Order Amidst Change: Work and Family Trajectories in Japan

Abstract

Substantial family and work macro-level change has been occurring in Japan. Examples include a decline in the availability of jobs that afford lifetime protection against unemployment, an increase in jobs that do not carry benefits such as a pension, an increase in age at marriage and at first birth, and an increase in marital dissolution. Using life history data from the 2000 National Survey on Family and Economic Conditions, young Japanese appear to have responded to these macro-level changes in a fairly orderly manner. Marriage and childbearing have been postponed, but marriage still precedes childbearing. Education is completed prior to starting work. For men, once work commences they continue working. For women, the classic conflict between work and family roles is evident. For men and women in both the family and work spheres Japanese young adults have more orderly life course trajectories than American young adults.

Introduction

Japan, along with all other countries that were both non-communist and industrialized by 1960, has experienced major change in both the work and family spheres. Painting with a broad brush, changes in the work sphere have included an increase in jobs that require more skills with a concomitant increase in educational attainment of young adults, an outsourcing of high-pay low-skill manufacturing jobs to countries with lower wage levels, an increasing globalization of economies, a reduction in jobs that have an explicit or implicit promise of lifetime employment, an increase in service jobs, an increase in employment opportunities for women including career-type jobs along with an increase in the proportion of women in the paid labor force, and a reduction in the likelihood that a parent will posses job-specific skills and wisdom that can be passed on to sons and daughters. While the details vary, all of these changes can be found in Japan, Western Europe, North America, Australia and New Zealand.

In the family sphere, the changes, sometimes termed the “second demographic transition” (van de Kaa 1987; Lesthaeghe 1995), have included, increasing participation of women in the paid labor force, delayed marriage and parenthood, increasing levels of non-marriage or never becoming a parent, increasing levels of non-marital fertility, cohabitation, divorce and single parent families, and increased use of child care services to minimize the strain between mother and worker roles.

These changes have been widely discussed (e.g. Bumpass 1990; Caldwell 2004; Inglehart 1977, 1990; Lesthaeghe and Moors 2000; van de Kaa 1987, 2001). Their implications are profound for both the family and the economy. For example, because of increases in unmarried childbearing and divorce in the U.S. a higher proportion of children spend time in single-parent families, and there is now widespread agreement that children in a single-parent family are at a disadvantage relative to their peers in two parent families (Brown 2004; DeLeire and Kalil 2002; Hill, Yeung and Duncan 2001; McLanahan and Sandefur 1994), though interpretations differ concerning the magnitude and mechanisms of the effects. Or to take another example, low levels of fertility, related to both timing and number changes (Bongaarts and Feeney 1998; Billiari and Kohler 2004; Frejke and Sardon 2004; Lesthaeghe and Wilson 1999), as well as changing marriage propensities, have led to population aging (Dang, Antolin and Oxley 2001; Keilman 1997; Lee and Miller 2002; Lutz, O'Neill and Scherbov 2003; National Research Council 2000; Tuljapurkar, Li and Boe 2000), which, in turn, threatens old-age support systems and public health care expenditures that are based on intergenerational transfers (Bongaarts 2004; Gruber and Wise 1999; Lee, Mason, and Miller 2003; McDonald and Kippen 2001). Low levels of fertility combined with migration restrictions have led to the onset of population decline in Japan and imminent decline in several European countries (Sardon 2004), but opinions differ with respect to whether this is a problem or a benefit (e.g. Morgan 2003).

These broad work and family changes have affected the life course of young adults: expanding the time spent in school, delaying entry into full-time employment, delaying marriage and parenthood. The changes have also increased the opportunity for changes in the order in which events occur in the life courses of young men and women. In some countries, for example, many young adults are now in school at ages when marriages and childbearing traditionally occurred, changes in economic opportunities allow young women to become mothers before becoming wives, and returning to school for additional education has become commonplace.

Rooted in socio-cultural theories of age and social relations (Elder 1998; Elder, Johnson and Crosnoe 2004; Ryder 1965), the life course perspective refers to a sequence of socially defined, age-graded roles that the individual occupies over time. A set of components of the life course framework revolve around the timing of roles and role transitions, as well as the order in which they occur. Most societies have expectations about the timing of the adoption of such adult roles as worker, spouse and parent, and hence the timing of role transitions can be thought of as early, on time, or late (Elder 1998; Riley, Johnson and Foner 1972; Rindfuss 1991; Settersten 2004).

This paper examines the timing and patterning of young adult roles for one country, Japan, and makes comparisons with existing work on the United States, and to a lesser extent Norway. In the next section we describe why Japan is of interest in terms of the changing work and family situation. This description is at the macro level, using published, aggregate statistics. In the micro-level empirical work in this paper, the timing of work and family transitions is examined, followed by a demonstration of the diversity of young adult trajectories. We then show that the life course trajectories of young men and women tend to be orderly, that is, they exhibit a pattern that follows normative expectations. To preview our results, we find the work and family life-course trajectories of young adult Japanese to be very orderly, with “order” being defined by traditional expectations in the work and family spheres, and much more so than in either the U.S. or Norway.

Japan

Japan differs strikingly from many Western post-industrial societies. Consider the work sphere first. Schools play a more important role in obtaining a job (Inui 2003; Ishida 1998; Ishida, Spilerman and Su 1997). For most of the past 40 years, the “new graduate recruitment system” has been the principal mechanism whereby young Japanese men and women obtain a job. Under this system, schools (both high schools and universities) act as middle-men in the recruitment process for “regular” jobs, that is jobs that offer fringe benefits and fall under the lifetime employment model. If a young person finishes school in a year when economic conditions are such that there are few full-time job openings, he or she might be forced to take a job without benefits, security and job advancement possibilities. This can have long-term implications because when the economy rebounds employers will prefer hiring new graduates.

The nature of the Japanese economy and the types of jobs available to young people has been changing. Growth slowed markedly in the 1990s largely due to the after-effects of over-investment during the 1980s (popularly called the "bubble economy") and domestic policies intended to wring speculative excesses from the stock and real estate markets (Cargill, Hutchison and Ito 1997: 91–116). Initially, government efforts to revive economic growth were met with little success and were further hampered by the slowing of the global economy in 2001–2002 (Cotis 2003; Yoshitomi 2004). Beginning 2003, the Japanese economy showed signs of a strong recovery until the recent worldwide economic crisis triggered by problems in the U.S. financial system.

Concomitant with rapid economic growth in postwar Japan, educational levels and paid employment (as opposed to employment in family enterprises) have increased dramatically, along with the educational requirements of most jobs. Such gains have been especially dramatic for women: the proportion of women age 25–29 with more than senior high school education increased from 10 percent in 1970 to 49 percent in 2000 (National Institute of Population and Social Security Research 2005: 149). During the same period, the labor force participation rate of women age 25–29 increased from 45% to 70% (National Institute of Population and Social Security Research 2005: 137).

In an attempt to increase their competitiveness and profitability, employers have begun to move away from the lifetime employment model (Boling 2008). This resulted in the rise of employees working in temporary positions as keiyaku (those hired under a fixed-term contract with limited provisions of social insurance and other fringe benefits) and haken (contract work whereby an agency sends workers to a corporation for a specified period) (Statistics Bureau 2001). These non-regular workers tend to be young: 61% of those employed as keiyaku or haken in 2005 were aged 15–34 (Statistics Bureau 2005).

Another phenomenon in the Japanese labor market is an increase of so-called “freeta,” young people (age 15–34) who do not hold a stable job and hop from one short-term temporary job to another (Statistics Bureau 2005). The proliferation of temporary employment among young adults not only harms their job prospects, but also poses profound implications for marriage and family building (Oppenheimer 1994; Oppenheimer, Kalmijn and Lim 1997; Tsuya 2009).

A final, related but different, trend has been an increase in young adults who are neither in school nor in the labor force. Borrowing a term that originated in the United Kingdom, the Japanese government began measuring the number of “NEETs” (Not in Education, Employment or Training) in 2000. The number of NEET, defined as those aged 15–34 who are not employed, not in school, and never married is increasing rapidly. Based on the population censuses, it is estimated to have increased 2.6 times from around 290,000 in 1995 to 750,000 in 2000, constituting a little over 2% of the population of that age group. The number is projected to rise to around 1.2 million by 2020 (Kadokura 2004).

It is important to recognize that Japan shares a Confucian familial-cultural heritage with other countries in East Asia, such as South Korea, China and Taiwan that have been undergoing rapid industrialization in recent decades (e.g. Brinton and Lee 2001; Chang and Lee 2006; Goldstein and Feng 1996; Mason, Tsuya, and Choe 1998; Peng 1991; Thornton and Lin 1994; Tien 1984). A significant component of the Confucian influence is the emphasis on filial piety and the importance of the patrilineal line (Aruga 1943; Otake 1982; Tsuya et al. 2004). The traditional Japanese family system, called ie, dating from at least the Meiji era (1868–1911), is based on patrilineal descent, patriarchal authority, and patrilocal residence (Fukutake 1989; Ishihara 1981). The government's system for registering legal residence (koseki) keeps track of lineages, and in Japan's birth registration system children born to unmarried mothers are still officially referred to as “illegitimate children” (hi-chakushutsu-shi), a stigma with negative consequences for children. Both this tradition and the koseki system likely influence the probability of non-marital fertility.

With respect to marriage, it is helpful to think of post-war marriage as a “package” to highlight the culturally defined linkage of marriage, childbearing, childrearing and frequently care of elderly parents of the husband (Rindfuss 2004; also see Coleman 1983), in what was expected to be a life long commitment. Further, the burden of household tasks falls overwhelmingly, and some would say “exclusively,” on the wife (Ishii-Kuntz and Maryanski 2003; Tsuya et al. 2005). With increases in education and employment opportunities for young women, the attractiveness of this marriage package has likely declined because upon marriage, traditionally, wives were expected to cease their employment and were to devote themselves full-time to household tasks. The very dramatic decline in marriages in the young adult years reflects the relative attractiveness of employment opportunities, the concomitant unattractiveness of the traditional marriage package, and, as we discuss next, the advantages of remaining single (Raymo and Iwasawa 2008). For example, among women aged 25–29 in 1975, only 21 percent were never married; by 2005, the comparable figure was 58 percent (National Institute of Population and Social Security Research 2007a: 109).

Among never-married young adults, co-residence with parents is very common. In 1994, 82 percent of single men and women aged 20–28 lived with their parents (Tsuya, Mason, and Bumpass 2004); the corresponding proportion was 86 percent in 2004. Since young men and women are often not expected to financially assist the household even when they are employed, these unmarried young adults tend to have substantial disposable income to use for travel, shopping or whatever else they desire, which also likely decreases the attractiveness of the traditional marriage package.

Also indicative of a questioning of the marriage package is evidence that divorce probabilities have been increasing. They are now at implied cumulative levels of about 30% within 20 years of marriage (Raymo, Iwasawa and Bumpass 2004).

Unlike the pattern found in Western countries when age at marriage increased, Japan has not witnessed a large increase in nonmarital fertility. Nonmarital births account for approximately 2 percent of all births (National Institute of Population and Social Security Research 2007b: 67), which is far lower than any Western European nation (Sardon 2004). With an increase in age at marriage and no substantial increase in nonmarital fertility, levels of period fertility have declined substantially. Japan experienced a dramatic decline in fertility shortly after World War II. In one decade the TFR dropped from 4.5 to 2.0. It was then stable in the 2.0 to 2.2 range for nearly two decades. In 1975 it began to decline again, reaching a low of 1.26 in 2005 (Statistics and Information Department 2006). Further, the proportion of couples who had their first child by the end of their second year of marriage has been decreasing since the mid-1970s, indicating a weakening of the tendency of couples to have their first child soon after marriage (National Institute of Population and Social Security Research 2006: 34).

To summarize, in Japan, there has been considerable macro change in the educational and employment environments, with young men and women staying in school longer, starting their careers later and facing structural changes in the workplace, along with a pattern of delayed marriage and childbearing in the family sphere. In the face of such macro change, how are young women and men constructing their life courses, within the work sphere, the family sphere, and across the two? To answer this we turn to micro-level data.

Data and analytic issues

The micro-level data for this paper are from the National Survey on Family and Economic Conditions (NSFEC) in Japan which was conducted in November 2000. The NSFEC is a national, two-stage stratified probability sample of Japanese men and women aged 20–49. First, three hundred fifty locales were randomly selected based on 2000 census data, and then individuals aged 20–49 were randomly selected within each locale based on the basic registration of its residents. Because the general objective of the survey was to collect information on role transitions over the early adult life course, individuals aged 20–39 were selected at twice the rate of those aged 40–49. (Further details on the data set can be found in Rindfuss et al. 2004.)

Life history data was collected for a number of domains, including education, employment, marriage and childbearing. The fertility data was collected via standard questions that asked if the respondent had any children, and if yes, their dates of birth. Unfortunately, only ever-married women were asked the fertility questions. This means that there may be some respondents in our data set who have never married and have had one or more children. Since one of the issues we address is the order of events in the family sphere (marriage and childbearing), this means that we are underestimating non-marital childbearing. How much might this affect our results? We expect the answer is “a little, but not enough to influence our overall conclusions.” Based on Japanese vital registration data, non-marital childbearing is exceedingly rare relative to other low fertility countries; since the early 1950s, only about 1–2 percent of all births have been non-marital (National Institute of Population and Social Security Research 2006: 67). Ever-married women could report non-marital births on the NSFEC, and they do so at a level one would expect given the low proportion of non-marital births in Japan. So, in interpreting our results it will be important to remember that non-marital births are underreported, but the magnitude of underreporting is unlikely to affect our broad conclusions.

For the other domains, data comes from a table that the respondents filled out. The rows in the table were activities such as going to school, working full time, and so forth. The columns were calendar years, beginning with 1980 and ending with 2000. For each activity they were instructed to “circle the year the activity began, and the year it ended. Then draw a line connecting the two circles. Do this for as many times as you started and ended the indicated activity. If you stopped one activity for a while and started it again in the same year, draw two circles in the year. Leave the row blank if you have not done the activity or behavior.” Our pretests indicated that respondents had no trouble understanding the instructions and filling out the matrix, and experience in the actual survey confirmed this.

The data available for analysis are illustrated in Chart 1, where the columns indicate birth cohort (or age in 2000) and the rows represent activity years from 2000 back to 1980. The entries in Chart 1 show the age of a respondent in a particular activity year given their birth cohort. We are interested in the sequences or trajectories of roles occupied during the young adult years. It can be seen that if we were to look at ages 20 through 40, there is only one cohort (1960) in our data set that spans the entire 21 years. On the other hand, if we wanted to examine just the sequences of roles occupied between age 25 and 29, a span of 5 years, we would have complete information for cohorts 1955 to 1971, 17 cohorts. We are interested in the young adult years, which are demographically dense when people are making the transition from student to worker and from family of orientation to family of procreation. Balancing our substantive interests and our methodological preferences for sample sizes as large as possible, we look at the experience of two different groupings of cohorts (1960–1970 and 1955–1965). While largely overlapping, these two groupings allow the observation of different segments of the young adult life course: 20–30 and 25–35 respectively. This experience can be seen in the shaded portions of the top and bottom panels of Chart 1.

Chart 1.

Cohorts, periods and ages included in the analyses

Two aspects of the data warrant further discussion. First, with such a limited set of cohorts we cannot carefully examine cohort change; we simply do not have data for enough cohorts for long enough portions of their life course. Hence the “change” in the title of this paper (Order admidst change) does not refer to cohort change; rather the “change” refers to the macro-level, transformational changes in the work sphere that have occurred in Japan as discussed above. These are changes that, other things being equal, would be expected to influence life course event, timing and sequence. So what we examine is the behavior of individuals as they live their lives during a period of rapid change at the macro level. In future work we plan to explicitly examine cohort change.

Second, it is important to note that our data are yearly rather than monthly², and this has both advantages and disadvantages. The principal advantage is that it allows us to “see” the main features of young men and women without being distracted by short-term fluctuations. And these main features are probably the ones best remembered by respondents. The disadvantage is that yearly data probably produces more orderly life course trajectories, and this will need to be remembered in interpreting our results – a topic to which we return below.

Timing of events

We begin by examining the timing of events: when do young adults finish school, first marry and first become a parent. Table 1, based on life table estimates, shows the cumulative proportion having left school, married and a parent for ages 18, 21, 25, and 30. Overall, people leave school at a relatively early age while marrying and becoming parents at a relatively late age, leaving, on average, a relatively large number of years between the start of adult employment commitments and the start of adult family responsibilities. Put differently, young Japanese men and women have many years of earning a salary or wage before they take on spouse and family responsibilities. This reduces the demographic density of the young adult years, and also is related to the relatively low level of fertility that currently exists in Japan.

Table 1.

Cumulative proportion who have left school, married and became a parent, at ages 18, 21, 25 and 30, by gender

	left school		married		became a parent
Age	male	female	male	female	male	female
18	7	3	0	1	1	0
21	59	74	3	6	2	3
25	95	99	24	47	10	21
30	100	100	53	73	37	56

Women finish school, marry and become a parent at younger ages than men. For example, at age 21, three-quarters of the women had left school compared to approximately three-fifths of the men. Given that education has increased more rapidly for women than men in recent cohorts, this gender difference in school-leaving age has probably declined in more recent cohorts. By age 25, almost half of the women had married compared to a quarter of the men. And by age 30, 56 percent of the women had become a parent compared to 37 percent of the men.

Trajectories

The life table results in Table 1 provide an opportunity to examine the timing of events, but they do not provide a clear overview of the life course experience of young adults, and to better understand the flexibility or rigidity of the young adult life course one needs a more holistic approach. Put differently, the life table approaches, as well as its multivariate variations such as event history analysis, allow one to focus on an event (e.g. a birth, a job start) and the reoccurrence of that event, but even with strong assumptions about causality they do not show how the event in question relates to other events in the person's life. (For discussion of the issues and some methods to overcome limitations of the life table approach see: Billari and Piccarreta 2005; Taris and Feij 1999; Wu 2000.) Our approach to obtain an overview or holistic understanding of the life course of young Japanese men and women is what Billari and Piccarreta (2005) term “frequency distributions of sequences.” We first divide events and states into the family sphere (marriage and parenthood) and the work sphere (education and employment). Consider the family sphere first, and the eleven years of the life course between ages 20 and 30 (panel A of Chart 1).

In the family sphere, we have three marital statuses, never married, married and post-married (separated, divorced or widowed) and two parenthood statuses, not a parent and parent³. Cross-classifying these two statuses results in six possibilities:

1. never married, not a parent

2. never married, parent

3. married, not a parent

4. married, parent

5. post-married, not a parent

6. post-married, parent.

In each of the eleven years, a person is classified into one of these six possibilities. Stringing together the eleven classifications results in a sequence or trajectory that captures, with a yearly rather than a monthly or daily brush, that person's family experience during the time they were 20 to 30. So, for example, a person who remained unmarried and not a parent throughout the eleven years would have this sequence:

$$11111111111$$ (1)

Or to take another example, a person who married when they were 23, became a parent when they were 27, and stayed married would have the following sequence:

$$11133334444$$ (2)

These two examples represent very different experiences. Someone who married at age 24, became a parent at age 27 and stayed married would have the following sequence:

$$11113334444$$ (3)

Sequences 2 and 3 are different, but quite similar to one another. In the next section we examine one aspect of similarity, but it is important to recognize that there are numerous other aspects of similarity that we do not examine.

In the work sphere, we have in school or not, employed part-time or not, and employed full-time or not. Since these can be combined in every possible way (for example simultaneously working a full-time and a part-time job), we have eight categories:

1. just in school

2. in school and employed part-time

3. in school and employed full-time

4. in school, employed part-time, and employed full-time

5. not in school and not employed

6. not in school and employed part-time

7. not in school and employed full-time

8. not in school, employed part-time, and employed full-time

We begin our analyses of these work and family trajectories by asking a few simple logical questions which are contained in the column headings in Table 2. The first thing that is evident is that there are a relatively large number of unique sequences. The average number of respondents per trajectory ranges from 3 to 14. For every possible comparison, in the family sphere and the work sphere, at ages 20–30 and 25–35, there are more sequences for women than men. Part of the reason is that there are more women in the sample than men, itself a function of higher response rates for women. But differences in number of respondents are only a small piece of the explanation. The bigger issue, to which we return below, is the gender differences in the way that work and family spheres impinge on one another, an indication of which is the very large disparity in the number of sequences in the work sphere for men and women.

Table 2.

Aspects of life course trajectories by age component of the life course, family or work sphere, and gender

	Number of respondents	Number of unique trajectories	Number of respondents per trajectory	Number of one person trajectories	Percent of trajectories with one person
20–30
Family sphere
Total	1,835	127	14	60a	34%
Male	857	78	11	31	40
Female	978	105	9	45	43
Work sphere
Total	1,854	450	4	338a	75
Male	866	120	7	92	77
Female	988	371	3	287	77
25–35
Family sphere
Total	1,508	133	11	62a	47
Male	692	94	7	34	36
Female	816	105	8	43	41
Work sphere
Total	1,528	260	6	191a	73
Male	702	51	14	38	74
Female	826	230	4	168	73

^aNote: the total is less than the sum of the male and female numbers because some trajectories have one male and one female occupant so that when males and females are combined there are two people in the trajectory.

For every paired comparison in Table 2 except one, there is a greater diversity of sequences in the work than family sphere. This is partly a function of more categories in the work sphere, but that is not the complete story. It is also partly a function of delayed marriage and childbearing past age 30 and 35, but, again, that is not the complete story. For women, again, the likely explanation is balancing work and family roles.

Table 2 also shows the number of sequences that are occupied by only one person. Similar patterns are seen. Women have more one-person sequences than men, and the work sphere has more one-person sequences than the family sphere. In the work sphere, approximately three-quarters of the sequences are occupied by one person.

Chart 2 shows the trajectories with the highest percentage of respondents for each gender and age combination. In the work sphere, employed the entire 11 years is most common for both ages and genders, but the percentage is substantially higher for men than for women, with the difference especially striking for the 25–35 age group. Also for women for ages 25–35, almost as many are not in school and not employed the entire 11 years, again reflective of the conflict between work and family roles for women.

Chart 2.

Most common trajectory by age component of the life course, family or work sphere, and gender

How do these patterns compare to experience in other countries? The only comparison of which we are aware is for the United States using data from the National Longitudinal Study of the High School Class of 1972 (Rindfuss 1991). The ages covered were 18–29, and there were three work categories: school, employed and other, with the analysis forcing respondents to be in just one of these three possibilities in any year. So the comparison involves numerous differences. Given these differences, it is interesting that the most common sequence is being employed all 12 years and that about three quarters of the work sequences have but one person in them – results identical to those in Chart 2.

Order and disorder

Parents, teachers, employers, and society in general have expectations about desired or preferred employment and family trajectories for young adults. These expectations partly stem from concerns that some trajectories might have negative consequences for the young man or woman. An example might be a young unmarried person who has stopped their education but is not employed, either full- or part-time. The parent would worry about the implications for the person's career and perhaps his or her marriage prospects. The teacher might worry whether the person is living up to his or her potential. A prospective employer might worry about the work habits of such a person. Expectations also stem from everyday observations of friends, relatives and acquaintances who successfully navigate the young adult years. Expectations can also be influenced by various media reports.

The various expectations of individuals in a society combine at the macro level to form norms about preferred young adult sequences. We recognize that norms about preferred sequences are notoriously difficult to define and measure (Hogan and Astone 1986; Neugarten and Datan 1973; Settersten 2004), and we do not attempt to do so here. Instead, we present disaggregated categories.

We use the term “order” to describe those sequences that follow the expected or normative path, and “disorder” for those that do not. Unfortunately, in everyday use, “disorder” has a judgmental connotation. We are not using it in a judgmental fashion, but simply refer to the extent to which the sequence is the expected path from the perspective of either work or family spheres. Indeed, as we discuss below, for women, what might be described as a disorderly work sequence from the perspective of the work sphere might also be judged as a preferred sequence from other perspectives.

In the family sphere, the normative sequence is to become a spouse before becoming a parent. The reverse is disorderly. In Japan, marriage has long been expected to be a life-long commitment, for the sake of the marriage and for the children. We have three categories of disorder or possible disorder: non-marital birth, a marital dissolution that occurs prior to becoming a parent, and a marital dissolution that occurs after becoming a parent. If someone has both a premarital birth and then a marriage followed by a marital dissolution (that is, two forms of disorder), they are left in the non-marital birth category because the traditional normative expectations against non-marital births are thought to be stronger and more strict than against marital dissolution in contemporary Japan.

From the perspective of the work sphere, it is normative to finish school, start working, and work throughout the young adult years. Spending time neither working nor in school would be considered disorderly. Having an interrupted educational career would generally be considered disorderly. But taking some type of advanced schooling while simultaneously holding a job might be considered orderly, especially if the additional schooling is at the request of the employer. Thus, we distinguish between two types of educational breaks. The years between a break in education might also be years in which the person is not working; in this case not-working takes precedence in classifying the sequence.

“Disorder” here is defined in terms of the work sphere, and the logic is that sequences which might have a negative impact on a person's occupational or income trajectories, or make it more difficult for the person to find the type of employment they desire are considered “disorderly.” For example, dropping out of the labor force for a few years and then returning tends to result in penalties in terms of wage rates and promotion to higher level jobs. Indeed, this is a major finding and issue in the gender stratification literature in economics and sociology (e.g. England and Folbre 2005). Or to take another example, in Japan schools play an important role in placing individuals in jobs (Inui 2003; Ishida 1998; Ishida, Spilerman and Su 1997; Kerckhoff 1995). Schools frequently act as go-betweens in the recruitment process for “regular jobs” – jobs that are considered full-time, offer fringe benefits and fall under the lifetime employment model. Schools help in getting a first job, but if someone drops out of the labor market and then wants to return, schools will be unlikely to help and the person returning will likely have a more difficult time obtaining the job they desire.

For men, from almost any perspective, labeling dropping out of the labor force and then returning as a “disorderly” young adult sequence would seem reasonable. For Japanese women, however, from the perspective of her spouse, her parents, her other relatives, or her friends, dropping out of the paid labor force when she has young children probably would be considered the expected, preferred or “normal” sequence. This terminological issue with respect to a common employment sequence for Japanese women goes to the heart of the difficulty Japanese women have in balancing work and family expectations/responsibilities. This is an issue to which we return below. For now, remember that we distinguish between orderly and disorderly work sequences from the perspective of the work sphere.

Table 3 shows, for men and women, the extent of order or disorder in the family sphere for both the 20–30 and 25–35 age components of the life course. The most striking aspect of Table 3 is the dominance of order in family sequences. In every age and gender group it is over 90 percent. This is a function of low rates of non-marital fertility and divorce. It is also increasingly a function of later ages at marriage and parenthood. If you are not married and not a parent, by our definition it is an orderly sequence, but remember that never marrying and never becoming a parent would be frowned upon by traditional family standards.

Table 3.

Order or disorder in the family sphere by age component of the life course and gender

	20–30			25–35
	Total	Male	Female	Total	Male	Female
Order	96	97	95	94	96	92
Disorder—non-marital birth	2	2	2	2	2	2
Possible disorder—post-married, not a parent	1	0	1	1	1	1
Possible disorder—post-married and a parent	2	1	2	3	2	5
Totala	100	100	100	100	100	100
N	1,835	857	978	1,508	692	816

^aTotals may not add to 100 because of rounding error.

How might this compare to other countries? Again, we do not have data sources that provide an exact comparison, but for women we can compare our results to results from Upchurch and her co-authors (2002) with respect to the order of leaving school, marrying and becoming a parent. The data sets, time period and restrictions are not strictly comparable, but close enough to permit a coarse comparison. They use data from the National Longitudinal Survey of Youth (NLSY). They use women who were aged 26–33 in 1991, examining their behavior from the teen years until 1991. The NLSY data are monthly rather than annual. Monthly data are more likely to reveal disorderly sequences than annual data, and so we would not want to interpret small differences between NLSY and NSFEC. The contrast between the United States and Japan in Table 4 is stark, and very unlikely to be entirely the result of differences in the nature of the two data sets. Among Japanese women, 99 percent are in the three categories that might be considered orderly sequences involving school, marriage and parenthood: 1) stop school, marry and have a child, 2) stop school, marry, but not yet a parent, and 3) stop school, not yet married and not yet a parent. The comparable number for the United States is 61 percent. We also note that the total fertility rate in the United States has been hovering around the replacement level while Japan's has only been about two-thirds of the replacement level. As we discuss below, we expect that the family sequencing flexibility in the United States facilitates higher levels of period fertility, allowing women to have children at younger ages.

Table 4.

Life course pattern of stopping school, marrying and becoming a parent, Japan and the United States, females

Order of Events	United Statesa	Japan
Stop School, First Baby, Marry	4	1
Stop School, First Baby, Never Marry	4	0
Stop School, Marry, First Baby	30	47
Stop School, Marry, No First Baby	12	11
Stop School, Never Marry, No First Baby	19	41
First Baby, Stop School, Marry	2	0
First Baby, Stop School, Never Marry	3	0
First Baby, Marry, Stop School	3	0
Marry, Stop School, First Baby	7	0
Marry, Stop School, No First Baby	6	0
Marry, First Baby, Stop School	11	0
Totalb	100	100
N	5,825	2,299

^aThe United States numbers are from Upchurch, Lillard and Parris (2002).

^bTotals may not sum to 100 because of rounding errors.

Order or disorder in the work sphere is shown in Table 5. There are two distinctive features of Table 5. First, breaks in schooling, that is, leaving school, doing something else for a while and then returning, are exceedingly rare. This reflects the tracked and ordered nature of the Japanese educational system, and the importance of entrance exams (Brinton 1990, 1992; Tsuya and Choe 2004). Once one leaves school, it is very difficult to be readmitted. In this way, the Japanese educational system is similar to Germany's and contrasts with the open system found in the United States and Norway, to pick two examples (UNESCO 1996; Wickremasinghe 1992).

Table 5.

Order or disorder in the work sphere by age component of the life course and gender

	20–30			25–35
	Total	Male	Female	Total	Male	Female
Order	64	91	40	56	91	27
Not in school or working	36	9	59	43	9	73
School break—not working while back in school	0	0	0	0	0	0
School break—working while back in school	1	0	1	1	0	1
Totala	100	100	100	100	100	100
N	1,854	866	988	1,528	702	826

^aTotals may not sum to 100 because of rounding error.

The second distinctive feature of the table is the gender difference, which is large for the 20–30 age segment, and enormous in the 25–35 age segment. As noted above, when defining “order” and “disorder,” the definition is from the perspective of the work sphere. The reasons for the gender differences, of course, involve the mother and wife roles. Indeed it can be argued that it is perfectly normative in the Japanese context of the last 20 years or more for women to drop out of the labor force when they marry and start having children, and that this is the expected sequence for adult women. But the view is quite different from the perspective of the work sphere. As the sex role literature noted long ago (e.g. Corcoran and Duncan 1979; Polachek 1975; Weiss and Gronau 1981; Wolf and Rosenfeld 1978) the labor market punishes, in both wages and job advancement, those who drop out of the labor market even for a short period.

We also note that 82 percent of the men in the 25–35 age segment were working full-time during the entire 11-year period (not shown in Table 5). Depending on the birth cohort, this spans the years 1980–2000. During at least 10 of those years Japan experienced severe economic slump, and yet more than four-fifths of the young men were steadily employed the entire period⁴.

In Table 5, nine percent of the men in both the 20–30 and 25–35 age groups experienced a spell of being neither employed nor in school. While relatively small compared to women, it is higher than one might expect from examining government cross-sectional figures. We examined what factors available in our data set might distinguish them from the 91 percent with orderly work sphere trajectories. The only consistent factor was education. Specifically, those with the least education, that is they did not finish high school, are the most likely to have experienced a spell of not being in school or in the labor force, and this is not surprising.

Again the question arises as how the patterns in Table 5 might compare to other countries, and again precise comparisons are not possible. However, some comparisons are possible with results from the United States using the National Longitudinal Study of the High School Class of 1972 (Rindfuss 1991). Using a similar definition of order as used here, for the 12 years from age 18 to 29, only 41 percent of the males and 17 percent of the females could be classified as orderly in the work sphere. Clearly Japanese men and women have far more orderly young adult life courses from the perspective of the work sphere.

Another contrast is possible with Norway, a country where it is quite easy to return to school after having dropped out and to do so multiple times. For the Norwegian 1964 birth cohort, Kravdal and Rindfuss (2008) used data from the Norwegian population registers and they report that for those who became parents, 21 percent of the women and 20 percent of the men had a higher level of completed education at age 39 than when they first became a parent – again a radically different picture than we see in Japan.

Discussion and conclusions

Our results show an interesting pattern of ordered young adult lives within a time period when macro structural change has been substantial. The sequences we examined covered the period 1980–2000. This is a period when educational levels were increasing, resulting in later ages at leaving school and starting employment/careers. Labor force participation rates of young women were increasing markedly, and the economy was restructured in a variety of ways. Childbearing was being postponed and total fertility rates steadily declining. Similarly, marriage was also being postponed and never marrying increasing. For example, among men aged 30–34 in 1975, 14 percent were never married, and by 2005 this had reached 47 percent (National Institute of Population and Social Security Research 2007a: 11).

Yet surrounded by this macro change in education and the economy, as well as marriage and childbearing patterns, the life course sequences of young Japanese men and women are very orderly. They finish school and then work for the rest of their young adult years, unless they are a woman and become a wife and mother. They seldom have breaks in their education; once they first leave school, they do not return. They seldom spend a year not in school and not working, simply hanging out. In the family sphere, having a child outside marriage is exceedingly rare, and divorce, while increasing, did not have a major impact on the young adult life course. Instead, the dramatic change in the family sphere has been the orderly postponement of both marriage and employment.

In describing the work and family trajectories of these cohorts as orderly, we need to point out that there is the possibility that the extent of orderliness might be overstated. Our life history data are yearly rather than monthly, and retrospective rather than from a prospective panel design. So short spells of unemployment, for example, might be missed. Cohabitation information was not collected⁵. In Japan, and especially for these cohorts, cohabitation would be considered disorderly. Yet, even if we had monthly data and information on cohabitation, it is likely that a picture of orderly postponement of both marriage and employment would have emerged, and especially in comparison to what we see in North America and many European countries.

What about change since 2000? There is evidence that cohabitation and divorce have been increasing (Raymo, Iwasawa and Bumpass 2004; Tsuya 2006). Further, the recent world-wide economic downturn has likely affected Japanese young adults. Are we likely to see an increase in disorder in the Japanese life course? This is a question we plan to address with a new wave of data being collected in 2009.

We were able to make some comparisons with the United States and one with Norway. The comparisons were less than perfect in that they covered different ages and calendar years. There were also differences in mode of data collection, sample sizes, and specifics of the available information. Nevertheless, the contrasts are instructive. In both Japan and the United States, the most common sequence is working the entire young adult period. Further, both countries showed considerable diversity in the range of sequences. This is where the similarity between the United States and Japan end. With respect to order and disorder, the two countries are qualitatively different. With the exception of the wife and mother roles interrupting the orderly work trajectories, Japanese experience orderly work and family young adult trajectories. In the U.S., on the other hand, disorderly family trajectories are common, and disorderly work trajectories are the majority experience. And in Norway, unlike Japan, obtaining more education after the birth of a child is common.

It is unfortunate that we do not have comparable indicators and patterns for the Mediterranean countries (Italy, Spain, and Greece). These countries are frequently grouped together with Japan on a variety of factors including fertility patterns (Caldwell and Schindlmayer 2003) and welfare regimes (Esping-Anderson 1999). Or to take another example, Reher (1998) characterizes the Mediterranean countries as having strong family ties – a description that would also be used for Japan. And Italy, like Japan, had low levels of cohabitation and divorce compared to Western Europe and North America, but levels of cohabitation and divorce have been rising in both countries (Castiglioni and Dalla Zuanna 2009). Given some of these broad similarities, it is possible that the trajectory patterns we see in Japan might also be present in the Mediterranean countries.

What might be the implications of the Japanese trajectory patterns for future fertility trends in Japan? Here we can only speculate, but given the trend in Japanese fertility, and its implications, speculation seems justified. Consider several elements in the current situation in Japan. The now traditional marriage package is one in which the wife is expected to quit work, have children, care for the husband's parents especially if he is a first-born male, have essentially total responsibility for child care and household maintenance, and not complain if the husband comes home late at night. This is not an appealing bundle if women have other options, and, given the macro changes in education and labor market, they now have other options. Their education levels have increased (National Institute of Population and Social Security Research 2007a: 175–176). Even if they still experience gender discrimination in the labor market, jobs are available. If they live with their parents, and most do, they have considerable disposable income. Both young men and women have resolved the competing pressures involving a marriage package that does not appeal to women, the normative imperative to marry and have a child, pressure to either be in school or employment (unless one is a mother), and the norm of conformity by postponing marriage and parenthood. Taken together, this would suggest continued low fertility.

Further, as we have argued elsewhere (Rindfuss and Brauner-Otto 2008), other things being equal, the greater the flexibility in the young adult life course, that is, the ease with which such roles as student, worker, spouse and parent can be sequenced in any manner to suit the convenience of the young adult, the higher the level of fertility in a country. Given that the Japanese educational system makes it difficult to drop out of school and then later return, and given that the labor market structure makes it difficult to drop out of the labor market and then return to a “regular” job (full-time, with fringe benefits and falling under the lifetime employment model), the implication is continued levels of relatively low fertility.

Continued postponement, and eventually not marrying or becoming a mother, has also received media attention. For example, a book published in 2003, Makeinu no Tōboe (Howlings of a Loser), by Sakai Junko, argued that childless single women over 30 should be proud of the freedom and financial independence they have as a result of not marrying. The book became a Japanese best seller and was widely discussed, favorably and unfavorably, by other media. While unlikely to persuade anyone, such discussion might ease further postponement, perhaps facilitating a long-term realignment of family and work roles such that a significant fraction of Japanese women are either in school or working their entire childbearing years, without marrying or becoming a mother.

Now let's return to the issue of whether, for women, dropping out of the labor force to become a wife and mother is “disorderly”? The issue is more than simply finding the right term. Rather it goes to the core of the macro level changes that have occurred in industrial and post-industrial societies. The macro, transformational changes in the work sphere discussed at the beginning of this paper have tended to produce for women (but not for men) a divergence between the normative expectations in the education and labor market institutions on the one hand and in the family institution on the other hand. This divergence has been accommodated and worked out differently in various countries. The nature of this divergence is that from the perspective of the education and work institutions, individuals (women and men) are expected to stay in school until they finish their formal education and to work continuously thereafter until retirement. Sanctions for dropping out of school or work tend to involve labor force re-entry difficulties, and then, if a re-entry is successful, sanctions include poorer options, less advancement and diminished rewards. In Japan, these sanctions tend to be larger than in most of Europe and North America. From the perspective of the family, women are expected to marry and then take major responsibility for rearing children, and typically this means dropping out of school or the labor force. Doing so often brings the rewards of praise from parents, other relatives and one's broader social network; and failing to do so can be sanctioned by expressed disapproval by the same social network. While we have no empirical evidence on the point, again in Japan, sanctions for a woman to simultaneously be the mother of a young child and a full-time worker tend to larger than in most of Europe and North America.

The result of the divergence of normative expectations in the labor market and family arena for women is that, within a given society, the work and family spheres can have conflicting norms on whether women should drop out of school and/or the labor force to rear children. Different countries have taken different paths to resolving these conflicting norms, and the resolutions can involve monetary child support, child care services, part-time employment, flexible work hours and the like. In this paper, we have addressed the issue of whether a woman dropping out of school/work is “orderly” or not from the perspective of the work sphere. It is not absolutely necessary to do so – a family sphere perspective could have been used. The important point, though, is that women and men are following very different trajectories in the work sphere in Japan. In the work sphere, women are essentially sanctioned for doing so and in the family sphere they are likely receiving positive feedback. Moving forward, an important question is whether the normative conflicts over education/work trajectories will be resolved, and, if yes, the nature of the resolution. The question is important because its resolution will have substantial implications for Japan's educational, economic and family institutions.