Asset-based Welfare and Social Investment: Competing, Compatible, or Complementary Social Policy Strategies for the New Welfare State?

Abstract More recently, two key developments have been observed in the comparative social policy literature: on the one hand, the implementation of a proactive “social investment strategy”, either alongside or in replacement of their established social security programs, and on the other hand, a concomitant shift towards the increased macroeconomic and political importance of private property assets – and homeownership in particular – in defining the economic well-being of individuals. At first glance, it seems that we are dealing with two conflicting policy paradigms here; after all, social investment relies on stable or even expansive welfare states, while the accumulation of private property wealth as a welfare resource seems to realign better with the notion of welfare state retrenchment. This contribution aims to illustrate that the fault lines between the two policy paradigms are, however, not that clear-cut. Based on comparative national-level statistics for all OECD member states in the 1995–2007 pre-crisis period and theoretical reasoning the paper argues that the two approaches may be understood as compatible welfare readjustment strategies, which have opened out into a more radical form of productive welfare capitalism, particularly in the liberal and social-democratic welfare states in North-Western Europe.


Introduction
Up until the mid-2000s, welfare state change has been primarily understood as a process of retrenchment, deregulation and withdrawal from the traditional social protection functions of the state (e.g. Gilbert 2002;Pierson 2001; for an overview see Starke 2006), targeting, inter alia, employment protection, social benefits and public pension systems (e.g. Hemerijck 2013). More recently, however, social policy researchers have become increasingly aware of two other salient transformations. First, often placed in the wider context of an emerging "New Welfare State" (e.g. Taylor-Gooby 2008), it has been argued that many governments have implemented a strategy of proactive "social investment" (SIS) (Morel, Palier, and Palme 2012), which combines social policies that aim to increase labour market participation through human capital investments with policies that seek to facilitate a better balance of work and family life (Bonoli and Natali 2012). Second, there has been a concomitant shift towards an increased macroeconomic and political emphasis on ownership of private property assets in defining the economic well-being of individuals (Ronald and Doling 2012;Ronald, Kadi, and Lennartz 2015). Indeed, facilitated by the rapid financialization and boom of housing markets prior to the Global Financial Crisis, increasing numbers of households have been expected to and supported by policy-makers in taking on high levels of private debt against the promise of higher levels of residential property wealth in the future; the rationale being that such (housing) assets may function as a private welfare resource at all life stages, particularly in enduring periods of economic need and reduced income (Ansell 2014;Lowe, Searle, and Smith 2012).
At first glance, it seems as if we are dealing with two conflicting policy paradigms here. After all, the social investment strategy (SIS) relies on the continuation of the welfare state, while the accumulation of private property wealth as a welfare resource realigns better with the notion of collective welfare withdrawal. This paper, however, promulgates the inverse position. Using comparative national-level statistics for 26 OECD member states, it aims to empirically support as well as explicate theoretically the notion that, instead, they may be understood as mutually reinforcingcompatible welfare readjustment strategies, potentially bringing about a more radical form of productive welfare capitalism. By doing so, this paper will contribute to two key contemporary debates. It speaks to the social policy literature, providing a more nuanced understanding of diverging trajectories of welfare state change and the question of why the social investment agenda has been only becoming meaningful in specific countries. On the other hand, it addresses the housing studies literature by offering a new perspective on the significance of property ownership and individual housing equity wealth for welfare state restructuring rather than retrenchment processes.
The paper proceeds as follows: the next section examines the theoretical foundations of the "social investment strategy" and "housing asset-based welfare", giving a more precise account of what they mean in social policy practice and how they compare to the more traditional policy approaches. It goes on, based on aggregate data on social spending and housing finance, respectively, to investigate empirically where and to what extent the policy approaches have actually materialized, showing that there is considerable regional overlap. The fourth section then discusses why these empirical links appear to be far from being coincidental, arguing that they should be considered as mutually reinforcing, compatible policy approaches instead. Finally, we will summarize the argument and discuss the main limitations to the "compatibility thesis" that we set out in the paper.

Social Investment and Asset-based Welfare -Two Emerging Policy Paradigms
The SIS In the 1980s and up until the early 2000s, many social policy commentators predicted that the dismantling of European welfare states would converge around a residual, low-expenditure regime reminiscent of the US social policy model (Pierson 2001). And indeed such a view has been supported by various policy shifts that have typically been qualitative in nature and have addressed the reduction of the encompassing character of some social programs. To name but few key markers: first, here has been the convergence towards multi-pillar pension systems in almost all welfare states and a rapid shift from benefit-defined to contribution-defined pension plans (e.g. Van Kersbergen and Hemerijck 2012). Meanwhile, labour market policies were reoriented towards stronger activation measures, stricter conditionality for unemployment benefits and the recommodification of labour through, inter alia, the promotion of temporary, de-unionized contracts, particularly in low-skilled, low-income sectors, ultimately fabricating a "dualizing" division between labour market insiders and outsiders (see Emmenegger 2012). Finally, on a macroeconomic policy level, governments have followed the belief that balanced budgets, the reduction of public debt, hard currencies and low inflation are the only way to economic growth and sustainable market economies (Starke 2006).
However, more recently, various authors have noted that despite these retrenchment policies, welfare states have endured and rather a convergence to the top in terms of public social spending has taken place (see Obinger and Starke 2014). One explanation for this development is the emerging reorientation of social policies towards a the more productive functions of the welfare state, in which it is considered an inherent part of the wider socio-economic settlement and may have a supporting role vis-à-vis economic policies (Hudson and Kühner 2009). The key to understanding the multi-faceted nature of welfare state change thus is to fathom that social policy is not only geared towards income protection and poverty reduction (i.e. addressing "old social risks"), but it may be equally concerned with economic development and the maximization and growth in high-quality, high-skilled employment (Bonoli and Natali 2012).
Many prominent social policy scholars have subsumed these transformation processes under the label "Social Investment Strategy" (SIS)where "new risk policies", "pre-distribution", "Third Way" or "active welfare" have been proposed as alternative terms (for an overview see Taylor-Gooby, Gumy, and Otto 2015). The foundation of this social policy paradigm is a life course perspective on social problems and risks, where the focus lies on the prevention of the materialization of old and new social risks rather than protecting citizens once an adverse social or economic event has occurred (Esping-Andersen 2002; Morel, Palier, and Palme 2012;Vandenbroucke and Vleminckx 2011). Building on public investments in the current and future employability of risk bearers, the SIS integrates two key elements: first, it seeks to endow the (potential) workforce with skills and human capital that can prepare them for the insecurities and constantly changing demands of deindustrialized, knowledge-intensive and service-based economies. Second, it aims to prevent the materialization of new social risks by giving citizens, and in particular women, opportunities to utilize their human capital in the labour market and maximize their employment efforts, through, inter alia, of more abundant and better childcare and eldercare facilities (Bonoli 2012;Hemerijck 2013).
One of the key debates in the social policy literature has revolved around the question of what constitutes the SIS in policy practice. Hemerijck (2013), amongst others, has proposed the distinction of, on the one hand, service-oriented capacitating welfare policies as more productive social investment spending, and on the other hand, income-oriented compensating policies as the more protective old social risk policies. Where the former comprises public expenditures on active labour market policies, family policies, childcare, education and in-kind rehabilitation services for the elderly, the latter is constituted of public spending on old age, unemployment, survivors, as well as disability pensions transfers (for similar approaches see Nikolai 2012; Van Vliet and Wang 2015). However, such a clear-cut policy field distinction certainly is not without problems (Nolan 2013). For example, unemployment benefits do not only serve the purpose of cushioning a household's path to lower income, but in its aggregate function, also retains the purchasing power of the society in the short run, particularly in times of mass unemployment. Similarly, in the sphere of family policies, generous child benefits and parental leave schemes clearly protect the income of young parents against the risk of having a child, and in effect often discourage mothers from returning to work on short notice. On the other hand, the provision of childcare facilities and the availability of parental leave arrangements that do not discriminate between mothers and fathers predominantly serve as employment activation measures. In short, even though the main idea and the enthusiasm for the SIS are easy to discern in parts of the welfare state literature, a watertight definition for empirical investigations has not been offered yet.

Asset-based Welfare -Housing Property as a Welfare Resource
The second broader transformation process concerns the fact that in almost all welfare states in the OECD a shift in responsibilities from public to private domainsin regards to private providers of welfare services as well as private households as welfare consumershave been observed over the past two decades. In particular, in the context of demographic change towards social ageing, relatively smaller workforces and the resulting diminishing public funding and resources, a fundamental discussion about how to make best use of the existing private means has been on the agenda of policy-makers more recently. Here, a relatively new important variable in the equation of private welfare security has emerged in the form of private property ownership and a related asset-based welfare (ABW) paradigm.
The principle assumption of ABW is that in times of curtailed unemployment benefits and defined contribution pensions, middle-as well as lower income households can be encouraged to accept greater responsibility for their own welfare needs by supporting their investments in property assetstypically an owner-occupied home, but in some cases also additional properties that could be rented outwhich augment in value over time. In theory, the accumulated wealth can be tapped into at any stage of the life course to supplement consumption and welfare needs when income is reduced or becomes insufficient. Using this wealth has been linked to the financing of private health care and care needs, to the supplementation of public pension entitlements and to the financial support for children and other family members to pay for educational qualifications or house purchase (Ronald and Doling 2012).
A first crucial consideration here is that wealth and income do not necessarily have to match throughout the life course. It might be true that higher incomes open up possibilities to accumulate assets, and vice versa; yet, in a dynamic perspective, the appreciation of assets can, at least temporarily, be much more rapid than the growth of real income: a development which on the macro level could be specifically observed in the late 1990s and early 2000s leading up to the onset of the GFC (see Aalbers and Christophers 2014). This potential disconnection then means that, in theory, private property wealth can be used as a "buffer stock" against increased risk of income poverty, where, in its most extreme form, it implies a shift from "employment-lead" to "asset-dominated" welfare stratification in Western societies (Ansell 2014). Given the astounding rise of housing wealth in the Eurozone, for instance, from €3.7 trillion in 1980€3.7 trillion in , to €13.2 trillion in 1999€3.7 trillion in , to €24.2 trillion in 2006€3.7 trillion in (ECB 2006, a period in which income from wages hardly increased at all, the numbers behind this shift seem to support this assessment. Another important consideration is that in reality, the ABW paradigm mainly relies on homeownership assets, simply because these make up the largest share in most households' wealth portfolio (for Eurozone data, for instance, see Arrondel, Muriel, and Savignac 2014). Given this assumption, how then does homeownership function as a welfare resource for households exactly? First of all "homeownershipbased welfare" derives from the imputed income in-kind achieved by paying off a mortgage debt and living rent-free. While homeowners commit to a significant deposit and usually but not necessarily high relative repayment costs in early lifeafter all, housing property could be inherited or partially be paid for through inter vivos transfers, as it is most often the case in southern and eastern Europe as well as many Asian societies (Forrest and Hirayama 2014) -, wage increases and inflation over time imply significantly reduced housing costs at old age. A mortgage-free home may thus provide some alternative, or supplement, to a pension (Kemeny 1981). A second mechanism is downsizing, whereby the owner-occupied home is sold and the household moves into cheaper accommodation (or may even "doubleup" with other kin), thus releasing part or all of the capital value that was accrued over time. The third means of providing utility from the home is achieved through "equity release products" (e.g. Ronald, Kadi, and Lennartz 2015). In this case, the home-owning household can release some of the value held in the home via a financial product. This may be achieved by extending the mortgage or by the purchase of a lifetime mortgage product, allowing the occupiers to stay on in the home until death, with any outstanding debt balanced by the final sale of the property. It is primarily these productstogether with the downsizing mechanismthat would make the house a welfare resource at all phases of the life course and not only in old age.
Key to understanding the rise of property ownership and ABW in the Western world is the financialization of the banking industry and the adherent (political) liberalization of mortgage markets since the mid-1980s. Although there are different ways through which house purchase can be accomplished, taking a mortgage loan has become the main financing channel (Forrest and Hirayama 2014). Since the 1990s, the growing availability of easy credit offered households the possibility to take on large amounts of private debt with the promise that this debt would turn into a valuable, and highly tangible asset. Crouch (2009) argues that the rise of private mortgage debt became an economic model in itself. Where governments have largely withdrawn from the Keynesian model of using public investment to stimulate economic growth, often resulting in significant public debt, the primacy of balanced budgets in the Great Moderation period since the mid-1980s has meant that this obligation has been transferred to private consumers. Watson (2009) points to the crucial role of appreciating property values in this model. Where house prices rise fast and continuously, debt appears to be risk free and the appreciation of the initial investment seems certain. And this is where the financialization of the housing market becomes politicized. Whereas general price inflation is perceived to be counterproductive to economic growth in neoliberal economies, a rise in asset prices is seen as corresponding rise in their value (and thus, does not contribute to general inflation).
Consequently, governments have come to support the appreciation of property assets in political practice, be it through the deregulation of credit markets, mortgage guarantees, mortgage tax deductibility or, more indirectly, through the dismantling of affordable rental housing to limit tenure choice (Aalbers 2015;Forrest and Hirayama 2014;Lennartz 2011;Schwartz 2012).

The Homeownership Assets and Welfare State Change Trade-off
For a longer time, housing studies have sought to position homeownership and property assets in the welfare realm, alluding that it is one cornerstone of the new welfare state (Malpass 2008). The discussion actually started with Kemeny's hypothesis (1981) of a trade-off between owner-occupied residential property and the quantity and quality of public welfare provision, the argument being that smaller welfare states help sustain larger homeownership sectors, and vice versa. Kemeny posits that due to a temporal distribution of housing costs over the life course owneroccupiers face a front-loading of housing expenses through mortgage deposit requirements and high transaction costs at an early life stage. As a result, homeowners tend to have an aversion to higher taxes and consequently generous welfare states, with voters favouring conservative and market liberal parties. From a welfare state change perspective, this would then result in households becoming more inclined to invest in property and home ownership in a policy retrenchment environment (see also Ansell 2014). And conversely, individual asset accumulation may be utilized as a political vehicle for welfare state retrenchment, particularly since private assets may ameliorate the impact of state withdrawal on individual welfare conditions (Malpass 2008).
Much has been said about the logic and empirics behind the trade-off thesis, where various studies have shown thatif existent at allthe trade-off only takes place between homeownership rates and the structure of public pensions (Castles 1998;Delfani, De Deken, and Dewilde 2014;Schwartz 2012). However, we contend that the whole trade-off literature appears to be imprisoned by the notion that welfare state change may only take the retrenchment route. As a result, it has neglected the multidimensional character of welfare state transformation processes. Indeed, if one limits the directions of welfare state change to an expansion/retrenchment dichotomy, then seeing private property-based welfare and state-provided welfare as mutually exclusive categories seems inevitable. Yet, if one takes into account that welfare state change may imply the readjustment towards employment-oriented social investment, then the expansion of homeownership-based welfare would at least theoretically not necessarily imply a smaller, but merely a different welfare state. In the following, we will first depict empirically and then discuss theoretically the potential links and how they play out differently to what the "really big trade-off" literature has argued.
An Empirical Investigation of Welfare State Change -Where have Social Investment and ABW Materialized?
Though the theoretical conceptualizations of the two policy paradigms and how they would function as practical policy approaches have become salient in the literature, the question of where exactly and to what extent a reorientation towards SIS and/or ABW has actually taken place is less clear. Hence, in a first analytical step, we investigate OECD data on aggregate social spending on productive versus protective spending in 26 countries from 1995 to 2007. Here, we generally follow Hemerijck's (2013) distinction between social investment-oriented capacitating welfare spending and more protective, old risk-oriented compensating (see discussion above). 1 Before showing the empirical results, a nuanced discussion of potential shortcomings is at hand. First, selecting specific years for measuring change over time is not without problems. The year 1995 was selected as the starting point because it is the first year in which Eastern European countries are present in the data-set, which in turn would guarantee the largest possible variety in the sample. 2 However, taking single years or averages for short periods as starting and end points is exposed to the risk that changes in spending are not necessarily the outcome of explicit policy, but they might be artefacts of macroeconomic cycles instead. 3 Second, an isolated public spending analysis inevitably misses the multidimensional character of welfare state change through, for instance, regulatory modifications and is thus susceptible to the "dependent variable problem" (Green-Pedersen 2004). Ideally, a quantitative study on the transformation of the welfare state integrates spending data, with measures of welfare generosity, policy stratification and regulatory change, simply because cutbacks or expansion in expenditure are possible without changing the institutional setting and vice versa (Seeleib-Kaiser 2008). Moreover, Adema, Fron, and Ladaique (2011) have shown that public social expenditure is in itself a, to say the least, imperfect measure, since it does not incorporate welfare provision through taxation, mandatory private spending and voluntary private spending, resulting in a relatively skewed picture of welfare state size across countries. 4 Keeping these issues in mind, Table 1 suggests the following structural diversities and transformation processes across countries and welfare regimes in the given time period. Within the OECD as a whole, we can see that in relative terms (i.e. spending on productive social policies as a share of total social expenditure, excluding health care) a minor shift towards social investment spending had occurred between 1995 and 2007. Depicting the country-specific differences, there appears to be substantial divergence within and between different types of welfare regimes. Generally, most conservative continental countries, the Eastern and Central European new member states and particularly Southern European nations retained a prevalence of their traditional protective social policies. The main distinction here is that this had happened in the context of a broader (public spending) retrenchment or welfare protectionism process in the Continental and Central/Eastern European regimes, and within a welfare expansion environment in the Mediterranean countries (see also Häusermann 2012). By contrast, the already social investment-oriented Northern European nations, as well as the liberal welfare states, the Netherlands, Luxemburg and Spain, readjusted their social policies (further) towards the SIS (see columns 5 and 6 in Table 1). Interestingly enough, in the Northern European countries, this shift has occurred in a retrenchment context, meaning that they were spending relatively less on both productive and protective policies in 2007; nonetheless, within this context we can observe a relative shift towards education, care and family support spending, mostly so in Denmark and Sweden.
In a second step, we take a closer look at national differences and divergences in the accumulation of housing assets in the 26 OECD countries in the 1995-2007 period. Starting with a comparison of homeownership rates, both mortgaged and owned outright, across countries, we can see that (outright) ownership is more widespread in the Southern and Eastern European rim, as well as in Japan, with some "outliers" in the other regimes (e.g. and most notably Norway, Belgium and Ireland) (see  Table 2). Accordingly, one could argue that the highest potential for ABW through low (monthly) housing costs in old age exists in these contexts. However, high homeownership rates may be based on historical legacy and intergenerational transmission, such as in the Southern European countries, or historical coincidences, as is signified by the radical privatization policies after 1990 in Eastern Europe. And as such, aggregate homeownership rates tell us very little about other mechanisms of ABW and the role housing finance plays herein. It then seems more useful to compare countries on the way new and future homeowning generations access housing property in the first place and, subsequently, how they can access their accrued housing assets when needed. Here, aggregate outstanding mortgage debt as a percentage of GDP has become popular indicator in internationally comparative research (e.g. Fernandez and Aalbers 2016; Schwartz and Seabrooke 2008): certainly, it should merely be seen as a proxy for the potential asset wealth of citizens in the future. And moreover, it is also exposed to developments in economic policy, interest rates, house prices, housing demand, fiscal policy and a host of other interconnected influences. Nonetheless, it clearly captures governments' roles in simplifying access to mortgages as well as the disposal of housing assets through credit market deregulation policies (see Schwartz 2012).
Table 2 then illustrates that although aggregate mortgage debt had increased substantially across most countries (except for Germany, Finland and Japan) between 1995 and 2007, we can see that the rise in indebted private households sector could be found in most of Northern Europe (including the Netherlands) and all Englishspeaking countries. Indeed, increases in mortgage debt as measured by percentage point change are particularly high in Ireland, the Netherlands, Australia and Denmark, all of which also experienced rapid increases of homeownership rates in the same period (Ronald and Doling 2012). Meanwhile, although some of the largest increases in debt levels can also be found in Spain and Greece, a pattern emerges in which most Continental, Eastern and Southern European countries have comparably low debt levels in 2007. In short, homeownership has had a long-standing function as a crucial welfare resource for individuals and families alike in many (if not in most) advanced capitalist economies. However, focusing on the issue of how financialized (property) asset markets have become implanted in the wider production of societal wealth and economic growth, we can see that the liberal English-speaking nations as well as the Northern European countries had become forerunners in the potential application of a ABW model before 2007 because they made it relatively easy for future owner to gain initial access to the sector and release housing equity (see column 5 in Table 2) later on.
Taking the analyses in the previous sections, one step further and bringing the separate results together, we can say that there seems to be a strong overlap between the countries were the debt-driven property access model (but not homeownership per se) had been pursued most explicitly between 1995 and 2007 and those countries where the SIS had featured most prominently in the same period. From a geographical viewpoint, the scatterplot in Figure 1 indicates a strong positive association between the two variables, meaning that nations with a relatively high share of social investment-oriented spending in 2007 also had relatively high amounts of private outstanding mortgage debt in the same year (Pearson's correlation coefficient is 0.69, with a statistical significance of p < 0.01). This pattern includes all liberal regimes, the Northern European nations (except for Finland, but including the Netherlands as the country with the highest mortgage debt to GDP level). Conversely, we can see that the Continental regime, as well as the Eastern European new member states and some Southern European countries are oriented towards welfare protectionism in their broader social policy models, which is associated with a more traditional, less financialized approach to homeownership access. Portugal and Spain, both countries with relatively high levels of mortgage debt and low levels of capacitating spending, and Estonia, which shows the inverse pattern, are the only major exceptions within the sample of 26 OECD countries. A similar, albeit slightly weaker positive relationship between the two variables exists if we consider them in a dynamic perspective.
Countries that have readjusted their spending capacities towards SIS in the period 1995-2007 the most, also tended to have the strongest increases in private mortgage debt in the same period (Pearson's correlation coefficient = 0.48 at p < 0.01 level; scatterplot cannot be shown here due to space limitation but are available on request).

Linking Social Investment and Debt-driven ABW -Two Compatible Emerging Social Policy Paradigms
Although the empirical links between SIS and debt-driven ABW appear to be quite clear, the question remains if we are not just dealing with spurious empirical residue and two detached developments that happened to occur at about the same time. Or is there actually a political and ideational link between the financialization of housing property market and emerging welfare state transformations across welfare regimes? A useful starting point for better understanding the relationship between the SIS and the debt-driven ABW approach is to ask which types of social risks the two paradigms are dealing with. Although they both have emerged from the same ruptures in the labour markets and shifts in the socio-demographic structures experienced by all advanced economies, they should be understood as reactions to two different aspects of these developments. Whereas the social investment paradigm is primarily addressing the emergence of new social risks, ABWand particularly the idea of bringing new households into the homeownership sector through early life debt accumulationis basically a response to the (budgetary) strains when keeping old risk social policies in place. Simply said, this means that they are not competing policy logics per se, but they may exist alongside each other, with the relative balance between the two reflecting, or even reaffirming, socio-historic contingencies, welfare regime legacies and path-dependence. Seemingly, the distinct policy shifts towards SIS and ABW might be considered as two sides of the same coin, where some governments may have used citizens' increased ability and willingness to invest into private assets (mainly housing property) in order to redirect public resources to other welfare dimensions. Or in a more radical reasoning, one might even claim that the establishment of a welfare system in which income protection is to an increasing extent achieved through the accumulation of private debt that eventually amortizes into asset wealth, is one necessary condition to further develop the SIS. Taken together, and if realized in policy practice, the two paradigms could then be seen as the cornerstones of a more productive-oriented and highly financialized type of welfare capitalism.
In the following, we want to discuss some possible explanations how debt-driven homeownership access and relatively high social investment spending are related, leading to our final argument that that they may be seen as compatible, yet not complementary social policy ideas. First of all, there is a political rationale for the compatibility thesis, which mainly comes down to the question of why governments do what they do. Political parties are first and foremost interested in electoral success. Since the late 1980s, an increasing success factor for being re-elected has become the combination of robust economic growth with the ideal of balanced government budgets (Seeleib-Kaiser 2008). This does, however, not mean that high government spending and social policy programs have become deal breakers per se; quite the opposite is true, as public welfare provision is still highly popular in most Western democracies (Vis, Van Kersbergen, and Hylands 2011). In trying to overcome this dilemma, governments in Northern Europe (again, this includes the Netherlands and excludes Finland) and in the liberal welfare states were keen to use a political strategy of "affordable credit claiming" (Bonoli 2012). Combining policies that would allow citizens to build long-term assets to secure oneself against lower income across the life course with social investment policies that come at relatively small costs and could be sold as some kind of replacement for retrenching income protection-oriented old risk policies, policy-makers, particularly the authors of New Labour and European Social Democracy in the late twentieth century (see Ryner 2003), had seemingly found the holy grail of election-winning strategies by linking balanced budgets, with retained social spending and rapid economic growth (see also Hay 2013).
Our second argument for the compatibility thesis revolves around the structure and functionality of labour markets in the productive, investment-oriented model of welfare capitalism. Here, the tertiarization of national economies has led to a meaningful rise in female labour participation and the demise of the male breadwinner model. Dual income couples and families have attained a stronger market position and can cope with higher levels of private debt: i.e. they are better able to satisfy recurring mortgage repayments and to afford relatively high house prices. Conversely, this means that those who are burdened by high levels of mortgage debt need to maximize their employment efforts permanently. Consequently, in order to sustain their dual income position, the logical behaviour by younger to middle-aged couples and families would be to support political parties that ensure public investments in care facilities for children and elderly alike. However, a side note here is that the upsurge of the flexicurity model that underlies the social investment approach has lead to more limited mortgage borrowing capacities of younger adults, making their access to property assets more difficult and often more dependent on intra-family support.
The final argument is a more conceptual reasoning and refers to the potential weaknesses of the SIS and ABW paradigms when translated into a coherent set of social policies. As they both take a life course perspective on the manifestation of social problems and their solution, poverty, social exclusion and inequality are primarily defined as individual rather than collective problems. Personal responsibility, risk-taking behaviour, long-term planning and treating life itself as a series of investment decisions (Langley 2008) have become key ingredients in securing one's own welfare position. Social policy in this framework is reduced to the function of adapting people to the needs of the market instead of diminishing their dependency on it. This arguably reflects the neoliberal foundation of both the SIS and the ABW model. As a result of this embedded market logic, both policy approaches share the same fundamental limitations in regard to guaranteeing welfare security. Firstly, employment-and human capital-oriented spending, as well as policies that support homeownership tend to cater for middle and higher income groups rather than the lower classes. Ample studies have illustrated how opportunities to gain access to high quality education and to accumulate housing wealth are distributed unevenly across social classes (e.g. Dewilde and Lancee 2013;Vandenbroucke and Vleminckx 2011). Hence, the two policy approaches tend to exclude the unproductive and do not offer a solution to those who find themselves unable to undertake the necessary initial investments in education, vocational training or property assets for future welfare security. Consequently, this may result in the perpetuation of the dualization of insiders and outsiders, which might generate adverse distributional outcomes. Given the strong influence of family background on initial opportunities to invest in education and property assets, both models, henceforth, appear to reinforce rather than limit the intergenerational transmission of poverty risks and life chances (see also Cantillon and Van Lancker 2013;Nolan 2013;and Köppe (2015) for ABW in particular).
Furthermore, both the SIS and ABW models rely on an inherent belief that investments will pay off in the future. However, the aftermath of Global Financial Crisis in 2008 build a clear case against continuous house price appreciation, while persistently high unemployment rates in Europe in the post-crisis period illustrate that human capital investments (quite literally) may not always be put to work. Hence, returns on human capital and property investments are conditional on time and space, or to put it more bluntly, having a job and a good education, as well as owning a houseparticularly when it is debt financed (see Lowe, Searle, and Smith 2012increases the likelihood but is not always enough to satisfy all welfare needs (Elsinga and Hoekstra 2015;Toussaint and Elsinga 2009). Hence, SIS and ABW may be compatible policy strategies, but they should not be considered as complementing policy models, simply because they face very similar limitations in providing a solution to universal welfare needs.

Conclusions
Welfare states in Western societies are not on the retreat but they have taken different forms through various transformation processes in the past two decades. Some policy dimensionsmost notably public pension systems and labour protectionwere targeted by retrenchment and deregulation, while other dimensions have been expanded, including family policies and the provision of care services. Numerous social policy researchers have identified that the welfare state is shifting from its traditional function of income protection towards a more productive social investment statea policy paradigm that primarily aims at employment maximization through the support individual skill formation and the reconciliation of work and family life. Although this development has been widespread, it is far from uniform across countries and welfare regime types. Indeed, the analysis of aggregate data on public social expenditure from 1995 to 2007 in 26 OECD countries revealed that the reorientation of policy priorities towards capacitating social investment spending had been most pronounced in Northern Europe (excluding Finland but including the Netherlands) and the liberal welfare states; in contrast, the conservative, Mediterranean and New EU Member welfare states have retained or even expanded their predominantly protectivist social policy approaches.
The key argument put forward in this paper then was that divergence in welfare policy priorities across different types of welfare states is strongly influenced by the extent to which property assets have become an essential private welfare resource in a nation's political economy. In many countries, policy-makers have implemented an ABW approach in that they enabled and encouraged all types of households to use mortgage debt as a means to invest in residential property assets, which can be tapped into in times of need or when facing lower income in general. The fact that this development has been most pronounced in English-speaking contexts and Northern Europe suggests that the two policy developments are interlinked in various ways. Our contention is that through the implementation of an ABW approachwhich, in effect, is a homeownership-based welfare approachgovernments in these countries were seemingly able to channel freed-up public funds to policy dimensions which they perceived to be more eminent threats to welfare security. Stimulating the accumulation of housing property wealth could thus be seen as an important component, if not necessary condition for realigning the welfare state towards a SIS. To conclude, ABW and social investment appear to be compatible social policy concepts, which intertwine at the political, labour market, as well as ideological level, and together advance the establishment of a more productive, investment-oriented form of welfare capitalism. However, they are not complementary policy strategies, mainly because they face very similar limitations as potential welfare resources and are geared towards a limited social stratum, i.e. those who can access high-quality education, jobs, and housing initially.
We would like to end this paper by discussing two potential limitations to this notion and the empirical findings that it relies on. On the one hand, this paper seems to suggest that liberal and social democratic welfare states have converged to become similar if not identical regimes. But, of course, the Northern European countries have the most redistributive and protective welfare states in all advanced economies. In the same vein, capacitating social policies such as universal childcare and gender-equal employment policies have a long history in the Nordic countries (Esping-Andersen 2002), and have actually existed well before the turn towards financialized housing markets occurred in the early 1990sin contrast to the light and late social investment approach in the English speaking world (see also Deeming and Smyth 2015). Conversely, in contrast to its Anglo-Saxon counterpart, the rhetoric around ABW has not entered the political discourse explicitly, meaning that the turn towards higher homeownership rates has merely been discussed as a salient change within the housing system rather than the formation of a new welfare paradigm. What this implies is that the links between social investment and housing property wealth might function differently in the two country clusters. Where the liberal countries stage a much clearer case for our argument of ABW as a source of welfare state restructuring towards a more productive social policy model, rising property wealth and the political support thereof in the Nordics should, maybe, be understood as a case of welfare state resilience. Where maintaining both protective and productive welfare policies at very high levels has become too expensive, the implementation of private property accumulation as a personal welfare safety net might be used to keep social investment policies alive. However, we are very much aware that only a more detailed analysis of the differences between the two country clusters can shed light on these issues.
In a similar vein, the fact that we have discussed the links between SIS and ABW mainly in the context of Liberal and Social Democratic welfare regimes, does not mean that the links between homeownership and welfare state restructuring are non-existent in other settings per se. Certainly, the more protective form of welfare capitalism had been fortified in Continental Europe, Southern Europe (excluding Spain) and Japan in the 1990s and early 2000s. Here, social policy still primarily addressed old social risks and aimed to preserve a more familialistic, male breadwinner oriented approach, which in fact made the financialization of housing markets appear to be less pressing and imminent issue. Nonetheless, housing market commodification and welfare state change are imminent issues here as well, requiring more detailed analysis and case by case examinations. A good starting point for grappling with all of these issues would be deeper analysis of specific country cases, possibly contrasting the experiences in liberal and Nordic countries. A particular concern could then be an investigation of the role of the agents of Social Democracy in bringing about a convergence towards a more productive form of welfare capitalism.
On the other hand, the argument presented here might be challenged by the specific time frame chosen for the empirical analysis. One might argue that it is only relevant as a pre-crisis evaluation of welfare state changewhich would not be a problem per se, but would give the paper more of a historical sensebecause postrecession policies in Europe since 2010 have been more strongly geared towards welfare state retrenchment and austerity (Lennartz, Arundel, and Ronald 2015). As a result, one might expect a complete retrenchment-guided turn towards private welfare settlements in which property assets take on an ever-more prominent role, while collective social investment spending is curtailed effectively. However, as Van Kersbergen, Vis, and Hemerijck (2014) argue, "retrenchment is not the only game left in town", making it worthwhile to further examine how exactly mature welfare states have developed in the post-crisis period, and whether we might actually be able to observe the perpetuation of the dichotomous welfare state transformations described here.

1.
Health care spending is difficult to place in this dichotomy. While it could be seen as a desirable social investment since at least parts of it are prevention measures and occupational health care that augment labour market attachment, it is evenly about the protection of people from income drops in times of sickness and/or have little to do with labour markets as such. Whereas Nikolai (2012) has proposed to subsume it under traditional compensating spending, this paper follows Hemerijck's (2013) approach to omit health care from the analysis of welfare state change, as we cannot clearly assign it to one function or the other. 2.
All OECD countries for which data on social spending and mortgage debt (see below) was available in 1995 and 2007 were included in the empirical analysis -Iceland was the only exception here since it was excluded due to being a major outlier in the data. 3.
For one thing, 1995 marked the end of the Nordic financial crises as well as the early 1990s European recession, implying cyclically high unemployment rates and unemployment benefit spending. By the year 2000, and for that matter also 2007 unemployment rates had fallen significantly, which had caused a default decline in compensatory spending as people moved back into employment and stopped taking early retirement. Hence, to make our analysis more robust, we thus investigated changes in spending patterns for different periods (e.g. 2000-2007; 1995-2009; 1990-2007 for a smaller number of countries; average 1995-1997 to average 2005-2007). The results cannot be reported here; however, no significant differences to the 1995-2007 period were found. The analysis can be obtained from the authors on request. 4.
Our analysis would ideally have looked at net public spending for each variable of productive and protective social expenditure. However, net public spending data are only published for total public spending and retirement spending. Again, to make our approach more robust, we tested different measures as far as they were available for all different policy domains, in all countries, for the whole period. One example here is that we replaced public expenditure as % of GDP with public expenditure on, inter alia, unemployment benefits as % of total social expenditure. The differences between the results for these approaches were very similar and statistically insignificant (data can be requested from the authors).