A descriptive analysis of three-generation households and mothers ’ employment in Japan, 2002 – 2019

Purpose – ThisarticleaimstoexplorehowJapanesewomenwithyoungerchildrenchangedtheircommitment to the labour market between 2000 and 2019 by comparing mothers in three-generation and nuclear family households. Design/methodology/approach – Japan currently has the highest ageing rate in the world at nearly 30%. Since the 1990s, employment flexibilization and women ’ s labour market participation have proceeded in parallel, and the conservativefamily values of the patriarchyand gender division oflabour thathave provided intergenerational aid for care within households have been shrinking, by conducting a descriptive analysis of the Labour Force Survey (LFS). Findings – Thisstudyidentifiedthataconspicuousincreaseinpart-timeemploymentamongmothersinboth household types and a decrease by half in the working mother ’ s population in three-generation households. These results suggest that the function of inter-generational assistance by multi-generation cohabitation, which was once thought to be effective in helping working mothers with younger children, is declining. Originality/value – A study examining the transformation of mothers ’ employment behaviour differences between three-generation households and nuclear family households is rare. This paper makes a new contribution to the research regarding the grandparents ’ caregiving, household types and mothers ’ employment.


Introduction
In 2019, Japan had the "highest-high" elderly population percentage (28.4%) in the world, and the annual number of births reached a record low in the last half century. In 2040, the old-age population ratio is estimated to reach 35.3% (MHLW, 2020a). Between 1986 and 2019, the number of households that included the population aged 65 years and over increased 2.6 times, and the number of single-person households that included those aged 65 years and over increased 5.8 times, while the number of three-generation households decreased by nearly half during this period (MHLW, 2020b).
These population changes have a great impact on Japan's socioeconomic structure. Japan currently has the third-largest GDP and the highest debt-to-GDP ratio in the world. Its total net social spending for GDP is 23%, which is not the highest but exceeds the Organisation for Economic Co-operation and Development (OECD) average (20.9%) (2015). Due to this everincreasing elderly population and social security expenditures, Japanese finances and social security systems are facing urgent issues (Ministry of Finance, 2020).
Japan has long been known as a familistic welfare state with a strong male breadwinner and female care responsibility model (Gottfried and O'Reilly, 2002). Even in the 21st century, Japanese men spend the least time on unpaid family work, such as domestic work and care, and the most time on paid work of all developed countries (Fuwa, 2004;Treas and Drobnic, 2010;OECD, 2011). However, the employment rate of women reached 70% in 2020, which is considerably above the OECD average (57.3%) and is not very different from the Nordic countries where the most gender-equal labour market participation has been achieved. In addition, both the employment rate and the female active labour force population have been rapidly increasing since 2000, especially since mid-2010. The increase in women's employment rate during prime age (25-54 years) in Japan from 2000 to 2019 was the highest among OECD member countries, from 63.6% to 78.2% (OECD, 2020a, b).
Considering the above factors, the gender division of labour and intergenerational mutual assistance based on cohabitation under the patriarchal system, which were considered the foundation of the Japanese family-oriented welfare regime model, are in steady decline, although they remain strong compared to Western countries where individualism is more prevalent.
Mainly in European countries that face ageing and austerity welfare policies, physically and economically independent older citizens are "re"-considered valuable childcare providers (Hank and Buber, 2008;Thomese and Liefbroer, 2013;Ho, 2015;Craig and Jenkins, 2016), and multigenerational cohabitation "is also seen as being advantageous for the younger generation, since it enables grandparents to help with childcare" (J€ onsson, 2003, p. 243). Family and household types have been observed and identified as factors that impact women's/mothers' employment and their work-care life reconciliation and conflicts (Kr€ oger and Sipil€ a, 2005; Lewis 2006Lewis , 2009Crompton, 2006;Lewis et al., 2008;Mahon, 2006).
In contrast, the Japanese government, seeking to strengthen the welfare function of the family, and specifically to reduce the social cost of childcare and the work-childcare conflict through intergenerational assistance, provided tax incentives and support measures for threegeneration cohabitation in the 2010s. These public measures were not successful (Sanpei, 2019). Overall, the impacts of changed labour market participation and household type of working mothers with younger children in recent decades have not been frequently studied.
The aim of this study is to explore the changes that occurred in three-generation households and mothers' employment between 2002 and 2019 in superageing Japanese society. It begins with a literature review of grandparents' care and multigenerational cohabitation in a cross-national context and then overviews the labour market and family changes in Japan in the 2000s and 2010s. Then, using data from the Labour Force Survey (LFS) from the nationally representative Statistical Bureau of the Ministry of Internal Affairs and Communications, the later sections conduct descriptive analyses and identify changes in women with younger children regarding their employment, working hours and reasons for part-time employment. These changes are examined for two types of households, namely, nuclear families and multigenerational families. The final section summarises the main findings of this research.

Conceptual framework
Previous evidence on intergenerational aid and mothers' employment The rise of female employment was a common trend among almost all postindustrial countries in the 1990s and 2000s, and work-(child)care arrangements have differed depending Three generation family and mothers' employment on welfare regimes and welfare states (Crompton, 2006;Daly and Klammer, 2005;Lewis, 2006).
To facilitate work-family reconciliation, especially for mothers with younger children, the workload within and outside the family must be reduced (Craig and Powell, 2011). To reduce work outside the home, flexible employment, such as part-time employment, has been considered key to work-family life reconciliation by "activation" policymakers (Takasaki, 2008;OECD, 2007) and by many female workers with family responsibilities (Fouarge and Muffels, 2008;Plantenga and Remery, 2009). However, women's part-time employment has also been criticised as a cause of social exclusion, social inequalities and polarisation of employment, especially for single-parent households with younger children (Hakim, 1989;Webber and Williams, 2008;Warren, 2008;Burgoon and Dekker, 2010;Blossfeld et al., 2011).
In contrast, within the family, childcare provision by grandparents and multigenerational cohabitation have been reidentified as having a positive impact on mothers' employment, although public childcare services and parents are still the two main providers of childcare (Jappens and Van Bavel, 2012;Thomese and Liefbroer, 2013;Hank and Buber, 2008;Greisen et al., 2009;Brandt et al., 2009;Igel and Szydlik, 2011;Kanji, 2018). From a macro perspective, childcare configuration differs according to the welfare state regime (Campbell, 2009;Aboim, 2010) and grandparents' role in relation to their younger family members also differs according to the welfare state regime.
Regarding liberal welfare states, a comparative analysis of nationally representative surveys of the US, the UK and Australia identified that the share of three-generation households in threegeneration households with younger children was 7-8% in Australia and the UK, almost half that of the US, at 15% (Pilkauskas and Martinson, 2014). Additionally, in the US, the share of three-generation households that include children increased from 5.7% to 9.8% between 1994 and 2008 (Pilkauskas and Cross, 2018). In fact, among liberal welfare states, the US has the most studies and the most diverse research fields, which range from ethnic studies to developmental psychology and criminology. Moreover, because there is less provision for affordable public childcare, coresident grandparents have been identified as important childcare providers for their grandchildren, especially in households with lower incomes, single parents, specific ethnicities and problems such as addiction and illness (Luo et al., 2012;Vandell et al., 2003;Kropf and Kolomer, 2004;Hayslip et al., 1998;Peek et al., 2000). In contrast, in the UK, with its relatively generous social policies and public childcare provision, grandparents' care for grandchildren has been more focused on their important supporting role in mothers' employment rather than on the type of residence, such as three-generation cohabitation and families with great difficulties (Grey, 2005;Wheelock and Jones, 2002;Tan et al., 2010).
In Europe other than the UK, interest in grandparental childcare tends to be the highest in Nordic countries and the lowest in Mediterranean countries (J€ onsson, 2003;Daatland and Lowenstein, 2005;Daly, 2011;Arber and Timonen, 2012). Grandparental childcare in Nordic countries is not based on intergenerational cohabitation (Sundstr€ om, 2009) and a higher share of intergenerational cohabitation does not imply a higher share of intergenerational aid, especially for childcare provision. In fact, three-generation households were most common in the Eastern European countries except for the Czech Republic, where they accounted for over 5% of all householdsand for over 10% in Romania, Latvia and Bulgaria. Across southern Europe, as well as in the Czech Republic and Estonia, three-generation households accounted for 1% to 5% of all households. They were least common in the Nordic and most of the other northwestern European countries, where they accounted for less than 1% of households (Iacovou and Skew, 2011).
Interestingly, in Nordic countries, which have been characterised by a high level of defamilization and decommodification of care service provision, the strongest adult worker model and a low rate of living with grandparents, a higher frequency of grandparents' support for their adult children and grandchildren was identified than in Southern and other IJSSP 41,13/14 European countries. This result seems odd given the general notion of defamilization/ familization in the existing comparative welfare studies (Hank and Buber, 2008;Igel and Szydlik, 2011). Similarly, even in the Netherlands, which is closer to the Scandinavian regime and has both well-developed formal childcare and part-time employment systems, grandparents' childcare support for their grandchildren has a positive impact on parental employment and fertility rates (Thomese and Liefbroer, 2013;Geurts et al., 2015).
The Southern European welfare regime, which has characteristics that contrast with those of Nordic countries, belongs to the "familism by default" care model, where public and market childcare services are relatively limited and intensive caregiving by family members living in the same household or in the neighbourhood has frequently been considered a natural family responsibility (Knijn and Saraceno, 2010). However, Italy has one of the lowest labour force participation rates of prime-aged women in Europe (OECD, 2020b), and as mentioned above, intergenerational assistance in childcare is not always frequent.
In East Asian countries, unlike Western countries, patriarchal and senior-oriented social values (referred to as Confucian traditions) still influence relatively strong intergenerational mutual aid and welfare familism; however, the tradition of multigenerational cohabitation has continuously diversified and declined since the second half of the 20th century (Yi, 2006;Sun, 2008;Matsuda and Hori, 2006;Tsai, 2007;MHLW, 2014;Izuhara and Forrest, 2012;Matsuda and Hori, 2006).
In Japan, a positive impact of coresident grandparents' support for domestic and care work on mothers' employment was identified in the 1970s (Hiroshima, 1981). Even in the 21st century, working mothers are often supported by their coresident grandparents in "threegeneration family households" (MHLW, 2003;Shu, 2012), and the provision of childcare by relatives is still "somewhat effective" for dual-career couples (Ochiai, 2009). The number of three-generation households is rapidly declining, but even in the 2010s, 16.3% of Japanese households with children under 18 years of age belonged to three-generation households (MHLW, 2014). Three-generation households remain one of the major household types and one of the major means of intergenerational assistance, especially in rural areas, and the share of three-generation households (as a percentage of all households) varies greatly depending on the region. The largest share was in Yamagata at 16.7% and the lowest in Tokyo at 1.8% of all 47 prefectures in 2018 (MHLW, 2018).
Regarding grandparents' caregiving role for their grandchildren, in liberal welfare regimes, more attention overall has been given to three-generation families with greater challenges in specific ethnic and income groups. In the Nordic welfare regime, the focus has been on the role of non-coresident, more self-reliant and financially stable grandparents in promoting mothers' employment. The UK has had intermediate characteristics between these two regimes. In East Asian countries, the traditional family system and family view, with cohabitation and intergenerational assistance as the norm, is very strong compared to in Western countries, although the regional differences are likely very large, and the norm is continuously shrinking.
In this study, we clarify the changes in mothers' employment in three-generation households in Japan from 2000-2019, which has not received much attention in international studies.
Trends in family changes, the labour market and public policies for three-generation households, 2000-2019 Family changesdownsizing households. Japanese families have traditionally respected intergenerational assistance and cohabitation based on large patriarchal family systems in which the older patriarch is dominant. This view remains stronger in Japan than in Western countries, especially in rural areas (Nagai, 2005;Cabinet Office, 2020d). From a historical perspective, the strict old Japanese patriarchal system based on multigeneration Three generation family and mothers' employment cohabitation, which typically starts with the marriage of the eldest son and continues until the cohabiting parents pass away, has been a typical living style and family type in Japan. This was true even in the mid-20th century (Kamiko and Masuda, 1976). Since then, however, three generation households have been continuously declining.
In fact, in 1953, the average number of Japanese household members was 5.0, decreasing to 2.39 in 2019 (MHLW, 2014(MHLW, , 2020b. In addition, 16.9% of all Japanese households were threegeneration households in 1975 (MHLW, 2014) compared to 5.1% in 2019 (MHLW, 2020b). Between 2000 and 2019, the number of three-generation households decreased by almost half, from 4.8 million to 2.6 million, and the average household size declined from 2.75 to 2.39. As shown in Table 1, the proportions of "one-person households", "couples without children" and "single parents with child(ren)" all increased in contrast to the decrease in "couples with children" and the halving of "three-generation family households" during this period. Additionally, the number of births decreased from 1.19 million to 0.86 million (MHLW, 2019).
These trends in family changes are also driven by the late marriage of adult children and the avoidance of intergenerational cohabitation by younger generations. The average age at first marriage increased by 2.4 years, from 27.0 to 29.4, and the average age at first birth increased by 2.7 years, from 28.0 to 30.7, between 2000 and 2018 (Cabinet Office, 2020c). In a survey by the Cabinet Office (2015) on ideal living styles for multigenerational families (N 5 1639), the most frequently reported preference was for "different houses placed separately" among the youngest group aged 20-29 years (43.8% for males and 42.9% for females). Multigenerational coresidency was most preferred by the oldest age group for females (30.5%) and by the second-oldest age group for males (31.1%). Furthermore, for married persons aged 20-49 years (N 5 420), the most ideal form was to live close to one's parents (not their spouse's parents) regardless of gender or working style. The proportion of individuals who considered three-generation cohabitation the most ideal living arrangement remained one-third of those who considered living close to their grandparents to be their ideal living arrangement (Cabinet Office, 2015).
Notably, the ever-declining number of three-generation households has tended to appear in higher proportions in the much older population: among those aged 65 years and over, 32.4% of men and 38.9% of women lived with their single or coupled children in 2019, and among those aged 80 years and over, this proportion reached 37.3% for men and 48.5% for women in 2019 (MHLW, 2018). This result can be explained by the younger generation's avoidance of cohabitation and the still-strong family care responsibility, both of which are mentioned above, as well as increased health problems, especially for the elderly aged 75 years and over (Cabinet Office, 2018).
Labour market. Since 2000, the Japanese economy has continuously languished at near-or under-zero growth and has thus been mostly stagnant (Rosenbluth and Thies, 2010 the employment situation of prime-age citizens aged 25-54 years in Japan did not deteriorate at all but rather seemed to improve. Unemployment also remained very low compared to the OECD averages (OECD, 2020a) and, if anything, has decreased a few points further in the last two decades. As shown in Table 2, between 2000 and 2019, the male labour force participation rate in Japan went from 97.1% to 95.5%, the employment rate went from 93.4% to 93.1%, both exceeding 90%, and the changes were limited. In contrast, the female labour force participation rate increased from 66.5% to 80.0%, and the employment rate increased from 63.6% to 78.2%. As a result, the consistent gender gaps, which were approximately 30 points in the labour force participation rate and the employment rate in 2000, were almost halved in 2020. The unemployment rate of the prime-age population decreased for both men and women during the period, reaching 2.5% for men and 2.3% for women in 2019.
Simultaneously, during this period, the share of prime-aged part-time employees who worked less than 35 hours per week increased by approximately 10 points for both men and women, and the gender gap in the proportion of part-time employees expanded from 21.9 points in 2000 to 24.9 points in 2019. In fact, in Japan in the late 2010s, women in their 30s and older were still less likely than men to participate in the labour force, and non-regular employment was particularly common. In 2019, women constituted 44.5% of all employed persons, and 56.0% of all employed women were atypical workers (mostly part-time) (Cabinet Office, 2020a). As background to Japanese women's employment situation, the Cabinet Office (2020b) has noted that there appear to be employment adjustments for women who are considering household income and family circumstances such as childrearing and longterm care.
Furthermore, as shown in Table 3, for middle-aged and older populations, especially women in their early 50s to early 60s-which corresponds to the age of younger grandparents-both labour market participation and the employment rate increased by Three generation family and mothers' employment approximately 20 points during this period, and the gender gap in employment and labour force participation rates, which were remarkable in 2000, consistently narrowed in 2019. Even in the group aged 65-69 years, more than half of men and approximately 40% of women were in the labour force in 2019. In contrast, the nonworking population increased significantly only for those aged 70 years and over.
Public policy and its impact on mothers' employment and three-generation households. Two main institutional reforms contributed to the consistent increase in middle-aged and older labour force participation between 2000 and 2019. First, the 2013 amendment to the Elderly Persons Employment Stabilisation Act raised the retirement age, and second, the pension reforms in 1994 and 2000 raised the starting age for pension payments from 60 to 65 years.
Japan currently has an occupation-based two-pillar pension system, which has undergone long-term reforms. Both its current system structure and the rules for retirement age are complicated. In general, in 2020, some people can receive pensions by 62 years of age for women and 63 years of age for men. This retirement age will increase to 65 for all men in 2025 and for all women in 2030 (MHLW, 2011).
In contrast, no public policies have been identified that clearly and directly contributed to the significant increase in women's labour force participation between 2000 and 2019. Raymo and Fukuda (2016) indicated that one-third of the factors behind the increase in Japan's female labour force participation rate from 1980-2010 can be explained by an increase in the number of unmarried women, which was particularly pronounced among those aged 25-39 years. At the same time, MHLW (2020a) pointed out that between 1989 and 2019, among households that included married male employees, double-income households increased significantly from 42.3% to 66.2%, in contrast to the decrease in households that included full-time housewives.
To promote mothers' employment, public childcare services, childcare leave and equal employment opportunities for men and women were at least legally established by the early 1990s. However, dependents' working income restrictions, which provide a disincentive for the active labour participation of married women, also continued to exist in the 2010s (Adachi and Kaneda, 2016).
Japan has a locally based public childcare system nationwide, and childcare policy has been consistently regarded as a priority issue for the government since the 1990s. In 2019, public childcare was partially made free, and the government set a goal of securing childcare for 320,000 children by 2020 (MHLW, 2020a). In fact, currently, 10% of children receive day care services such as nursery schools at the age of 0, but this figure rises to 86% at age 3 if kindergarten and other educational services are included (MHLW, 2016). In contrast, the coverage rate for the youngest children in Japan is not high among OECD countries, mainly due to the rapid increase in women's workforce participation (Bonoli and Reber, 2010). Therefore, grandparents' support for grandchildren, even if it is only supplementary, is effective, particularly for working mothers with younger children.
In addition, a budget of 5.5 trillion yen (50.1 billion USD) was added in 2018 to fund low birth-rate countermeasures (Cabinet Office, 2020c). The expansion of public childcare between 2013 and 2020 was remarkable, with a plan to secure a total capacity of 855,000 posts in day care centres. However, this did not precede the rapid increase in working mothers during this period, and in fact, social issues have always existed regarding children on day care waiting lists (MHLW, 2020a). Furthermore, for the elderly, the National Long-Term Care Insurance System (NLTCIS) has existed since 2000, and the number of users of public care services increased from 1.49 million to 4.87 million from 2000 to 2019 (MHLW, 2020a). The NLTCIS is thought to have provided some incentive for the defamilization of long-term care for the elderly and the labour force participation of some middle-aged and older adult women who had previously provided family care.

IJSSP 41,13/14
In sum, it can be said that the changes in women's employment in Japan over 2000-2019 were brought about by changes in employment and gender roles during this period. These include the decline of employment practices such as lifetime employment and seniority-based wages, which have resulted in a shift from a single male breadwinner and a fully dependent full-time female housewife to a male breadwinner and a part-time female breadwinner rather than a full-time dual-income couple.
Regarding public support for three-generation households, the 2006-2015 National Basic Plan for Living Life by the Ministry of Land, Infrastructure, Transport and Tourism pledged support for three-generation families who live together or near each other to assist childrearing households amid a severely declining birth rate. Then, in 2016, the Cabinet Office set tax credits (from 2016 to 2019) for both income tax and public subsidies for people renovating owned houses for the purpose of cohabitation by three generations. Although the government established these measures not only to support families raising younger children but also to promote family care for the elderly to reduce social security costs, they basically provided a maximum of 500,000 yen per case, the subsidies were a one-time payment, and the affected households were limited (Sanpei, 2019).
Additionally, the series of public support measures for three-generation cohabitation appear more targeted to family care for older people who already own their homes than to childcare support for families with younger children who have formed new families. Sanpei (2019) also pointed out that these measures do not address the rapidly declining number of three-generation households or the needs of younger generations who need childcare support.
Mothers' employment in nuclear and three-generation families, 2002-2019 Data and method. By descriptively comparing mothers in nuclear families and threegeneration family households, this section aims to reveal how women with younger children changed their commitment to the labour market during 2000-2019 in these two different household types. To this end, this study uses data for 2002 and 2019 from the LFS performed by the National Statistics Bureau of the Ministry of Internal Affairs and Communications to conduct a descriptive analysis. The LFS was used for the following reasons. First, it has been conducted every month since 1947 at the national level and publishes monthly and yearly results. The minimal changes in the survey methods over this period allow for long-term comparability. Second, it is a major national survey that comprises at least 100,000 nationally representative samples surveyed on employment status by age group and household type. Third, it surveys mothers' employment status and working hours by different household types and age groups of the youngest child.
The mothers surveyed here are classified into two types of households: "couples living with child(ren)" as nuclear households and "couples living with children and parent(s)" as three-generation households. These two household types are further divided into four groups based on the age of their youngest child (0-3, 4-6, 7-9 and 10-12 years). Based on this classification of mothers, the population, labour force population and participation rates, and numbers and share of part-time work were compared for 2002 and 2019. The comparison is limited to these years because the data from the detailed tabulation of annual average results of the LFS were published only after 2002 in the official statistics, e-stat.
Changes in population and labour force. According to the LFS, the total number of women who belong to all four household types slightly decreased but remained almost unchanged between 2002 and 2019. However, if we examine this female population by child age group and household type, a clear change during this period is identified. As shown in Figure 1, the population drastically decreased by less than half in all four age groups in three-generation Three generation family and mothers' employment households; this contrasts with nuclear households, which increased for all age groups except for those with children aged 0-3 years. Figure 2 shows mothers' labour force population and participation rates by household type and the age group of the youngest child between 2002 and 2019. During this period, in nuclear households, mothers' labour force population increased more than 1.5 times for all four ages and increased 1.8 times in the group with children aged 0-3 years, from 870,000 to 1.57 million. In contrast, in three-generation households, the number was halved for all four age groups, with some differences in each age group.
The labour force participation rate increased during this period in all age groups for the two types of households, with the largest increase of 31.2 percentage points in the 0-3 age group in nuclear family households and the lowest increase of 4 percentage points in the 10-12 age group in three-generation households. For both types of households, the increase in the age 0-3 age 4-6 age 7-9 age 10-12 age 0-3 age 4-6 age 7-9 age 10-12 Mothers' Population (thousand) 1600 1800 age 0-3 age 4-6 age 7-9 age 10-12 age 0-3 age 4-6 age 7-9 age 10-12

Note(s):
Part-time work is defined as working less than 35 hours per week. The mothers' age in the labour force is 15 years or older, and there is no upper age limit; The mothers' age is considered to be 65 years or younger because these are mothers of younger children labour force participation rate during this period was largest for the 0-3 age group and smallest for the 10-12 age group. Additionally, the labour force participation rate tended to increase as children grew older, similarly to the tendency in OECD countries (OECD, 2007), in both years and in both household types. Changes in part-time/full-time work participation. The changes in mothers' part-time employment between 2002 and 2019 are shown in Figure 3. During this period, the changes in the number of part-time workers for all age groups in both households appeared to roughly match those in the labour force population examined above, with a large increase in nuclear family households and a halving in three-generation households. In contrast, two clear differences were identified between the trends of changes in labour force participation and part-time employment.
First, the part-time employment rates for nuclear households remained almost unchanged except for a 9.9-point decrease in the group with children aged 0-3, while these rates increased in all age groups in three-generation households. Second, in contrast to the labour force participation rate, the ratio of part-time employment was higher in nuclear family households than in three-generation households.
Furthermore, in 2002, the difference between the two households was at least 10 points; however, it clearly narrowed in 2019, and the ratio of nuclear households exceeded that of three-generation households for children aged 0-3 years, who should represent the greatest care burden. That is, the much higher shares of full-time employment for mothers in threegeneration households decreased more than for mothers in nuclear family households during this period. Moreover, these results suggest that the provision of grandparents' care through the coresidence of three generations, which was thought to support mothers' full-time employment, no longer had a consistent positive impact on the employment of mothers with the youngest children, who need the most care.
Reasons for women working part-time. What are mothers' reasons for preferring for parttime work? The LFS presents people's reasons for working less than 35 hours per week according to sex and age group. The survey question was "If you have worked less than 35 hours in the past week, please explain why". Excluding "other", the three main answer options were "originally engaged in part-time job", "slack in business" and "own/family condition". Of these, the "own/family condition" category was further broken down into the Three generation family and mothers' employment four possible responses of "maternity or childrearing", "caring for aged or sick family members", "vacation" and "other". Figure 4 shows the reasons why women workers chose to work part-time in 2002 and 2019 according to the three age groups of 25-34, 35-44 and 45-54 years. These groups are mostly of working age and are responsible for raising children and taking care of their families. The corresponding total number of samples was 670 in 2002 and 778 in 2019. Between 2002 and 2019, in all age groups, the number of part-time female workers who chose part-time work for the convenience of "employers" decreased, while those who made this choice for the convenience of "family" increased. In particular, in the 35-44 age group, the number of females working part-time because of family care increased 5.6 times, from 0.58 million to 2.8 million. The proportion of women working part-time because of "maternity or childcare" increased significantly by 18.5 points for women aged 25-34 years and 16.4 points for women aged 35-44 years during this period. In fact, in 2019, 38.9% of women aged 25-34 and 23.1% of women aged 35-44 chose to work part-time to care for family members. The population of women who chose part-time work "for family care" tended to decline as they grew older in 2002, but this trend disappeared altogether in 2019. Similarly, in 2002, the number of females working part-time for "employer reasons" substantially exceeded those doing so for "family care reasons" in all the age groups; however, this trend also disappeared in 2019.

Conclusion
This study examined the changes in the labour market, the family and the labour participation of working mothers in three-generation households in Japan over the two decades since 2000. From the macro perspective, during this period, both the labour force and the employment rate of women increased significantly, while these rates remained almost unchanged for men among the prime-age population (aged 25-54 years). These shares for middle-aged and older people aged 55-69 years increased more than for the prime-age Note(s): In the Labour Force Survey, the reasons for part-time employment consist of the four major items and six substrata. Both "Maternity or Childcare" and "Long-term Care or Nursing" are substrata, and the share of long-term care or nursing is calculated by "Maternity or Childcare" and "Long-term Care or Nursing"/Part-time workers -Originally engaged in PT work and other However, between 2002 and 2019, among nuclear and three-generation households including women with children under the age of 12 years, more varied changes can be identified, and the following three were notable. First, the population in the labour force and the number of part-time workers consistently increased in nuclear households, in contrast to the decline by almost half in all age groups for three-generation households. Second, the labour force ratio in three-generation households, which was higher for all age groups than in nuclear households in 2002, narrowed in 2019. Third, the part-time employment ratio, which was consistently lower in three-generation households than in nuclear family households in 2002, narrowed in 2019. These three changes indicate that grandparental assistance in threegeneration households, which has long been thought to benefit the employment of women with younger children, shrank over this period.
Interestingly, the decline in the function of coresident grandparents' support to encourage mothers' full-time employment seems to have been caused not only by the decrease in both the total population and labour force population of mothers in three-generation households but also by the decrease in the proportion of full-time mothers in three-generation households. Although this study did not aim to identify the factors in the decline in grandparents' support function for their grandchildren and working children, which are complex, it identified two likely factors. First, in addition to the significant decrease in three-generation households during this period, when the grandparents were older, the rate of three-generational coresidency was higher. Second, not limited to three-generation households, the employment rate of middle-aged and older grandparents increased, especially for women aged 55-69 years, who were more likely to be active caregivers. Furthermore, after 2000, the NLTCIS for the elderly was established and public childcare posts grew significantly. These social (care) policies may have somewhat reduced the need for intergenerational mutual assistance and family responsibility for caregiving.
In contrast, the promotion policy measures for three-generation cohabitation were limited to smaller financial support packages for the partial renovation of owned houses, which were for middle-aged and older grandparents who already had their own houses rather than for younger families with younger children.
Therefore, in today's three-generation households in Japan, which continue to decline and are unevenly distributed in rural areas, intergenerational support, such as caregiving or domestic work, is no longer exchanged equivalently between generations. It is transferred less from grandparents to working children and grandchildren needing care and transferred more from children and grandchildren to much older grandparents who have difficulty living independently. In addition, women's work-family care balance appears to involve the choice to reduce working hours rather than to live with their parents (who are often old or busy with their own work) to obtain their help. This is suggested by the larger increase between 2002 and 2019 in both the proportion of part-time workers in three-generation households and the number of part-time female workers who care for their families in the two age groups.
These results may also confirm the mismatch between the Japanese government's threegeneration cohabitation support policy packages and the expectation of an equivalent exchange of caregiving between children's families and grandparents. To encourage intergenerational cohabitation, such as three-generation households (whether or not this is an effective measure to counter the declining birth rate and promote mothers' labour force participation), generous or at least sufficient cash benefits to adult children's families with younger children rather than to older grandparents may be more effective in promoting threegeneration households, especially in rural areas.

Three generation family and mothers' employment
What distinguishes Japan is that the labour market and social policy systems have scarcely contributed to a gender-equal adult worker model (Shoppa, 2006;Osawa, 2007;An, 2017), and both labour market (re)participation and part-time employment have dramatically increased since 2000 for women with the youngest children, for whom the care burden should be the heaviest. Compared with previous European comparative studies (Crompton et al., 2007;Lewis et al., 2008), this study finds that this set of circumstances in Japan conflict not only with progress towards gender-equal patterns of individualisation but also with the care patterns in traditional large families.
Finally, although the findings above are strictly in line with the aim of this study, there are limitations and implications for future research. First, as discussed earlier, this study was a descriptive analysis that compared the employment status of mothers and household types, and it did not show statistical significance between the assistance of grandparents and mothers' employment in the same household. To identify the strict relationship between them, new data and more rigorous statistical analysis, such as multivariate and/or multipleregression analyses, are needed. Second, the higher proportions of and contributions from maternal grandparents who live close to mothers should not be ignored (Kanji, 2018;Cabinet Office, 2015). Further attention should also be given to intergenerational proximity and reciprocity, which differ from traditional styles, such as patriarchal three-generation households.