Explaining the Cross-National Pattern of Policy Shift toward Childcare Deinstitutionalization

Abstract Why do some countries become early policy adopters and fast policy implementers? We investigate this question through the analysis of cross-national policy shift toward childcare deinstitutionalization, i.e., the transition from institutional to community-based provision for children without parental care. The article presents the newly collected data on the adoption of childcare deinstitutionalization policy by 15 countries – previously republics of the Soviet Union. Qualitative comparative analysis is employed to explore the role of national-level attributes affecting the timing of policy adoption and the rate of implementation. Expectations from political-economy and the institutional world-society perspectives on cross-national patterns of public policy adoption and implementation are incorporated in the hypotheses. Two respective groups of factors are considered: (1) the national economic system and the number of children in institutional care; policy implementation capacity (gross domestic product and government effectiveness); (2) world society ties to international organizations (IOs), conferences and European institutions; and local receptivity to world culture, operationalized as the contraceptive preference for pill. Results suggest that countries’ ties to policy-relevant international organizations are an important condition for earlier policy commitment, which is in line with institutional arguments. The findings point at the need for a more refined conceptualization of cross-national policy patterns for cases where world cultural drivers interplay with political-economic factors.


Introduction
Under what conditions do countries become early or late policy adopters? What conditions lead to faster or slower policy implementation? This paper investigates these questions, using the case of childcare deinstitutionalization. The term deinstitutionalization of childcare (DI) refers to a shift in policy and practice of caring for orphans and children deprived of parental carefrom a residential institutional environment to a family environment in the community (Cantwell et al. 2012). With DI policy, one acknowledges the suboptimal nature of institutional care, compared to family-like residential care and to family-based care by the extended family, adoptive or foster parents, and kafalah (the practice of guardianship in Islam). Advocates of DI emphasize that deinstitutionalization should not be reduced to the closure of institutions but understood more broadly as a complex change in childcare provision, aimed at downsizing, transformation and closure of institutions, as well as development of services in the community that would prevent separation of children from their parents (UNICEF 2010: 52).
The deinstitutionalization of childcare is an increasingly important issue on the international agenda, as millions of children continue to be brought up in institutional care (Desmond et al. 2020;Petrowski et al. 2017). Child institutionalization is often seen as a major global problem, primarily due to its multiple adverse impacts on the child's wellbeing and life-long consequences. Exposure to institutional care has been associated with slow physical, cognitive and socio-emotional development of children, whereby longer time of exposure diminishes the chance of recovery (Johnson et al. 2006;van IJzendoorn et al. 2020).
The issue is particularly relevant to the countries of the former Soviet Union, which inherited a legacy of extensive child institutionalization. At the fall of the USSR, the total number of children in residential care in the region was estimated at about 800,000 (Carter 2005: 20). Despite joining the global deinstitutionalization movement, the level of institutional care in these countries has remained high, with the level of child institutionalization per 100,000 of child population in the Central and Eastern Europe and the former Soviet Union was nearly fifteen times higher than in the United Kingdom in the early 2000s (Carter 2005: 20). Today, Europe and Central Asia have the second highest number of children living in institutions in the world, estimated at over one million (Desmond et al. 2020). In other words, DI policy in this region affects a large target population, with major social implications.
A growing number of studies discuss childcare deinstitutionalization reforms in the former Soviet region. However, they do not analyze policy patterns and factors in a systematic way, as they tend to be qualitative single case or small-N studies (e.g., An and Kulmala 2021;Babington 2015;Bindman et al. 2019;Huseynli 2018;Kuuse & Toros 2019;Pivoriene 2020;Ulybina 2020). It is common to focus on hindrances to effective policy implementation, rather than on explaining patterns of DI policy adoption and implementation. This paper will address this gap by asking two questions: Under what conditions do countries become early versus late DI policy adopters? And what conditions lead to faster versus slower DI policy implementation? Answering these questions may help us better understand the driving forces behind DI policy commitments and explain the extent to which these commitments are realized. By doing so, the paper contributes to both applied comparative policy research, as well as broader theoretical debates about conditions, drivers and nature of social change. We address these questions by drawing on both politicaleconomic theories as well as neo-institutional world society theory, to analyze data for 15 ex-Soviet countries through qualitative comparative analysis. Below, we explain our theoretical framework, data and method, followed by the presentation of results and discussion.
Theoretical framework: Political-economy and neo-institutional approaches to DI Existing empirical research points at some possible drivers of childcare deinstitutionalization: (1) DI is an internationally promoted policy; (2) DI policy has a two-fold nature which may make it attractive to policy makersit is a policy grounded in human rights and increasing child welfare awareness, as well as associated with reducing the cost of childcare service provision (e.g., An and Kulmala 2021;Babington 2015;Knapp et al. 2011;Kuuse & Toros 2019;Carter 2005: 64;Johnson et al. 2006;Shen et al. 2017;Tobis 2000;UNICEF 2018). The latter two facets of DI policyits affinity with ideas of children's rights advocates and alignment with neoliberal interests and cost-benefit considerations of governmentsconstitute the grounds on which the policy is commonly promoted (e.g., UNICEF 2010).
Current research into isomorphic public policy trends offers two broad perspectives, with different predictions. Firstly, we consider DI in the context of political-economic theory and secondly through the lens of world society theory.

Political-economic theory
Although our focus is specifically on childcare deinstitutionalization (which usually covers children with and without (mental) health issues), we also draw on studies of adult mental health deinstitutionalization, as an immediately related and extensively researched policy area. Both child and adult deinstitutionalization have been linked to the wider rise of neoliberalism (Kuuse & Toros 2019;Scull 2021;Caplan and Ricciardelli 2016: 33). Neoliberalism here is understood as 'the extension of market principles (e.g., self-interest, competition, efficiency, profit) to all areas of life' (Mladenov 2015: 445), in our case meaning that cost-benefit calculations are deemed applicable to the provision of childcare, a reduction in government spending on out-ofhome childcare, deregulation, and other measures designed to increase the importance of the non-state sector in care provision.
The political-economic context and the relationship of the government to neoliberalism is likely to affect the timing of DI policy adoption. Firstly, as DI policy is associated with neoliberalism and efforts to reduce costs of childcare provision for the state, we expect that countries with more liberal economic systems are more likely to be early adopters of DI policies. As such countries are likely to already have experience with similar DI policies in say, mental health care, we may also expect them to implement the policy faster.
Secondly, although all analyzed countries inherited the Soviet system of children's institutions, they had different rates of child institutionalization. For example, Turkmenistan had on average 53 children in residential care per 100,000 child population in 1990, whereas Georgia had 1150 (UNICEF Europe and Central Asia Regional Office, TransMonEE Database, www.transmonee.org). Some DI studies suggest that the high cost of residential care was the primary concern of senior policy makers in the region (Tobis 2000: 30). The number of children in institutions may have therefore acted as an important motivation to reduce the institutional care provision.
We will further study the effectiveness of DI policy implementation. Previous studies have shown that the rate of implementation of international norms regarding children without parental care varies by level of economic development (Nicklett and Perron 2010). Some DI studies suggest that not only DI implementation, but also the very DI policy turn occurs under favorable macro-economic conditions, e.g., Georgia (O'Brien and Chanturidze 2009).
Deinstitutionalization is a complex multi-sectoral, umbrella policy, which often requires an overhaul of child protection, social assistance, modernization of education and health care (Cantwell et al. 2012;Jones 2019). Where the required family-and child-support services are lacking and there are no alternatives to institutions, adoption of a DI policy requires extensive capacity building on multiple levels, along with potentially high initial physical and management set-up costs (Carter 2005: 66;Csaky 2009). This means that the speed of DI implementation may vary across countries, due to significant differences in governments' effectiveness, and economic capacities to reform (for similar analysis in other policy areas see Collier and Messick 1975;Sugiyama 2011;True and Mintrom 2001).

World society theory
In contrast to political economic theory, the neo-institutional world society theory (WST) sees the roots of DI policy in the major ideational and cultural shift experienced by Western societies in the 20 th century, described as the spread of modernity (Novella 2008: 312-313). The explanatory power of WST in relation to the cross-national spread of child-related policies has been tested on several other policies, e.g., concerning child marriage (Ebet€ urk 2021).
WST explains the growing transnational isomorphism and policy convergence through a shared world culture, which is transmitted by world society actors (Meyer et al. 1997). According to WST, identities and choices of national actors, including policy makers, are shaped by the exogenous cultural environment. Internationally legitimated ideas and externally constructed policy models, commonly referred to as 'modernization', penetrate nation-states and may lead to major institutional and policy changes. National policies are thus not necessarily a reflection of domestic circumstances but rather the incorporation of external social constructs, policy models that are internationally perceived as appropriate, often irrespective of major differences in national resources, institutions and traditions (Meyer et al. 1997).
The concept of international legitimacy is central to WST and its explanation of public policy patterns. For WST scholars, policies are primarily adopted not because of their technical efficiency and functionality in terms of solving particular issues in the relevant policy area, nor as a means for securing particular economic interests, as could be argued by political-economy scholars. Rather, policy adoption acts as an instrument to strengthen the international legitimacy of the government. In other words, policy adoption is treated not a result of efficiency seeking or lobby by economically interested parties but rather as outcomes of policy makers' conformist behavior.
World culture compatible policy models spread 'at several levels and through a variety of linkages' (Meyer et al. 1997: 154), one major channel being international organizations (Fernandez and Lutter 2013;Frank et al. 2000). Ties to world society structures, for example through international organizations and international agenda-setting events, such as major international conferences, have been shown to have significant impact on policy adoption (Meyer et al. 1997;Frank et al. 2000;True and Mintrom 2001). World society theory appears to be particularly relevant to explaining the cross-national spread of DI policy. DI is deeply grounded in the modern, Western world culture: it is based on the ideology of individualism and celebrates the child as an individual with their own rights and interests (General Assembly resolution 2010, 64/142/UN Guidelines; EC 2009). In its essence, DI incorporates the world culture of a modern, agentic individual, who has legitimate 'capacity and responsibility to act' for their individual benefit (Meyer and Jepperson 2000: 102), which is incompatible with restrictive institutional culture of children's homes.
Further, DI is often promoted through typical world society mechanisms, such as international intergovernmental and non-governmental organizations. The deinstitutionalization of childcare has become a European and international policy norm. The United Nations Convention on the Rights of the Child (UN CRC) (1989) recognized the right of the child 'to grow up in a family environment'. Since then, the United Nations Committee for the Rights of the Child, along with a number of international advocacy groups, have been encouraging governments to reduce their reliance on institutional care (Ulybina 2020) (Table A1).
Empirical studies of DI reforms in the former Soviet region find that international organizations, part of the international children's rights community, have played a major role in DI policy adoption and implementation, for example in Russia (Bindman et al. 2019) and Georgia (Greenberg and Partskhaladze 2014;O'Brien and Chanturidze 2009;Ulybina 2020).
Although all the countries studied here signed the UN CRC and were exposed to international influences since the fall of the USSR, they did so to varying degrees. National governments and other key actors in the childcare sectors have not been equally connected to international pro-DI communities. There are three main sets of international interactions that may be of importance for the adoption of DI policies. Firstly, international organizations are a major source of world society pressure, advancing the spread of world culture (Boli and Thomas 1999;Lim and Tsutsui 2012). In the case of DI, we also expect international intergovernmental and non-governmental organizations to diffuse ideas about child protection, children's rights and development, and to encourage domestic actors to adopt and implement DI policy. This is in line with applied DI studies, which point at the importance of international organization in the cross-national spread of DI (e.g., Legrand 2015). We particularly expect to find this effect for organizations with a child-related agenda.
Secondly, international conferences dedicated to out-of-home childcare and child protection can be viewed as another mechanism of world society pressure. When domestic actors attend these events, they are more likely to become socialized to world cultural norms, and experience greater pressure to comply with these norms.
Thirdly, the European Union is a major actor in the modern world society, using a variety of tools to disseminate the individualistic world culture of human rights, including in its neighboring countries, which often seek political partnership and economic integration with the Union (Muradyan 2019). In other words, the European Union can serve in many ways as a crude proxy for the 'Western' modern world society and an advocate of the modern world culture.
Relevantly to our study, in relation to out-of-home childcare, the EU has been explicitly promoting deinstitutionalization, both within and beyond the borders of the European Union, especially in the new member states, accession countries, but also in neighboring countries such as Ukraine, Moldova and other (EEG 2012;Herczog 2021;Ivanova and Bogdanov 2013;Pivoriene 2020;Hadjiev 2018). It is widely recognized that social policies of countries like Estonia were visibly affected by EU accession and the following use of European structural funds. Estonian DI policy is viewed as a result of commitment to integration with the EU, and socialization into 'the Western-value system' (Kuuse & Toros 2019).
The European Union was a key development partner supporting child welfare reforms in Georgia, including childcare deinstitutionalization. In particular, the European Union's Food Security Programme, worth e20 million for 2005-2006, had DI-related conditions attached. Importantly, the EU used not only coercive tools, such as conditionality, but also (and probably much more so) various instruments of soft governance, to encourage countries make the DI turn and reinforce existing DI initiativesthrough recommendations, declarations and guidelines (Herczog 2021;Ulybina 2020). The European Union provided funding and technical assistance, supporting the development of new legislation and standards of childcare, as well as raising public awareness about the need to move away from institutions (Greenberg and Partskhaladze 2014;Pivoriene 2020;Ulybina 2020). In relation to the effectiveness of EU's external policy concerning child protection, there remains controversy. For example, some researchers argue that the advancement of EU's children's rights agenda in some countries of the Eastern Partnership has been slow, because they can enjoy relatively high legitimacy due to their role in the geopolitical situation in the region, and the role they play in the political-economic interests of the European Union (Muradyan 2019). We expect that stronger ties to Europe drive countries to make DI commitments earlier (Table A2).
Even if actively promoted by international actors, policy models do not travel without 'friction'. Rather, their adoption and implementation in contingent upon domestic context, such as capacity of the state or domestic cultural orientations (Fernandez and Lutter 2013;Frank et al. 2000;Swiss and Ilonze 2021). In particular, domestic context shapes the extent of decoupling between symbolic policy commitments and actual practices on the ground (Hafner-Burton and Tsutsui 2005;Meyer et al. 1997). Under intense international pressure, governments may ratify conventions, sign treaties and pass new laws, as a way 'to display low-cost legitimating commitments to world norms', thus signaling the adoption of new international norms and policy standards (Hafner-Burton et al. 2008: 115). However, these innovations often remain 'ceremonial' and poorly enforced, being adopted for their symbolic value. This happens when national actors are not willing or do not have the capacity to implement the policies they committed to. National linkages to world society actors may not be effective in importing world cultural norms if domestic cultural orientations contrast sharply with these world cultural norms (Fernandez and Lutter 2013). In other words, local context shapes the extent of local receptivity to the modern world culture. Therefore, we assume that the more a society is penetrated by world culture, i.e., the more it respects individual agency, the stronger domestic pressures for the rights of the child and DI will be.

Hypotheses
There is no reason to assume that both early and late DI policy adopters are driven to commit to DI by the same mechanisms. How do the above considerations affect the timing of DI commitments and the rate of implementation, for early versus late adopters? Similar to Shen and Snowden (2014), a cross-national analysis of mental health system reform, we base our hypotheses in different assumptions about the motives behind policy commitments, and respective predictions concerning the timing of commitments and rate of policy implementation, i.e., decoupling. Similarly to their study, we assume that early versus late policy adoption may signify differences in countries' motivations (Table A3).
From a political-economy perspective, one would view both early and late policy commitments as related to efficiency concerns, driven by particularly high child institutionalization (and presumably related high public expenditure) and neo-liberalization. From this point of view, decoupling of implementation from policy is predicted for early policy adopters, rather than late adopters. While both early and late adopters commit to the policy for efficiency reasons, late adopters have the advantage that they can learn from other countries and apply knowledge gained by earlier adopters to their own reforms (Clark 1985;Shen and Snowden 2014). With sufficient resources, late adopting countries have the potential to achieve a faster rate of implementation.
Hypothesis 1 (based on political-economy arguments): A more liberal national economic system, with a higher number of children in institutions, leads the country to adopt the DI policy early and implement it slower than late adopters with high capacity.
The world society approach would view early policy commitments a result of efficiency considerationseither as an attempt to address existing technical/economic problems with the care provision, such as a large number of children in institutions and growing cost of provision, or a response to political-economic interests, such as the rise of neoliberalism. Later commitments would be seen, however, as largely conformist behavior by national policy makers, adopting internationally legitimated and promoted policy models. The WST perspective predicts decoupling between policy commitments and practice, for many late adopters, who commit to world models for reasons of international legitimacy. A WST approach expects decoupling mostly among (a) low-capacity countries, which do not have the required resources and/or capabilities to implement the policy (Meyer et al. 1997), (b) countries with low local receptivity to world culture, with which the adopted model was associated.
Hypothesis 2 (based on neo-institutional arguments): Countries with stronger world society ties (whether through governmental or non-governmental linkages) adopt DI policy earlier, however they implement DI slower in conditions of low capacity and/or low receptivity to world culture.

Method
Since multiple factors, or combinations thereof, may be relevant to the adoption and implementation of DI policy, we use a qualitative comparative analysis (QCA) approach which is explicitly formulated to extract the dependence of an outcome on multiple initial conditions (Berg-Schlosser et al. 2012;Baumgartner and Thiem 2020). In other words, we do not expect individual conditions to explain the observed policy patterns and use QCA, in order to explore the potential interplay between conditions. QCA is an appropriate approach, as it focuses on using coded case data from a relatively small number of cases with a specific aim of understanding causality. Instead of seeing whether there is a continuous relationship between the inputs and the outcomes, the approach asks which set of inputs, i.e., necessary or sufficient conditions, are required to get a certain output.
Another reason for employing the QCA is that we aim to go beyond contrasting two alternative perspectives (in our case political-economy versus neo-institutionalism) but seek to understand whether some combinations of domestic political-economy and international institutional factors can explain DI commitments and implementation. Scholars turn to QCA as an approach that does not assume causal homogeneity but allows revealing multiple combinations of conditions that lead to policy (non-)adoption and (non)implementation and is thus appropriate for understanding complex causality behind the cross-national spread of different international norms and policy models (Thomann 2020;Zori et al. 2017).
Various forms of QCA analysis are available, 'crisp set' which uses binary levels, multi-valued QCA and fuzzy-set QCA which allows inputs and outcomes to partially belong to a set. With our data we use a crisp set approach. We code the data into binary categories. We derive the thresholds directly from the case data by using the midpoint of the range of each variable as the threshold, with the exception of the economic system where the threshold is set as '3 0 , meaning a left-wing government is coded as 1, and the rest is coded 0. The data lack the clear theoretical or case driven reasons that could justify the use of multiple thresholds as required for other forms of QCA analysis (Arts and De Koning 2017). In Table A4, we give the thresholds used for each condition. The resulting truth table can be found below (Table A5).

Operationalization and data
Below we explain how the conditions were operationalized and data collected. For our analysis, several datasets were collected or sourced from existing databases on 15 ex-Soviet republics in Eastern Europe and Central Asia: Armenia, Azerbaijan, Belarus, Estonia, Georgia, Kazakhstan, Kyrgyzstan, Latvia, Lithuania, Moldova, Russia, Tajikistan, Turkmenistan, Ukraine, and Uzbekistan.

The time of DI policy adoption
There can be various ways of operationalizing the time of policy adoption, particularly in the case of such a complex policy as childcare deinstitutionalization. In this study, we are interested in the explicit policy shift, or policy commitment, manifested in the first-time official declaration of the goal to reduce and/or prevent child institutionalization. So here, the term 'policy adoption' is synonymous to initial 'policy turn', indicated by an explicit DI commitment in an official national document. This approach is justified, given the complex nature and the history of DI in the region. Although 'deinstitutionalization' is an internationally recognized term, many countries effectively adopt this policy without making an explicit reference to 'deinstitutionalization' (Ulybina 2022), and if they do, not necessarily at the time of the initial policy adoption. Hence, a basic document search with keywords like 'deinstitutionalization of childcare' is likely to return many falsely late dates. Second, DI involves a wide range of measures to prevent family separation, establish and quality-control out-of-home childcare services like foster care, day care homes; improve health care provision, mainstream disability-inclusive education, and support family re-integration. DI reforms can have different 'entry' points: one country may take radical steps to close existing institutions, while another country may start with establishing foster care or with a broader reform of social services, both countries effectively adopting a DI policy. Countries in the ex-Soviet region introduced various DI-relevant measures at least since the late 1980s (Burke 1995;Tobis 2000: 30), whereby childcare centers for disabled children were opened, foster care was legally introduced, payment of benefits to child guardians. Therefore, it is difficult to nominate one particular type of legal, regulatory or policy document to be used for dating childcare deinstitutionalization policy. To address this issue and ensure consistent and replicable data collection, 'DI policy adoption' was operationalized as explained above. It allows dating DI on the grounds of a wide range of policy documents, using one consistent markerthe goal of reducing child institutionalization.
In order to establish the year of DI policy adoption, we carried out a snowball study, going back in time, looking for the earliest mention of DI, or statement which could be qualified as a DI commitment in national-level documents. First, we read through all national reports submitted in compliance with the United Nations Convention on the Rights of the Child, which have a dedicated section 'Family environment and alternative care' (https://www.ohchr.org/en/hrbodies/crc/pages/crcindex.aspx). To complement this information, we read through annual UNICEF country office reports for the period 2010-2019 (earlier reports unavailable, as confirmed in personal communication with UNICEF HQ and several country offices), plus all relevant (related to out-of-home child care) documents from the Better Care Network Library of Documents (https://bettercarenetwork.org/) (here, we first filtered the documents using country name as a filter and then browsed the documents). These sources were mostly accessed prior to April 2020. Having read the above documents for a particular country, we used snowball method, locating, where possible national and international documents, mentioned in the above sources. To do so, we searched national government websites, as well as national law profiles compiled by the Child Rights International Network, in order to trace the earliest possible documents indicating a DI policy shift. In total, more than 500 documents were analyzed. The process was iterative for each individual country, in order to maximize consistency of interpretation. Nevertheless, there still remains a risk that a) some relevant documents were missed (including because they are not (or no longer) available in the electronic format), b) their interpretation was inconsistent. A large share of national policy documents are in national languages only, and therefore not accessible for the author. Therefore, we started with internationally available information, and used references in English-and Russian-language documents to national documents, which were located to verify their content, where possible, including with the help of google translate. Since policy development is usually a continuous process, these dates have to be treated not as an exact time of policy change but rather an approximate indication of the time when deinstitutionalization objectives started to be explicated in official documents. These data were captured in the format: Country -Year -Document (Table A3).

Rate of deinstitutionalization of childcare
The deinstitutionalization of childcare is operationalized as a reduction of the number of children in institutional care. The rate of DI policy implementation is operationalized as a change in the rate of children in residential care. The rate of children in residential care (at the end of the year, per 100,000 population aged 0-17) is retrieved from the database of Transformative Monitoring for Enhanced Equity programme run by UNICEF Innocenti Research Center http://transmonee.org/database/ . While TranMonEE data have a number of limitations and do not fully reflect the progress of DI reforms (e.g., UNICEF 2010: 10), it is the only available cross-national dataset for the required time period. For the TransMonEE data the annual percentage rate of DI is calculated using the average change in the number of children in care per 100000 per year in the latest available year (2014) divided by the equivalent number in the year of DI policy adoption multiplied by 100. The exception is Turkmenistan where data is only available up to 2006.
As an indicator of national economic system, following Lim and Tsutsui (2012), we use a measure of country's ruling political party orientation with respect to economic policy. The data are retrieved from the Inter-American Development Bank's database of political institutions (Cruz et al. 2021). We use this measure as a crude proxy for the extent to which a national government can be considered neoliberal. We code left-wing as 1 and everyone else as 0.
The rate of institutionalization in the year of adoption of DI policy is determined using the TransMonEE database for the policy adoption year as explained above.
Policy implementation capacity is operationalized as per capita gross domestic product (GDP) and government effectiveness. Government effectiveness indicator from the Worldwide Governance Indicators Database (the World Bank, the Worldwide Governance Indicators project, http://info.worldbank.org/governance/wgi/#home) measures the quality of public services and is a standard proxy for administrative capacities. The per capita GDP data were retrieved for the year of DI policy adoption, from the World Bank database at https://databank.worldbank.org/source/jobs/Series/NY.GDP. PCAP.PP.KD.
World cultural pressures are operationalized in three ways, allowing us to capture different channels of influence of DI-relevant transnational actors and policy communities: national connections to international organizations with out-of-home childcare agenda; national participation in international conferences dedicated to childcare; and national ties to Europe.
The first variable we use to capture the national inclusion in world society, and in particular the international community with interests in childcare policies, is national connections to international organizations with out-of-home childcare agenda. National embeddedness in world society is commonly measured through country memberships in international organizations (Boli and Thomas 1999;Fernandez and Lutter 2013).
Studies tend to either consider generic world society ties (Frank et al. 2000), or focus on countries' ties relevant for a specific sector (e.g., True and Mintrom 2001). This study follows the latter approach and considers ties to DI-relevant international community. For this study, we chose to use not the generic membership numbers but more focused data, because many children's NGOs do not run membership schemes. Instead, we gathered cross-national information about connections to international organizations with an out-of-home childcare agenda. 'Connection' means that an organization either has a member in the analyzed country or has had operations (run projects) in the country. To compile the list of international organizations with out-of-home childcare agenda, we included full members of the European Expert Group on the Transition from Institutional to Community-Based Care, and organizations identified through three further sources -The Children without Appropriate Parental Care Working Group (part of the bigger coalitionthe NGO Group for the Convention on the Rights of the Child), which was founded to create and promote the Guidelines for the Alternative Care of Children; the website dedicated to the implementation handbook for the Guidelines -'Moving Forward: Implementing the 'Guidelines for the Alternative Care of Children''; and the Moving Forward project steering group. From this combined list, we excluded organizations with a geographically irrelevant focus, such as Latin America and Africa. The resulting list of 19 organizations is not comprehensive but includes key actors in the field of out-of-home childcare. We then searched the websites of these organizations for information about members and their operations by country. Where necessary, lists of members were requested from organizations through personal contact (see Annex). Analysis of these data should have the following caveats in mind: historic information about NGO activity is often hard to locate; the data show the extent of country connectedness to international networks as of 2019, which does not necessarily mean that all these organizations were active in these countries when DI policies were adopted. It is assumed that country's inclusion in international networks is path-dependent and that these data give an indication of the country's openness and connectedness in the recent past. For the analysis, country connections to individual organizations were used to produce a cumulative score, where each connection would equal one point.
The second variable we use to capture the national inclusion in world society, and in particular the international community with interests in childcare policies, is conference participation. The list of major international conferences dedicated to childcare was put together from the following sources: United Nations webpage 'Past Conferences, Meetings and Events for 1992-2008'; Better Care Network website; the website of the Council of Europe 'Children's rights: Events' (https://www.coe.int/en/web/children/ events). The key criteria for including a conference in the data were geographic relevance (e.g., pan-African and South Asian were excluded), and availability of the list of participants. The final list contains information for 17 conferences that took place in 1990-2019 (see Annex). Generally, names of conference participants are not disclosed for data protection reasons. So, comprehensive data collection of this sort is difficult. Most identified conferences occur only after countries have made the DI policy turn, so no direct causality can be implied. Rather, conference participation, similarly to ties with transnational actors, serves as a proxy for country's inclusion in world society, and in particular in international community with childcare agenda.

National ties to Europe
The third variable we use to capture the national inclusion in world society, and in particular the international community with interests in childcare policies, is national ties to the European Union and the Council of Europe. Here, we take a generic, rather than DI-specific, measure of political ties to the European Union and the Council of Europe, since we treat both institutions as world society actors promoting modern world culture and respective policy models. In other words, we focus on generic ties in the same way as WST scholars focus on generic ties to INGOs, without looking in more detail at the exact agenda or cooperation projects that links a country with that INGO. Having said that, both the European Union and the Council of Europe have long had out-of-home childcare on their agenda.
In order to measure these ties, we compiled information about several aspects of countries' relationships with the European Union and the Council of Europe, by using EEAS EU External Action Country Factsheets. In particular, the following information was gathered: whether the country is a member of the Council of Europe; an EU member or a Neighborhood country; whether the country ever had EU membership aspirations; intensity of the political dialogue with the EU. The intensity of political dialogue was measured on the basis of agreements signed with the EU: a country was assigned a 'high' score if it has become an EU member or signed an Association Agreement or a Comprehensive and Enhanced Partnership Agreement, or a 'low' score if none of the above occurred. An association agreement is a bilateral agreement between the EU and a third country, which provides ground for implementing the accession process, and therefore signifies the existence of relatively strong ties to the EU. The Comprehensive and Enhanced Partnership Agreement provides a framework for the EU and a third country to collaborate in a range of policy areas, including human rights, and to deepen the EU's relationships with the neighborhood countries. The 'ties to Europe' index was constructed as a cumulative score, where one point was allocated for each positive value in the above categories (see Table A2).
Local receptivity to world culture is operationalized as the contraceptive preference for pill. Contraceptive use rates are associated with the spread of the modern world culture, more so than countries' sociocultural and economic features (Cole and Geist 2021). Unlike the World Values Survey, commonly used as an indicator of prevalent social norms and attitudes (e.g., Fernandez and Lutter 2013; Pandian 2019), it is a 'revealed preference' indicator, i.e., it is more likely to reveal the actual strength of world-cultural norms and agentic behavioral patterns in society.
The data for the contraceptive preference for pill are retrieved from the UN database on contraception (UN 2019). The contraceptive prevalence for pill is measured as 'a percentage of married or in-union women of reproductive age'. Compared to indicators such as the number of women's, children's or LGBT organizations, it is less prone to politically induced distortions, where temporary political suppression may result in the low numbers of formal organizations, missing the informal manifestations of actorhood. Importantly, it is not strongly wealth-dependent, which allows avoiding the problem of confounding variables. Compared to other contraception methods, it comes closest to manifesting the woman's individual childbearing preferences and demonstrating the extent of women's individual actorhood. It is easily employed by a woman herself, does not require partner's consent (as for male condoms), nor a third-party participation (as with sterilization, implant or vaginal barrier methods), is invisible, relatively cheap and widely available. Where the data were not produced annually, we linearly interpolate the available data.
Below, we present our findings from the qualitative comparative analysis. We present the resulting truth table showing possible pathways leading to early or late DI adoption and fast or slow implementation (Table A5).

Results
We investigate the data in two ways, firstly through standard QCA techniques and then through the lens of the above hypotheses.
To investigate the truth table using standard QCA methods we use the R package 'QCA' (Dusa 2019). There are several levels of complexity that can be investigated. At the simplest level where a connection to a single variable is investigated, we find that a high IO connections index score forms a necessary condition for earlier adoption of DI policy, in that the four countries (Belarus, Russia, Latvia and Lithuania) that adopted DI early, have extensive connections to international organizations. Three other countries have a high IO connections score but have a later than the threshold adoption date (Estonia, Moldova and Ukraine). Interestingly, the connections to international organizations are also relevant to the reverse outcomes: all eight countries with a low IO connections index have a later adoption date for DI policies. This makes a low IO connections score a sufficient condition for a later DI policy adoption date. Overall, the IO connections index correctly predicts twelve out of 15 cases. No other single variable is able to explain a similarly wide range of outcomes.
Moving to a higher level of complexity, reasonably simple solutions giving sufficient outcomes to explain slow adoption of DI can be found. An example of such a solution is $ IOIndex þ (Ties to Europe Ã $GDP), where the $ indicates a negation of the condition. Four such solutions exist, all of which contain $ IOIndex as a separate term, with the other term generally containing a mix of the world society and political-economy characteristics. These solutions can explain ten out of eleven cases in this group. No solution to explain all four early adopters is found, although a large number of possibilities to explain three of the cases is obtained.
For the second outcome, the annual rate of reduction of child institutionalization, no such simple relationships between the conditions and outcomes could be found. To obtain a solution that covers half the cases at least five conditions are required. Whilst the QCA approach is designed to investigate multiple conjunctural causation (Berg-Schlosser et al. 2012), the lack of success in explaining a large set of the cases suggests that these results do not constitute a reliable ground for robust conclusions. This means that the success of policy implementation, i.e., the rate of decrease in institutionalization, is more likely to depend on idiosyncratic factors that cannot be captured by any of the whole-country metrics we investigate here.
Turning to the hypotheses, Hypothesis 1 is derived from political economy considerations and we can break it down as follows. Firstly, is there evidence for early adoption driven by a neoliberal national system or a high number of children in institutions? Four countries adopted the policy before 1999: Belarus, Latvia, Lithuania and Russia. Of the early adopters, half have a high rate of child institutionalization, and three out of four have liberal economic systems, the exception being Belarus. The majority of countries with high institutionalization rates or liberal economic systems are late adopters, however. Turning to policy implementation, of the early adopting countries three out of four are slow implementers. Only one late adopting country, Estonia, has both high GDP per capita and high government effectiveness and it indeed is a fast implementer, as expected from the first hypothesis. Looking at the inverse, of the nine countries that have both low GDP per capita and low government effectiveness four are slow implementers. Clearly, the data is not explained in a consistent way with the hypothesis drawn from a political economy perspective.
A neo-institutional argument, in particular made by the world society theory, is tested in hypothesis 2. Starting with early adopters, we see that all of the early adopters have strong links to international organizations, three have strong links to international conferences, the exception being Belarus, and two (Latvia and Lithuania) have strong European ties. For the late adopters, we see that only three of eleven have strong links to international organizations, six to international conferences and seven to Europe. Looking at the implementation, Belarus, Latvia and Russia are slow implementers. These data do not support our second hypothesis, since Belarus has low contraceptive preference for pill and low implementation capacity, whilst Latvia has high capacity and high contraceptive preference for pill, whilst Russia has high contraceptive preference for pill but low government effectiveness, one of the two measures we use for policy implementation capacity. Whilst there is a strong indication of the relevance of world society ties, the decoupling is not clearly driven by the factors as predicted by world society theory.

Discussion
Our study makes several important contributions. First, we collected new data on the timing of national commitments to childcare deinstitutionalization for 15 countries of the former Soviet Union, which is crucial for understanding international patterns and nature of social policy change.
Second, our analysis addresses a major gap in international comparative public policy research. To our knowledge, it is the first medium-N comparative study of childcare deinstitutionalization policy adoption and implementationa major international policy area, affecting millions of children around the world. It is also the first study to bring out-of-home childcare policies of post-Soviet countries, including Central Asian countries, into systematic comparative analysis.
The collected data show that all countries in the region committed to the deinstitutionalization of childcare within a historically short time period 1993-2005 (Table A3). The countries did so, despite major differences in the levels of child institutionalization, policy implementation capacity, domestic orientations, and international ties. For example, the contraceptive preference for pill is around 1% in Armenia, Azerbaijan, Georgia and Turkmenistan, while it is over 10% for Estonia, Latvia, and Ukraine, around the time of DI adoption. The GDP per capita per year is below $3,000 for Kyrgyzstan, Moldova, and Tajikistan, while the same measure for Estonia, Kazakhstan, Lithuania, and Russia is over $10,000. Further, countries are not equally connected to international actors, the key carriers of DI ideas and advocates of DI policy model. The index measuring country's ties to DI-relevant international organizations stood at 3 or fewer for Kazakhstan, Turkmenistan and Uzbekistan, while it was 12 for Estonia, Latvia, Lithuania, and Russia.
Importantly for human rights and social policy scholars, we find that this crossnational shift toward DI at the level of policy commitments was accompanied by noticeable reduction in child institutionalization. For instance, Georgia, Moldova, and Azerbaijan reduced child institutionalization by more than 50% in the studied period. In other words, these high-level policy commitments do go beyond 'window-dressing' and bring real structural change, providing ground for an optimistic view of human rights policy commitments and outcomes by low-and middle-income countries. While human rights are a well-known decoupling-prone policy area, our study suggests that certain commitments to children's social rights do indeed translate into substantive changes on the ground. It is true, of course, that reduction of the number of children in institutions reflects only one part of DI story, as it does not show the development of alternative community-and family-based services, nor does it reflect organizational changes within institutions. Therefore, this simple statistical measure cannot capture the complexity of DI transition, and may in fact be not indicative of improvements (or lack thereof) in children's rights and children's well-being. Therefore, future studies may benefit from employing several alternative measures of DI policy implementation.
Our third contribution is to theoretical debates about drivers and conditions of public policy adoption, as well as their impact on substantive social change. We observe noticeable variation in the time of policy turn, as well as rate of the following policy implementation, between countries. In this paper, we asked: what conditions affect the timing of policy commitments and the rate of implementation? In other words, how can we account for the spread of DI policy commitments and the resulting patterns of implementation in the former Soviet countries? We tested two alternative hypotheses based on political-economy and neo-institutional perspectives concerning drivers of public policy adoption. Neither hypothesis finds support in our data.
At the same time, our results point toward some explanatory power of the neo-institutional world society theory. We find that there is a single conditionnational ties to international organizationswhich is clearly relevant to the timing of DI policy adoption. Having strong ties constitutes a necessary condition for early adoption, while having weak ties constitutions a sufficient condition for late adoption. We cannot demonstrate a direct causal effect here because the data on country ties to international organizations was gathered for 2019, i.e., after the analyzed countries adopted their DI policies. However, it seems reasonable to assume that most countries are rather pathdependent in the degree of their involvement in international networks, and so to use these data as a rough proxy for earlier connectedness of countries to world society. This result confirms earlier findings of world society scholars, who argued that country's ties to international organizations are likely to have significant impact on the speed of policy adoption.
We find that national ties to DI-promoting international organizations and conferences are highly correlated, although ties to conferences do not appear a similarly important condition as ties to organizations. This seems to be due to one case, which is Belarus, which has a high IO index and low conferences index.
This pro-WST evidence finds additional support in our related finding, related to cost-benefit calculations by DI-committing governments. One indication of DI reforms being carried out for political-economy reasons, such as cost optimization purposes, would be government-commissioned cost-benefit analysis of out-of-home care options in the country. Although family-and community-based services are generally found to be more cost-effective than institutional care (Mansell et al. 2007;Tobis 2000), countries in the region differed greatly in their spending on institutions (Tobis 2000: 29), as well as the extent and quality of existing social services and infrastructure. Besides, community services may only be cheaper if services for people with severe disabilities are not included (Mansell et al. 2007: 43). However, as detailed below, cost-benefit analysis was sometimes completed at a later stage, after the governments had already committed to childcare deinstitutionalization. For example, Georgia made the DI policy turn in 1999, and the earliest costing report we could locate is dated 2008 (Gugushvili & Arganashvili 2008). Costs of Child Care Services in Georgia: Report of the Unit-Costing Exercise Conducted for the TACIS Support to Child Welfare Reform). Azerbaijan adopted its DI policy in 2005, and in 2010 the Department on De-Institutionalization and Child Protection under the Ministry of Education initiated the project within which the department would 'try to determine the minimum amount required for foster families, for families with children with disabilities and to cover the costs of services for children from residential care facilities who are to be reintegrated with the families' (Azerbaijan NGO Alliance for Children's Rights 2011: 22). This finding may be an indirect indication that DI commitments were indeed driven primarily (or initially) by world cultural imperatives, rather than economic interests. It is possible though that governments did not commit to DI as a result of rational calculations as in comparison of alternative costs, but rather DI policy was a reaction to the reductions in expenditures on residential institutions during the economic crisis of the 1990s, accompanied by the drop in foreign aid, which was also relevant for children's homes (Laur 2002: 13;Tobis 2000: 27). In other words, the reduction of available funding for institutions possibly made conversion of children's homes and boarding schools into general schools and community-based services a feasible policy response.
Another argument in favor of the world cultural explanation of DI commitments is the observation that the general drift toward DI policy started already in the USSR, i.e., well before neoliberal policies were adopted in the analyzed countries (Burke 1995;Tobis 2000: 30).
Lack of further results in support of the institutional argument (world society ties through conferences and proximity to Europe) is probably related to the way we operationalized world society ties. World culture spreads not merely through international organizations and major policy-focused institutions, but also through multiple microlevel channels, such as transnational learning from peer professionals; difficult-to-quantify recommendations from international institutions; individual cross-border travel (Dion 2008;Pegram 2010;Ulybina 2020). So arguably, findings may point at the diffuse, permeating impacts of international culture and institutions, whereby global norms spread through a variety of channels, at multiple levels including not only government officials and administrators, but also care providers and parents (Schofer and Hironaka 2005: 26). Hypothetically, the observed DI commitments and reduction in child institutionalization could be associated with the increasing exposure of the population of these countries to world culture, through a variety of channels beyond immediate ties with key agenda-setting events and world society agents such as the European Union. Future research may benefit from taking this into account and identifying more refined ways to operationalize exposure to world culture.
A related issue to do with operationalization is that while the role of Europe is routinely highlighted in relation to DI reforms, one does not always clearly delineate between different contributions of the European Union/Council of Europe and other European actors at different stages of DI shift, such as agenda-setting, support with implementation, etc. For example, in the case of Georgia, the European Union played a very important role in moving forward the DI reform, however, it was not the European Union and not the Council of Europe, but ground-level actors from Europe who drove the early stages of DI transition (for more details see for example, Ulybina 2020).
Another interesting direction for future research is to explore the regional effect of DI policy spread. Remarkably, all countries of the former Soviet Union have moved to deinstitutionalize their childcare, despite significant variations in domestic conditions and even where ties to the relevant international organizations were weak. Besides, the shift took place, on a historical timescale, quite rapidly. Such cross-national policy shifts are sometimes explained by world society scholars through a 'regional' effect similar to 'peer pressure', whereby world culture and associated policy models spread faster among neighboring countries, which learn from and imitate each other, leading to isomorphism despite different initial conditions . Applying this argument, one would expect post-Soviet countries to introduce similar changes to their national childcare policies and within a historically short period of time, which we indeed observe in our data. The next step would be to identify reference groups of each country, and test whether early versus late DI commitments are driven by DI policy adoption in the relevant reference groups.
Another important theoretical implication of our study concerns the opposition of neo-institutionalism (and in particular world society theory) with other, such as functionalist or political-economic theoretical perspectives. As discussed above, childcare deinstitutionalization is an interesting example of a policy which is promoted both on human rights and economic grounds. DI policy can thus be viewed an arena where world cultural human rights norms and neoliberal political-economic interests simultaneously play out. Furthermore, world society actors have been promoting DI policy on these two grounds. Depending on which of these two aspects of DI policy dominates policy implementation in each particular country, this may lead to drastically different outcomes for children and society and may fundamentally change how the observed changes are interpreted and conceptualized. Like many other social policies, DI policy has been rapidly spreading around the world (Ulybina 2022). Since DI is a human rights-embedded policy, world society scholars may be tempted to use this evidence to proclaim the triumph of the Western individualistic world culture and the power of related human rights discourses. Such conclusions can be questioned if one takes into account the economic aspect of DI. Indeed, none of our findings can be used to rule out political economic explanations. Therefore, world society analysis needs to carefully consider the grounds on which human rights policy models are promoted, which may show that human rights policies are 'sold' on efficiency grounds, and world cultural influences are difficult to distinguish from neoliberal interests. For this, comparative analysis would need to be preceded by more qualitative analysis, looking in more depth at the nature of policy promotion. This would allow refining the comparative analysis, by including not only factors of the demand-side (i.e., conditions describing the policy adopting country) but also factors of the supply side (i.e., conditions describing the nature of policy promotion).
A related limitation of world society theory is the conceptualization of world culture. The case of childcare deinstitutionalizationa policy promoted by world society actors on human rights and cost-benefit groundssuggests that neoliberalism and a concern for economic efficiency, as a legitimate measure of successful childcare policy, may in fact constitute part of the same individualist and rational world culture, with which human rights are associated. In other words, the current analytical distinction between different aspects of the same world culture is rather blurred, and a clearer conceptualization is required, in order to test more refined models of policy adoption and decoupling. We hope that our paper will pave the way for more robust and theoretically pluralistic investigation of the globalization of childcare policies and their effects on children.

Declaration of interest statement
No potential conflict of interest can be reported by the authors.

Funding
This work was supported by the European Commission through the Marie-Curie Fellowship no. 742981 'EUCHILD'.

Olga Ulybina
http://orcid.org/0000-0003-0186-5658  Deinstitutionalization working group set up, also: The Orphans and Neglected Children (Adoption Procedure) Act, adopted in 1999. 'The purpose of the act is to safeguard the preeminent right of the child to be raised in a family environment, to bring down the levels of abandonment of children, to help build the character of children and to promote their socialization and adaptation to community life.' (from UN CRC report 2001) UN CRC report 2001: 'The Georgian Ministry of Education is currently continuing work on the implementation of a project begun in 1999 on the welfare of children and families, designed to promote alternative forms of childcare for abandoned or orphaned children. The project has as its priorities: first, support (returning the child to its biological family); second, prevention (reducing the risk that the child will be placed in the children's home); and, third, placement in a foster family.' (continued) 'The measures listed above testify to the fact that the State is doing everything possible to address the problem of the decline in the number of children sent to boarding establishments, and to restore a family environment for them.' (UN CRC report 2001: 42) Kyrgyzstan 1999 CRC report 1999: 'Adoption and guardianship remain the priority means of providing for children deprived of parental care. There has been some development of family-type children's homes in which from 5 to 10 orphans are brought up. An advantage of these forms of care for children deprived of parental support is that they are raised in a family environment with excellent social adaptation to the community and to work. At present, however, family-type children's homes, being on the budget of the local State administrations, are experiencing considerable financial difficulties.' UN CRC 2 nd report,