Transnational agency and domestic policies: The case of childcare deinstitutionalization in Georgia

This article describes a policy adoption case study about deinstitutionalization of childcare in Georgia since independence. It highlights the evolving and non-homogeneous nature of transnational agency in the area of childcare deinstitutionalization, and offers insights into the complex relationship between transnational agency and national policymaking. The analysis draws on national policy documents, reports of United Nations agencies, the European Union, the United States Agency for International Development (USAID) and non-governmental organizations that contributed to the evolution of childcare deinstitutionalization in Georgia. We trace several developments: evolution of Georgian domestic policy versus the changing role of childcare deinstitutionalization in activities of various transnational actors. We find that Georgian childcare was shifting towards deinstitutionalization at the same time as global policy actors were developing their interventions in this policy area, showing how a lower middle-income country can develop its domestic social policies in conditions of an incoherent external environment.


Introduction
The plight of Romanian orphanages in the 1990s shook the world and brought the issue of child institutionalization onto the global agenda. An important role in the childcare deinstitutionalization (hereafter, DI) reform in Romania was played by the European Union, which implemented child protection conditionality for EU accession, as well as other transnational actors including agencies of the United Nations (UN), multilateral and bilateral donors and international non-governmental organizations (NGOs). The developments in Romania, in turn, prompted a major shift in European children's rights agenda (Iusmen, 2014;Stalford, 2012), and helped mobilize international resources to address the issue of child institutionalization. The EU involvement in Romanian childcare reform 'created a precedent in terms of the extent and depth of EU action in this policy area' (Iusmen in Stalford, 2012: 219). This article looks at another country in the European neighbourhood which followed Romania in adopting a childcare DI policy -Georgia. We analyse the case of childcare DI in Georgia and ask: what was the nature of EU and other transnational actors' involvement there? In the immediately post-'Romanian orphanages' world, who were the key transnational actors influencing domestic alternative childcare policies? To what extent was their agency purposive and coherent? And how did their agency relate to the evolution of domestic childcare policies in Georgia? DI policy aims to guarantee that every child grows up in a family, or in a family-like environment (Costa and Giraldi, 2014; UN Guidelines for the Alternative Care of Children, 2009). This includes closing down large childcare institutions, preventing child abandonment, family reintegration, providing high-quality and inclusive health care, education and social support services, as well as alternative childcare services (foster care, adoption, day care homes). For children without parental care, post-independence Georgia had a Soviet system of residential institutions. Between 2003 and 2014, Georgia reduced the number of children in residential care from 8155 to 1100, that is, by about 86% (calculated on the basis of TransMonEE, 2016), becoming one of the first lower middleincome countries to deinstitutionalize childcare. While formal policy adoption and related statistics are rarely indicative of the actual depth of reform and quality of policy implementation, or impact, we leave a more comprehensive evaluation of policy outcomes outside the scope of this article, as such analysis would require a methodological approach and source material different to that employed here.
Below, we trace how DI policy developed in Georgia, how transnational actors contributed to this development, and whether this reflected the mainstreaming of childcare DI in European and global policies. We discuss what this tells us about the nature of transnational agency in the Georgian case, and more broadly about domestic social policy formation in conditions of emerging transnational agency. Looking at social policy changes through the lens of transnational agency is particularly justified in countries like Georgia, one of the post-communist countries, which have been strongly shaped by the 'pervasive influence of transnational actors' (Orenstein et al., 2008: 1). Transnational actors can be individual or collective, including, for example, international governmental and NGOs, foundations, activist groups and networks, as well as 'individual policy advocates and translators' (Deacon and Stubbs, 2013: 15). Since the 1990s, the European Union has arguably been the most prominent transnational actor in the region (e.g. Orenstein et al., 2008). The case study method (e.g. Gerring, 2016) will provide space for sufficiently detailed analysis, based on national policy documents, reports of UN agencies, the European Union, the United States Agency for International Development (USAID) and relevant NGOs. The article first discusses the emergence and development of DI policy in Georgia, in the context of diverse transnational actors and their activities in the area of alternative, out of home childcare. The discussion section considers the shifting transnational DI agency and roles played by different global policy actors in the DI process in Georgia.

How policy reform unfolded
The DI reform unfolded in conditions of globalization, economic growth and targeted support of Georgian child welfare and DI reform by transnational actors. At independence in 1991, Georgia was a small, conflict-ridden and poor country (World Bank Group, 2018), with an inadequate system of public social assistance. Between 1997 and 2016, however, the Georgian economy grew at an annual average rate of 5.5% (World Bank Group, 2018: 5). Favourable economic ground for reform was created by significant foreign aid, which Georgia received in the wake of repeated crises.  (World Bank Group, 2020). In these conditions, Georgia carried out governance and public sector reforms (The World Bank, 2012;World Bank Group, 2018). One of these reforms was the DI of childcare.
The DI reform in Georgia can be viewed as a two-stage process. In the late 1990s-early 2000s, the government made its first attempts at DI, including the first DI pilot projects (UNICEF Regional Office for CEE/CIS, 2006), setting up a DI working group (1999), training social workers (OCHA, 2003b) and adoption of the Georgian National plan of action for children 2003(Government of Georgia, 2003. In 2003, the Rose Revolution, a peaceful power change, removed President Shevardnadze, associated with Georgia's Soviet past and installed a pro-reform, pro-Western government. From this point, the DI reform moved to the next stage, with allocated public funding (Irimia and Shatberashvili, 2015;UNICEF, 2009). In 2005, the Georgian government introduced 'a moratorium on entries in large size institutions under reorganization' (Irimia and Shatberashvili, 2015: 60). The Government's Plan of Action for Child Protection and Deinstitutionalization 2005-2008 included optimization of childcare institutions; administrative and legal reform of childcare and registration of children with disabilities (UN Convention on the Rights of the Child, UN CRC, 2007: 6). The whole child welfare system was overhauled and integrated with the system of social assistance (O'Brien and Chanturidze, 2009). Social assistance became needs-based, allowing better targeting of vulnerable families. Alternative childcare services, such as foster care and day care centres, were expanded; a new gatekeeping policy and family reintegration support were introduced. With the adoption of the new Government's Child Welfare Plan of Action 2008-2011, the child welfare reform broadened its scope from mere DI and reducing the number of children in institutions to ensuring welfare of all children (Government of Georgia, 2008). The DI reform continued a new law on adoption 2008; a national campaign to promote foster care and introduction of a family reunification package (Cantwell et al., 2012: 45;European Commission (EC), 2010;Greenberg and Partskhaladze, 2014;O'Brien and Chanturidze, 2009: xi). By the late 2010s, Georgia was on a par with Western European countries in terms of childcare DI, the DI reform expanded to include more children with disabilities and non-state children's homes (Public Defender of Georgia, 2015;UNICEF, 2015).

Transnational agency: individuals and NGOs
Georgian DI policy shift was enabled by newly emerging local, non-state DI initiatives. As the country became more open after independence, childcare professionals from countries, which were more 'advanced' in DI, started arriving in Georgia and bringing in new ideas and practices of working with children. Often, individuals and organizations came initially with humanitarian aims to help meet the basic survival and health needs of the population. As the economic situation improved and immediate needs of food, shelter and basic health care were addressed, those actors switched their attention to other problems, including the issue of child institutionalization.
In 1996, child psychologists Dr Jane-Mary Castelfranc-Allen Rawls (registered clinical psychologist, the New Zealand Psychological Society) and Dr Barry S. Parsonson (registered clinical psychologist, the University of Waikato, NZ) visited Georgia and stayed for nearly two decades to teach hands-on skills to professionals working with special needs children. To publicize the plight of children, in 1998, they set up the Children of Georgia Project, which involved international knowledge exchange trips to and from Georgia, to train personnel in family-and community-based care for disabled and special needs children, with a view to removing them from large childcare institutions. The project later developed into the NGO 'Children of Georgia', which in 2005Georgia', which in -2012 carried out a number of projects to support the DI of disabled children and provide technical support to the government for the child welfare reform. They advocated for the DI reform with the government, worked with ministries and UNICEF, assisted in developing policies and programmes to deinstitutionalize disabled children. So, although their activity was mostly visible during the intensive reform period since the mid-2000s and the NGO itself was officially registered only in 2006, the foundation for their work was laid much earlier in the late 1990s.
In 1998, the First Step charity was founded in Georgia by Jane Corboy, wife of the first EU Ambassador to Georgia and Nino Kidigidze Zhvania, wife of the to-be prime minister of Georgia. After their first attempts to improve the services for children with disabilities within existing institutions through staff retraining programme proved unsuccessful, they turned to international experience and set up a new community-based programme. They set up pilot programmes to provide care for children with special needs, introducing the new professions of occupational therapy and social work to Georgia. They supported family reunification, prevention of abandonment and opened the First Step Village, initially for 12 children, in 2002. In 2003 Step launched integrated education in one of Tbilisi's schools, one of the first steps towards inclusive education in Georgia. In 2001, the First Step social workers engaged in the state-run DI programme, 'Deinstitutionalisation, children back to their biological families', and partnered with the government on the DI reform after the change of government in 2003. In 2005, they opened a day care centre for children with special needs in the village. By 2008, the First Step developed into a major player in DI implementation: it ran training programmes for caregivers upon UNICEF request, and organized DI policy trips for Georgian policy makers and social work practitioners to Ireland. Ten years after they started, in 2009, the First Step reached their goal of closing the Kaspi children's home. Since 2010, they closed their housing cottages, focussing instead on family support services, early intervention programmes and staff training. Importantly, although now the First Step, as an innovative provider of training and services for children with special needs, is a key player in the DI reform in Georgia, back in 1990s, it started without a clear DI idea, but rather with an idea to merely improve the conditions within institutions. Moreover, their first attempts to bring about change date back to 1995, when they partnered with UNICEF, Echo, Save the Children and Caritas and set up a programme to improve the lives of children in four orphanages. Their programme ran for a year and cost US$200,000, but with no result. 'There was little or no difference to the lives of the children. They still lived in a vacuum. Staff attitudes had not changed' (Corboy, 2009).
The NGO New Life started operating in 1998: Nata Rostomashvili and Mariam Dagundaridze, supported by the American Friends of Georgia, started a school for children with mental and physical disabilities in Telavi, Kakheti Region. The objective was to provide sensitive day care and education to these children and an alternative to institutionalization. So, by the late 1990s, a number of NGOs had started to provide services without which effective DI would not be possible in Georgia -alternative to institutions childcare settings, integrating children with disabilities into society and so on.
The 1990s was thus a time when various new actors and initiatives emerged, working to improve the situation of children in institutions. These initiatives led to the wide realization that the old approach of trying to improve the quality of life in institutions was not effective, and a more radical solution, that is, DI, was needed. These organizations later supported the government's DI reform and in the 2010s received foreign funding from the EU, UNICEF, Save the Children and so on. However, in the 1990s, they were not part of any organized transnational Georgia-targeted activities.

Major transnational actors
Georgia embarked on the DI reform at a time of increased international interest in children's rights and welfare, with a series of major agenda-setting events, such as the 1990 World Summit for Children, the International Year of the Family 1994, the UN General Assembly Special Session on Children 2002, the Stockholm Conference on Children and Residential Care 2003 and the rise of new DI-focussed transnational actors, such as Hope and Homes for Children (1994) and LUMOS (2005). After the plight of children in Romanian orphanages was revealed in the 1990s, child policies attracted international attention (e.g. 1999 World Report of Human Rights Watch with its special section 'Orphans and Abandoned Children'; 'Towards an EU Human Rights Agenda for Children' [Save the Children, 1998]; 'A World Fit for Children' [UNICEF, 2002]; 'Children on the Brink' [UNAIDS, UNICEF and USAID, 2002]). From around 2000, UNICEF focussed on supporting the reforms of childcare, including DI, in Central and Eastern Europe (see 'The Budapest Statement'). In other words, the DI reform in Georgia occurred at a time of increased global attention to child policies.
Although children in institutions were on the international agenda, there was no significant pressure on Georgia to reform from outside the country in the early stages of DI reform. For example, the above-mentioned Human Rights Watch report focussed on the 'big' cases of Romania, China and Russia, but did not mention Georgia (HRW, World Report 1999), and neither did any of its other reports from the 1990s. There is no evidence that the above-mentioned external events directly led to reforms in Georgia, but they possibly helped the transmission of the idea of DI and some local DI-related projects into Georgia.
In 1994, Georgia acceded to the UN CRC. The convention (UNHR, 1989) provided a generic framework for reform, and asserted a child's right to grow up in a family environment. Starting in 2000, the UN Committee on the Rights of the Child repeatedly highlighted the need for DI reform in Georgia and provided more detailed guidance in its Concluding observations. The list of issues to be taken up in connection with the consideration of the initial report of Georgia encouraged Georgia to make institutions more child-friendly and increase its DI efforts, although no direct recommendation to reduce the number of children in institutions was made: ensure the periodic review of placements in institutions; . . . introduce measures to guarantee and protect the human dignity of children living in institutions and to make these institutions more child friendly . . . to increase the level of resources allocated for the protection and care of children deprived of a family environment . . . increase its efforts to prevent institutionalization and pay special attention in that regard to vulnerable families by, for example, providing them with adequate support services . . . provide adequate resources, both financial and human, for effective implementation of the new law on foster care; introduce programmes to raise awareness and promote foster care . . . (UN CRC, 2000: 7).
In 2003, the Committee recommended to 'continue measures to strengthen support for families to enable them to care for their children at home', 'strengthen and intensify the programme of de-institutionalization' and so on (UN CRC, 2003: 7-8).
UNICEF has been the main partner of the Georgian government in DI reform since independence. UNICEF coordinated the activities of other partners, leveraged additional funding, provided policy advice and expertise and was involved in all key aspects of the reform (Irimia and Shatberashvili, 2015;UNICEF, 2013). Between 2000 and 2005, UNICEF was expanding its activities on the ground across the region and at the same time, building their own capacity to support the DI reform in the countries. In 2003, 'to guide country offices on a common agenda for child care reform', UNICEF and the World Bank developed 'toolkits for reformers' in the following three key areas of DI reform: on improving the gatekeeping system; on the standards of services and on how to reallocate resources from residential care to community-based services (UNICEF, 2013).
The first national-level attempts at DI mentioned above were joint efforts of national and transnational actors. For example, the DI working group included UNICEF, EveryChild, World Vision International, the First Step Foundation, the World Bank, representatives of the Georgian government, Ombudsman's office and a few others. Georgia has maintained relationships with the EU since the 1990s, and became a member of the Council of Europe in 1999. However, to our knowledge, their representatives were not part of the DI working group. Unlike Romania, the European Union did not impose child welfare conditions on Georgia in the 1990s. Rather, at the initial stage, the European Union indirectly influenced Georgian childcare institutions by providing budgetary and technical support, primarily through EC Technical Assistance to the Commonwealth of Independent States (TACIS) and the EC Food Security Programme (EC FSP), which started with generic aims of poverty reduction, financing family allowance benefits and ensuring basic social provision, such as food and health care. It also funded existing childcare institutions, helped set up their inventory and supported their refurbishment (Tasker and Ispiryan, 2010: 10-11). No DI conditions were initially attached to EU funds.
From 2003 Thomas et al., 2006: 41). The EC supported child welfare reform through FSP and TACIS until 2010, after which EC assistance related to DI only indirectly -through projects in health care, vocational education, justice, issues of internally displaced peoples and so on (Tasker and Ispiryan, 2010).
In the post-revolution period, other foreign donors also started actively support DI work in Georgia. In 2004, the World Bank funded a pilot School Feeding Project for Deinstitutionalization (UN WFP, 2004). UN agencies and NGOs like the British charity EveryChild were actively helping the government to run pilot projects, building the necessary human resources, to enable the realization of DI reform on the ground (OCHA, 2003b; UNICEF Regional Office for CEE/CIS, 2006).
From 2005, USAID's Displaced Children and Orphans Fund provided funding for the DI reform in Georgia (Civil Georgia, 2004;United States Congress 109-101, 2006: 16, 64), through UNICEF and NGOs (Thomas et al., 2006: 45;UNICEF and USAID Georgia, 2014). At the time when USAID started supporting the DI reform in Georgia in 2005, DI promotion was not yet systematically part of the US government's external action. It was only at that point, in 2005, that the United States adopted the US Assistance for Orphans and Other Vulnerable Children in Developing Countries Act, which focussed US development programmes on family support and community-based programmes (U.S. Government, 2005Government, : 2113. Since 2010, USAID increased its support to the DI reform in Georgia, upgrading the infrastructure, supporting family reunifications, developing foster care and small group homes.

Discussion: transnational agency and domestic policy development as co-evolving processes
Transnational DI advocates and supporters, such as the EU, the UN and USAID, were not all equally involved in Georgian DI reform from the start. The emergence and development of DI policy in Georgia was shaped by various global and national policy actors who played different roles in this policy area at different times.
From the mid-1990s, individual, grassroots initiatives emerged with the general aim to improve conditions in childcare institutions. It was not uncommon for these initiatives to be set up by foreigners, in conditions of intensified globalization and cross-border migration, which brought new people with high agency potential and possibly different views on what constitutes acceptable childcare. These initial DI activities have been little considered in previous literature about DI reform in Georgia (e.g. Calinescu and Calinescu, 2016;Greenberg and Partskhaladze, 2014), with more emphasis placed on the Rose Revolution and international donors. The Rose Revolution undoubtedly provided a strong impetus for reform, however, the foundation for policy change (knowledge, ideas, skills, etc.) started developing earlier, including through immigrants' transnational agency. The potential of such transnational individual agents to transform domestic social policies is little studied. The global social policy literature, although putting international actors in focus and recognizing the importance of individual actors (Deacon and Stubbs, 2013), has often tended to interpret transnational agency as international organizations, NGOs or supranational organizations (Foli, 2016;Kaasch et al., 2019;Kaasch and Martens, 2015), rather than individual migrants. While a strong focus on major organizational actors is unsurprising, it ignores the roles of individual, independent transnational actors, who in their daily lives and activities re-assert and reproduce social ideas and practices they brought with them from other social systems, and are likely to turn to international experience in search of solutions (as in the example of the First Step NGO). Furthermore, when individual agents become the focus of analysis, it often concerns individuals in position of power or within prominent existing organizational structures, for example, civil servants and staff of international organizations (Deacon and Stubbs, 2013). Our case highlights the need to better understand the policy-transformative role of 'ordinary', non-elite transnational migration. It raises the question to what extent and under what conditions such individual agents transform the social space around them, with knock-on policy effects. While international organizations can act as norm entrepreneurs that formulate policies and guidelines, individual transnational agents outside established organizational structures can 'make a difference' in other ways -through iterative and adaptive practice of social transformation, rather than policy formulation. Such was, for example, the case of the founders of the First Step charity in Georgia, whose activities saw an evolution from improving existing services in institutions to a more radical DI, when initial efforts failed to deliver. The Georgian story suggests that 'ordinary' individual transnational actors, as they moved from ad hoc initiatives to institutionalized partners of the government and key international organizations, through meetings, delivering training and organizing international learning programmes, were in a position to transmit new policy ideas to national policy makers and practitioners. Our case, therefore, raises questions about the contribution of knowledge, skills and ideas transfer by expatriates to national social policy transformations and the conditions when expatriates' agency can become a leverage for policy reform.
For many years, state funding of Georgian children's homes and boarding schools was inadequate and children's institutions relied on the assistance of international aid agencies (Implementation of the Convention . . ., 1999: 14). In the early 2000s, there were no strong international advocacy coalitions to lobby for the DI of childcare in Georgia. In fact, it would have been probably 'easy' for the government to obtain more aid for children's institutions and maintain the status quo, rather than embark on reform. As late as 2004, reports highlighted the need for 'better coordination of efforts among donors and NGOs and the relevant government ministries involved in care and assistance to children in institutions to avoid donor initiatives which could be counterproductive for the deinstitutionalization process' (OCHA, 2004). So, the earlier lack of DI reform in Georgia was not against the trend, and the initial lack of political will for reform in a way mirrored actions of international donors, who were maintaining institutions rather than unanimously lobbying for DI reform. Since about the mid-2000s, donors and humanitarian partners started explicitly urge the Georgian government to embark on the DI reform, specifying the concrete steps to be made (OCHA, 2003a(OCHA, , 2004, and provided funding for it. This situation, when international donors funded residential care, creating barriers to a transition to community-based care, was not unique to Georgia. Until very recently, in various countries (e.g. Flagothier, 2016;LUMOS, 2015), international donors continued to prop up institutions, by funding and trying to repair and 'fix' them, rather than radically reform or dismantle.
UN agencies played an important role in the 1990s and early 2000s. In 1990s-2000s, the Georgian DI reform was unfolding with encouragement and guidance from the UN Committee on the Rights of the Child. However, it was only a decade later, in 2009, that the UN issued more specific guidance -the UN Guidelines for the Alternative Care of Children, aimed at enhancing the realization of the UN CRC, particularly concerning the provision of a family environment. By this point, Georgia was already well into DI reform. Georgian DI shift was thus unfolding in parallel to the UN developing its governance tools for alternative care of children.
At the time of initial DI policy emergence in Georgia, the EU could be described as a transnational agent for DI, only in terms of being a cluster of DI advocates, first disjointed and later turning into an advocacy coalition. Changes on the ground were propelled to a large extent by not by the EU as an organization with well-defined child welfare policies or child-oriented foreign policy, but rather by various individual actors and charities from Europe, in particular, the United Kingdom: EveryChild/European Children's Trust, First Step and Save the Children.
The EU provided ample assistance to the DI reform in Georgia, however, the nature of their involvement changed over the years, and the clear focus on DI was absent during the early stage of reform. The EU was not directly involved in the DI until mid-2000s.
And even when it got involved, its pressure was not very firm. In addition to finance conditionality, the EU had other potential 'hard' tools to influence Georgian child welfare policy, for example, through its trade scheme, the Generalized System of Preferences Special Incentive Arrangement for Sustainable Development and Good Governance (GSP +). Georgia benefitted from GSP + additional trade preferences in 2005-2016, in return for ratifying and implementing certain conventions, which included the UN CRC (Report on assessment of the applications . . ., 2013: 2). The progress of child welfare and DI reform was explicitly mentioned in the GSP + reports for Georgia (Joint Staff Working Document . . ., 2016: 134-135; Report on assessment of the applications . . ., 2013: 14). The potential leverage of this tool was likely to be significant, given that the EU was Georgia's main trade partner (European Commission (EC), 2018). However, we could not find any evidence that the EU ever applied this leverage.
The initially limited role of the EU is unsurprising given when the DI of childcare became an important and recognized part of EU internal and external policies. The risks of child's separation from the family and deprivation of maternal care, have been known since the studies in 1950s (Council of Europe, 1996: 118), and in particular, the work of John Bowlby and his 1951 report 'Maternal Care and Mental Health'. However, this did not lead immediately to policy change in Western Europe. Even though DI of childcare has been a policy trend in some European countries at least since World War II, as well as part of many recommendations of the Council of Europe, until very recently, it was deemed a domestic matter and was not regulated through EU internal or external policies. In 2009, the European Expert Group on the transition from institutional to communitybased care was set up, as a pool of expertise and advocacy coalition of major NGOs (EEG, 2018). The 2009 report of this group (European Commission (EC), 2009) was the first EU high-level document which offered not only a rationale for DI, but also practical implementation advice for EU Member States. It recommended using EU structural funds for DI, and ensuring that no funds were invested in maintaining institutions (EC, 2009). In 2013, Eurochild, an advocacy network co-funded by the EU, started the Opening Doors for Europe's Children campaign, to support DI efforts in Europe. The EU's Structural Funds Regulations for 2014-2020 made the transition from institutional to community care, an investment priority (EC, 2013a(EC, , 2013b and the (European Commission (EC), 2014: para 98) banned the use of structural funds for investment in residential institutions as some EU Member States had previously used it (EEG, 2016;Parker and Bulic Cojocari, 2016). The gap between evidence of the detrimental effects of institutionalization and the mainstreaming of DI policies within the EU turned out to be more than 50 years. It is not uncommon, however, for the EU to retain controversial social policies among Member States, while promoting these policies externally. For example, similarly to the early days of EU external action in the area of DI, the European Union promoted minority protection norms beyond the EU, although it 'lacked a foundation in EU law and remained controversial, even in the "old" member states' (Sasse, 2013: 47).
In the 1990s, the EU did not have a coherent strategy for external action with regards to the rights of the child. The two notable instances were the ad hoc DI interventions in Romania and Bulgaria, where EU accession negotiations were made conditional on the reform of childcare system and DI (Accession Partnership, 1999: 8;European Commission (EC), 2003b: 9, 2006b: 20; IP/99/751). In the Romanian case, the EC accession agenda to a great extent resulted from the individual initiative of the EU rapporteur on Romania, Emma Nicholson (Iusmen, 2013), who suggested setting up a high-level group to support the reform in Romania (Save the Children Sweden, 2001: 15). The Romanian DI model, later translated to other countries, was also largely a 'random' event, rather than systemic EU effort: a British social worker Georgette Mulheir came to work in Romania in the early 1990s and designed the model. Overall, however, children were still mostly invisible in EU policies, and the European agenda for children still needed to be established (Ruxton, 1999).
In the early 2000s, children's needs and rights were gradually gaining prominence on the EU external agenda. In 2005, the EU endorsed the UN Framework for the protection of orphans and vulnerable children living in a world with HIV and AIDS, which stressed the highest priority of providing children with a family environment (UNAIDS and UNICEF, 2004). Children's rights, 'the needs of orphans and vulnerable children' were moving towards becoming part of political dialogue with third countries (European Commission (EC), 2005a: 3). The EC increasingly promoted country strategies that gave preference to 'family-based and community-based care', 'as an alternative to institutional care for orphans and vulnerable children' (EC, 2005a: 12). However, it was not until 2006 that the EC issued its Communication 'Towards an EU Strategy on the Rights of the Child', which laid down an objective to mainstream children's rights in EU actions and promote the rights of the child in external relations. Children were a central theme of the new EU development policy 'The European Consensus on Development' (2006/C 46/01), although it did not specifically mention children in institutions. An accompanying inventory of past EU actions highlighted the 'weak visibility of EU actions in favour of children's rights' as well as the urgent 'need for a coherent and comprehensive Community strategy on the Rights of the Child' (European Commission (EC), 2006a: 4, 15). It admitted that 'the EU's approach had, to date, lacked a clear coordination and overriding objectives' (European Commission (EC), 2006a: 16), there was a need for more consistency between EU's internal and external policies, between EU and other international actors' actions. Strikingly, this recent document from 2006 mentioned only 'a potential European Union Children's Policy' (European Commission (EC), 2006a: 12).
The Lisbon Treaty (2007) signalled an increase in the EU's commitment to the protection of the rights of the child in its external action (Article 3). In the same year, the first EU Guidelines for the Promotion and Protection of the Rights of the Child came out, focussing on operational tools to promote children's rights in third countries. In 2008, the European Commission issued the Communication 'A Special Place for Children in EU External Action' (COM (2008) 55 final), which introduced an EU Action Plan on Children's Rights in External Action. The current version of the EU Guidelines for the Promotion and Protection of the Rights of the Child guides the EU to encourage partner countries to review their legislation with a view to 'promoting alternative care for children and providing them with appropriate support to participate in community life and to access mainstream services' (European Commission (EC), 2016: 19). The above shows that EU external action for children, and in particular, the strong focus on DI, developed only recently, in the late 2000s.
It is well known that the EU's agency is not limited to formal guidelines and conditionality. The EU uses different instruments to exercise soft influence on other countries, and enhance compliance with 'less formally institutionalized EU preferences' (Lavenex, 2013). The not-yet formalized ideas and norms have multiple ways to spread, for example, through socialization and internalization of European cultural elements (e.g. Flockhart, 2010;Schimmelfennig, 2012). The scope of this article does not allow us to cover all these potentially DI-relevant modes of EU agency and mechanisms of influence. However, with regard to formal provisions, our case suggests that the DI policy in Georgia was developing in parallel to regulations and programmes concerning child protection policies of the EU, not in consequence thereof. The Georgian DI working group and the first DI pilots were taking place in Georgia when issues relating to the well-being of children were within the competence of EU Member States, and although the concept of children's rights was increasingly gaining international importance, the EU was not a strong international actor in this field. The DI policy developments in Georgia in the late 1990s-early 2000s cannot, therefore, be ascribed to well-coordinated efforts of the EU with other international actors, nor to internally consistent, targeted external action of the EU.
USAID also institutionalized its support for DI reforms in other countries only after the Georgian reform was already well under way. In 2012, in the wake of the US Government Evidence Summit: Protecting Children Outside of Family Care, the US Government Action Plan on Children in Adversity came out -'the first-ever whole-ofgovernment strategic guidance for US Government (2018) international assistance for children'. The plan directed U.S. Government (2012: 25) assistance to families and prevention of 'unnecessary family-child separation'. In 2013, to promote the implementation of the Action Plan in low-and middle-income countries, a public-private partnership Global Alliance for Children was established.
It was not until the 2010s, when Georgian reform was already at an advanced stage, that DI became a major part of cooperation programmes with other foreign partners. Transnational agency for DI in Georgia was a work-in-progress, evolutionary process, whereby foreign intervention co-evolved with DI reform in Georgia, rather than being a streamlined package of actions, pre-made policies or solutions to copy. The first institution-related initiatives funded by international organizations tended to be ad hoc and sometimes mutually contradictory -some projects supporting DI and some projects helping improve the conditions in existing institutions. When the first DI-related projects started in Georgia and even when DI accelerated after the Rose Revolution, the international development community was far from being a united, multi-vocal chorus of DI advocates. Not only was transnational agency for DI inconsistent and immature, but actors were also promoting the ideas of children's rights, inclusion and DI, before Western European countries (EEG, 2016;Parker and Bulic Cojocari, 2016) and the United States (ACF, 2015) themselves completed their DI reforms. During the second, 'organized' stage, transnational actors built a united advocacy front, with mainstreamed DI policies, aimed at national actors and accompanied by targeted resources and issue-focussed action. It was not a case where donors came with clear ideas and solutions and 'fixed' the problem, but rather supported the reform through iterative adaptive intervention.

Conclusion
The DI of childcare in Georgia demonstrates some well-known features of transnational agency (e.g. the variety of soft and hard influence mechanisms at work), and also shows the possibility of significant policy transformations in a low-and middle-income country before the policy innovation is mainstreamed and institutionalized by powerful Western actors. It is a case of social policy modernization, in conditions when transnational agency in the relevant policy area is not homogeneous or coherent.
The Georgian case shows that the roles and approaches of transnational actors can vary. This observation agrees with existing research (e.g. Foli, 2016), which suggests that transnational actors can employ multiple strategies to influence social policies in third countries, including 'ideational, institutional, and material incentives', going beyond coercion and conditionalities. In the case of out-of-home care policies, international organizations like the United Nations Committee on the Rights of the Child and UNICEF can use their knowledge, authority and legitimating power; the European Union and USAID can use the leverage of material incentives as tools of policy change. Individual agents can build on their past experiences and expertise (such as a background in child psychology in the discussion above), and use social capital, soft skills (e.g. to set up innovative projects, organize cross-border cooperation networks, professional training), to change practices bottom-up and to make policy change feasible. Individual agents are likely to use persuasion as a tool of influencing policy change when they work with the government and international organizations on policy design, especially at later stages, once their agency has been institutionalized, as, for example, in the case of 'Children of Georgia' NGO. The Georgian case suggests that some approaches can be more relevant than others at different stages of policy change, especially in the situation of relatively early policy adoption. In the early, 'unorganized' stages of policy reform, transnational DI agency was to a large extent about the introduction and legitimation of new ideas, and non-systematic, disjointed initiatives, aimed at finding practical ways to implement these ideas. Naturally, the collective transnational agency in the area of childcare was not streamlined for some time, however, this did not prevent a lower middleincome country like Georgia from adopting the policy.
Remarkably, Georgia started its DI reform at the time when DI was not yet mainstreamed on the international agenda. A lower middle-income country did not simply copy policies from high-income countries as a matter of coercion or emulation, but rather they developed their respective policies in a parallel and to some extent cooperative learning process. Now, DI can be rightfully called a global public policy, with a growing number of national and international commitments. However, it was not so when Georgia first stepped on the path of DI reform.
Georgia embarked on the DI reform at an interesting moment, when several global trends intersected: the rise of child welfare on the European and global agenda; ideological shift in international assistance -from needs-based charity to rights-based development, from philanthropy to rights (Slim, 2002). But these trends were only maturing at the time of DI reform in Georgia, and the international pressure for DI was only yet consolidating. Major global governance actors, including UN agencies, the EU and USAID, were not univocal on DI at the time when DI reform started in Georgia. The EU developed its foreign (and to a great extent domestic) policies on DI in parallel to Georgia. In the case of Georgia, adoption of DI policy was not only about domestication of the DI idea, but also domestication of the idea in conditions of a certain 'ideational' uncertainty. The European Union, even though a key enabler and origin of some key actors in DI reform in Georgia, was not a global policy frontrunner. Neither did other global governance actors form a united front of DI advocates, with a clear roadmap for action from the early days of reform. Transnational agency for DI and domestic DI policy development in Georgia were parallel, co-evolving processes, rather than one strictly following the other. The nature of transnational agency for DI evolved from patchy initiatives to a unified front of actors with a relatively consistent agenda and actions. And it did so during the same time period as the DI policy was implemented in Georgia.
The DI-relevant transnational agency was exercised by multiple policy actors with various agendas at different periods, ranging from high-level UN-and UNICEF-led path-breakers -to smaller scale, but probably, no less important grassroots, strengthened by individual transnational migration. Our case shows transnational transformative agency as multi-form, inter-related even if often incoherent, continuous and evolving. It suggests the importance of analysing policy change and evolving surrounding agency as a continuum.

Funding
The author disclosed receipt of the following financial support for the research, authorship and/or publication of this article: Olga Ulybina acknowledges funding from the European Commission through the Marie-Curie Fellowship no. 742981 'EUCHILD'.