AMERICAN SOCIOLOGICAL REVIEW, 2005, VOL. 70 (October: 801 –822 )
 802—– AMERICAN SOCIOLOGICAL REVIEW


 alized societies have consistently found women’s   labor upon which neoclassical economic and
 educational attainment to be positively or         functionalist theories of marriage are predicat-
 insignif icantly related to marriage (e.g.,        ed (Oppenheimer 1997). This view of the “shift-
 Blossfeld and Huinink 1991; Bracher and            ing economic foundations of marriage”
 Santow 1998; Goldstein and Kenney 2001;            (Sweeney 2002) suggests that the association
 Sweeney 2002; Thornton, Axinn, and Teachman        between women’s educational attainment and
 1995). A similar pattern of cross-national vari-   marriage is most likely to be negative when rel-
 ation is observed with respect to other indica-    ative improvements in women’s economic
 tors of women’s economic resources such as         opportunities are not accompanied by conver-
 earnings and employment. Women’s economic          gence in men’s and women’s economic roles
 resources are positively or insignificantly asso-  within the family.
 ciated with marriage in the United States (e.g.,      This is an intuitively appealing interpreta-
 Sweeney 2002; Xie et al. 2003), Sweden (Ono        tion of the empirical evidence, but it is impor-
 2003), and Australia (Santow and Bracher           tant to recognize the limitations of the research
 1994), but negatively associated with marriage     upon which it is based. To date, most relevant
 in Japan (Ono 2003). Italy is one of the few other work has focused either explicitly or implicit-
 industrialized countries in which an inverse       ly on the relationship between women’s educa-
 relationship between women’s educational           tional attainment and the economic desirability
 attainment and marriage has been documented        of marriage, paying little or no attention to
 (Pinelli and De Rose 1995).                        potential changes in the demographic feasibil-
    Sociologists have attributed these cross-       ity of marriage. If, however, relative improve-
 national differences in the relationship between   ments in women’s educational attainment are not
 women’s education and marriage to gender con-      accompanied by convergence in the criteria that
                                                    men and :
 text, arguing that women’s economic inde-by Ingenta to women use to evaluate the education-
                                        Delivered
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
 pendence may be most relevant for                  al attainment of potential spouses, the demo-
     2606), Kyushu University Central Library
 understanding changing marriage behavior in(2); Zasshi (cid 52024142), Kyushu University to
                                                    graphic feasibility of marriage is expected
                                                    decline for highly educated 46009061),
    Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cidwomen. More
 societies where gender asymmetry in the divi-
                              Kyushu University Library (cid 43009622)
                                                    specifically,
 sion of domestic labor makes it difficult Jan 2006 03:55:30 if women’s desire to marry men of
                                       Tue, 24 for
 women to balance work and family (e.g.,            higher socioeconomic status than themselves
 Blossfeld 1995). For women in societies such       (female status hypergamy) remains strong, con-
 as Japan, where gender specialization remains      vergence in the educational attainment of men
 a basic feature of marriage (Tsuya et al. 2005;    and women will decrease the relative size of the
 Tsuya and Mason 1995), marriage typically          pool of “attractive” partners for highly educat-
 entails either a reduction in market employ-       ed women (and for less educated men). 1
 ment or a burdensome “second shift.” As            Because gender asymmetry in spouse pairing
 women’s educational attainment and associat-       patterns (i.e., female status hypergamy) is
 ed economic opportunities increase, both of        thought to reflect gender differences in work and
 these alternatives may become progressively        family roles, with men’s role as primary eco-
 less attractive.                                   nomic provider generating competition in the
    In most other industrialized societies, how-    marriage market for economically “attractive”
 ever, convergence in husbands’ and wives’ eco-     men (Kalmijn 1994), it is reasonable to expect
 nomic roles (Gershuny 2000) is thought to          that gender convergence in spouse pairing pref-
 increase the importance of women’s economic        erences is least likely to occur in societies where
 potential as a spouse selection criterion          gender differences in work and family roles
 (Sweeney and Cancian 2004), thus contributing      remain most pronounced.
 to an increasingly positive relationship between      This “marriage market mismatch hypothesis”
 women’s educational attainment and rates of        thus predicts a pattern of change identical to that
 marriage (e.g., Goldstein and Kenney 2001).
 Although women continue to perform the
 majority of domestic work (Shelton and John           1 We focus on women’s preferences for the sake of
 1996), the terms of marriage have shifted such     convenience. The same logic applies to men’s desire
 that American families are no longer charac-       to marry women with lower socioeconomic status
 terized by the gender-asymmetric division of       than themselves (male status hypogamy).
                                       EDUCATIONAL ATTAINMENT AND MARRIAGE IN JAPAN—– 803


 of the economic independence hypothesis—            approach allows us to decompose changes in
 relatively large decline in the marriage rates of   education-specific marriage rates into (a)
 highly educated women. The observed nega-           changes in the propensity to marry and (b)
 tive relationship between women’s educational       changes in marriage market composition.
 attainment and marriage in relatively gender        According to the economic independence
 inegalitarian societies such as Japan may there-    hypothesis, relatively large declines in the mar-
 fore reflect reductions in either the economic      riage rates of highly educated women are due
 gains to marriage or the availability of suitable   primarily to changes in their propensity to
 spouses, or some combination of both. Although      marry. In contrast, the marriage market mis-
 both theoretical scenarios are consistent with rel- match hypothesis suggests that growing edu-
 atively large decline in the rates of marriage for  cational differentials in marriage are largely
 highly educated women, they imply very dif-         due to changes in marriage market composition.
 ferent processes of social change. The basic        Before describing these analyses, we briefly
 theoretical premise underlying the economic         summarize recent trends in marriage and edu-
 independence hypothesis is that the gains accru-    cational attainment in Japan and further dis-
 ing to marriages based on gender specialization     cuss the relevance of the two alternative
 decline as women’s economic opportunities           theoretical explanations for understanding the
 improve relative to those of men. In this sce-      inverse relationship between women’s educa-
 nario, later marriage among highly educated         tional attainment and marriage.
 women reflects their active avoidance of mar-
 riages characterized by gender specialization.      WOMEN’S EDUCATIONAL
 The marriage market mismatch hypothesis, in         ATTAINMENT AND MARRIAGE
 contrast, suggests that relative improvements       IN JAPAN
                                        Delivered
 in women’s economic resources reduce the poolby Ingenta to :
                                                     TRENDS IN KYUSHU
Kyushu University Central Library; Mr Beppu (cid 35001789),MARRIAGE SHIKA DAIGAKU (cid
 of attractive partners for highly educated women
                                                     AND E UCA 52024142), Kyushu
 because strong preferences for female status (2); ZasshiD(cid TIONAL ATTAINMENT University
     2606), Kyushu University Central Library
    Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
 hypergamy (male status hypogamy) persist when       Age-specific first marriage rates presented in
                               Kyushu University Library (cid 43009622)
 economic roles within the family remain sharply 2006 03:55:30
                                        Tue, 24 Jan  Figure 1 clearly show the trend toward later
 differentiated by gender. Simply stated, grow-      marriage among Japanese women. At ages
 ing educational differences in women’s mar-         20–29, marriage rates in the year 2000 were less
 riage rates reflect increasing economic             than half of their 1970 values. Small increases
 independence from men in the first scenario         in marriage rates at ages 30–39 reflect the shift
 and continued economic dependence on hus-           in women’s mean age at first marriage from
 bands in the second scenario. Interpretation of     24.2 in 1970 to 27.0 in 2000. The very low mar-
 the observed negative relationship between          riage rates among teenagers have remained rel-
 women’s educational attainment and marriage         atively stable over time, whereas the low
 thus hinges on the ability to distinguish between   marriage rates of women in their forties appear
 these two theoretical scenarios. Unfortunately,     to have declined somewhat. Table 1 provides fur-
 however, previous studies have not produced         ther evidence of changes in marriage timing
 the empirical evidence needed to make this dis-     over the past 30 years by showing trends in the
 tinction.                                           proportions of 25–29 and 35–39 year-old
    We begin to fill this gap in the literature by   women who have never married. Between 1970
 examining the extent to which changes in the        and 2000, the proportion yet to marry tripled
 educational composition of the marriage mar-        from .18 to .54 at ages 25–29 and doubled from
 ket have contributed to later marriage among        .07 to .14 at ages 35–39. Education-specific
 highly educated women in Japan, one of the          figures show that the proportion never married
 most gender-inegalitarian industrialized coun-      at ages 25–29 increased by at least 25 percent-
 tries (Brinton 1988, 1989; Fuwa 2004). We do        age points for all educational groups over this
 this by employing the methodology used in           30-year period, with the largest increase (33
 Qian and Preston’s (1993) influential study of      percentage points) observed among university
 U.S. marriage behavior to analyze a unique          graduates. The proportion never married at ages
 source of data on the marriage behavior of men      35–39 increased by only a few percentage points
 and women born between 1942 and 1979. This          among women with at least a high school degree
 804—– AMERICAN SOCIOLOGICAL REVIEW


                      300


                      250


                      200
  Marriage Rate (‰)




                      150


                      100


                      50


                        0
                            15-19   20-24   25-29        30-34             35-39   40-44          45-49
                                                          Age
                                              1970    1980   1990    2000


                                           Delivered by Japanese Women, 1970–2000
 Figure 1. Trends in Age-Specific First Marriage Rates ofIngenta to :
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
     2606), Kyushu University Central Social Security Research. 2005. Latest Demographic Statistics.
 Source: National Institute of Population and Library (2); Zasshi (cid 52024142), Kyushu University
 Tokyo, Japan: author.
    Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
                                 Kyushu University Library (cid 43009622)
                                          Tue, 24 Jan 2006 03:55:30
 but more than doubled among those with less               in 1980 and 1995. These figures (not shown)
 than a high school education. By 2000, nearly             indicate that decline in marriage has been most
 20 percent of both the most highly educated               pronounced among less educated men, with the
 and the least educated 35–39 year-old women               proportion of 35–39 year-old men never mar-
 had yet to marry. Because census publications             ried increasing by 29 percentage points among
 do not provide tabulations of men’s marital sta-          junior high school graduates, by 13 percentage
 tus by age and education before 1990, it is not           points among high school and junior col-
 possible to examine trends in men’s marriage in           lege/vocational school graduates, and by 7 per-
 the same way. We can, however, use the survey             centage points among university graduates.
 data described below to estimate age-educa-                  Figure 2 depicts trends in the progression
 tion-specific proportions of men never married            from high school to higher education. Through


 Table 1. Education-Specific Proportions of Women Never Married at Ages 25–29 and 35–39 in the 1970 and
          2000 Censuses

                                                      Ages 25–29                           Ages 35–39
                                               1970                 2000           1970                 2000
 Totala                                         .18                 .54            .07                  .14
 Junior High School                             .15                 .42            .07                  .19
 High School                                    .20                 .45            .08                  .12
 Junior College/Vocational School               .29                 .57            .11                  .14
 University                                     .36                 .69            .15                  .18
 Source: 1970 and 2000 Population Census of Japan (Statistics Bureau—Management and Coordination Agency).
 a Total includes those still in school.
                                             EDUCATIONAL ATTAINMENT AND MARRIAGE IN JAPAN—– 805


           50

           45

           40

           35
                                                                                                    32
           30                                                                   15
 Percent




                                                             12
           25                                                                                48
                                                     39
           20
                                                                          33
           15                   27
                                        7
                                                             21                 22
           10
                14                                                                                  17
            5             3             11
                1         3     2                    2                    2                   2
            0
                M
                M        FF    M
                               M      F              M
                                                     M     FF           MM       FF          M
                                                                                             M      FF
                     1960        1970                 1980               1990                  2000
                                     Junior College/Vocational School   University
                                           Delivered by Ingenta to :
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
 Figure 2. Trends in Progression from High School to Higher Education, 1960–2000 Kyushu University
     2606), Kyushu University Central Library (2); Zasshi (cid 52024142),
    Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid Statistics.
 Source: National Institute of Population and Social Security Research. 2005. Latest Demographic 46009061),
 Tokyo: author. Note: M = male, Kyushu University Library (cid 43009622)
                                 F = female.
                                           Tue, 24 Jan 2006 03:55:30

 1980, women were less likely than men to                 versities increasing nearly three-fold between
 continue their education beyond high school              1980 and 2000. Perhaps reflecting declining
 and the nature of higher education differed              family size, improved occupational opportu-
 sharply by gender. Throughout the forty-year
                                                          nities for women, and increasing parental
 period, nearly all men entering post-second-
 ary education attended four-year universi-               aspirations for daughters’ education, the pro-
 ties, whereas the majority of women, until               portion of female high school graduates enter-
 very recently, attended two-year junior col-             ing four-year universities reached 32 percent
 leges. With curricula emphasizing education,             in 2000. Because university education is asso-
 home economics, literature, music, and arts,             ciated with substantially higher earnings
 junior colleges have been known less for seri-           potential across women’s life course in Japan
 ous academic or vocational training than for
 producing “good wives and wise mothers”                  (Ogawa and Clark 1995; Raymo 2003), the
 (Brinton 1988, 1993). These gender differ-               educational trends depicted in Figure 2 sug-
 ences in the function of higher education                gest an increase in women’s economic inde-
 reflect the fact that university education has           pendence. Although per vasive gender
 typically not provided women with access to              discrimination in the labor market continues
 better jobs (Brinton and Lee 2001) and has               to limit women’s occupational opportunities
 been viewed as a potential handicap in the
                                                          (Brinton and Lee 2001), passage of the Equal
 marriage market (Brinton 1993).
    This two-tiered system of higher educa-               Employment Opportunity Law in 1986 also
 tion is changing rapidly, however, with the              suggests growing opportunities for econom-
 proportion of women entering four-year uni-              ic independence among Japanese women.
 806—– AMERICAN SOCIOLOGICAL REVIEW


 THE ECONOMIC INDEPENDENCE HYPOTHESIS                   THE MARRIAGE MARKET MISMATCH
                                                        HYPOTHESIS
 While other industrialized countries have expe-
 rienced similar changes in marriage and the          It is not difficult, however, to describe a very dif-
 educational and occupational opportunities of        ferent theoretical scenario linking these same
 women, Japan is one of the few societies in          contextual characteristics to the observed neg-
 which the relationship between women’s edu-          ative relationship between women’s education-
 cational attainment and marriage is negative.        al attainment and marriage. Although increasing
 Theoretical interpretations of this distinctive      economic independence surely allows some
 relationship emphasize gender-asymmetry in           highly educated women to avoid marriages char-
 the division of domestic labor, arguing that         acterized by gender specialization, it is impor-
 women’s economic resources reduce the bene-          tant to recognize that career interruptions and
 fits (either perceived or real) of marriage when     part-time employment increase economic
 it is difficult for women to balance work and        dependence on husbands for the large majori-
 family (Blossfeld 1995; Ono 2003; Raymo              ty of women who do marry (Hobson 1990;
 2003). This explanation is intuitively appealing     Sørensen and McLanahan 1987). Assuming a
 in the Japanese context, where married men           desire to maintain or improve socioeconomic
 spend significantly less time on domestic labor      status via marriage (Yamada 1996), Japanese
 than their counterparts in most other industri-      women thus have a strong incentive to marry
 alized countries (Kamo 1994; Tsuya and Mason         men of similar or higher status than themselves.
 1995). In 1994, for example, Japanese husbands       Conversely, the difficulty that women face in
 spent an average of 2.5 hours per week on            combining career and family provides men with
 housework—less than 10 percent of total house-       little incentive to emphasize the earnings capac-
                                                      ity of
 work hours (Tsuya et al. 2005). The corre-by Ingentapotential wives. These gender-asymmet-
                                        Delivered              to :
                                                      ric spouse-selection criteria, in combination
 sponding f igure in 2000 was Mr hours,
Kyushu University Central Library; 2.8 Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
      2606), that the highly asymmetric division      with rapid improvements in women’s educa-
 suggestingKyushu University Central Library (2); Zasshi (cid 52024142), Kyushu University
                                                      tional attainment, Library (cid 46009061),
     domestic labor characterizing Japanese Kyushu University Medical have potentially important
 of Ropponmatsu Library (cid 46003719), mar-
                                                      implications for marriage behavior.
                              Kyushu University Library (cid 43009622)
 riages has changed very little in recent years.
                                       Tue, 24 Jan 2006 03:55:30 baby boom (bust) will result in a
                                                          Just as a
    This asymmetric division of domestic labor
                                                      female (male) marriage squeeze if there is a
 limits women’s ability to remain in full-time
                                                      persistent tendency for women to marry men
 employment after marriage, with only 18 per-
                                                      older than themselves (e.g., Akers 1967), con-
 cent of women with children under the age of
                                                      vergence in men’s and women’s socioeconom-
 three in the labor force in 1995 (Shirahase
                                                      ic status will, all else being equal, result in
 2003). Because continuous job tenure and work
                                                      marriage market mismatches for higher-status
 experience are heavily rewarded in the labor
                                                      women and lower-status men if there is a per-
 market (e.g., Brinton and Ngo 1993), and             sistent tendency for women to marry up (hyper-
 because opportunities to reenter the labor force     gamously) and men to mar ry down
 are largely confined to low-paying jobs in small     (hypogamously) with respect to characteristics
 companies or part-time work (Tsuya et al.            such as education and occupation (Lichter,
 2005), career interruptions associated with mar-     Anderson, and Hayward 1995; Lichter et al.
 riage and childbearing have profound implica-        1992; Wilson and Neckerman 1987).2 Although
 tions for women’s earnings potential in Japan.       “marriage squeeze” and “marriage market mis-
 For married women who do return to the labor         match” refer to the same phenomenon, the term
 force, tax policies also provide a strong incen-     “marriage squeeze” is typically used in discus-
 tive to work part-time and limit earnings in         sion of changes in the age composition of the
 order to qualify for dependent status (Ogawa and
 Ermisch 1996). In this context, the economic
 independence hypothesis suggests that the dif-          2 Here, we are referring to the tendency for women
 ficulty of remaining in full-time employment         to marry men of higher status, net of marginal dis-
 should lead highly educated women to post-           tributions. That is, women are more likely to marry
 pone or avoid entering “the onerous status of the    up with respect to indicators of socioeconomic sta-
 Japanese wife and mother” (Tsuya and Mason           tus, net of their lower levels of socioeconomic status
 1995:156).                                           relative to men.
                                         EDUCATIONAL ATTAINMENT AND MARRIAGE IN JAPAN—– 807


 marriage market. Research on age-based mar-           ment, both junior-college- and university-edu-
 riage squeezes has shown that fluctuations in         cated women are most likely to marry men with
 fertility during the second half of the twentieth     a university degree (Suzuki 1991). The tenden-
 century have actually had relatively little impact    cy for university graduates to marry homo-
 on the availability of potential mates (Schoen        gamously not only reflects the role of education
 and Baj 1985) and that the demographic feasi-         as a signal of men’s earnings potential but also
 bility of marriage is more strongly impacted          points to the effectiveness of universities as
 by excess male mortality associated with wars         marriage markets (Kalmijn 1991) and the
 (see Anzo 1985 on Japan) and sex-selective            importance of family background as a spouse-
 migration (Goodkind 1997).                            selection criterion (Brinton 1993). To the extent
    There are several reasons to treat this focus      that women’s educational attainment is posi-
 on marriage market mismatches as a compelling         tively correlated with family class background,
 theoretical alternative to the economic inde-         highly educated men will tend to marry simi-
 pendence explanation for the inverse relation-        larly educated women even if women’s educa-
 ship between women’s educational attainment           tional attainment itself is not highly valued in
 and the transition to marriage in Japan. For          the marriage market. Although the association
 example, attitudinal survey data highlight large      between husbands’ and wives’ educational
 gender differences in the criteria by which           attainment appears to have weakened over time
 potential spouses are evaluated. Among unmar-         in Japan (Raymo and Xie 2000), it is not clear
 ried respondents to the 11th National Fertility       that such changes have been large enough to off-
 Survey (conducted in 1997), women were twice          set the rapid convergence in men’s and women’s
 as likely (50 percent) as men (24 percent) to cite    educational attainment.
 educational attainment as either “important” or          The impact of relative improvements in
                                         Delivered     women’s :
 “a consideration” in choosing a spouse. Genderby Ingenta to educational attainment on marriage
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
 asymmetry in the importance of “earnings              market composition is clear from the educa-
      2606), Kyushu University with 91 percent         Zasshi (cid 52024142), ages 25–29 and 35–39
 potential” is even more striking,Central Library (2); tion-specific sex ratios atKyushu University
                                                       presented Medical Library (cid 46009061),
    Ropponmatsu Library (cid 46003719), Kyushu Universityin Table 2. Over the past thirty years,
 of women but only 31 percent of men, citing this Library (cid 43009622)
                                Kyushu University
                                                       the ratio of
 as a spouse selection criterion (NIPSSR 1999). 2006 03:55:30 men to women increased at the
                                         Tue, 24 Jan
 More detailed measures of the criteria by which       lower end of the educational spectrum for both
 women evaluate potential spouses are not avail-       age groups, with men in the lowest category
 able from scientific surveys, but informal sur-       (junior high school) outnumbering similarly
 veys and anecdotal evidence highlight the strong      educated women by 2000. A similar pattern
 emphasis on men’s earnings capacity. In the           holds for 25–29 year-old high school gradu-
 late 1980s, for example, the mass media popu-         ates, with 109 men for every 100 women in
 larized an explanation for delayed marriage that      2000. Over the same period, the representation
 emphasized women’s desire for marriage to a           of highly educated women increased dramati-
 man with “three highs”—a salary of more than          cally. While 25–29 year-old male university
 10 million yen (roughly $80,000), graduation          graduates outnumbered their female counter-
 from a prestigious university, and over 5’7” tall     parts by nearly five to one in 1970, this ratio was
 (Yamada 1996).                                        less than two to one in 2000. Decline in the sex
    It is therefore not surprising that analyses of    ratio for university graduates is even more pro-
 assortative mating have shown educational             nounced at ages 35–39, with men outnumber-
 attainment to be the most important socioeco-         ing women by more than eight to one in 1970
 nomic dimension of spouse selection in Japan.3        and by less than three to one in 2000.
 Net of gender differences in educational attain-         Among junior college/vocational school
                                                       graduates, women outnumbered men by a sub-
                                                       stantial margin at both points in time. Because,
    3 Previous analyses using the same data that we    however, a large proportion of female junior
 examine in this paper have shown the association      college graduates marry university graduates
 between spouses’ educational attainment to be         (Suzuki 1991), the marriage-market opportu-
 stronger than the association between spouses’ pre-   nities of highly educated women are more
 marital occupations or between the occupations of the appropriately measured by sex-ratios that com-
 husband and the wife’s father (Raymo 2000).           bine both types of post-secondary education.
 808 —– AMERICAN SOCIOLOGICAL REVIEW

 Table 2. Education-Specific Sex Ratios at Ages 25–29 and 35–39 in the 1970 and 2000 Censuses

                                                     Ages 25–29                           Ages 35–39
                                              1970                2000             1970                2000
 Totala                                         99                103              101                 102
 Junior High School                             91                145               93                 148
 High School                                    84                109               84                  92
 Junior College/Vocational School               42                 41               76                  32
 University                                    487                186              859                 277
 All Postsecondary education                   198                 88              321                 106
 Source: 1970 and 2000 Population Census of Japan (Statistics Bureau—Management and Coordination Agency).
 a Total includes those still in school.




 This combined measure, presented in the last           individual-level survey data are essential for
 row, shows that there were two 25–29 year-old          our analyses because the necessary informa-
 men for every similarly aged woman with high-          tion is not available from any other source.
 er education in 1970 but women actually out-           Vital statistics data provide cross-tabulations
 numbered men by 2000. In other words, for              of spouses’ ages for all marriages occurring in
 each 25–29 year-old woman with post-second-            a given year, but they do not contain infor-
 ary education, there was one fewer man of sim-         mation on brides’ and grooms’ educational
 ilar age and education in 2000 than there was          attainment. The most likely source of data for
 in 1970. For 35–39 year-old women with post-           defining the population at risk of marriage by
 secondary education, the ratio of over three           educational attainment, the national census,
                                         Delivered      does to :
 men for every woman of similar age and edu-by Ingentanot provide age-specific tabulations of
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
                                                        marital status and educational University
 cation in 1970 declined to a ratio of roughly one-(2); Zasshi (cid 52024142), Kyushu attainment for
     2606), Kyushu University Central Library
 to-one by 2000. The marriage market mismatch           men prior to 1990.
    Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
 argument suggests that this markedUniversity Library (cid 43009622) appropriate data from the
                              Kyushu decline in            In the absence of
                                                        census or
 the relative supply of economically “attractive” 2006 03:55:30administrative records, we use the
                                        Tue, 24 Jan
 men has made it more difficult for highly edu-         large sample of pooled JNFS data to recon-
 cated women to locate a suitably educated              struct the population at risk of marriage, clas-
 partner.                                               sified by educational attainment and five-year
                                                        age group, for each year between 1957 and
 DATA                                                   1997. We accomplish this in three steps. First,
                                                        we generate observations for married men based
 To provide the first rigorous empirical evalu-         on the responses provided by married women
 ation of the marriage market mismatch hypoth-          about their husbands’ educational attainment,
 esis, we use pooled data from the Japanese             birth date, and date of marriage. Second, we
 National Fertility Surveys (JNFS) conducted            merge the data for the married and unmarried
 in 1982, 1987, 1992, and 1997. These surveys           samples of each sex. Third, we expand these
 provide information on age, educational attain-        merged data into person-year record form, cre-
 ment, and age at marriage for nationally rep-          ating one observation for each year that a
 resentative samples of married women and               respondent was between the ages of 15 and 49
 unmarried men and women age 18–34 in 1982              during the period 1957–1997.5 Using these per-
 and 1987 and age 18–49 in 1992 and 1997.               son-year data, we examine change over time
 Dropping observations with missing data and            by limiting our focus to two five-year periods,
 pooling data from the four surveys results in
 a total sample of 24,860 married women,
 13,749 unmarried women, and 16,181 unmar-
                                                        7,884 in 1997. Sample sizes for surveys of unmar-
 ried men born between 1942 and 1979.4 These
                                                        ried men and women are 4,958 in 1982, 6,038 in
                                                        1987, 9,576 in 1992, and 9,358 in 1997.
                                                           5 The oldest respondents (born in 1942) were 15
   4 Sample sizes for surveys of married women are      years old in 1957 and 1997 is the year of the most
 3,806 in 1982, 3,506 in 1987, 9,664 in 1992, and       recent survey.
                                          EDUCATIONAL ATTAINMENT AND MARRIAGE IN JAPAN—– 809


 1978–82 and 1993–97.6 Several considerations           tion of a national marriage market. This is obvi-
 motivate the choice of these two time periods. A       ously not a realistic portrayal of marriage markets
 practical consideration is the need to aggregate       in Japan (or elsewhere), and it is possible that
 data over a period of years in order to limit the      results would differ if we were able to calculate
 number of empty cells for uncommon pairings. A         marriage rates at a lower level of regional aggre-
 methodological consideration is that, by using         gation. Unfortunately, however, there is no avail-
 periods whose midpoints correspond to census           able information that would allow us to evaluate
 years, we can compare marriage rates construct-        the potential impact of this assumption.
 ed from these survey data with marriage rates             We assess the extent to which changes in mar-
 based on aggregate data from the vital statistics      riage market composition and spouse-pairing pat-
 and the census. A substantive consideration is         terns have contributed to declining marriage rates
 that the most dramatic changes in marriage tim-        by analyzing these data in two different ways.
 ing have occurred since the 1980s. For the sake of     First, we replicate Qian and Preston’s (1993) analy-
 simplicity, we refer to these two periods by their     sis to provide a general picture of changes in mar-
 midpoints (i.e., 1980 and 1995).                       riage rates by age and educational attainment. By
    Although these pooled JNFS data are the most        comparing changes in observed marriage rates
 appropriate for our purposes, they are not with-       with changes in counterfactual marriage rates cal-
 out limitations. Because we are using sample sur-      culated by holding either marriage market com-
 vey data rather than registration data, we must        position or the propensity to marry constant over
 assume that our calculations are not affected by       time, we are able to evaluate the relative impor-
 differential nonresponse and differential mortal-      tance of changes in the availability of spouses for
 ity with respect to sex, age, marital status, and edu- understanding educational differences in the trend
 cational attainment to an extent that alters           toward later and less marriage. These figures also
                                                        facilitate :
 substantive interpretations. A similar assumptionby Ingenta to comparisons with changing patterns of
                                            Delivered
                                                        marriage in KYUSHU SHIKA DAIGAKU (cid
Kyushu University Central Library; Mr Beppu (cid 35001789),the United States over a similar peri-
 must be made with respect to the absence of
                                                        Zasshi (cid 52024142), Kyushu methods to
     2606), Kyushu University for the 6Library (2); od of time. Next, we use life-table University re-
 information on first marriages     Central percent
    Ropponmatsu Library (cid 46003719), pre-
 of men and women in our sample who were Kyushu University Medical Library (cid 46009061),
                                                        express observed and counterfactual marriage
 viously married. AssumingKyushu University Library (cida43009622)
                                 that differences in    rates in way that provides a clear visual repre-
                                           Tue, 24 Jan
 marriage timing and spouse selection with respect 2006 03:55:30the substantive importance of changes
                                                        sentation of
 to marital history are negligible, we impute first-    in marriage market composition and marriage
 marriage characteristics for these respondents         propensities. By describing educational differ-
 using mean ages at marriage and modal educa-           ences in observed and counterfactual marriage
 tional pairings for observed first marriages.7 Our     trajectories, we are also able to assess the extent
 concern about the impact of these assumptions is       to which explicit consideration of changes in mar-
 mitigated by the similarity of age-specific mar-       riage market composition may alter conclusions
 riage rates based on our data with corresponding       based on standard techniques for evaluating the
 marriage rates constructed from vital statistics       economic independence hypothesis.
 and census data collected in 1980 and 1995.8 A
 second limitation is that we have neither the sam-     CHANGES IN MARRIAGE RATES,
 ple size nor the residential history information       1980–1995
 required to construct region-specific rates. We
 are thus forced to make the simplifying assump-        METHODS
                                                        The harmonic mean function proposed by
                                                        Schoen (1988) allows us to assess the impor-
                                                        tance of marriage market mismatches in Japan
   6 The number of person-year observations is 70,700
                                                        by decomposing changes in marriage rates into
 for 1978–82 and 22,317 for 1993–97.                    changes in marriage market composition and
    7 Results do not change when all formerly married
                                                        changes in a composition-independent measure
 respondents and formerly married spouses are           of the likelihood of marriage. Previous appli-
 removed from the analyses. Results from these sup-
                                                        cations of this approach using data from the
 plemental analyses are available from the corre-
 sponding author upon request.
                                                        United States have examined the relevance of
    8 These comparisons are available from the corre-   changing marriage-market composition for
 sponding author upon request.                          understanding racial differences in marriage
 810 —– AMERICAN SOCIOLOGICAL REVIEW


 rates (Schoen and Kluegel 1988) and changes               are thus expressed as the product of the force of
 in the relationship between educational attain-           attraction and an “availability ratio” (i.e.,
 ment and marriage (Qian and Preston 1993).                    Mtjl
 The harmonic mean model is expressed as                     t + Mt ). If the distribution of men’s educa-
                                                           F ik       jl

                               F t Mt                      tional attainment is relatively stable over time,
                Ntijkl = atijk t ik jl t ,           (1)   changes in these availability ratios will be large-
                              Fik + M jl
                                                           ly determined by changes in women’s educa-
 where Ntijkl is the number of marriages between           tional distribution, with spouse availability
 women age i with educational attainment k and             decreasing for groups that grow in relative size
 men age j with educational attainment l in peri-          (i.e., highly educated women) and increasing for
 od t (i, j = 15–19, 20–24, 25–29, 30–34, 35–39),          groups that decline in relative size (i.e., less-edu-
 (k, l = less than high school, high school, jun-          cated women). With five age groups (j) and
 ior college, university), (t = 1980, 1995).9 Ftik is      four educational categories (l) used to classify
 the number of women age i with educational                potential spouses, we calculate 20 separate mar-
 attainment k at risk of marriage in period t, and         riage rates for women in each of the 20 age-edu-
    t is the corresponding number of men age j
 M jl                                                      cation combinations in each five-year period
 with educational attainment l. For each of the            (i.e., a total of 20 20 = 400 marriage rates for
 two five-year periods, the age-education-spe-             each time period).
 cific population at risk of first marriage is                 Because age-education-specific marriage
 defined as the sum of the never married mid-              rates are the product of two components, we
 year populations in each of the five years (i.e.,         can examine the relative importance of change
        5                                                  in each component by calculating counter-
 Ftik = Ft*) where Ft* is the number of unmar-             factual marriage rates for women of a given
       t*=1 ik           ik
                                                           age and :
                                            Delivered by Ingenta to education level. For example, by
 ried women of age i and educational attainment
                                                Beppu
Kyushu University Central Library; Mr’80, ’81,(cid 35001789), KYUSHU SHIKA DAIGAKU (cid
 k at the middle of year t* (t*=1978, ’79,                 replacing measures of availability from 1995
                        University Central compo-
      2606), Kyushu ’96, ’97). The other Library (2); Zasshijl(cid 52024142), Kyushu University
 ’82, ’93, ’94, ’95,
                                                           (    M 95
                                                                       )
                                                               95 + M95 with those from 1980 F80 + M80
    Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
 nent of equation (1), atijkl, isKyushu University Libraryik(cid 43009622)
                                   the “force of attrac-
                                                           and by summing marriage rates across values of
 tion” between women age i with education k and 2006 03:55:30
                                                             F          jl                        (    ik
                                                                                                          M80
                                                                                                           jl
                                                                                                              )
                                                                                                              jl
                                           Tue, 24 Jan
 men age j with education l in period t. As                j and l, we are able to answer the counterfactu-
 described by Qian and Preston (1993:483), the             al question, “What would women’s age-educa-
 force of attraction is a composition-independ-            tion-specific marriage rates be in 1995 if
 ent measure of marriage propensity reflecting             marriage market composition had not changed
 both the rate of encounters in the marriage mar-          since 1980?” Similarly, we can replace forces
 ket and the proportion of such encounters that            of attraction from the later period (a95 ) with cor-
                                                                                                  ijkl
 lead to marriage. It thus reflects multiple influ-        responding values from the earlier period (a80 )   ijkl
 ences on the desirability of marriage, including          to examine what women’s age-education-spe-
 women’s economic circumstances, men’s eco-                cific marriage rates would be in 1995 if these
 nomic circumstances, attitudes toward mar-                composition-independent measures of the
 riage, and preferences for particular spousal             propensity to marry had not changed since 1980.
 pairings.
     To calculate marriage rates for women age i           RESULTS
 with education k to men age j with education l,
                                                           Table 3 compares observed and counterfactual
 we divide both sides of equation (1) by Ftik.
                                                           age-specific marriage rates for women at each
                                Ntijkl t          t
                                                Mjl
                  (
 Marriage rates MRtijkl = t =a ijkl t
                               F ik                 )
                                            F ik + Mtjl
                                                           level of educational attainment. These age-edu-
                                                           cation-specific rates are calculated by summing
                                                           marriage rates (MRtijkl) across all combinations
                                                           of men’s age (j) and educational attainment
                                                           (l).10 Ratios of observed marriage rates in 1995
    9 We adopt the conventional approach of combin-

 ing junior college graduates and vocational school
 graduates into one category. Because the former are
 far more numerous than the latter, we refer to this          10 These age-education-specific marriage rates are

 group as junior college graduates.                        slightly lower than actual marriage rates because we
                                           EDUCATIONAL ATTAINMENT AND MARRIAGE IN JAPAN—– 811

 Table 3. Comparison of Observed and Counterfactual Marriage Rates, by Educational Attainment and Age

 Educational Attainment and Age          Observed (MR95/MR80)          Counterfactual 1a   Counterfactual 2b
 Junior High School
 ——15–19                                        1.43                      1.28                   1.08
 ——20–24                                        0.49                      0.44                   1.03
 ——25–29                                        0.33                      0.29                   1.10
 ——30–34                                        0.66                      0.54                   1.10
 ——35–39                                        3.57                      3.33                   0.94
 High School
 ——15–19                                        1.61                      1.14                   1.48
 ——20–24                                        0.51                      0.46                   1.19
 ——25–29                                        0.58                      0.63                   0.94
 ——30–34                                        0.90                      0.89                   0.99
 ——35–39                                        1.49                      1.55                   0.95
 Junior College
 ——15–19                                        0.60                      0.68                   1.37
 ——20–24                                        0.41                      0.44                   1.01
 ——25–29                                        0.58                      0.75                   0.80
 ——30–34                                        0.65                      0.78                   0.82
 ——35–39                                        1.77                      2.48                   0.77
 University
 ——15–19                                      0.00                        0.00                   0.90
 ——20–24                                      0.27                        0.32                   0.88
 ——25–29                                      0.43                        0.52                   0.87
 ——30–34                                      0.88                        1.00                   0.87
 ——35–39                                Delivered by Ingenta to :
                                              0.11                        0.12                   0.87
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
 a Availability ratios held constant at 1980 values.
      2606), Kyushu University Central Library (2); Zasshi (cid 52024142), Kyushu University
 b Forces of attraction held constant at 1980 values.
     Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
                                    Kyushu University Library (cid 43009622)
                                             Tue, 24 Jan 2006 03:55:30
 to observed rates in 1980 in column 1 describe             sponding observed rates for 1980. Here, ratios
 the age pattern of decline in marriage rates and           less (greater) than one indicate that change in
 allow for comparison across educational levels             marriage market composition has contributed to
 and comparison with Qian and Preston’s (1993)              lower (higher) rates of marriage.
 figures for the United States. The second col-                Ratios of observed rates in the first column
 umn presents ratios of counterfactual rates for            show the trend toward later marriage very clear-
 1995 calculated by holding the availability of             ly. For all educational groups, marriage rates at
 potential mates
                   (    F ik   )
                            Mtjl
                          t + Mt
                                  jl
                                      constant at 1980
 values to the corresponding observed rates for
                                                            ages 20–24 and 25–29 declined by at least 42 per-
                                                            cent between 1980 and 1995. Consistent with pre-
                                                            vious studies of marriage in Japan (e.g., Raymo
 1980. Here, counterfactual ratios of less than one         2003), declines in the marriage rates of women in
 indicate that decline in marriage propensities             their late twenties have been greatest among the
 (independent of changes in marriage market                 most highly educated. Marriage rates of universi-
 composition) has contributed to lower rates of             ty graduates declined by 73 percent at ages 20–24
 marriage. Conversely, ratios greater than one              and by 57 percent at ages 25–29. The correspon-
 indicate that change in the composition-inde-              ding declines in the marriage rates of high school
 pendent propensities to marry has contributed              graduates were 49 and 42 percent, respectively.
 to higher rates of marriage. The third column                 In contrast to the United States, there is lit-
 presents ratios of counterfactual rates for 1995           tle evidence that marriage rates of highly edu-
 calculated by holding the forces of attraction             cated women in Japan have increased at older
 (atijkl) constant at 1980 values to the corre-             ages. In 1995, the marriage rates of junior col-
                                                            lege and university graduates are lower than in
                                                            1980 through ages 30–34. The large apparent
 are unable to include the small number of marriages        decline in the marriage rate for 35–39 year-old
 to men age 40 and above.                                   university graduates (ratio of 0.11) should not
 812—– AMERICAN SOCIOLOGICAL REVIEW


 be given too much emphasis as marriage rates             market composition on marriage rates. These
 for these women were based on a very small               ratios are less than one at all ages for universi-
 number of marriages.11 The same is true of the           ty graduates and beyond ages 25–29 for junior
 very large increase in the marriage rate of 35–39        college graduates, indicating that decline in the
 year-old junior high school graduates (ratio of          supply of suitable mates has contributed to lower
 3.57). Although these irregularities in the very         rates of marriage for women with postsecondary
 low rates of marriage at ages 35–39 have rela-           education. In contrast, the counterfactual rates
 tively little impact on the life-table analyses          calculated by holding marriage propensities
 discussed in the following section, results              constant at their 1980 values are either higher
 beyond age 35 should be interpreted cautious-            than or very similar to the observed rates for
 ly.                                                      1980 at all ages for women with a high school
     In general, the ratios of observed marriage          education or less. For example, at ages 30–34,
 rates in column 1 indicate that university grad-         the marriage rate of high school graduates in
 uates have experienced relatively large declines         1995 would have been no different than the
 in marriage rates at younger ages and no increas-        observed marriage rate in 1980 if forces of
 es in marriage rates at older ages. This pattern         attraction had not changed (i.e., the ratio of
 of change contrasts with Qian and Preston’s              counterfactual to observed rates is 0.99). For
 (1993) description of changes in American mar-           women in the lowest educational category, the
 riage and is ostensibly consistent with the              corresponding rate would have been 10 percent
 hypothesis that increasing economic opportu-             higher than observed. The observed declines
 nities (as proxied by educational attainment)            in the marriage rates of less-educated women in
 should be associated with later and less marriage        their twenties and early thirties are thus due
 in societies such as Japan, where gender asym-           entirely to changes in the propensity to marry.
 metry in the division of labor makesDelivered by Ingenta to :
                                           it difficult
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
 for women to combine work and family.                    OBSERVED AND COUNTERFACTUAL
      2606), Kyushu University Central Library                                       Kyushu
     Ratios in the second column are, not sur-(2); Zasshi (cid 52024142),TORIES University
                                                          MARRIAGE TRAJLibrary (cid 46009061),
     Ropponmatsu Library (cid 46003719), Kyushu University Medical EC                         FOR
 prisingly, similar to those in the first column. For LibraryNTH43009622) ORTS
                                Kyushu University         SY (cid ETIC COH
                                          the force of
 all age-education groups, decline in Tue, 24 Jan 2006 03:55:30
 attraction is the primary reason for declining           To describe the impact of changes in market
 marriage rates. At the same time, however, the           composition on marriage timing and the likeli-
 counterfactual ratios in the second and third            hood of marrying by age 40, we use the
 columns suggest a somewhat more complex                  observed and counterfactual marriage rates
 story. For example, ratios in the second column          summarized in Table 3 to construct synthetic
 are, in most cases, greater than the observed            cohort first-marriage trajectories for women at
 ratios in the first column for more highly edu-          each level of educational attainment. We first
 cated women but lower for women with a high              use standard procedures to convert the rates
 school education or less. This pattern indicates         into probabilities (see, e.g., Preston, Heuveline,
 that changes in spouse availability have con-            and Guillot 2001:76) necessary for the con-
 tributed to lower rates of marriage among high-          struction of multiple-decrement marriage tables.
 ly educated women while partially offsetting             We then construct a total of 16 marriage
 decline in the forces of attraction among less-          tables—eight tables based on observed rates
 educated women.                                          (i.e., four levels of education multiplied by two
     This pattern of change is more immediately           time periods) and eight tables based on the
 evident in the third column. Because the coun-           counterfactual rates calculated by holding mar-
 terfactual rates used to calculate the numerator         riage market composition and forces of attrac-
 assume no change in forces of attraction, these          tion constant at their 1980 levels. In each of
 ratios describe the effect of changing marriage          these marriage tables there are 20 possible
                                                          decrements representing husbands’ age and edu-
                                                          cational attainment (i.e., four levels of education
    11 The only women in our sample ages 35–39 in         multiplied by five age groups). We ignore mor-
 1978–82 were 45–49 year old respondents to the           tality in all marriage table calculations.
 1992 survey. This group contained very few univer-          Calculation of cumulative first marriage
 sity graduates.                                          probabilities based on both observed and coun-
                                              EDUCATIONAL ATTAINMENT AND MARRIAGE IN JAPAN —– 813


 terfactual marriage rates produces visibly intu-     These two marriage curves clearly demonstrate
 itive descriptions of the extent to which increas-   the trend toward later marriage. At each level of
 es in the age-specific proportions of Japanese       educational attainment, the synthetic cohort
 women who have never married reflect changes         proportions ever married beyond age 25
 in marriage market composition—as suggested          declined substantially between 1980 and 1995.
 by our theoretical emphasis on changes in the        The largest observed change is the 36-point
 demographic feasibility of marriage. This com-       decline in the synthetic cohort proportion of
 parison of marriage trajectories across levels of    university graduates married by age 30 (Figure
 educational attainment also allows us to assess      6). Large declines in life table proportions ever
 the extent to which standard methods for eval-       married by age 40 suggest increases across the
 uating the economic independence hypothesis          educational spectrum in the proportion of
 may generate misleading conclusions by ignor-        women who will never marry. Whereas the pro-
 ing the impact of changes in the availability of     portion never married at age 40 in 1980 ranged
 potential mates.                                     from 3 percent among university graduates to
     Figures 3–6 present the four cumulative first-   7 percent among junior college graduates,
 marriage curves for each educational group.          roughly 20 percent of women with a two-year
 Because we have calculated marriage proba-           college degree or less and 27 percent of uni-
 bilities for five-year age groups, the points on     versity graduates had yet to marry by age 40 in
 these marriage trajectories represent the pro-       1995.
 portion ever married at the beginning of the             The dashed lines marked with represent
 five-year age group (i.e., at exact age 15, 20,      counterfactual cumulative first marriage curves
 .|.|.|., 40). We use solid, bold lines to depict     calculated by holding availability ratios (i.e.,
 cumulative first-marriage curves based on               Mtjl
                                        Delivered by Ingenta to ) :constant at their 1980 values. These
                                                        t + Mt
 observed marriage probabilities for 1980             F ik
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
                                                               jl
 (marked with ) and 1995 (marked with ). (2); curves (cid 52024142), Kyushu University
       2606), Kyushu University Central Library       Zasshi thus indicate what the synthetic cohort
     Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
        100                    Kyushu University Library (cid 43009622)
                                       Tue, 24 Jan 2006 03:55:30
                   90

                   80

                   70
  % Ever Married




                   60

                   50

                   40

                   30

                   20

                   10

                    0
                        15          20                 25 Exact Age 30              35          40
                                  1980          1995         FOA=1980       AR=1980

 Figure 3. Observed and Counterfactual Marriage Trajectories for Junior High School Graduates
 Note: FOA = force of attraction; AR = availability ratio.
 814—– AMERICAN SOCIOLOGICAL REVIEW


                     100

                         90

                         80

                         70
    % Ever Married




                         60

                         50

                         40

                         30

                         20

                         10

                          0
                               15      20             25 Exact Age 30             35        40
                                     1980      1995          FOA=1980      AR=1980
 Figure 4. Observed and Counterfactual Marriage Trajectories for High School Graduates
                                          Delivered by Ingenta to :
 Note: FOA = force of attraction; AR = availability ratio.
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
    2606), Kyushu University Central Library (2); Zasshi (cid 52024142), Kyushu University
   Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
                                Kyushu University Library (cid 43009622)
    100                                  Tue, 24 Jan 2006 03:55:30

             90

             80

             70

             60

             50

             40

             30

             20

             10

                     0
                              15      20              25           30             35        40

                                    1980      1995           FOA=1980      AR=1980
 Figure 5. Observed and Counterfactual Marriage Trajectories for Junior College Graduates
 Note: FOA = force of attraction; AR = availability ratio.
                                          EDUCATIONAL ATTAINMENT AND MARRIAGE IN JAPAN—– 815


 marriage trajectories would have looked like in   lege graduates and university graduates, respec-
 1995 if marriage market composition had not       tively. For both groups, changes in the avail-
 changed. The dashed lines marked with rep-        ability of potential partners are responsible for
 resent the counterfactual marriage patterns cal-  a nontrivial proportion of the overall decline in
 culated by holding forces of attraction (i.e.,    the synthetic cohort proportion ever married.
 atijkl) constant at their 1980 values. These      This is particularly true for junior college grad-
 curves thus indicate what the synthetic cohort    uates beyond age 30, for whom the counterfac-
 marriage trajectories would have looked like      tual curve calculated by holding availability
 in 1995 if propensities to marry men of a given   ratios constant at their 1980 levels is roughly
 age and educational attainment had not            seven percentage points above the observed
 changed over time.                                curve for 1995. In other words, if marriage mar-
     For women with less than a high school        ket composition had not changed over the 15-
 education (Figure 3), the marriage-promot-        year period, the decline in the synthetic cohort
 ing effect of changes in marriage market          proportion of two-year college graduates yet to
 composition described earlier is clear. The       marry would have been seven percentage points
 counterfactual marriage curve calculated by       less than was actually observed. For university
 holding availability ratios constant at their     graduates, the difference between the observed
 1980 values falls below the observed 1995         and counterfactual curves is five percentage
 curve at all ages, indicating that age-specific   points beyond age 30. These differences repre-
 proportions married would have been even          sent approximately one-third of the total decline
 lower than observed had marriage market           in the synthetic cohort proportion of two-year
 composition not changed over time. Beyond         college graduates unmarried beyond age 30 and
 age 25, the cumulative proportion married         one-fourth of the observed decline for univer-
                                      Delivered
 would have been four to six percentageby Ingenta to :
                                                   sity graduates.12
 points University Central Library; Mr Beppu          Overall, KYUSHU SHIKA DAIGAKU (cid
Kyushu lower for these women had marriage(cid 35001789),it is clear that changes in the avail-
      2606), Kyushu University Central Library     Zasshi (cid 52024142), Kyushu University
 market composition remained unchanged (2); ability of potential mates account for a signif-
     Ropponmatsu Library of 46003719), Kyushu University Medical Library (cid 46009061),
 from 1980. Examination(cidindividual avail-       icant proportion of
                             Kyushu University Library (cid 43009622) the decline in marriage
 ability ratios (not shown) suggests that this 2006 03:55:30 educated women but have had lit-
                                     Tue, 24 Jan   among highly
 is largely due to the increasing prevalence of    tle effect on the transition to marriage among
 older unmarried men, especially in the two        high school graduates. Conclusions regarding
 lowest educational categories. For high           the relevance of the economic independence
 school graduates (Figure 4), a similar, but       hypothesis may therefore depend upon the treat-
 less distinct, pattern is observed at younger     ment of changes in marriage market composi-
 ages. At older ages, however, the decline in      tion. We demonstrate this in Figure 7 by
 marriage is due entirely to changes in the        presenting ratios of the synthetic cohort pro-
 force of attraction. There is no evidence that    portions of junior college and university grad-
 changes in marriage market composition            uates ever married by a given age to the
 have affected the marriage behavior of high       corresponding proportions for high school grad-
 school graduates beyond age 30. To a large        uates. Age-specific figures for 1980 (striped
 extent, this is due to the relatively small       bars) and 1995 (black bars) indicate that the
 decline in the availability of high school        likelihood of marriage for junior college grad-
 graduates (see Table 2). It also reflects rel-    uates (left) and university graduates (right) rel-
 atively high rates of educational heterogamy      ative to high school graduates has declined at
 among high school graduates and the fact          all ages. For example, at ages 30–34, the ratio
 that declines in the relative availability of
 junior high school graduates have been off-
 set by increases in the relative availability of     12 We calculated the percentage contribution of
 men with postsecondary education.                 changing marriage market composition as (Pik   95(AR=80)
     A very different pattern of change is             95
                                                   – Pik)       80 - P95), where Pt is the proportion of
                                                              (Pik ik               ik
 observed among women with higher levels of        women with education k ever married by age i in
 education. Figures 5 and 6 indicate that          period t and Pik95(AR=80) is the counterfactual propor-
 changes in marriage market composition have       tion calculated by holding availability ratios con-
 contributed to later marriage among junior col-   stant at their 1980 levels.
 816—– AMERICAN SOCIOLOGICAL REVIEW


 100

       90

       80

       70

       60

       50

       40

       30

       20

       10

               0
                                     15                20                25              30                 35            40

                              1980                         Ingenta to
                                          Delivered byFOA=1980 :
                                           1995                             AR=1980
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
    2606), Kyushu University Central Library (2); Zasshi (cid 52024142),
 Figure 6. Observed and Counterfactual Marriage Trajectories for University GraduatesKyushu University
   Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
 Note: FOA = force of attraction; AR = availability ratio.
                                Kyushu University Library (cid 43009622)
                                         Tue, 24 Jan 2006 03:55:30
                                          Junior College Graduates                                 University Graduates
                               1.1

                               1.0

                               0.9
 % HS Graduates Ever Married




                               0.8

                               0.7
      % Ever Married /




                               0.6

                               0.5

                               0.4

                               0.3

                               0.2

                               0.1

                               0.0
                                      25          30         35          40     Exact Age     25           30      35          40


                                                                  1980   1995    Availability Ratio=1980

Figure 7. Observed and Counterfactual Proportions Ever Married, Relative to High School Graduates
                                        EDUCATIONAL ATTAINMENT AND MARRIAGE IN JAPAN —– 817


 of the synthetic cohort proportion of universi-      SUMMARY AND DISCUSSION
 ty graduates ever married to that of high school
                                                      We have shown that changes in marriage mar-
 graduates was .95 in 1980 and .69 in 1995. This
                                                      ket composition account for one-fourth to one-
 increasing educational differential in marriage
                                                      third of the decline in the synthetic cohort
 has typically been interpreted as evidence of an
                                                      proportion ever married among Japanese
 independence effect, that is, highly educated        women with post-secondary education. This is
 women have become less dependent upon men’s          a substantial contribution and is consistent with
 economic resources and thus marry at increas-        the hypothesis motivating this study. At the
 ingly lower rates.                                   same time, however, it is important to recognize
    A rather different picture emerges, however,      that the majority of change in marriage timing
 when we look at counterfactual marriage tra-         at all levels of education is due to change in the
 jectories calculated by holding marriage market      propensity to marry. Large declines in the
 composition constant at its 1980 values (dotted      propensity to marry presumably reflect a vari-
 bars). These ratios indicate that the growing        ety of mechanisms including reductions in the
 differential between university graduates and        gains to marriage and changing attitudes toward
 high school graduates is due, in part, to changes    marriage and family formation.
 in marriage market composition. For example,            At the same time, our results highlight a fun-
 the relative likelihood of marrying by age 35        damental limitation with the conventional
 declined from .97 to .87 using observed rates but    approach of interpreting relatively large decline
 from .97 to .94 using counterfactual rates.          in the marriage rates of highly educated women
 Beyond age 30, differential decline in the pro-      in relatively gender-inegalitarian societies as
 portion of university graduates ever married         evidence of increasing economic independence.
 would have been 40 to 70 percent lower than          We find that roughly half of the “excess” decline
                                        Delivered by Ingenta to : observed among female university
                                                      in marriage
 observed had marriage market composition not
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
 changed over the 15-year period. This counter- (2); graduates (typically interpreted as support for
      2606), Kyushu University Central Library        Zasshi (cid 52024142), Kyushu University
                                                      the economic independence hypothesis)
 factual pattern of change suggests that, Kyushu University Medical Library (cid 46009061), is
     Ropponmatsu Library (cid 46003719), to a
                                                      explained by changes
 large extent, apparent support for the econom- Library (cid 43009622) in marriage market com-
                              Kyushu University
                                                      position. Changes in the supply of potential
                                       Tue, 24 Jan 2006 03:55:30
 ic independence hypothesis in Japan (Ono 2003;
 Raymo 2003) actually reflects differential           partners brought about by relative improve-
 changes in the availability of potential mates. It   ments in women’s educational attainment have
 also suggests that mechanisms of change may          contributed to lower rates of marriage among
 vary by age, with economic independence more         highly educated women and somewhat higher
 relevant for university graduates at younger         rates of marriage among women with a high
                                                      school education or less. Shifting marriage mar-
 ages (20–29) and marriage market mismatch-
                                                      ket composition accounts for all of the change
 es more relevant at older ages (30–39).
                                                      in the relative proportions of two-year college
    The impact of shifts in marriage market com-
                                                      graduates and high school graduates ever mar-
 position is even more striking in the corre-
                                                      ried by a given age. The composition-inde-
 sponding figures for junior college graduates.       pendent propensity to marry among junior
 For these women, counterfactual ratios are high-     college graduates has actually increased relative
 er than the observed ratios for 1980 beyond age      to that of high school graduates.
 30, indicating that the decline in the marriage         Our findings do not imply that economic
 rates of junior college graduates would have         independence is unimportant for understanding
 been less than that of high school graduates if      the trend toward later marriage in societies such
 there had been no changes in the availability of     as Japan, where asymmetric gender relations
 potential partners. Indeed, marriage trajecto-       make it difficult for women to combine work
 ries of junior college graduates are identical to    and family. Indeed, our results are consistent
 those of high school graduates beyond age 30         with predictions of the economic independence
 under this counterfactual scenario. The increas-     hypothesis in that changes in the composition-
 ing differential between the proportions of jun-     independent forces of attraction account for
 ior college graduates and high school graduates      roughly half of the differential decline in mar-
 ever married beyond age 30 is thus due entire-       riage of university graduates relative to high
 ly to changes in marriage market composition.        school graduates. At the same time, however, the
 818 —– AMERICAN SOCIOLOGICAL REVIEW


 strong marriage-inhibiting effect of shifts in        highly educated men continues to decline, it is
 marriage market composition points to the             likely that increasing proportions of highly edu-
 importance of women’s continued economic              cated women (and less-educated men) will
 dependence on men.                                    marry at later ages or not at all. The plausibil-
    These contrasting influences on the marriage       ity of this scenario is suggested by evidence
 behavior of highly educated women in Japan are        that highly educated women in the United States
 consistent with earlier emphases on attitudinal       appear to prefer to remain single rather than
 heterogeneity (Tsuya and Mason 1995).                 marry a man with less education than them-
 Marriages characterized by a highly asymmet-          selves (Lichter et al. 1995).13 Although there is
 ric gender division of labor may indeed be            some evidence that the association between
 increasingly unattractive to highly educated          spouses’ educational attainment in Japan has
 women with higher levels of career investment         weakened over time (Raymo and Xie 2000;
 or career ambitions while other similarly edu-        Suzuki 1991), this change has clearly not been
 cated women with less commitment to work              sufficient to offset the large changes in marriage
 and a gender-egalitarian division of labor            market composition. In fact, supplemental log-
 between spouses may be marrying later (and            linear analyses of our data (results not shown)
 less) as a result of the increasing numerical dif-    indicate no change in educational pairing pat-
 ficulty of locating a suitably educated spouse.       terns between 1980 and 1995.14 It is conceiv-
 This focus on attitudinal heterogeneity also sug-     able, however, that changes in men’s
 gests a compelling interpretation for the strik-      contribution to housework and the implemen-
 ing impact of shifting mar riage-market               tation of policies designed to facilitate work-
 composition on the behavior of junior college         family balance may result in more symmetric
 graduates. As more women choose to attend             spouse-pairing preferences that offset declines
                                                       in the to :
 four-year universities, those enrolling in juniorby Ingenta relative supply of highly educated men.
                                       Delivered
 colleges may be Central Library; Mr Beppu
Kyushu Universityan increasingly self-selected(cid 35001789), KYUSHU SHIKA DAIGAKU (cid
                                                       In the absence of such changes, however, con-
     2606), respect to career ambition Library         tinued improvements Kyushu University
 group withKyushu University Central and atti-(2); Zasshi (cid 52024142), in women’s economic
                                                       opportunities may Library (cid 46009061),
    Ropponmatsu Library (cid 46003719), Kyushu University Medicalactually strengthen the role
 tudes toward a “traditional” division of house-
                               Kyushu University Library (cid 43009622)
                                       subsequent      of men’s economic resources as a spouse-selec-
 hold labor. To the extent possible, Tue, 24 Jan 2006 03:55:30
 analyses of the relationship between women’s          tion criterion (i.e., because the opportunity costs
 educational attainment and the transition to          of career interruption would increase), thus
 marriage should attempt to incorporate meas-          exacerbating the marriage market mismatches
 ures of family attitudes and career aspirations.      that we have described here.
    Our results contrast with those of Qian and           It is also important to consider the theoreti-
 Preston (1993) and other studies of changing          cal implications of economic changes charac-
 marriage patterns in the United States. We find       terizing the “lost decade” (Kelly and White
 no evidence of relatively large increases in mar-     2005) of the 1990s. During this period, the pro-
 riage rates at older ages among highly educat-        portion of young men and women unemployed
 ed women in Japan. Studies of marriage in the         or employed temporarily or part-time on fixed-
 United States have suggested that relative            term contracts (popularly referred to as freeters)
 improvements in women’s economic opportu-             increased across the educational spectrum
                                                                          -
 nities may facilitate marriage by enabling cou-       (Kosugi 2001; Oishi 2004). The fact that these
 ples to pool resources or by enabling women to        trends have been more pronounced for women
 marry men who do not yet have the financial           than for men at higher levels of education
 resources to marry but are otherwise attractive       (Genda and Kurosawa 2001) calls into question
 (Oppenheimer 1988). This marriage-facilitating        the relevance of theoretical emphasis on increas-
 effect of women’s economic resources may be           ing economic independence for highly educat-
 less relevant in societies such as Japan, where
 the highly asymmetric gender-division of work
 and family roles reinforces highly educated              13 See Lewis and Oppenheimer (2000), however,
 women’s desire for status-homogamous or sta-          for evidence that women adjust their spouse-selec-
 tus-hypergamous marriages.                            tion criteria in response to limited mate availability.
    If these asymmetric spouse-pairing prefer-            14 Results of these supplemental analyses are avail-

 ences remain strong while the relative supply of      able from the corresponding author upon request.
                                          EDUCATIONAL ATTAINMENT AND MARRIAGE IN JAPAN—– 819


 ed women. A more compelling economic expla-            market composition, in conjunction with rela-
 nation for the relatively large decline in the         tively stable spouse-pairing patterns, have played
 marriage rates of highly educated women might          an important role in declining rates of marriage
 focus on the very limited economic opportuni-          among highly educated women in countries
 ties for women with a high school education or         such as Italy, Spain, Korea, and Taiwan would
 less (Kosugi 2001). It is possible, for example,       provide further reason to question the conven-
 that increasing economic dependence among              tional interpretation that women’s economic
 women with lower levels of education has lim-          independence is the primary reason for declin-
 ited decline in their marriage rates to a greater      ing rates of marriage in societies where asym-
 degree than increasing economic independence           metric gender-relations within the family make
 has accelerated the trend toward later marriage        it difficult for women to combine work and
 among more highly educated women.                      family.
 Alternatively, the fact that the 1990s recession          At the same time, such findings would sug-
 has had relatively limited impact on the eco-          gest the general relevance of a scenario in which
 nomic circumstances of highly educated men             highly educated women in societies character-
 (Genda and Kurosawa 2001) points to the poten-         ized by structural and normative impediments
 tial relevance of theoretical emphasis on shift-       to successful work-family balance face increas-
 ing marriage market composition. If women              ing numerical difficulty in locating a spouse
 respond to the combination of growing eco-             upon whom they can be economically depend-
 nomic uncertainty and unchanging family roles          ent. With women’s educational attainment
 by placing greater emphasis on the earnings            increasing rapidly in many developing coun-
 power of potential spouses, increasing compe-          tries (United Nations 2000), the potential rele-
 tition for highly educated men should contribute       vance of such a scenario is broad. If the time lag
                                                        between :
 to delays in marriage for the womenDelivered by Ingenta to growth in women’s educational and
                                         most like-
 ly to marry these Central Library; Mr                  occupational opportunities and DAIGAKU in
Kyushu Universitymen (Yamada 1996). Beppu (cid 35001789), KYUSHU SHIKA convergence(cid
      2606), Kyushu University this study high-         Zasshi (cid 52024142), Kyushu University
     We believe that the results ofCentral Library (2); the gender division of household labor observed
     Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
 light an important gap in cross-national research      in Western industrialized countries (McDonald
                               Kyushu University Library (cid 43009622)
                                        increasing
 on the relationship between women’sTue, 24 Jan 2006 03:55:30
                                                        2000) is a relatively universal pattern of social
 educational attainment and marriage, and we            change, shifting marriage market composition
 hope that subsequent research will provide a           may be particularly relevant for understanding
 fuller understanding of this relationship by           changes in the marriage rates of highly educat-
 addressing several important limitations of this       ed women in rapidly developing Asian and Latin
 study. The most obvious limitation is the nature       American countries. Indeed, preferences for
 of the data that we use. In the absence of regis-      female hypergamy and associated marriage mar-
 tration data, we have relied upon retrospective        ket mismatches have been offered as an expla-
 data pooled across multiple cross-sectional sur-       nation for relatively low rates of marriage among
 veys. Such data not only preclude examination          highly educated women in Thailand and
 of change in marriages that were rare in one or        Indonesia (Jones 1997). Because the process of
 both time periods (e.g., marriages involving           social change implied by the marriage market
 35–39 year-old women with post-secondary               mismatch hypothesis contrasts fundamentally
 education), but they also cannot be used to            with the notion that women’s increasing eco-
 examine more than two dimensions of spouse             nomic independence enables them to “buy out”
 pairing. The number of observations is not suf-        of marriages characterized by gender special-
 ficient to examine pairing with respect to a           ization, there is clear sociological relevance in
 wider range of characteristics (e.g., parental
                                                        developing a better understanding of how the
 background). It is unlikely that subsequent
                                                        relationship between women’s economic
 research will be able to address these limitations
                                                        resources and marriage may depend upon gen-
 by using registration data from Japan. It is both
                                                        der context.
 feasible and important, however, that our find-
 ings from Japan be evaluated in other settings         James M. Raymo is Assistant Professor of Sociology
 where a negative relationship between women’s          at the University of Wisconsin-Madison. His current
 educational attainment and marriage has been           research focuses primarily on trends in marriage
 observed. Finding that changes in marriage             and divorce in Japan. In other research, he is exam-
 820 —– AMERICAN SOCIOLOGICAL REVIEW

 ining relationships between work and family at older     from School to Work in Japan.” Journal of the
 ages in Japan and the United States.                     Japanese and International Economies 15:465–88.
                                                        Gershuny, Jonathan. 2000. Changing Times: Work
 Miho Iwasawa is Senior Researcher in the
                                                          and Leisure in Postindustrial Society. New York:
 Department of Population Dynamics at the National
                                                          Oxford University Press.
 Institute of Population and Social Security Research
                                                        Goldstein, Joshua A. and Catherine T. Kenney. 2001.
 in Tokyo. Her research focuses on various aspects of
                                                          “Marriage Delayed or Marriage Foregone? New
 family formation in Japan, including contraception,
                                                          Cohort Forecasts of First Marriage for U.S.
 premarital pregnancy, unintended fertility, and cohab-
                                                          Women.” American Sociological Review
 itation. She is an investigator for the Japanese
                                                          66:506–19.
 National Fertility Surveys and a member of the
                                                        Goodkind, Daniel. 1997. “The Vietnamese Double
 research team that produces Japan’s population pro-      Marriage Squeeze.” International Migration
 jections.                                                Review 31:108–27.
                                                        Hobson, Barbara. 1990. “No Exit, No Voice: Women’s
 REFERENCES                                               Economic Dependency and the Welfare State.”
                                                          Acta Sociologica 33:235–50.
 Akers, Donald S. 1967. “On Measuring the Marriage      Jones, Gavin W. 1997. “The Demise of Universal
    Squeeze.” Demography 4:907–24.                        Marriage in East and South-East Asia.” Pp. 51–79
 Anzo, Shinji. 1985. “Measurement of the Marriage         in The Continuing Demographic Transition, edit-
    Squeeze: Its Application.” Japanese Journal of        ed by G. W. Jones, R. M. Douglas, J. C. Caldwell,
    Population Studies 8:1–10.                            and R. M. D’Souza. Oxford, England: Clarendon.
 Becker, Gary S. 1991. A Treatise on the Family.        Kalmijn, Matthiijs. 1991. “Shifting Boundaries:
    Cambridge, MA: Harvard University Press.              Trends in Religious and Educational Homogamy.”
 Blossfeld, Hans-Peter. 1995. “Changes in the Process     American Sociological Review 56:786–800.
    of Family Formation and Women’s Growing             ———. 1994. “Assortative Mating by Cultural and
    Economic Independence: A Comparison of Nine           Economic Occupational Status.” American Journal
                                           Delivered
    Countries.” Pp. 3–32 in The New Role of Women:by Ingenta to :
                                                          of Sociology 100:422–52.
    Family Formation in Modern Societies, Beppu
Kyushu University Central Library; Mredited by(cid 35001789), KYUSHU SHIKA DAIGAKU (cid
                                                        Kamo, Yoshinori. 1994. “Division of Household
      2606), Kyushu University Central
    H.-P. Blossfeld. Boulder, CO: Westview.Library (2); Zasshi (cid 52024142), Kyushu University of
                                                          Work in the United States and Japan.” Journal
    Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
 Blossfeld, Hans-Peter and Johannes Huinink. 1991.        Family Issues 15:348–78.
                                Kyushu University
    “Human Capital Investments or Norms of Role Library (cid 43009622)and Merry I. White. 2005.
                                                        Kelly, William W.
                                          Tue, 24 the
    Transition? How Women’s Schooling Affects Jan 2006 03:55:30 Slackers, Singles, Seniors, and
                                                          “Students,
    Process of Family Formation.” American Journal        Strangers: Transforming a Family-Nation.”
    of Sociology 97:143–68.                               Forthcoming in Japan and Asia, edited by P.
 Bracher, Michael and Gigi Santow. 1998. “Economic        Katzenstein, and T. Shiraishi. Ithaca, NY: Cornell
    Independence and Union Formation in Sweden.”          University Press.
    Population Studies 52:275–94.                       Kiernan, Kathleen. 2000. “European Perspectives
 Brinton, Mary C. 1988. “The Socio-Institutional          on Union Formation.” Pp. 40–58 in The Ties that
    Bases of Gender Stratif ication: Japan as an          Bind, edited by L. J. Waite. New York: Aldine de
    Illustrative Case.” American Journal of Sociology     Gruyter.
    94:300–34.                                          Kosugi, Reiko. 2001. “The Transition from School
 ———. 1989. “Gender Stratif ication in                    to Work in Japan: Understanding the Increase in
    Contemporary Urban Japan.” American                   Freeter and Jobless Youth.” Japan Labor Review
    Sociological Review 54:549–64.                        1:52–67.
 ———. 1993. Women and the Economic Miracle:             Lewis, Susan K. and Valerie K. Oppenheimer. 2000.
    Gender and Work in Postwar Japan. Berkeley,           “Educational Assortative Mating across Marriage
    CA: University of California Press.                   Markets: Non-Hispanic Whites in the United
 Brinton, Mary C. and Sunhwa Lee. 2001. “Women’s          States.” Demography 37:29–40.
    Education and the Labor Market in Japan and         Lichter, Daniel T., Robert N. Anderson, and Mark D.
    Korea.” Pp. 125–50 in Women’s Working Lives in        Hayward. 1995. “Marriage Markets and Marital
    East Asia, edited by M. C. Brinton. Stanford, CA:     Choice.” Journal of Family Issues 16:412–31.
    Stanford University Press.                          Lichter, Daniel T., Diane K. McLaughlin, George
 Brinton, Mary C. and Hang-Yue Ngo. 1993. “Age and        Kephart, and David J. Landry. 1992. “Race and the
    Sex in the Occupational Structure: A United States-   Retreat from Mar riage: A Shortage of
    Japan Comparison.” Sociological Forum 8:93–111.       Marriageable Men?” American Sociological
 Fuwa, Makiko. 2004. “Gender and Housework in             Review 57:781–99.
    22 Countries.” American Sociological Review         McDonald, Peter. 2000. “Gender Equity in Theories
    69:751–67.                                            of Fertility Transition.” Population and
 Genda, Yuji and Masako Kurosawa. 2001. “Transition       Development Review 26:427–39.
                                         EDUCATIONAL ATTAINMENT AND MARRIAGE IN JAPAN—– 821

 National Institute of Population and Social Security   United States.” Pp. 19–39 in The Ties that Bind,
   Research (NIPSSR). 1999. Dokushin Seinenso no –      edited by L. J. Waite. New York: Aldine de Gruyter.
   Kekkonkan to Kodomokan – Dai 11kai Shussho       – Raymo, James M. 2000. Spouse Selection and
     – –             –
   Do ko Kihon Cho sa [Attitudes toward Marriage        Marriage Timing in Japan. Ph.D. dissertation,
   and the Family among the Unmarried Japanese          Department of Sociology, University of Michigan,
   Youth—The Eleventh National Fertility Survey].       Ann Arbor, MI.
   Tokyo, Japan: National Institute of Population and ———. 2003. “Educational Attainment and the
   Social Security Research.                            Transition to First Marriage among Japanese
                             –               –
 ———. 2002. Nihon no Sho rai Suikei Jinko : 2001-       Women.” Demography 40:83–103.
   2050 [Population Projections for Japan:            Raymo, James M. and Yu Xie. 2000. “Temporal and
   2001–2050]. Tokyo, Japan: National Institute of      Regional Variation in the Strength of Educational
   Population and Social Security Research.             Homogamy (Comment on Smits, Ultee, and
                        – –            – –
 ———. 2005. Jinko To kei Shiryo shu [Latest             Lammers, ASR 1998).” American Sociological
   Demographic Statistics]. Tokyo, Japan: National      Review 65:773–81.
   Institute of Population and Social Security        Retherford, Robert D., Naohiro Ogawa, and Rikiya
   Research.                                            Matsukura. 2001. “Late Marriage and Less
 Ogawa, Naohiro and Robert L. Clark. 1995.              Marriage in Japan.” Population and Development
   “Earnings Patterns of Japanese Women:                Review 27:65–102.
   1976–1988.” Economic Development and Cultural      Santow, Gigi and Michael Bracher. 1994. “Change
   Change 43:293–313.                                   and Continuity in the Formation of First Marital
 Ogawa, Naohiro and John F. Ermisch. 1996. “Family      Unions in Australia.” Population Studies
   Structure, Home Time Demands, and the                48:475–96.
   Employment Patterns of Japanese Married            Schoen, Robert. 1988. Modeling Multigroup
   Women.” Journal of Labor Economics                   Populations. New York: Plenum.
   14:677–702.                                        Schoen, Robert and John Baj. 1985. “The Impact of
 –                                 – –                  the Marriage Squeeze in Five Western Countries.”
 Oishi, Akiko S. 2004. Jakunen Shugyo to Oya to noby Ingenta to :
                                          Delivered
     –                                                  Sociology and Social Research 70:8–18.
                                              Living
   Do kyobetsu [Employment and Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
Kyushu University Central Library; Mr                 Schoen, Robert and James R. Kluegel. 1988. “The
     2606), Kyushu University Central People].
   Arrangements of Unmarried Japanese Library (2); Zasshi (cid 52024142), Kyushu University
         –                –                             Widening Gap in Black and White Marriage Rates:
   Jinko Mondai Kenkyu [Journal of Population
    Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
                                                        The Impact of Population Composition and
   Problems] 60(2):19–31. Kyushu University Library (cid 43009622)
                                                        Differential Marriage Propensities.” American
                                         Tue, 24 Jan
 Ono, Hiromi. 2003. “Women’s Economic Standing, 2006 03:55:30
                                                        Sociological Review 53:895–907.
   Marriage Timing, and Cross-National Contexts
                                                      Shelton, Beth A. and Daphne John. 1996. “The
   of Gender.” Journal of Marriage and Family
                                                        Division of Household Labor.” Annual Review of
   65:275–86.                                           Sociology 22:299–322.
 Oppenheimer, Valerie K. 1988. “A Theory of           Shirahase, Sawako. 2003. “Wives’ Economic
   Marriage Timing.” American Journal of Sociology      Contribution to the Household Income in Japan
   94:563–91.                                           with Cross-National Perspective.” Luxembourg
 ———. 1997. “Women’s Employment and the Gains           Income Study Working Paper Series, no. 349.
   to Marriage: The Specialization and Trading Model    Syracuse University, Syracuse, NY.
   of Marriage.” Annual Review of Sociology           Sørensen, Annemette and Sara McLanahan. 1987.
   23:431–53.                                           “Married Women’s Economic Dependency,
 Parsons, Talcott. 1949. “The Social Structure of the   1940–1980.” American Journal of Sociology
   Family.” Pp. 173–201 in The Family: Its Function     93:659–87.
   and Destiny, edited by R. N. Anshen. New York:                                            –
                                                      Suzuki, Tohru. 1991. Nihon no Tsu konken (2)
   Harper.                                              Shakaiteki Tsu  – konken [Social Intermarriages in
 Pinelli, Antonella and Alessandra De Rose. 1995.                       –                  –
                                                        Japan]. Jinko Mondai Kenkyu [Journal of
   “Italy.” Pp.174–90 in The New Role of Women:         Population Problems] 46(4):14–31.
   Family Formation in Modern Societies, edited by    Sweeney, Megan M. 2002. “Two Decades of Family
   H.-P. Blossfeld. Boulder, CO: Westview.              Change: The Shifting Economic Foundations of
 Preston, Samuel H., Patrick Heuveline, and Michel      Marriage.” American Sociological Review
   Guillot. 2001. Demography: Measuring and             67:132–47.
   Modeling Population Processes. Oxford, England:    Sweeney, Megan M. and Maria M. Cancian. 2004.
   Blackwell.                                           “The Changing Importance of White Women’s
 Qian, Zhenchao and Samuel Preston. 1993. “Changes      Economic Prospects for Assortative Mating.”
   in American Marriage: 1972–1987.” American           Journal of Marriage and Family 66:1015–28.
   Sociological Review 58:482–95.                     Thornton, Arland D., William G. Axinn, and Jay D.
 Raley, R. Kelly. 2000. “Recent Trends and              Teachman. 1995. “The Influence of School
   Differentials in Marriage and Cohabitation: The      Enrollment and Accumulation on Cohabitation
 822 —– AMERICAN SOCIOLOGICAL REVIEW

   and Marriage in Early Adulthood.” American         United Nations. 2000. The World’s Women 2000.
   Sociological Review 60:762–74.                       New York: United Nations.
 Tsuya, Noriko O., Larry Bumpass, Minja Kim Choe,     Wilson, William J. and Kathryn Neckerman. 1987.
   and Ronald Rindfuss. 2005. “Is the Gender            “Poverty and Family Structure.” Pp. 63–106 in
   Division of Labor Changing in Japan?” Asian          The Truly Disadvantaged, edited by W. J. Wilson.
                                                        Chicago, IL: University of Chicago Press.
   Population Studies 1:47-67.
                                                      Xie, Yu, James M. Raymo, Kimberly Goyette, and
 Tsuya, Noriko O. and Karen O. Mason. 1995.
                                                        Arland Thornton. 2003. “Economic Potential and
   “Changing Gender Roles and Below-Replacement         Entry into Mar riage and Cohabitation.”
   Fertility in Japan.” Pp.139–67 in Gender and         Demography 40:351–67.
   Family Change in Industrialized Countries, edit-   Yamada, Masahiro. 1996. Kekkon no Shakaigaku
   ed by K. O. Mason and A.-M. Jensen. Oxford,          [The Sociology of Marriage]. Tokyo, Japan:
   England: Clarendon.                                  Maruzen.




                                   Delivered by Ingenta to :
Kyushu University Central Library; Mr Beppu (cid 35001789), KYUSHU SHIKA DAIGAKU (cid
    2606), Kyushu University Central Library (2); Zasshi (cid 52024142), Kyushu University
   Ropponmatsu Library (cid 46003719), Kyushu University Medical Library (cid 46009061),
                          Kyushu University Library (cid 43009622)
                                   Tue, 24 Jan 2006 03:55:30

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:10/24/2012
language:English
pages:22
tongxiamy tongxiamy http://
About