Migrant care workers and care-migration policies: a comparison between Italy and Japan

Japan and Italy are the most aging societies in the developed countries and they both face the rapid increase of the social cost and the demand in manpower for long-term care. Both countries have the common welfare state trajectories and characteristics. In contrast, both the care and migration policies and the role of migrant care workers between Italy and Japan are consistently different. The paper compares the welfare state characteristics, care configurations, and the care-migration policies in 2000s, and reveals how the role of migrant care workers in Italy and Japan would be influenced by the migrant and care policies reciprocally. The paper concludes that the combination of the restricted migrant policies for unskilled migrant care workers and the in-kind-based national care policies is significant in efforts to maintain a qualified and regulated care work. Conversely, it verifies that the in-cash-based national care policy with the weak migration control for unskilled migrant workers brings the spread of marketized care with weak control and low professional skills, mostly in the gray market.


Introduction
Japan and Italy are the most aging societies in the developed world. Japan has the highest share of population aged 65 and over (compared to the labor force population aged 15-64), and Italy ranks second. In 2020, the figures for these two countries will be almost coincident, exceeding 55% (OECD 2014).
In addition, great similarities are found in their welfare configurations: that is, the Bismarckian social insurance-based and the household (and not individual)-based welfare system with Bfamilistic welfare^. These welfare configurations have long been formulated by the typical BOld Europe^in Italy (Palier 2012) and in Japan (Shinkawa 2005). Indeed, Italy and Japan have had the narrowest historical processes of modernization and industrialization since the second half of the nineteenth century (Ferrera 1993;Ferrera and Maino 2006;Tada 2014). Thus, in the twenty-first century, they are facing unprecedented welfare challenges, mostly due to their hyper-aging populations.
At present, one of the greatest challenges in highly developed and aged societies is ensuring the provision of adequate, qualified long-term care for elderly citizens. Italy and Japan have long been classified as having a typical familistic welfare/care model, which is sustained by uncompensated care for family members provided exclusively by women not in the labor force. However, since the 2000s, a relevant divergence from the typical familistic care model has been identified in both countries. In Italy, migrant domestic/care workers have increased drastically; this phenomenon has been called the Bmigrant-in-the-family model^ (Bettio et al. 2006) and/or the Bmigrant-based care model^ (Da Roit et al. 2013). In contrast, in Japan, the National Long-Term Care Insurance (NLTCI) has been in force since 2000. Approximately 4.68 million citizens, which amounts to 14.2% of the population aged 65 and over, and more than 80% of older citizens who were certified as persons in need of care by the NLTCI received formal care services in 2014 (MHLW 2014a). In 2014, there were 1.71 million care workers under the NLTCI system (MHLW 2015c).
These changes in Italy and Japan in the 2000s are considered a certain kind of Bdefamilization^from the traditional familistic care model. Put differently, the family(based) care model is changing in both countries in contrasting directions: This change is occurring due to the marketized migrant care model in Italy and the nationalized formal care model in Japan.
Considering the major commonalities of their aging populations and welfare systems, on the one hand, and the great differences in their eldercare configurations, on the other hand, this study focuses on the divergence of the current defamilized orientations of eldercare provision between Italy and Japan and proposes the following research question: Why has the migrant-based care model been largely diffused only in Italy and not in Japan, even though the highest levels of population aging, the institutional welfare configuration, and the strong social norms regarding the family have been commonly identified in both countries?
In the following section, the theoretical framework for this research question is discussed in detail. Next, the commonalities of the welfare systems and the contrasting distinctions in eldercare policy between Japan and Italy are surveyed. Then, the fourth section discussed both the migrant and the public care policies that influence the provision of eldercare by migrant care workers and their relevance in the field of eldercare in Italy and Japan. The final section summarizes the findings and presents the implications for future research.

Migrant care model and intra-model variations by professional skills and evaluation schemes
In the 1990s, Esping-Andersen's works (1990,1997,1999) revealed the complexities of care configurations between the state, the family, and the market (and the quasimarket). Italy was classified as representing a conservative welfare state typology, and Japan was classified as liberal-conservative, with neither of them belonging to the typical three major welfare state classifications. Since then, care models have been analyzed by cross-national comparison (Scott 1998;Sipilä & Korpinen 1998;Ungerson 1997;Daly and Lewis 2000;Daly 2002;Evers 2005), but Italy and Japan have often been excluded from comparative social care (model) studies or simply classified as representing familistic care or the family care model due to the scarce (information on the) national long-term care system and the countries' strong single-male breadwinner model, in which housewives mainly engage in uncompensated care work for family members.
Since the middle of the 2000s, also influenced by fiscal austerity, the financial crisis, and population aging, studies have further investigated the division of care provision between the state, the market, and the family in detail, focusing on different ways of care distribution. In the state, the public transfer for care was further analyzed in terms of being in-kind and in-cash, revealing that the increasing trend of cash transfer was more related to the family sphere, utilizing it as Brouted wages^ (Ungerson and Yeandle 2007). Regarding the family, the familistic welfare regime was revised from its traditional definition, and intra-regime variations were identified (Bambra 2007;Calzada and Brooks 2013;Gal 2010;Saraceno 2008;Saraceno and Keck 2010). Hank and Buber (2008) identified that intra-family care provision from grandparents to grandchildren was more frequent in Nordic countries than in Mediterranean countries such as Italy.
Regarding Italy and Japan, a stronger familistic aspect of the welfare system has been identified in both Mediterranean countries and East Asian countries. The European comparative analysis of Saraceno and Keck (2010) classified Italy as Bfamilism by default^, with the strongest care/welfare model; nevertheless, they also regarded the Italian long-term care scheme, with the combination of unbounded cashfor-care benefits and unregulated marketization (of migrant workers), as Bimplicitsupported familialism.^Similarly, based on its indicators, the Japanese long-term care configuration would correspond to Bexplicit-supported familialism^because the NLTCI system, which provides exclusively in-kind service for the elderly, was founded in 2000 but Japan's familistic welfare characteristics continued to strongly benefit from several welfare institutions, the breadwinning role in the labor market, and the family role for care work and domestic work (Bambra 2007;Calzada and Brooks 2013;Fuwa 2004;Gal 2010;Ochiai 2009).
In the same period, studies that regarded migrant care and domestic workers as the relevant care providers in welfare states have rapidly increased (Anderson 2012;Bauer and Österle 2013;Howe 2009;Shutes & Chiatti, 2012;Williams 2012). Moreover, the study that introduced migrant care/domestic workers into cross-national comparative analyses of welfare and care regime models was a well-known Italian study that pointed out that the Italian care model in the 2000s had already changed from the family model to the migrant-in-the-family model (Bettio et al. 2006). This work of Bettio et al. (2006) revealed not only that migrant domestic workers have rapidly increased or been made visible by regularization, but also that the interrelation among the repeatedly conducted regularization programs, the increase in unauthorized migrants, the hyper-aging population, the welfare familism and the domestic work sector creates this Italian care model.
In contrast, the Japanese migrant care model has scarcely been discussed and compared with the models in other countries. The primary reason was that there are very few migrant care workers in Japan, mainly due to Japan's restricted migration control system, which is a totally different mobility regime from those in EU states. Rather, in that respect, the migrant-in-the-family model would likely be identified in other economically developed regions in East Asia, such as Singapore, Hong Kong, and Taiwan (Huang et al. 2012;Liang 2014;Michel and Peng 2012).
Thus, especially since the 2010s, in comparative social care studies, migrant care workers have been studied to specify their role not as a single subject in the triangular theoretical framework of state-market-family but as socially and/or institutionally stratified subjects. The reason is that Bthe same skills (of migrant workers) may be regulated and valued differently in different sites^ (Kofman and Raghuram 2015: 100) and migrant workers' and/or citizens' social and employment security conditions are differently stratified by their entry route, resident status, country of origin, and social and professional skills. In other words, a new theoretical framework for migrant care models is needed to focus on the stratification of migrant care workers and their different ways of labor participation, which are mostly regulated by both the migration and social policies in the receiving countries (Catarino et al. 2013;Van Hooren 2012;Williams 2012;Kofman and Raghuram 2015). Kofman and Raghuram (2015) focused on the reproductive skilled sectors for (female) migrant workers and suggested that skilled migrants have been relatively privileged as bearers of human and cultural capital (127). From this perspective, the higher professional skills of migrant workers, regulated and certified by national qualifications, should benefit from higher levels of social and employment security and settlement, on the one hand, in contrast to undocumented migrants and/or the lowest skills level of migrant workers, on the other hand. The skills level of migrant care workers is mostly evaluated by migration and care policies. In this respect, Catarino et al. (2013) demonstrated that migrant female domestic workers in Southern European countries with the migrant-in-the-family care model were much less institutionally recognized and their workers' rights, work formalization, and professional skills were evaluated at a much lower level compared to their counterparts in Nordic European countries. Additionally, Van Hooren (2012) analyzed the reciprocal impacts of care-migration policies on the eldercare labor force of migrant care workers by comparing their employment conditions and relevance in the social care provision for the elderly in Italy, England, and the Netherlands. The results showed that direct migration policy impacts on the role of migrant care workers for eldercare Bwere not found^and were Bmore ambiguous^(Van Hooren 2012:133,144), even though important impacts of migration policies on the conditions under which migrant care workers are employed and the legality/illegality of their employment status were identified. In this respect, Simonazzi (2009) reached almost the same conclusion as Van Hooren (2012) with respect to the characteristics of the Italian migrant care model, although migrant policies were excluded in Simonazzi's analysis. Indeed, Van Hooren's detailed elaborative analysis of the EU mobility regime and migrant workers' working conditions complicates revealing the relationship between migration policy and the role of migrant care workers.
The locus of migrant workers' care provision-the state, the market, and the family-is different in each country. Consequently, the policy impact on regulations and skill evaluations for migrant care workers is also different. Focusing on the Italian and Japanese cases, this study first clarifies the commonalities of the welfare states and the locus where migrant care workers are mainly engaged in providing long-term care.
Second, this study attempts to identify the interrelations of migrant-care policies and their impacts on the regulation, evaluation, and training for migrant care workers' employment conditions and professional skills.
State or market? Social expenditure and long-term care policy in Italy and Japan Common welfare state models but contrasting care policies The highest levels of population aging, old age-related social spending, and deficit financing in the developed world are found in both Italy and Japan. As shown in Table 1, the share of the elderly population in 2020 is projected to be 27.8% in Japan and 23.5% in Italy. Additionally, in regard to the share of the elderly population and the government debt-to-GDP ratio, Japan and Italy have the first and second highest amounts, and Italy maintains the highest level of public expenditure on pensions (15.4% of GDP) ( Table 1).
In addition to their common old age-oriented welfare systems, Italy and Japan have three phases of historical development in common (Ferrera 1993(Ferrera , 1998Tada 2014). Space constraints do not allow for a more detailed discussion, but the three phases are as follows: first, the era in which their backward industrial and modern nation states were founded and their experience of totalitarianism (1860s-1940s); second, the era in which employment-and social insurance-based welfare states were founded in democratic regimes with high levels of economic growth/recovery (1950s-1970s); and, third, the period in which the old age-oriented welfare systems were strengthened by Bexcessive path dependency^and the pension and care crises emerged in postindustrial aging societies (1980-2010s).
In contrast, the detailed systems and institutions of long-term care for the elderly in Italy and Japan have great differences in the following aspects (Table 2), their common welfare state characteristics shown above notwithstanding.
& In Japan, the NLTCI was implemented in 2000 to provide standardized services for dependent elderly (MHLW 2015a). In Italy, none of the national-level systems provides in-kind services, and there are remarkable gaps in providing care services among local governments (MLPS 2010). & In Japan, the NLTCI provides only services in-kind and not in-cash (MHLW 2015a). In Italy, in-cash services are much more prevalent than in-kind services, and the cash allowance for citizens with massive disabilities, called the BIndennità di accompagnamento,^is a unique national system available for older citizens in need of care (Pavolini and Ranci 2008: 222). & In Japan, under the NLTCI system, the agencies that directly employ care workers and the NLTCI system provide double controls for working conditions, regardless of whether care workers are migrants or not (MHLW 2015a). In Italy, both local and national in-cash-based systems scarcely have controls for their use. This gap facilitates the employment of care and domestic workers, mostly migrants, in the gray market (Pasquinelli and Rusmini 2008). & In Japan, 2.7% (2.1% for home care use and 0.6% for institutional care) of the population uses formal long-term care services, which is higher than the average for OECD countries (2.3%). In Italy, the share remains 1.4% (0.8% for home care use and 0.6% for institutional care) (OECD 2011).
The largest difference in long-term care policy between Italy and Japan is twofold: first, whether they have a universal national care service system or not and, second, whether their care policies are in-cash oriented or in-kind oriented. In Italy, the in-cashbased national care policy and the modest in-kind public care services stimulate the provision of care by (female) migrant domestic workers mostly in the gray market and by family members. In contrast, in Japan, the in-kind-based national system for longterm care (NLTCI) has been prevalent, and marketized home care services are much less common than they are in Italy.
National long-term care system in Japan: a disincentive for the marketization of care The public long-term care systems involve variations in the role of marketized care and migrant care. In contrast to Japan, where the NLTCI exists and marketized care is not very diffused, in Italy, the national/universal long-term care system is absent and migrant domestic workers are diffused. Their working conditions are frequently uncontrolled and undeclared (as shown in the next section). The Italian case has been identified as the Bprivate employment^migrant-in-the-family model (Pasquinelli and  The NLTCI is a compulsory national insurance system that has been in force since 2000, and it provides both residential and institutional care services. Space limitations do not permit further discussion on the historical and political process of the establishment of the NLTCI, but it marked a drastic change in Japanese long-term care policy and was prepared over a short period. As an institutional design, the NLTCI is largely coincident with the National Health Insurance (NHI) system, which has been in effect since 1961, with the idea that professional medical care services should be provided for all citizens and managed by the state, not by the market. Both the NLTCI and the NHI are social insurance systems, and 40-50% of their total spending is financed by general tax. Due to the stronger universalism of in-kind-based public healthcare and long-term care systems, a higher sense of entitlement as beneficiaries is shared among citizens.
The premiums of the NLTCI are set at nine levels in response to the income/pension level, and the benefit ceilings are set at seven grades based on the eligibility levels (two for the Bpreventive care^program and five for regular long-term care), which are evaluated by the national certification system. Certified citizens aged 65 and over can receive the care services they choose, equivalent to the ceiling of the benefit of their eligibility level (from $446 to $3214 per month).
For service recipients, the NLTCI system charges only 10% of fixed care costs as a co-payment fee. Therefore, for care recipients/purchasers, a competition between freemarket-based care provision and the care provision of the NLTCI would be scarcely realistic. This lower co-payment fee in the NLTCI serves as a deterrent to the diffusion of care provision in the market. Under the NLTCI system, for-profit enterprises are permitted as entitled care providers only in the institutional care, not in the field of home care. 1 To be care providers under the NLTCI, they must be approved by local governments and have adequate management and controls for both care workers/providers and recipients. Therefore, care work is under the tight control of both the employee and the NLTCI, 2 and the ability to diffuse irregular care work in the gray market without stay/work permits and/or a labor contract is scarce (MHLW 2015a). Overall, as a care policy system, the NLTCI itself constitutes the strongest disincentive for the diffusion of marketized care work, independent of the origins of care workers, and whether they are native or foreign-born.
Simultaneously, however, the current eldercare provision under the NLTCI is not sufficient (the fact that it exceeds the OECD average, as shown in the previous section, notwithstanding). There is a consistently insufficient number of care workers (MHLW 2015c). In 2015, nearly 70% of the major caregivers for the elderly were their family members. Nearly 70% of the major family caregivers were female, and nearly 70% of them were aged 60 and over (MHLW 2015b). The females of this generation mostly belong to the male full-time breadwinner and female full-time homemaker model. However, women in their 30s and 40s, who will become the major family caregivers in the near future, are no longer full-time homemakers. Indeed, due to caregiving for their older parents, a turnover figure of 101,100 workers between 2011 and 2012, with 80% of them being female, has already been identified. This caregiver turnover is recognized as a serious social issue by the Japanese government (MHLW 2015b). Overall, the Japanese care provision model may be called the Bfamilyand NLTCI-based^model.
Dualistic domestic work sector in Italy: a strong incentive for gray-marketized care Since the medieval and modern eras, the migrant-(domestic worker)-in-the-family model has long been spread in urban middle class families. The Bforeignness^of domestic workers is also a historical and conventional fact. They often come from different peripheral regions with lower socioeconomic status, for example, from rural area to cities in the medieval and modern eras, from the south to the north, from the second half of the nineteenth century to the 1970s, and from Asian (and South American) Catholic countries by the 1990s and from Eastern and Central European ex-communist countries since the 2000s (Sarti 2004(Sarti , 2010. Additionally, care work has been recognized as part of the wide variety of work in the family/private spheres. In the current national collective agreement for domestic workers in Italy, Bdomestic work (lavori domestici)^covers every kind of labor for private households, from Bstable keepers^to Bcaregivers for disabled persons^(Bettagno 2012).
The long history of the Italian domestic work sector contributes to the development of the sector. Italy became the first country in Europe to ratify ILO Convention 189 on Domestic Work in January 2013. This early ratification was achieved owing to the long history of labor union-management-government cooperation in the domestic work sector that began in the 1950s when the first national collective agreement for domestic workers was signed by these parties as a result of the historical legacy of the labor campaign for the protection of domestic workers' rights in Italy (Sarti 2010).
According to data from the National Institute of Social Security (INPS), 3 the number of domestic workers registered in the INPS was 886,125 (men, 108,328; women, 777,797), and 87.8% of them were women in 2015. In 2014, the number of migrant workers was 692,640, accounting for 77.1% of all domestic workers, and 87.8% of these migrant domestic workers were women. Additionally, 39.6% of all domestic workers were so-called badanti, domestic workers who assist elderly and disabled people, that is, care workers, in the fourth quarter of 2014.
As will be seen in the following sections, the drastic increase in immigrant domestic workers has occurred since 2002 when regularization was carried out, and this regularization and annual quota system have unexpectedly triggered the high share of irregular migrant workers. Here, however, it should not be ignored that the issue is not the migrant workers but the Italian domestic sector, which, since the beginning of the 1990s, has long been a sector with a non-regular employment rate of more than 70%. Indeed, in 1992, the share of foreign workers was 20.4%, and the share of irregular workers was 74.5% (Sarti 2004: 3).
In summary, the Italian gray-marketized care model was necessarily related to migrant domestic workers but, rather, to the characteristics of the Italian domestic sectors. In addition, it is the long tradition of the Italian domestic work sector that has enclosed domestic work, i.e., Bwork performed in or for a household or households( art.1 of C 189 of ILO), in the family and household economy and domestic workers in the private and family sphere.

Migration policy characteristics and historical backgrounds
Italy: restrictive migration system and neglected resident control for third-country nationals Italy had long been a country of emigration, and a comprehensive migrant policy was absent until the 1980s. After the Boil shock^in the 1970s, migrant workers in Northwestern Europe flowed into Mediterranean countries mostly as irregular workers due to their Bweak internal control^and the large gray market (Triandafyllidou and Ambrosini 2011;Einaudi 2007). Since 2000, large-scale regularizations and the enlargement of the EU have drastically changed the migration flow into Italy. In 2013, the number of foreign nationals in Italy exceeded 4.4 million, and the share of the foreignborn population accounted for 9.4% (Table 3). Italy has become one of the largest host countries for migrant workers in Europe.
In However, from a macro perspective, Italian migrant policy has always been characterized by a backward adoption of Bintegration,^excessively rigid (and sometimes unrealistic) migration control systems for the entry of economic migrants, and passive controls for foreign stayers already within the country's borders (Einaudi 2007;Triandafyllidou and Ambrosini 2011). In addition, its characteristics have not changed much since the second half of the twentieth century to the present. Indeed, before the Bossi-Fini Law in 2002, more than 60% of foreign residents in Italy had regular/legal status as a result of past regularizations (Blangiardo and Tanturri 2004). These circumstances allowed more migrant domestic workers to work in private families without legal documents and contracts.

Japan: disallowing/rejecting the system and policy dualism for migrant unskilled workers
At the end of 2014, the number of foreign residents in Japan was 2,121,831, and their share in the total population amounted to 1.67%. Compared to other developed countries, the share of foreign(−born) citizens in Japan remains the lowest (Table 3).  Similar to Italy, Japan had been an emigrant country rather than a host country until the middle of the twentieth century. Its migration policy has been a rigid and restricted system for the entry of economic migrants, especially for permanent and/or long-term stays, and the entry of low-skilled migrant workers has not been approved. The justification for the negative/restricted attitude of the Japanese government with regard to accepting (low-skilled) migrant workers has long been articulated in terms of concerns over Bdualization^in the Japanese labor market. In this discourse, a (lowskilled) migrant labor force, which could be considered to introduce a Bcheaper and worse^workforce, would be undesirable for the Japanese labor market (Suzuki 2009). Indeed, the Japanese government and official publications generally do not use the term Imin, migrant/migration in Japanese, and Prime Minister S. Abe recently stated on 28 January 2016 that BI absolutely will not consider introducing any migration policies (Imin seisaku)^in the congress of the House of Councilors (Sankei Shimbun 2016). In these two symbolic cases, for the Japanese government, the term Bmigrant/migrationî mplies the settlement of foreign citizens in the Japanese society, towards which the Japanese government has a negative attitude.
Conversely, it should also be noted that the official view is not always coincident with the reality. The BTechnical Intern Training Programme^(TITP) was introduced in 1993 as a means of transferring trade skills to foreign trainees. At present, 66 types and 126 categories of occupations are covered by this system, and the aim of this program is for trainees to acquire technologies and skills through on-the-job training. Nevertheless, the TITP has been criticized as a system that makes foreign TITP trainees under-valued low-skilled workers, causing their undocumented condition. In practice, the trainees of the TITP system provide low-skilled labor (or quasi-labor). In 2013, there were 137,000 foreign TITP trainees, accounting for 19% of all foreign residents/workers in Japan (MHLW 2014b; Suzuki 2009).
From a macro perspective, similar to Italy, Japan has a dualistic migration environment with restricted entry systems and a contradictory reality for low-skilled migrant workers. However, in contrast to Italy, employing private domestic care workers as low-skilled/non-qualified workers has not been regularized and diffused in Japan. This situation is also because of the rapid development of the welfare system and because since the second half of the twentieth century, under the patriarchal family system based on the strong single-male breadwinner model this sort of care and domestic work has been considered to belong to full-time female homemakers (Chou et al. 2013).

Immigration policy systems for care workers
Italy: the Bex post facto approval system^for non-qualified care workers As shown above, migrant workers in the Italian migrant-in-the-family care model are privately employed domestic workers and not qualified care workers such as their counterparts in Japan. The Italian migration policy system for domestic workers has been mostly covered by previous studies, and in general, there is a twofold migrant system for legitimate entry, stay, and work for domestic/care workers from non-EU member states.
The first is the annual quota system, which restricts the number of non-EU migrant workers who may enter based on their work status. While the quota is supposed to be the singular legal entry system, it functions and is recognized as a de-facto regularization, and/or Bmini-regularization^. The reason is that (1) there are remarkable gaps between the quota and the number of applicants and (2) small business owners and private families (mostly the employees of domestic workers) prefer to use the quota system to re-employ the migrant workers who they employed irregularly before and who are thus well known to them (Castagnone et al. 2013;Fasani 2013).
The second is the regularization system, which is supposed to be an exceptional and peculiar application of the legal entitlement to migrant workers who already stay and work irregularly in the country. Since the 1990s, regularization has been repeatedly enforced at least twice every decade and it has nearly become the ordinary system. The number of regularized cases in Italy between 1980 and 2004 is consistently higher than that in other European countries and more than 80 times the number of British cases (Barbagli et al. 2004). Furthermore, since the 2000s, the regularization systems have been enforced with several favorable treatments for domestic workers (such as the regularizations in 2002 and 2009).
In practice, there is no pre-control function for the entry and qualifications of migrant domestic/care workers in the Italian migrant policy. The Italian migrant policy for migrant care/domestic workers may be called an Bex post facto approval system^for the stay and working permit of non-qualified workers. Additionally, with the combination of the in-cash-based care system and the large spread of domestic work with the direct employment of migrant workers by families, the Italian model of care provision may be called Bfamily-based and partially regularized migrant care in the market.Ĵ apan: the pre-screening system for qualified care workers-economic partnership agreements The present Japanese migration policy requires the acquisition of a status of residence by job categories, but as a job category, Bcare worker^is inexistent. The implication is that Japanese migrant policy does not permit the economic migration of care workers as of this moment. However, the economic partnership agreement (EPA) represents an exceptional case. The EPA is an exceptional bilateral agreement for the flows of goods, services, and human resources between East Asian countries and Japan. Since 2008, 4 together with nurses, care workers have been targeted, and the candidates have come from three Asian countries: Indonesia, the Philippines, and Vietnam. These individuals train and work in institutions in Japan with a stay permit designated Bspecial activity.^Between 2008 and 2015, a total of 2078 candidates from these three countries entered Japan, and in April 2014, only 753 of them worked as EPA candidates along with those already certified as care workers (Uebayashi 2015).
As the limited shares of migrant care workers suggest (Table 3), the EPA system for migrant care workers has many rigorous requirements for them to work in Japan. In general, EPA care worker candidates are supposed to undergo the following three processes. First, they are supposed to have graduated from the faculty of nursing of a university with a 4-year course and have Japanese language training in their country of origin. They are then selected by public agencies (in their country of origin). Second, the care worker candidates come and stay in Japan as language and care work trainees, and the duration of the language training is at least 6 months (2.5 months for Vietnamese), aiming for candidates to reach the N3 level (the ability to understand Japanese used in everyday situations to a certain degree) of the Japanese Language Proficiency Test. Third, the foreign care worker candidates have to receive on-the-job training in institutions for at least 3 years, and then they have to take the National Test for Care Workers (NTCW) in the Japanese language under almost the same conditions as those for Japanese candidates.
In Japan, the NTCW is the qualification for care workers established by national law and in effect since 1988. The number of care workers who had national qualifications amounted to 660,546 in 2014, and the number of qualified workers was 1,189,979 in 2014. Both of these numbers have been continuously increasing since 2000 (MHLW 2015c). However, the number of EPA candidates from Indonesia and the Philippines decreased to less than one third between 2009 and 2011. In addition, more than half (approximately 60%) of the EPA candidates have returned to their countries exclusively due to failing the NTCW.
For EPA candidates, higher-order Japanese language writing skills such as writing work records and reports are indispensable for their qualified care work, and they are also one of the greatest obstacles to obtaining the national qualification and the settlement of the candidates in Japan. Since 2012, foreign care worker candidates can take the NTCW again when they have failed. Additionally, with the aim of increasing the number of successful foreign applicants, 5 the Japanese language wording of the NTCW has been made easier only for foreign candidates. 6 On the other hand, for the institutions that accept foreign care worker EPA candidates, there are the following obstacles. First, the institutions where the candidates work and are trained are supposed to have at least two EPA candidates per institution. Consequently, regarding the fees for agencies and administrations, the institutions are supposed to bear a cost of more than 1 million yen (which corresponds to more than three times the average monthly wage in Japan), only before the candidates make entry into Japan. Second, after their entry, the institutions are supposed to offer both monetary and human resources for their training. Especially in regard to the latter, veteran care workers in the institutions must take time to train the foreign candidates, which results in financial repercussions for the institutions (Uebayashi 2015).
Overall, the economic migrants who will work as care workers in Japan are required to clear multiple pre-screenings both before and after their entry. The Japanese migrant policy for migrant care workers can therefore be called a Bdual pre-screening systemf or qualified care workers.

Conclusion
This study discussed the dividing factors of two countries with similar welfare configurations but contrasting care models: the migrant-based care model in Italy and the nonmigrant-based care model in Japan. In short, the family-based and gray-marketized migrant care provision in Italy and the NLTCI-and family-based care in Japan bring contrasting outcomes for the conditions of the majority of migrant care workers in each country.
In Italy, migrant policy and care policy scarcely concern the regulation, training, and evaluation of migrant care workers' working conditions and professional skills. The Italian in-cash-oriented care policy does not have a universal national care service system, and these cash benefits are often unbounded by constraints-gray-marketized care can also be used. To complement their restrained in-kind care services, migrant care workers are indispensable to Italian care systems. Simultaneously, the Bex post facto approval systems^of Italian migrant policy have contributed to the wide spread of migrant care workers in the family, where the direct employment of domestic workers by private families has been largely diffused. This traditional Italian domestic work sector contributes to developing workers' rights, on the one hand, and to leaving the private sphere of the household as the breeding ground for under declaration (undeclared and slave labor), especially for unskilled, undocumented migrant workers, on the other hand. From this perspective, the Italian (elderly) care model has changed from Bfamilism by default^to Bimplicit-supported familialism,^with the (gray-marketized) migrant-in-the-family model remaining as a kind of a path-dependent option for the family model, rather than a paradigm shift to defamilization.
In contrast, the founding of the Japanese NLTCI system in 2000 led to a paradigm shift to some extent from Bfamilism by default^to Bexplicit-supported familialism,â lthough Japan's migrant policy orientation was largely unchanged until 2016. In this period, foreign care workers can enter and work in Japan only through bilateral agreements with three countries under the EPA system. This system assigns foreign care worker candidates several rigid requirements regarding their academic career, job, and Japanese language training, with the aim of allowing these candidates to pass the national test and acquire the national qualification of care workers by the time they return to their countries. In addition, the NLTCI system, which provides exclusively inkind service for older citizens, is a much stronger barrier to both (gray-)marketized care and the unskilled care workforce. In summary, Japanese dual control of care-migration policy systems currently prevent both the wider spread of care services in the market and a workforce composed of unskilled migrant care workers. Furthermore, this dual control of care-migration policy systems also prevent the spread of an irregular care workforce in the gray market that is workers with neither regular labor contracts nor regular work/stay permits.
Overall, is the combination of an in-kind-based national care system and rigid prior control of migration policy for care workers in Japan a sustainable and ideal policy model for the long-term care of the elderly? The answer is not necessarily yes. In 2016, the Japanese government lifted a prohibition on home care services by EPA migrant care workers. In 2017, legislative system reform, which allows the entry of migrant domestic workers and care workers as Btrainees^under the TITP, is planned. These drastic legislative changes in Japan may be considered a conversion from the country's current family-and NLTCI-based care model to a migrant care model, even if it has shown just the slightest sign of changing at this moment. Changes in the Japanese care model in the near future warrant continued attention, especially in regard to the reality of professional skill training and evaluation.