WoPSI-the World Politics of Social Investment

The “World Politics of Social Investment” (WoPSI) project aims at explaining variance in social investment agendas and social investment reforms across democratic countries in different regions of the World. Virtually all capitalist economies grapple with challenges of demographic change, slow economic growth, poor employment performance and increasing poverty rates. In dealing with these problems, a social investment strategy appeals to a wide audience, both political and academic. However, social investment reforms and performances in democratic countries around the globe are highly unequal and remain fragmentary: different countries have implemented different types of policy instruments, with different functions, at different points in time, and to different degrees. Despite a growing number of scientific contributions on social investment reforms and their effects, a systematic mapping of the design of social investment agendas and policies in different democratic countries around the globe is still lacking. Moreover, we lack an explanation for the variance in the development of social investment

Despite a growing number of scientific contributions on social investment reforms and their effects, a systematic mapping of the design of social investment agendas and policies in different democratic countries around the globe is still lacking. Moreover, we lack an explanation for the variance in the development of social investment policy reforms. This is where we locate the research interest of the project presented in this paper. In order to understand why social investment agendas and policies have developed differently across countries, we need to study the politics of social investment reforms. Thus, we ask: How do social investment agendas and social investment policy reforms vary across democratic countries around the globe? Under what political conditions do social investment agendas and/or reforms develop?
In this paper, we situate these questions in the existing state of the literature, and we outline a way to answer these research questions in the context of social policy reforms in Latin America, East Asia, as well as Western and Eastern Europe. We argue that political coalitions (actors and their interests), as well as the institutional embeddedness of social investment politics are key factors in explaining the high variety of social investment agendas and policies between countries.

Introduction: Social Investment Politics around the Globeresearch questions and goals of the project
The "World Politics of Social Investment" (WoPSI) project aims at explaining variance in social investment agendas and social investment reforms across democratic countries in different regions of the World.
Social investment has become more than a buzzword in political and academic discussions (Esping-Andersen 2002, Hemerijck 2012, Morel et al. 2012. Virtually all capitalist economies grapple with challenges of demographic change, slow economic growth, poor employment performance and increasing poverty rates as a consequence of structural change, dualized labour markets, as well as new social needs and demands that grow out of post-industrial patterns of family and labour market reconfiguration. In dealing with these problems, the language and core ideas of social investment appeal to a wide audience, both political and academic. However, despite the immediate appeal of this policy strategy, social investment reforms and performances in democratic countries around the globe are highly unequal and fragmentary (Beramendi et al. 2015, Bonoli 2013, Hemerijck 2012, Morgan 2012, Nikolai 2012.
The literature shows that over the past three decades, there has been a substantial amount of implementation of social investment policies in different regions of the world (Latin America, Asia, East-and Western Europe). However, the turn to social investment is highly diverse in its form and extent: Different countries have implemented different types of policy instruments, with different functions, at different points in time, and to different degrees (Bonoli 2013, Hemerijck 2013, Jenson 2012, Morel et al. 2012.
Despite a growing number of scientific contributions on social investment reforms and their effects, a systematic mapping of the design of social investment agendas and policies in different democratic countries around the globe is still lacking. Moreover, we lack an explanation for the variance in the development of social investment policy reforms. This is where we locate the research interest of this project. We argue that in order to understand why social investment agendas and policies have developed to different extents, in different forms, and at different points in time across countries, we need to study the politics of social investment reforms. Thus, we have the following main research questions:  How do social investment agendas and social investment policy reforms vary across democratic countries around the globe? How can we explain this diversity?
 Under what political conditions do social investment agendas and/or reforms develop?
We want to answer these research questions in the context of social policy reforms in Latin America, East Asia, as well as Western and Eastern Europe, and we are interested in social investment agendas and policy reforms in all social policy fields. We argue that political coalitions (actors and their interests), as well as the institutional embeddedness of social investment politics are key factors in explaining the high variety of social investment agendas and policies between countries (see section 4 of this background paper).
This working paper is a programmatic background paper to the project. It is aimed at situating our project in the existing literature, defining the key research questions and developing a theoretical framework that is formulated in a sufficiently general way so that it applies to the politics of social investment reform in all democracies around the world. The background paper also defines the key concepts of the chosen approach (e.g. "social investment", "agendas", "social investment policies", "policy legacies", and "social investment reforms"). It is structured as follows. In the following section 2, we discuss where the literature on social investment stands, i.e. we present empirical findings on social investment agendas and reforms in countries around the globe, as well as on the distributive effects of social investment policies. Section 3 consists in a broad literature review of theoretical arguments explaining variance in social investment policies, and welfare states more generally. In section 4, we provide our key definitions, research questions, and our main theoretical focus, as well as the contours of the planned research design.

What we know best so far: The emergence and effects of social investment policies
Existing research on social investment can be divided in two main strands of literature. A first strand explores the design and emergence of social investment policies in specific policy fields and/or specific countries and regions of the world (section 2.1). A second strand of literature analyses the effects of social investment policies on socioeconomic and political outcomes, such as social inclusion and poverty reduction (section 2.2). Many of the existing studies have focused on Western democracies, often even only on Western Europe, despite the fact that democracies around the globe have conducted similar reforms. With this project, we aim at gathering research from all parts of the world. Moreover, hardly any studies attempt to systematize and to explain the variety of social investment agendas and policies across countries, which will be a major focus of our project.

The emergence of social investment policies in different countries and world regions
In this section, we provide a first overview of the programmes and reforms adopted in the countries we are interested in. We naturally cannot provide an exhaustive account, but will rather illustrate social investment reform processes with empirical examples. As our account is necessarily incomplete, we hope for contributors to our project to fill these gaps, either by focusing on the politics of the mentioned reforms, or to focus on other debates/policies/programmes/reforms that we have missed here. The main purpose of this section is to underline the diversity in the development of social investment reforms around the world.
Later in this background paper, we will provide our own definition of social investment, adjusted to the purposes of our project. In order to identify the existing literature on social investment, we used as a preliminary definition the one proposed by Morel et al. (2012): "The social investment approach rests on policies that both invest in human capital development (early childhood education and care, education and life-long training) and that help to make efficient use of human capital (through policies supporting women's and lone parents' employment, through active labour market policies, but also through specific forms of labour market regulation and social protection institutions that promote flexible security), while fostering greater social inclusion (notably by facilitating access to the labour market for groups that have traditionally been excluded). Crucial to this new approach is the idea that social policies should be seen as a productive factor, essential to economic development and to employment growth." (Morel et al. 2012: 2).
Because we know this literature better, we start by discussing some findings for the Western European countries. Here we focus on active labour market policies (ALMPs) as well as family and education policies, even though we are aware that other policies in various policy fields can be associated with a social investment logic. We however hope that as a result of our project, the knowledge about existing social investment agendas and reforms will be much more shared and balanced amongst the various policy fields and regions of the world.
Western European welfare states have implemented social investment reforms to very different degrees (Beramendi et al. 2015, Bonoli 2013, Hemerijck 2012, Morgan 2012, Nikolai 2012. In Europe, the Nordic European countries (Denmark, Sweden, Norway and Finland) are usually mentioned as the most advanced in terms of social investments. These countries have known ample and largely successful programs of this sort for several decades (e.g. Ansell 2015, Nelson andStephens 2012), even though recently there seems to be a certain level of de-investment in those policies (Kraft and Van Kersbergen fthc.). For Sweden, where spending on ALMPs was exceptionally high already during the 1980s, Bonoli (2013: 30) observes a de-investment on ALMPs during the last twenty years. All Scandinavian countries invested heavily in all forms of education already during the 1950s, expanded post-secondary education and universal, high qualitative childcare much earlier than most other countries (Busemeyer 2015, Garritzmann 2016, Iversen and. In addition, compared to other West European welfare states, the Nordic countries spend most on ALMPs (Bonoli 2013: 28-46). However, only Sweden was a traditional high spender on ALMPs in the 1980s already. The other Nordic countriesstarting with equally low amounts of spending as continental European countries in the 1980sbecame top spender on ALMPs only during the 1990s (Bonoli 2013: 34f.).
Continental European countries (Bismarckian welfare states) (such as Germany, Austria, and France), in contrast, have historically been strongly consumption-oriented, focusing especially on male breadwinner insurance (Esping-Andersen 1990). Recently, however, several studies document an expansion of social investment policies, especially in the fields of family policy andto some extentactivation policy. Germany and Switzerland can be mentioned as examples of continental welfare states, where social investment reforms took the form of care infrastructure, fiscal incentives for childcare, as well as activation programmes with regard to the unemployed or disabled (Bleses and Seeleib-Kaiser 2004, Fleckenstein 2011, Gerlach, 2004, Häusermann and Kübler 2011, Häusermann and Zollinger 2014, Hieda 2013, Morgan 2013, Naumann 2005. Other continental welfare states also intensified social investment to different degrees and in different forms. Generally, continental European countries have intensified their spending on ALMPs over the past twenty years, albeit to a lesser extent than Northern Europe (Bonoli 2013: 43). Continental European countries have embarked on the social investment turn considerably later than the Nordic countries. They still display a lower level of spending on childcare or ALMPs (Bonoli 2013: 31), as well as a high internal variation. The same holds true for education spending: Despite the fact that education has featured increasingly saliently on the political and the public agenda (Ansell 2010, Busemeyer et al. 2013, in most continental European countries, education systems remain underfunded, especially higher and early childhood education (Busemeyer 2015, Garritzmann 2016, Iversen and. Liberal welfare states (e.g., the UK, Canada, the USA) are generally less generous when it comes to social compensation policies (Esping-Andersen 1990). Regarding some types of social investment, however, liberal welfare states have early on focused on education and skill development. In fact, Heidenheimer (1973) and others (Beramendi et al. 2015, Busemeyer and Nikolai 2010, Iversen 2005 argue that while conservative welfare states have focused on compensation and did not develop investment policies, liberal welfare states have focused on investment at the expense of compensatory policies. Thus, despite the fact that social compensation spending is at best average in liberal welfare states, education spending (both private and public) is comparatively high. Recently, this agenda has been broadened in some liberal welfare states. In the UK, for example, parents of young children gained the right to ask for a flexible work schedule, paid leaveschemes were introduced, and childcare expanded (Morgan 2013). In May 2004 the Brittish government announced a plan to offer middleincome families tax reliefs for hiring nannies (Morgan 2013: 94). However, the English speaking countries are low spenders on ALMPs in Europe, even though spending has increased during the last twenty years (Bonoli 2013: 31). English-speaking countries introduced the cheapest active labour market policies; in liberal welfare states ALMPs are based on work incentives rather than on investing in human capital (Bonoli 2013: 33, 43). Such reforms that strengthen the work incentives for job seekers, as benefit conditionality, do not necessarily impact spending, but put pressure on beneficiaries to reintegrate into labour market.
Southern European countries (such as Italy, Spain, and Portugal) have recently introduced several social investment reforms in different policy fields. Yet, spending in these fields increased only in the late 2000s, and remains on a comparatively lower level than in the other European welfare regimes (Bonoli 2013: 31). However, proportionally these countries have faced the biggest increase in spending in ALMPs during the last twenty years. They have also shown some structural change in their family policies. Southern European countries have long been characterized as welfare systems of 'unsupported' familialism (Léon and Pavolini 2014: 354). However, since the early 1990s a social investment turn in family policy can be observed in some countries, mainly in Spain. Most importantly, Spain introduced universal pre-school education in 1990 (ibid.: 358), while childcare provision for children started to grow substantially during the years before the outbreak of the financial crisis in 2007. In addition, there have been several improvements of maternity and paternity leave and the creation of a national law for long-term care (ibid.: 359). In Italy since the late 1990s there have been some social policy reforms, which promoted care services, but only limited resources were provided for its implementation (ibid.: 355, Morgan 2013: 100). Even though there have been many new family and care policies in Spain and some of them in Italy, during the recent financial crisis budgetary cuts have affected these policies considerably.
Whereas spending on ALMPs remains comparatively low among all Eastern European countries, social investment policies on family policies have gained relevance since the 2000s, however to very different degrees in several countries. Eastern European countries (Czech Republic, Hungary, Poland and Slovakia) spend even less on ALMPs than Southern European countries (Hemerijck 2013: 232, 259). However, in Czech Republic and Poland spending on ALMPs nearly doubled in the last ten years (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) whereas it stagnated or even decreased in Hungary and Slovakia respectively. In the formerly socialist Eastern European countries, women were encouraged to join the labour force by special incentives including publicly provided, affordable childcare services already before the 1990s (Szelewa and Polakowski 2008). After the collapse of the Soviet Union, the governments in the states of Eastern Europe closed many childcare centres and withdrew financial support (ibid.: 115). The beginning of the 1990s has been described as post-communist era of "refamilialization" Polakowski 2008, Saxonberg andSzelewa 2007). However, Szelewa and Polakowski (2008: 126) show convincingly that instead of an overall tendency towards familialization of policies in East European countries, over the last two decades (1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005) post-communist countries follow different types of familialization: Some of them have strengthened defamilializing policies and introduced social investment components in their family policies. After a decline of enrolment rates of children in childcare centres in the first half of the 1990s, several countries noted a considerable improvement since the late 1990s (ibid.: 122f.). The enrolment rates are for example comparatively higher in the Czech Republic, Estonia, and Hungary and rather low in Poland and Slovenia. The generosity of parental leave provision increased in all Eastern European countries over time, but with great variance between the countries. Hungary, which already had the most generous family support system in the world four decades ago, still has an earnings-related payment that lasts for two years (ibid.: 124). The Czech Republic introduced a universal parental leave scheme in the middle of the 1990s, which was extended to four years. Regarding higher education, some Eastern European countries have moved in the direction of "dualized systems", where the best students study free of charge, while students with worth grades are charged considerable tuition amounts (Garritzmann 2016). Thus, the relationship between public and private spending is also of particular interest here.
Taken together, the vast majority of West and East European countries have over the past twenty years faced a reorientation of their social policy in the direction of the social investment perspective. The expansion of social investment policies in general seems to be a universal trend and seems to happen across welfare regimes (Hemerijck 2013). Nonetheless, countries from all welfare state regimes haveeven though to different degreesincreased their spending on ALMPs. In addition, we can observe an expansion of work-family reconciliation policies most prominently in continental countries during the early 2000s and a catching up process of those policies in Southern European countries prior to the crisis in the late 2000s. In Eastern European countries the trend of an expansion of spending for social investment policies can be observed as well, even though those countries display the lowest spending level for ALMPs and a rather fragmentary and highly diverse development of work-family reconciliation policies. To summarize, several Western and Eastern European countries have introduced and expanded social investment policies in the field of labour market and family policy. However, we are not able to see clear empirical patterns of social investment agendas and policies yet, as we lack an encompassing overview of social investment agendas and reforms.
European countries are not the only ones in which we observe social investment policies. Since the 1990s, social investment reforms could also be observed in South East Asia and Latin America. During the 1990s, governments in Japan and South Korea, for example, have begun to reconfigure their welfare states, mainly by a new kind of social investment policy. Peng (2014) observes for both countries a shift in the targets of welfare state protection. "Old" social policies (e.g., unemployment and old age insurance), targeted at full-time male industrial workers, public sector employees, military personnel, and teachers, have been complemented by "new" social investment policies supporting labour market participation of outsiders. They incorporate more peripheral, marginalized, and vulnerable population groups such as women, the young, children, as well as the elderly. This shift has been accompanied by a political rhetoric of social inclusion and was framed in terms of enhancing intergenerational equality.
The type and degree of reform, however, also varies across countries and programs.
The social investment strategy in Japan seems mostly focused on reconciliation policies with a primary focus on child-and elderly-care (Peng 2014). Over the past two decades, the government incrementally enlarged maternity and parental leaves and expanded childcare and family support services. In addition, Japan significantly expanded elderly-care services and long-term care insurance. As a consequence of a sharpened insider/outsider problematic, there was also a reform of the employment insurance system. This reform entailed a focus on older workers, women, and youths who were increasingly engaged in precarious working conditions. The reform involved measures to keep workers over the age of fifty-five in the labour market by a special subsidy for employers. The government also institutionalized support programs to address high youth un-and underemployment.
In South Korea a more explicit discourse on social investment gathered much attention during the Kin Dae-Jung government (1997)(1998)(1999)(2000)(2001)(2002) and even more from the mid-2000s onwards under the Roh Moo-hyung government (2003-2007) (Seung-ho and Seung-yoon 2014). The later government presented the "Vision 2030 for Economic Growth and Welfare" where the social investment approach was central. In the Vision 2030 social welfare is seen as an investment in the future (Han 2007). In the mid-2000s the government adopted the "Basic Plan for Low Fertility and Aged Society (2006-2010)" (Seung-ho and Seung-yoon 2014). However, a high proportion of the expenditure on total family policy is allocated to cash transfers for those parents who stay at home (Seung-ho and Seung-yoon 2014: 12f.). Government investments on services, such as high quality public childcare centre, are substantially lower that these financial transfers. Parental leave was introduced in 1987 by the "Equal Employment Opportunity Law", but the coverage is low. In its second period (2011-2016) the "Basic Plan for Low Fertility and Aged Society" focused on work-life balance policies.
Taiwan, as a late industrialized country, needed to simultaneously deal with old and new social risks (Yeh and Lue 2014). Therefore traditional income maintenance policies and future-oriented social investment policies came on the political agenda at the same time (ibid.: 17). However, social investment policies were not supported by a majority in parliament, and social spending increases were concentrated on protective social policies, particularly pension policies.
Overall, these examples show that South East Asian countries are indeed concerned with social investment agendas; during the past two decades they have introduced social policy reforms particularly in the field of family and labour market policyhowever to very different degrees. While Japan and South Korea have introduced social investment policies in the field of family and old age policy, in Taiwan no majority coalition was established for social investment reforms. Whereas Japan and South Korea know universal public educationsince this has been the main growth strategy in those countries after WWII (Peng 2014)education in other South East Asian countries is almost all private. Regarding higher education, moreover, most Asia countries are still characterized by extremely high private and low public spending (Garritzmann 2016), so in this respect social investment is comparatively underdeveloped.
Latin America as a developing region faces high levels of poverty, income inequality, and a sizeable minority being permanently trapped in inferior life trajectories (Fenwick 2014). For this continent, the literature that relates to a social investment logic of welfare policies seems mostly focused on Conditional Cash Transfers (CCTs). CCTs are social assistance programs that provide cash transfers to poor households. However, cash transfers are conditional on the families sending their children to school or participating in health plans (Bourguignon et al. 2003, Fenwick 2014, Huber and Stephens 2014, Nelson and Sanberg 2014, Sandberg 2016, Soares et al. 2009). CCTs have the objective to address poverty in the present, while also investing in the future productive capacity of young people, thus reducing the intergenerational transmission of poverty and supporting economic competitiveness. Therefore, social investment in Latin America seems to have predominantly addressed poverty alleviation and investment in the life chances of the poor.
CCTs seem to have spread in Latin America as a compensation to the social cost of structural adjustment programs initiated by the World Bank and the IMF at the beginning of the 1990s (Nelson and Sandberg 2014). Among the first CCT programmes were those initiated in the Brazilian municipalities of Campnas and Brasilia in the mid 1990s, but also programmes in Mexico. Since then, 18 countries in Latin America have introduced CCT programs (Cecchini and Madariaga 2011). In Brazil, for example, a first program aimed at investing in human capital was initiated during the Cardoso Government (Fenwick 2014). This program began at the subnational level in 1995 and it sought to diminish child vagrancy by subsidies to parents who send their children to school. President Lula da Silva's Worker's Party initiated the Bolsa Familia (2003-2010), a national poverty alleviation program that delivers cash benefits to families in situations of extreme poverty (Fenwick 2014: 4ff.). Cash benefits of the program are conditional on school attendance, basic health checks, and prenatal care. In the strongly federalist country, program implementation was contingent on well functioning municipal administrations. In 2011, President Rousseff announced a new strategy called Brasil Sem Miséria ( , Fenwick 2014. The central objectives of this umbrella strategy, primarily targeted to the poor, were guarantee of income, access to public services (health and education) and inclusion in technical and vocational training. The strategy was designed to increase both citizen's individual capacity and equality of opportunity. To summarize, the policy focus of Brazils CCTs was on investing in children, investing in maternal and early childhood health, whereas gender equality and increasing women's labour market participation is not one of the central pillars of the agenda (Fenwick 2014: 8). More studies need to be gathered on the politics of CCTs in other Latin American countries, on the processes of diffusion of such policies throughout the Continent and the politics behind them, but also on other policies than CCTs that still can be associated with a social investment logic.
In sum, social investment policies have been adapted throughout democratic countries in different world regions. The kind and degree of adapted policies as well as the timing of the introduction and expansion of the respective policies, however, differs starkly across countries. Overall, however, we still only have eclectic knowledge on which social investment reforms have been implemented. We still lack a systematic and comprehensive overview of social investment agendas and policies across the democratic world. Also, given that existing research focuses on individual countries or regions, we lack a theorization and an empirical understanding of how institutional legacies and actor interests influence the kind of social investment initiatives that arrive on the agenda, as well as their chances of being adopted.

Effects of social investment reforms
A second, increasingly established strand in the social investment literature investigates distributive effects of social investment policies on social, economic, and political outcomes, such as social cohesion or poverty reduction. This researcheven though focusing on the effects rather than the determinants of social investment reformsmay be of importance for understanding the politics of social investment, which is the main reason why we refer to it here.
Several studies investigating childcare policies in West European countries have demonstrated the existence of a social bias in access to childcare services: high quality public childcare services is disproportionally often used by children living in higher income households (Abrassart and Bonoli 2015, Ghysels and Van Lancker 2011, Schlanser 2011, Van Lancker and Ghysels 2012). These findings show thatat least in Western European countriessocial investment in childcare is likely to support the better-offs more than the lower income quintiles (a finding that we know very well from the other potentially regressive educational sector, namely higher education [cf. Fernandez and Rogerson 1995;Garritzmann 2016]). However, if one includes beside pre-school care, elderly care and primary education as social investment policies into their estimation, new social policies perform better in vertical distribution than old social policies (Vaalavuo 2013). Newer studies show, that the effects of pre-school care services strongly depend on the design of those policies. A strong positive effect on social cohesionas targeted by the social investment perspectivecan only be expected if childcare use is universal or targeted to lower strata and if the costs for low-income families are moderate (Abrassart and Bonoli 2015, Zollinger fthc.). Cantillon (2011) and Taylor-Gooby et al. (2015) have shown that poverty rates in Western European countries have stagnated in the last 20 years, despite an increase of spending for social investment policies. Overall, these studies have deepened our understanding of the (so far) limited potential of social investment family policies to foster social cohesion and poverty reduction. The implication of this finding for politics is that at least in Europe (except in Nordic countries), the main social group benefitting from social investment is the (upper) middle class. This may be one reason why it is precisely this class that seems to be the main supporter of social investment (Beramendi et al. 2015, Gingrich andPalier forthcoming). Hence, it may not be surprising that at best modest effects of social investment reforms on socio-economic upward mobility have been foundthey may not have been intended in the first place.
To give a second example, Hui and Young (2014) argue that to be effective and to have a social and economic return, social investment policies in South Korea should go beyond family policy and actively address the issue of the labour market. Despite the increase in family policy expenditure in South Korea from 2002 onwards, female labour market participation has decreased (Seung-ho and Seung-yoon 2014). Seung-ho and Seung-yoon (2014) see this as a consequence of the design of family policies: The highest expansion of expenditures was on financial transfers for childcare at home and the quality of public services of childcare remains low. In addition, only few South Korean women are insured for parental leave (insurance is restricted to those employees who are insured by the Korean employment insurance for more than 180 days), since females are concentrated in certain occupational segments of the labour market with often precarious working conditions (Seung-ho and Seung-yoon 2014: 5f.). In these cases, the specific design of social investment reforms can also inform us on the expected support coalitions, especially regarding the support these policies can gather from different groups of women.
Thirdly, several evaluation studies show positive effectsat least in the short-termof Conditional Cash Transfer (CCT) programs (Nelson and Sandberg 2014: 4f.). CCT programs can indeed reduce poverty in the short term and the conditionality requirements increase school attendance and the number of health checks. This may be part of the explanation why this has been the major social investment policy field in Latin America. However, the effects of CCTs on the reduction of intergenerational transmission of povertyvia improving participant's chances of finding a jobremain weak (Cecchini andMadariaga 2011, Nelson andSandberg 2014: 5ff.). Levy (2008: 78) pointed to the fact that without more productive jobs, poor workers will need CCT transfers permanently, which may affect the middle class support these policies have initially benefited from. Numerous studies also underline that the effectiveness of CCTs is contingent on state capacity, as well as on the availability and quality of public services. This interaction effect will also shape the political support coalitions CCT reforms can rely on.

Summary: what we know and what we don't know
The literature shows that there has recently been a substantial amount of social investment agendas and policies in several countries in different regions of the world. However, the turn to social investment is highly diverse. Countries in different regions of the world have implemented different types of policy instruments with different functions and different target groups. Moreover, the timing and the degree of the adoption of these policies vary starkly across countries.
In addition, thanks to a growing number of studies investigating the impacts of social investment policies we begin to understand the effects of those policies, pointing to the winners (and the losers) of social investment policies, hence to the potential supporters of (or opponents to) social investment programmes. Taken together, our main take-away point from the literature on effects of social investment policies is to emphasize that different kinds of social investment policies can have very different socio-economic and political effects, as their redistributive dynamics vary. CCTs, for example, seem to benefit especially lower-income groups and therefore have the progressive effects that policy-makers hoped for (at least in the short run). The expansion of childcare, higher education, and ALMPs in most OECDcountries, however, seems to have had little effect on socio-economic upward mobility. Rather, the (upper) middle class seems to have benefitted from these policies. We believe that in order to understand the politics of social investment reforms around the democratic world, we need to keep these differential effects in mind.
What is still lacking in the existing literature, though, is a systematic, encompassing comparison of the varieties of social investment agendas and policies in different regions in the world, as well as an explanation of this variation in terms of the political conditions facilitating or impeding social investment reforms. Therefore, in the WoPSI project we first need to descriptively map which social investment policies countries have conducted. Secondly, to understand this variety, we need further understanding of the political conditions under which social investment agendas and reforms are discussed, elaborated, adopted, or rejected.

Towards a theory of variance in social investment agendas and reforms
Existing theories in comparative welfare state research and public policy analysis provide us with a large set of potential determinants that are relevant in understanding the cross-national and cross-regional variation sketched in section 2 above. In this section, we discuss these potential determinants as a theoretical foundation with which we can define the focus of the WoPSI project, and as a basis for our own theoretical framework.

Structural context conditions
Neo-functionalistic theories (of welfare state reform) assign the development of welfare policies to socio-economic forces such as deindustrialization, globalization, skill-biased technological change, or fiscal constraints. Hence, according to these theories we would expect an increase of social investment policies in post-industrial societies (Bonoli and Reber 2010), as they are cross-pressured by instable family structures, working poverty, low fertility rates, and the dualization of labour markets (Emmenegger et al. 2012a, b). However, crossnational variation within similar structural contexts (as in Western Europe, for example) is at odds with such a functionalist explanation. Therefore, structural context conditions certainly play a role in shaping the policy agenda, as well as the interests and ideas of actors, but we can only understand their effect in interaction with existing institutional legacies.
The context of "permanent fiscal austerity" (Pierson 1998, Scharpf 1991, Stephens et al. 1999) that has been observed in many (predominantly Western) welfare states is a context condition that deserves special attention, because of its political nature and its effect on actor dynamics: the higher the austerity pressure, the more difficult it is to expand social investment as yet another part of the welfare state. In other words, trade-offsor recalibrations and reallocations of public spendingbecome an issue on the political agenda.

Institutions: policy legacies and state capacity
The notion that policy legacies matter for the design of welfare state reforms is a key insight from historical institutionalism (Pierson 1993, 2000, 2001, Skocpol 1992. Historical institutionalists argue that it is not only the case that actors shape policies; it is equally true thatonce establishedthese policies re-shape actors' preferences, strategies, and power relations. In other words, existing institutions affect the menu of available policy options, i.e. existing policy legaciesin interaction with the evolving structural context (Beramendi et al. 2015, Häusermann 2010)influence the policy agenda and the range of discussed policy reforms.
Following Pierson's (1993Pierson's ( , 1996Pierson's ( , 2000 and Skocpol's (1992) seminal work, the bulk of existing studies has assumed that policies generate "positive" feedback effects, i.e. that policies reinforce themselves thereby making radical policy change increasingly unlikely. To give a classical example: In contribution-based pension systems, radical policy change becomes increasingly costly for political actors because workers contribute throughout their working lives to these systems and are highly likely to oppose any change from the contributionbased systems towards e.g. a universal entitlement system. Busemeyer (2015) and Garritzmann (2015Garritzmann ( , 2016, for example, have applied these arguments to the study of social investment policies in OECD countries, and Mettler (2002Mettler ( , 2005 and Mettler and Soss (2004) to the U.S., showing that education systems create strong positive feedback effects that over time lead to strong path dependencies. In this sense, positive feedback effects limit the range of issues that can make it to a policy agenda.
On the other hand, however, scholars have theorized self-undermining ("negative") feedback effects (Fernández/Jaime-Castillo 2012, Jacobs and Weaver 2015, Weaver 2010). To give an example, insurancebased welfare states (endogenously) generate welfare state outsiders over time. In interaction with structural change in families and labour markets, the share of these outsiders steadily increases (Häusermann 2010). In other words, institutional legacies may generate costs and -politically speaking -their own "enemies", just as they generate their own supporters (Mahoney and Thelen 2010). Hence, policy legacies both seem to contribute to define the boundaries of policy agendas, as well as the issues and proposals that make it to this agenda. In addition, they pre-structure political conflict lines.
Regarding social investment, especially two policy legacies are likely to affect social investment reforms: The makeup of the existing welfare state (in terms of the consumption/investment ratio), and the type of production regime. Regarding the makeup of the existing welfare state, it matters for the design and implementation of social investment policies whether extensive social compensation policies already exist and have existed for a long time (Beramendi et al. 2015).
In the former case, social investment policies might only be possible at the expense of other social policies, especially in times of permanent fiscal austerity, while expansive reforms are more likely in countries with a limited compensatory welfare state or a more favourable traditional ratio between investment and consumption.
Regarding the production regime, the "Varieties of Capitalism" approach (Hall and Soskice 2001) argues that differences in the institutional framework of the political economy generate differences in how much and what type of human capital countries strive for (Estevez-Abe et al. 2001, Iversen 2005. In coordinated market economies (CMEs) many firms employ production strategies that rely on a labour force with specific skills (Hall and Soskice 2001: 24f.). Hence, firms in CMEs depend on education and training systems capable of providing workers with specific skills. Skill formation in liberal market economies (LMEs) in contrast is argued to focus on general skills. Only if policy makers introduce complementary institutional reforms, policies will be effective (ibid.: 46). The expansion of formal education to enhance general skill levels for example will be a more effec-tive strategy in LMEs than in CMEs. For this reason Hall and Soskice (2001: 50) hypothesize a correspondence between types of political economies and types of welfare states (cf. also Estevez-Abe et al. 2001, Iversen 2005. Hassel and Palier (2015) argue in a similar vein. According to them, governments of all advanced economies seek economic growth. How they can achieve economic growth, however, depends on the institutional capacities of their states and their political economies. Based on institutional legacies, governments develop and implemented different kinds of "employment and growth strategies" (Hassel and Palier 2015). They differentiate between an "export-led growth strategy" and a "consumption based growth strategy". Both strategies imply the development of different kinds of social policies.
Based on this literature it can be argued that the form of social investment agenda and reforms depends on the growth strategy the country follows. If the growth strategy is based on innovation and high-skilled services (like finance in UK, or ICT in California or East Asia) the question of skill development and education will be at the centre of a social investment agenda (Häusermann and Palier fthc.). An investment in the creation of skills and capabilities of children is the "sine qua non for a sustainable, efficient, and competitive knowledge-based production system" (Esping-Andersen 2002: 28). For countries, which do not pursue a growth strategy based on highskilled services, but rather on manufacturing production, we can expect a social investment agenda, which focuses on skill activation, hence without an emphasis on skill creation or preservation.
There are two further aspects of political institutions that are relevant for our project. On the one hand, macro-electoral institutions (the electoral system, the institutions of government-formation, and powersharing) are relevant when theorizing how actor preferences are aggregated into actual reform coalitions (Häusermann and Palier forthcoming). For this reasons, we discuss them in section 3.5. below as conditioning factors. On the other hand, bureaucratic capacity and state capacity in a more fundamental sense are likely to affect social investment agendas and reform chances, especially in the context of Southern European and Latin American countries. Clientelistic and particularistic practices can play an important role for the expansion of social investment policies (Sandberg and Tally 2015). While lacking state capacity always negatively affects implementation and socially egalitarian outcomes, their impact on the politics and reform capacity is all but clear. They could either support an expansion, as a votebuying strategy of political actors (Hall 2007, Zucco 2008, or play against the adoption of social investment policies (lack of credibility/trust).

Ideas and framing
In addition to structural factors and institutional legacies, ideational factors shape the reform proposals that build a reform agenda. A lot of research has been done on the social investment ideas or perspective as constituting a new social policy paradigm (e.g. Hemerijck 2013, Jenson and Saint-Martin 2006, Morel et al. 2012. The role of these new paradigms in shaping the diagnoses and solutions to (new) social and economic problems is clearly of importance to understand the politics of social investment. In that perspective, it seems that international organizations (like UNICEF, the World Bank, and more recently the EU) as well as some international scholars (like Giddens or Esping-Andersen) have contributed to promote the adoption of social investment policies. Rihanne Mahon (2010) has for instance underlined the role of the World Bank and the OECD in promoting childcare policies. However, since these trends are international, and relatively similar at the global level, they cannot per se explain the diversity of social investment policies discussed and adopted at the regional and national level though.
But ideational approaches also emphasize the (instrumental) reframing of the political discourses as sources for policy change (Béland 2005, Häusermann and Kübler 2010, Kübler 2007. The main argument is that rationalesor framescan be strategically developed to justify (social) policy change. Framing is defined as the ability of actors to influence how a social problem should be interpreted and what the aim of a policy is (Benford andSnow 2000: 216, Hall 2010: 171f.). Policy frames emphasize specific aspects of a policy measure while fading out others.
In the case of social investment, ideas and frames are particularly relevant for understanding both reform agendas and actual reform outputs, becauseas we discuss in detail belowsocial investment is an ambiguous concept and provides a common-sense meaning, which is open to multiple interpretations and can link numerous policy communities (Jenson 2010: 73f., Mahon and Lewis 2006). Non-parental childcare, for example, pursues different goals at the same time: Childcare services substitute the female caregiving function in male breadwinner systems. From this perspective, it can first be framed as a gender-equality policy (Häusermann and Kübler 2010: 173). Second, the investment in women can be framed as a labour-market policy. By the activation of female human capital, it can supply labour markets with high skilled women. Moreover, perceived as an investment in children, childcare policies can contribute to the social integration of children from different socio-economic and socio-cultural backgrounds and can support social cohesion (Esping-Andersen 2002). Hence, it can thirdly be framed as a social-integration policy. In a similar vein, Jenson distinguishes three dimensions of the social investment discourse as it was developed within the relative bureaucra-tic and expert networks (Jenson 2012: 28-32). The Nordic countries focus on human capital formation by constant, life-long learning. In liberal welfare regimes the second dimensionorientation to the futureis important. Addressing child poverty was one of the big problems of New Labour. In corporatist regimes the idea that social investment policies benefit not only individuals but the whole society was preeminent: they supported mainly those social investment policies, which supported labour market participation and helped to raise fertility rates. The chosen policy instruments are according to Jenson dependent on which dimension is stressed over another dimension in the political discourse (Jenson 2012: 32).
Given this inherent multidimensionality of social investment, ideas and framing (as promoted by both domestic and international political actors) are supposed to be a key element for understanding the reform agenda, i.e. the set of diagnosed problems and proposed solutions to them. But ideas and framing are also relevant when it comes to politics: the ambiguity inherent in the social investment concept opens up room for modernising coalitions between actors who purse different goals with the same policy instrument (Häusermann and Kübler 2010), so-called "ambiguous agreements" (Palier 2005). Hence, we can expect political actors to reframe the political discourse strategically in order to build majority coalitions in favour of a social investment reform.

Actors and preferences
Social policies are the result of a distributive struggle between different societal interests (Esping-Andersen 1990, Huber et al. 1993, Korpi 1983, Stephens 1979. Political parties, trade unions, and employers have been identified as key representatives of these interests, which is why their preferences matter for understanding the politics of social investment. In addition, bureaucratic actors and experts are also relevant when explaining public policy variation.
Social policy outputs in the industrial age were generally conceived as the result of the distribution of power between collective actorsespecially political partiesrepresenting the interests of labour and capital (Alt 1985, Hibbs 1977, i.e. a one-dimensional axis with stateinterventionism on one extreme and market-oriented policies on the other. In this view of a distributive conflict, welfare state policies are conceptualized along such an economic left-right dimension. Whereas left political parties and trade unions are assumed to stand up for welfare state expansion, right-wing parties and employers are theorized to want to limit welfare state development (Boix 1998, Busemeyer 2009, Castles 1989, Esping-Andersen 1990, Huber and Stephens 2000, 2001, Huo et al. 2008, Schmidt 1996, Seeleib-Kaiser et al. 2008. Several studies have shown that the expansion of the welfare state in capitalistic democracies during the "Golden Age" can be reasonably well explained on the basis of this model (Esping-Andersen 1990, Huber and Stephens 2001, Korpi 1983, Myles and Quadango 2002. Several authors have transferred these arguments to the study of social investment, in particular with regard to new social risks. Some have remained highly sceptical as to the chances for substantial change in the direction of social investment policies, since the victims of postindustrial social needs and demandssuch as lone mothers, young families, low educated persons, or divorced parentsare a very heterogeneous and dispersed group that is not only weakly represented in parliaments, parties, and trade unions, but also unlikely to mobilize massively (Bonoli 2005, Ebbinghaus 2006). People concerned with new social risks lack the conscience of a "common cause", i.e. their specific needs diverge too much (Kitschelt and Rehm 2006), and there is no single actor likely to bundle these interests, as both parties and trade unions are conflicted between the preferences and interests of different constituencies (Häusermann 2010, Rueda 2007. Against this background, social investment policies (at least those targeted at new social risk constituencies) "are unlikely to be the result of the political mobilization of those who benefit from them only" (Bonoli 2013: 55).
Power resource theory thus allows developing hypotheses in both directions: Either social investment policies arejust like other social policiesintroduced and expanded by (left-wing) political actors, or parties and unions are unlikely to matter in the case of social investment because the affected groups are too marginalized. Several studies have tested these claims and came to diverging conclusions. Huo et al. (2008) find support for the importance of left parties for the expansion of active labour market policies (ALMPs). In their view, left parties favour ALMPs (and not only de-commodifying policies), since social democrats have a preference for full employment, and ALMPsas well as all other kind of social investment policiescan contribute to this end (Huo et al. 2008: 7). Boix (1998) and Iversen and Stephens (2008) confirmed this finding, but others found no effects and emphasize the fact that Social Democracy is torn between different goals (job creation vs. job protection) when it comes to labour market policy (Rueda 2005, Tepe and Vanhuysse 2013, Van Vliet and Koster 2011. Similarly, political conflict lines are contested when it comes to family policy and education policies. While many studies still find left-wing parties to be the most important advocates of progressive family and education policies (Boix 1998, Busemeyer 2009, Castles 1982, Garritzmann 2016, Häusermann and Zollinger 2015, Hieda 2013, Iversen and Stephens 2008, Naumann 2012, Schmidt 2007, several studies also emphasize that the political configuration of interests is not as clear-cut as in the past (recently for example Garritzmann and Seng 2016 for the case of education), especially as current welfare politics implies difficult trade-offs: tradeunions may privilege consumption instead of investment, whereas employers may prioritize investment over consumption (Beramendi et al. 2015). This is in line with Breunig and Busemeyer's (2012) finding that discretionary spending (such as education and many other social investments) are more retrenched in times of austerity than entitlement spending (such as pensions). In addition, market-liberal parties and employers have supported activation-oriented types of family policy expansion in many countries (Häusermann 2006, Hieda 2013. Theorizing the positions of collective political actors with regard to investment reforms is further complicated by the fact that the membership profile and constituencies of these actors have radically changed. While Social Democrats have become middle class parties and trade unions tend to represent workers with above-average wages Häusermann 2015, Becher andPontusson 2011), the traditional working class voters increasingly vote for the populist right (Rydgren 2012). Therefore, in order to understand actor interests and strategies, it has become more fruitful to theorize preferences at the level of social groups (rather than at the level of collective actors, Beramendi et al. 2015), because the collective actors who represent these societal interests vary across countries and contexts. In this vein, and based on what we know from the literature on the effects of social investment, we may hypothesize that the educated middle classes, as well as parts of business should be the most ardent supporters of social investment reforms (Beramendi et al. 2015, Häusermann and Palier forthcoming). Besides social classes, women can be seen as another important social constituency for social investment policies. Especially for the expansion of family policieswhich ease the conciliation of care and workpolitical power resources of women are seen as an important explaining factor. Several authors have pointed to the fact that women's movements as well as their representation have an influence on welfare state development (Huber and Stephens 2000, 2001, Naumann 2005, O'Connor et al. 1999, Orloff 1993, Sainsbury 1996, and the expansion of reconciliation policiesat least in OECD countriesis indeed positively correlated with the share of women in parliaments (descriptive and substantive representation) (Bolzendahl andBrooks 2007, Campbell et al. 2010).
Moreover, politics is not just a struggle between societal interests, but this struggle is shaped and guided by bureaucrats, civil servants, and experts. Jenson (2012: 23) argues that innovations concerning a social investment oriented welfare state emerged within the public administration. She argues that unelected civil servants played a crucial role in reconfiguring the universe of political discourse, i.e. the development of the social investment perspective. The public administration is engaged in puzzling over policy problems and in doing so civil servants are developers of social knowledge. The configuration of the political discourse is a power resource, since it provides greater representative legitimacy to some actors and their ideas than to others (Jenson 1989). In broad networks involving actors from international organizations or scientific actors, bureaucratic actors are able to promote ideas and control the discourse in the political space (Jenson 2012: 24). According to Jenson (2012: 33), the public administrationand especially the finance ministriesis a central actor in promoting the social investment idea as well as concrete policy proposals.

Conditional explanations of policy variation
Over the past decade, the comparative welfare state literature has emphasized that welfare politics can and should not be conceptualized in a purely linear way, as the outcome of interests and power relations only. Rather, both the interests and the interactions of actors are conditioned by important context factors (Häusermann et al. 2013). Hence, when explaining the success or failure of reform coalitions in favour of social investment, such conditional arguments are highly relevant. There are three crucial conditioning factors that have been analysed in the literature: 1) the institutional policy legacies that shape actor preferences, 2) the relative saliency of different political cleavages which condition coalition-formation between actors, and 3) the macro-electoral institutions that condition power relations and the likeliness of coalition-formation.
Let us first focus on policy legacies as conditioning factors of actor interests. The idea here is that the distributive effects of a particular reform depend on the pre-existing policy legacies. Therefore, the positions of actors will reflect this interaction of distributive preferences and policy legacies. For example, several authors argue that party positions in the field of education policy can only be understood if one takes into account the distributive consequences of the existing education system (Ansell 2010, Fernandez and Rogerson 1995, Garritzmann 2016, Gingrich and Ansell 2015, Iversen 2006. Consequently, left parties cannot be expected to be universally in favour of higher spending levels in all cases. Higher education, for example, is fiscally regressive, if enrolment rates are not universal and access to higher education institutions is stratified by family background (Fernandez and Rogerson 1995). Complicating things further, the redistributive patterns might change over time as enrolment levels increase (Ansell 2010) and they become even more difficult to disentangle when we simultaneously take public and private spending into account and differentiate between different kinds of spending (Garritzmann 2016). Thus, the preferences of parties are likely to vary depending on the type of spending, the status quo of the respective education system, the access conditions, and other institutional context factors. Similarly, Gingrich and Ansell (2015) show that the preferences of left-and right-wing actors in favour or against activating labour market reforms depend on the dualized or universal structure of the labour market. In other words, distributive effects of investments in ALMP depend on their interaction with pre-existing employment regulation. For our project, this implies that expected preferences of societal groups, political parties, and other collective actors will depend on the distributive consequences of specific policy proposals and the already established status quo of the respective systems. Hence, conditional on policy legacies, similar actors may support different social investment policies in different contexts.
A second conditioning factor that affects the theorization of actor interests and strategies (beyond a simple left-right dimension) is that the political space (in Western Europe) today is not one-dimensional, but at least two-dimensional, with distributive social and economic conflicts being complemented by a second, cultural and identity-based dimension of political competition (Häusermann 2010, Kitschelt 1994. This second dimension opposes advocates of cultural liberalism and libertarian values to advocates of conservatism and traditionalism. Because of the relevance of social investment for gender equality, equality of opportunity, and education, these values bear strong relevance also for the distributive politics of social investment (Fossati andHäusermann 2014, Häusermann 2010). In today's welfare politics, political parties position themselves in this two-dimensional policy space, which opens up opportunities for new actor coalitions. This two-dimensional actor configuration can provide a basis for modernizing coalitions (again "ambiguous agreements", cf. Palier 2005) among left-progressive actors and liberal actors, such as business interests or market-liberal parties (Beramendi et al. 2015, Häusermann 2010). According to this argument, the high diversity of social investment policy reforms is the result of new political coalition dynamics in a multi-dimensional policy space. This implies that simple, linear arguments for policy change (e.g. "the stronger the left, the more expenditures") will not be sufficient. Rather, we need to take into account the number and kind of conflict lines a policy reform generates, as well as the relative saliency of these conflict lines, in order to conceptualize the coalitional dynamics of social investment politics.
Finally, the macro-electoral institutions of a country influence the kinds of coalitions that are likely to form. While power-sharing institutions increase the number of veto points and therefore the obstacles for successful policy change (Huber et al. 1993, Lijphart 1999, Tsebelis 1995, they also allow for increased coalitional flexibility and for multidimensional reform politics (Häusermann 2010). Hence, while social policy reforms are clearly slowed down by decentralized state structures (Vatter 2002(Vatter , 2014, the influence of proportional power sharing on social policy reforms is less clear (Vatter and Freitag 2002). Lijphart (1999) argued that compared to majoritarian political systems, consensus oriented political systemscharacterized by proportional distribution of political power in legislative and executivecan have a positive influence on social spending (mostly via mechanisms of coalition-building and log-rolling, see Birchfield and Crepaz 1998 on this). Häusermann and Palier (forthcoming) argue that in a consensual institutional framework, coalitional dynamics are more flexible, and varying alliance potentials have better chances to result into actual social investment reform coalitions.
To summarize, the literature review showed that an analysis of the variety of social investment agendas and policies should consider a number of structural, institutional, and actor-based determinants, as well asimportantlytheir interactions. Regarding the structural and institutional context factors, socio-economic factors (such as globalization, deindustrialization, changing family structures, etc.), policy legacies (especially existing welfare state and social investment legacies as well as the type of production regime), and political institutions appear to be crucial. Regarding actors and power relations, studies of the politics of social investment need to consider the role of societal groups and political parties, of unions, employers, public administrations, and experts. We by now also know that not only the relative power of these actors, but even their policy preferences are contingent on a number of factors. In other words, we should not treat actor positions as exogenously given (anymore). A theoretical framework to explain variance of social investment agendas and reforms therefore needs to integrate these factors and their interactions into a model of political coalition-formation. This is what the WoPSI-project aims at.

The core research questions
Several studies have documented the existence of a high variety of social investment agendas and policies in different countries around the globe. However, there exists neither a systematic descriptive comparison of the existing variety of policy agendas and reforms nor a systematic analytic comparison on how this variety of social investment agendas and policy reforms can be explained. Our research project intends to address this research gap by studying the politics of social investment across democratic countries in Latin America, Asia, as well as Eastern and Western Europe from a comparative perspective. An analysis of the variety of social policies has to explain both social investment agendas and social investment reforms and it has to take complex conditional effects of contexts on coalition dynamics into account. We are interested in the following research questions:  How do social investment agendas and social investment policy reforms vary across democratic countries around the globe? How can we explain this diversity?
 Under what political conditions do social investment agendas and/or reforms develop?

What is social investment?
"Social investment" is a very broad concept. Accordingly, very different definitions and understandings of what social investment is (or should be) exist in the literature. In order to clarify our understanding of social investment and in order to allow systematic comparative empirical analysis, we provide here our definition of the concept. We find it helpful to think about social investment in terms of its goals, functions, and policies. We refrain from providing a simple policybased definition, because many (if not all) social policies can include elements of social investments, which might be overlooked when focusing too narrowly on a few specific policies only.
1. The goals of social investment. Social investment aims to prepare, support, and equip individuals in a way that increases their chance to participate in the knowledge-based economy and reduces their future risks of income loss and poverty. Hence, it is about investing welfare resources now in order to harvest returns in the future. This strategy involves avoiding social risks and overcoming intergenerational transfer of disadvantages and poverty (Esping-Andersen 2002, Jenson 2010. In other words, social investment aims to prepare individuals to have less to repair. The future-oriented social investment logic stands in contrast to purely transfer-oriented consumption policies (Beramendi et al. 2015, Hemerijck 2012, Morel et al. 2012. Whereas consumptionoriented policies provide social security immediately via decommodifying social policies such as income replacement in case of unemployment, old age, or sickness, social investment policies work differently, as they focus on future income-streams. The goals of the social investment strategy are Janus-faced: social investment policies serve both social and economic goals (Kvist 2016). In terms of social returns, social investment policies are supposed to contribute to decrease poverty levels in the longer run, to increase social cohesion, reduce inequalities, and contribute to human development. In terms of economic returns, the social investment perspective expects to increase labour market participation and employment performance. A better-educated and better-equipped population is expected to contribute to higher growth rate (since workers can better contribute to the knowledge based economy). Beyond education and training, social investment is also aimed at better helping people to mobilise, preserve, and improve their working capacities and skills. Increased labour market participation, less unemployment, and more growth canin turnenlarge the tax base, increase the volume of social contribution and taxes paid, thus contribute to the future viability of the welfare state. 2. The functions of social investment: create, mobilise, and preserve skills/human capital/capabilities. In a life course perspective social investment policies have the function to enable individuals to prevent and/or overcome difficult life events and life transitions without losing their earnings capacity (Kvist 2016: 6f.). According to this perspective, social investment policies help to create, strengthen, maintain, re-establish, and use individuals' capabili-ties and skills. One can thus distinguish three main functions of social investment programmes. Social investment policies can first refer to an investment in human capital formation (skill creation). Second, it can comprise an investment in the mobilization of human capital for labour market participation (skill mobilisation). Third, an investment in the preservation and improvement of human skills and capabilities to better handle life events and transitions (skill preservation). Skill creation focuses on the formation of human capital (creating capabilities), skill mobilisation focuses on the effective use of existing human resources, and skill preservation focuses on the maintenance or improvement of the capabilities. Social investment policies aim at fulfilling (one or more of) these functions. 3. Social investment policies. Many different policies can serve the social investment goals and fulfil social investment functions. Most immediately and obviously, the social investment perspective translates into strengthening certain education and social policies and into developing accessible and high quality welfare services. Starting from a Western European perspective, a key focus of our project will, of course, lie on social investment in the field of family, labour market, and education policies (early childhood education and care, school and post-secondary education, as well as lifelong learning). However, following our definition of the social investment conceptincluding an integrated life course approach to human capital enhancementother policy fields such as health care policy, housing, or disability policy may also be concerned with a social investment agenda and/or reforms. As pointed out by YM Kim (2007: 14-19, 18), a precondition for social investment to work is that various factors such as a decent home and community environments, good health care, and culture and physical fitness policies are present. Therefore, any strategy to enhance or use human capital needs to recognize the influence of the social setting (ibid.: 18). Next to the usual set of policy domains (early childcare and education policies, conciliation of work and life policies, active labour market policies, and training), this can also be obtained by labour market policies that support the move from school to work for young people, by general housing supporting low income groups, or by programs supporting physical training for elderly people (Kvist 2016: 7). Conditional cash transfer policies for examplewhich are prominent in Latin American countriesmake health insurance coverage for poor people conditional on sending children to a physician for a general check up.
So how can we identify social investment policies concretely? In order to be classified as a social investment policy in our understanding, a policy must aim to achieve (some of) the aforementioned goals and functions. If one looks at the goals and functions of social investment policies, there are sometimes ambivalent domains, such as active labour market policies. The activation of unemployed people for the labour market by negative incentives such as sanctions and benefit reductions only is not part of our understanding of the social investment logic (cf. also Bonoli 2013, Taylor-Gooby 2004. Only positive skill creation, mobilization, and preservation, corresponds to our understanding of social investment. The goal of social returns can be reached by social inclusion, by fostering capabilities (improving skills and empowering jobless people) and by securing quality jobs (Bonoli 2013: 23, Morel et al. 2012.

Social investment agendas
In order to analyse the politics of social investment, we are convinced that we first need to understand the dynamics leading to a social investment policy agenda. An agenda comprises a politicized problem, a diagnosis of the problem, and a proposed policy solution (Kingdon 2010). However, in the policy analysis literature there is a distinction between a public and a formal agenda (Howlett et al. 2009). A public agenda comprises all salient topics put forward in public debates, for example during election campaigns or in the mass media. A formal agenda more narrowly is defined as a problem, which is put on the parliamentary agenda. A formal agenda comprises all governmental bills discussed in parliamentary decision-making processes. To make agendas observable, we follow a two-step definition, which encompasses beside a formal agenda the public discourse (public agenda) as well.
First, we focus on the formal agenda. A formal agenda comprises all policy proposals that are discussed in parliaments. This includes proposals that became bills and laws in the end, but also non-adopted proposals (e.g., suggested by oppositional politicians). However, we argue that not everything that is politicized in public debates leaves a trace in parliament. We expect that there exist big and publicly visible social investment campaigns, which never make their way on the formal agenda, but which might still be of interest for understanding the politics of social investment. The fact that they comprise nondecisions is also telling for our project.
Therefore, we are also interested in visible public social investment campaigns for example during elections. Such campaigns putting social investment policies or strategies on the public agenda can consist in initiatives of interest groups and local movements, as well as major campaigns and/or manifestos of political parties or politicians, which may be salient during elections. To summarise, while public and formal agendas need to be kept analytically apart, we are interested in an analysis of formal and public agendas comprising social investment policy proposals.

Social investment reforms
Most of our interest lies in understanding the political processes leading to the adoption of social investment policies or reforms. In most cases, it will be reforms, in the sense that the policy decisions are embedded in already existing sets of polices that are changed and adjusted (reformed) by the decisions taken. This is why we prefer to use the "reforms" terminology (instead of "adoption" or "introduction"), as it suggests that reforms are more likely to happen than the establishment of entirely new policies. Social investment reforms always take place in a given (institutional) context. A policy reform is defined as a policy output that results from a parliamentary decision-making process. A policy reform comprises a formalized policy solution in the form of a law or a decree. One policy reform can include a range of different elements, whereby maybe only one of those elements complies with our definition of social investment. By a reform element we mean one single instrument of intervention. If one of the reform elements corresponds to our definition of social investment, this part of the policy reform should be defined as a social investment reform. It should be studied as such, in its interaction with the other parts of the reform since it may well be part of a policy package that is of interest to understand social investment.

The theoretical framework of the WoPSI project
To understand the high variety of social investment agendas and policy reforms in different democratic countries, we argue that political coalitions (actors and their interests), as well as the institutional embeddedness of social investment politics are key factors. Social investment strategies first need to be put on the public or formal agenda. In a second step, the parliamentary decision-making process is crucial in understanding success or failure of a proposal and the content it will eventually take. Hence we need to theorize both the determinants of agendas and reforms sequentially. We draw on the discussion of existing theories in chapter 3 to develop a theoretical framework that is both encompassing and analytically distinctive. We summarize our theoretical framework in figure 1 below.
In terms of the analytical focus of the project, we particularly emphasize the role of (1) institutions, i.e. existing policy legacies and (2) coalitional dynamics for explaining reform outputs. More concretely, we argue that the institutional embeddedness of social investment politics as well as political coalitions (actors and their interests) are key factors in explaining the high variety of social investment agendas and policies between countries (Häusermann and Palier fthc.). Policy legacies condition the way structural pressures appear on the reform agenda, and they also influence the actors, their interests and their ideas that are relevant at this stage. As for political coalitions, we consider them a key factor because of the well-established multidimensionality of social investment.
More specifically, we first argue that the social investment policy agenda of a country depends crucially on the pre-existing institutional context in the form of policy legacies, as well as on the interaction of these policy legacies with structural changes. In other words: socioeconomic conditions, such as high divorce or high child poverty rates do not directly explain the type of problems policy makers perceive as relevant and the respective policy agenda. The specific institutional context interacts with structural problem pressures and leads to different policy agendas for different types of welfare and production regimes (Iversen and Wren 1998;Scharpf and Schmid 2000). In addition to policy legacies, the presence and relative power of actor interests, as well as ideas promoted by national and international bureaucratic expert actors account for social investment agendas.
Second, we argue that the political success of these (agenda-specific) social investment policy proposals depends on the availability of a political majority coalition, which supports the reform proposals in the political decision-making process. Such a coalition does not require consensus or unanimity and based on the highly ambiguous character of the social investment concept different actors can support a policy reform for quite different reasons. However, a policy reform needs a political coalition that is broad enough to establish a majority. Both the configuration of actor interests and their power relations, as well as their framing strategies and the macro-institutional environment will shape the likelihood for agenda proposals to result in a sufficiently large support coalition in order to be adopted.
As the literature review illustrates, social policymaking is influenced by (several types of) actors at multiple stages. We distinguish the relevant actors in the process of agenda setting (interests I, see figure 1) from those actors relevant for coalition formation (interests II, see figure 1). As for "Interest I" (actors acting upon the agenda setting), we see a broad range of potentially relevant actors. Beside political parties, trade-unions, and employers, technocratic experts (oftentimes from international organizations) contribute to the advocacy of social investment in a country and can have a substantial influence in the process of agenda-setting. Not only public administration but producer and interest groups as well as social movements, scientific experts, and journalists can certainly contribute strongly to the shape of the social investment agenda in a country. As policy entrepreneurs (Kingdon 1995) they contribute to the diagnosis of problems, the formulation of solutions and the strategic coupling of the policy streams during critical moments.
However, in democratic countries policy reforms are the product of parliamentary decision-making processes. Policy reforms depend on the presence of relevant social forces that are able and willing to support (a reorientation of) a policy. Such social forces are mobilized and organized bycountry-specificcollective actors (political parties or interest organizations, trade unions etc.). We argue that thinking about support coalitions for social investment in terms of class coalitions makes sense politically. Hence, as relevant actors in the decision making process of social investment reforms (Interests II) we see all collective actorsrepresenting social classesinvolved in parliamentary decisions.

Figure 1: Theoretical model explaining social investment agendas and reforms
In sum, we are interested in an analysis of the agenda setting dynamics and/or the dynamics of reform-processes. Within these two areas, we are particularly interested in policy legacies as well as coalitional dynamics as important mechanisms explaining policy agendas and social investment reforms respectively. As illustrated in figure 1, we wanthoweverto keep a sufficiently open analytical perspective on further determinants, since we are aware that country-or region-specific specific idiosyncrasies may matter greatly.
The goal of the project is to answer our research questions on the basis of empirical studies of social investment agendas and reforms in Latin America, East Asia, as well as Eastern and Western Europe. We are only interested in democratic countries. The analyses should map and explain social investment strategies appearing on the formal or the public agenda, as well as social investment policy reforms in the period between 1990 and 2016. An obvious focus lies on social investment strategies and policies in the field of family, labour market, and education policy. However, according to our definition of social investment (see above), it is possible that social investment proposals or reforms can be observed in other policy fields such as health or housing policy, as well. Hence, we deliberately do not restrict our empirical design to specific policy fields. Rather, we are interested in social investment agendas and reforms in all social policy fields.
For the part of the project that will systematically trace the agendaand coalition-building processes around social investment reforms, our units of analysis are the proposals that form the social investment agenda, as well as individual social investment policy reform processes. Our key focus lies on formal agendas and legislative reform processes at the national level. However, if there have been important and major reforms and/or political debates at the subnational level, we include agenda-setting and/or coalitional dynamics at the subnational level as well. Sources for quantitative or qualitative data on social investment agendas and reforms can vary across countries, depending on the political systems. For the systematic tracing of the politics of social investment across world regions, we envisage the data collection in two steps. The objective of the first stage of data collection is to compile an inventory of social investment agendas and reforms between 1990 and 2016. The second stage requires data on factors explaining variance in social investment agendas and/or reforms. According to our theoretical argument, we choose to focus mainly on policy legacies as crucial determinants of social investment agendas and coalitional dynamics (interests) as determinants of policy reforms.