Housing, the welfare state and poverty: On the financialization of housing and the dependent variable problem

ABSTRACT The comparative study of housing systems enjoys a “disjointed” relationship with comparative literature on the welfare state and poverty, a separation which has been to the detriment of each of these fields. Such are its implications for the income distribution that the concept of “financializaton” developed in housing studies provides potentially rich raw material for welfare state and poverty literature. However, there are also limitations in the way in which “financializaton” has been operationalized in housing studies: definitions are often inconsistent, and it has proven challenging to identify indicators that measure convincingly the multiple dimensions of financialization. We argue that greater consideration of the “dependent variable problem”, extensively discussed in the welfare state literature, but generally neglected in housing studies, can advance understanding of housing market financialization and its effects. A greater integration of scholarship on the welfare state, housing studies and poverty would be to mutual benefit.


Introduction -Housing, Poverty and Moving beyond Disjointed Literatures
Housing systems in European countries and beyond are, according to many accounts, undergoing significant change, relating to the processes of commodification and financialization.Accounts of these processes take a variety of forms, but typically converge on rising house prices, leading to falling rates of homeownership and a "return" of the private rented sector.These processes are, it is claimed, leading to a deterioration in housing outcomes, most obviously in relation to housing affordability.
This initial observation raises some important questions.Are housing systems really changing?How do we know change when we see it?If housing systems are changing, what does this look like?Is change occurring in many countries and is it largely of the same type, or is it idiosyncratic, with different dynamics in different places?What are the consequences we should expect to observe when housing systems change and do these, in fact, materialize?These questions are ripe for examination in comparative studies and are, we argue, of relevance for scholars of the welfare state and poverty as well as those of housing studies.
This paper is part of an interdisciplinary project examining the links between housing and poverty (and welfare states more broadly), which seeks to move beyond the "disjointed literatures" of housing, welfare states and poverty (Hick, Pomati and Stephens, 2022a).This paper, and our wider work, falls within the intellectual project called for by Kemeny 1992Kemeny 13, 2005) ) in relation to the need to better-integrate housing studies within wider disciplines.
There has been something of a "turn towards housing" in the literature on political economy in recent years.A recent re-specification of Varieties of Capitalism focussing on "growth regimes" emphasizes the importance of housing market financialization as a critical variable in national growth strategies, with import-and export-oriented regimes differing in the extent to which this is treated as an "engine of growth" (Hassel and Palier 2021; Reisenbichler 2021).Political scientists, too, have started to turn their attention to the potential "ballot box" consequences of rising house prices (e.g.Ansell and Cansunar 2021).The time is right, we believe, to bring housing back into the study of the welfare state and poverty.
The comparative welfare state literature is itself wide-ranging with contributions from sociologists, social policy scholars and economists examining micro-level connections between welfare state institutions and social outcomes such as poverty and political economists and politics scholars examining macro-level determinants of welfare state differences and their relationship with political processes (Hemerijck and Plavgo 2021).While we recognize the importance of the latter, it is the former that is our interest here: that is, we are concerned with the welfarist outcomes that are anticipated to follow from particular sets of housing and welfare state arrangements and our hope is that, by better understanding these processes, in a form that is compatible with comparative study, we are able to identify the housing systems that best enable "a decent home for all at a price within their means" (Hills 2017: 1).
We start by offering a brief conceptual model of the relationship between welfare regimes, housing systems and poverty at the micro-level.The interaction of labour markets, tax systems and social protection systems has a significant bearing on rates of poverty and inequality.In Esping-Andersen's classic formulation, countries could be clustered in three welfare "regimes" that reflected the qualitatively different organizing principles that historical actors had struggled for and which were embedded in policy (Esping-Andersen 1990b: 118).Significant for the argument we present here, these welfare regime clusters are predictive of the degree of redistribution between market and disposable incomes and, thus, of poverty rates (e.g.van Kersbergen and Vis, 2014: 92).In these models poverty is often measured before housing costs, treating these costs as any other item of consumption.There are important limitations to this practice: these costs are often the first and largest item of a household's budget; they often vary significantly over place and time in ways that do not relate to quality, and they may be difficult to curtail in the short run in the event of a reduction or loss of income.Here housing institutions come in, mediating the relationship between disposable household incomes, shelter needs and post-housing cost income.At the risk of oversimplification, these institutions can reflect or moderate inequalities in disposable household income and can influence both the extent to which shelter needs are met and differences in poverty, when this is measured in ways that are sensitive to these costs -namely, when poverty is measured after housing costs (AHC) or, potentially, using a suite of material deprivation indicators or via subjective measures of poverty.While housing has long been neglected by scholars of the welfare state and by poverty researchers, housing institutions form an essential link between wider welfare state institutions and living standards.
The aim of this paper is to move beyond these disjointed literatures by considering the possibilities and -at present -the limits of scholarship on commodification and financialization for that of the welfare state and poverty.Our paper aims to give substance to the greater integration between these fields that we believe is possible, with a particular focus on the kinds of theory that can inform comparative, empirical analysis.But is disjuncture between academic fields necessarily problematic?Are there not gains to be made from specialization?Our claim is not that specialization is inherently problematic, but rather that when either advances in knowledge arising from specialization have potential value in fields adjacent to those they were created in or when new realities in politics, institutions or outcomes imply newfound empirical entanglements that are of relevance to multiple theoretical or substantive traditions, then the gains to be realized from interdisciplinary integration are, potentially, at least, high.In substantive terms, the growth in house prices in the period since the Global Financial Crisis has led to newfound interest in, and concern with, developments in relation to housing, including amongst scholars of the welfare state and poverty.The threat that rising house prices presents to living standards makes this an opportune moment for greater integration between these disjointed academic fields (see also Hick, Pomati and Stephens, 2022a).
Our core argument is that accounts of housing system change provide potentially significant raw material for scholars of the welfare state and poverty to make sense of developments in living standards, while these traditions offer lessons to housing studies scholars in how to render accounts of commodification and financialization operational (and thus some of the claims made in relation to them testable) in a comparative framework.To move towards a comparative approach in relation to the commodification and financialization of housing, we need greater clarity in relation to conceptualization and measurement, and about the outcomes we expect to follow from these processes.
The paper is structured as follows.In the next section we discuss the early comparative literature on welfare regimes and housing systems, which proved to be a rich seam that influenced a wide-ranging theoretical and empirical body of literature; in both fields, later theoretical accounts would work outside or beyond these typologies, though we show how this played out differently in the two literatures.The second and third sections contain the core arguments of the paper.We show that concepts of commodification and financialization remain the subject of much disagreement but refer to processes that potentially result in new forms of stratification and that they therefore should be of interest to scholars of the welfare state and poverty as well as housing studies scholars.In an attempt to bring out the relevant components, we seek to trace what the outcomes of the processes of commodification and, especially, financialization might be.The third section explores measurement problems, which relates to the key concepts themselves and the lack of agreement on the housing outcomes associated with them, while the conclusion summarizes the core argument.

Welfare Regimes and Housing Systems and the Possibilities of Comparative Analysis
Theoretical engagement between housing studies and scholarship on welfare states and poverty is most extensive and explicit in relation to the literatures on welfare regimes and housing systems (for recent discussions, see Lowe 2011;Blackwell and Kohl 2019;Stephens 2020a).Attempts to cluster countries by differences in their welfare or housing systems have produced a rich body of literature that is comparative, empirical and has opened-up important avenues of theoretical enquiry.In The Three Worlds of Welfare Capitalism, Esping-Andersen's (1990) classification of 18 nations into Liberal, Corporatist and Social Democratic welfare-state regimes on the basis of their differences in decommodification and stratification have proven to have enduring influence in the comparative literature on welfare states.While Esping-Andersen's primary focus was on institutional differences and their causes, subsequent empirical studies found the three clusters to successfully predict cross-national differences in poverty rates (e.g.Fouarge and Layte 2005), confirming the significance of institutional differences in welfarist terms.Despite understanding his project as entailing deepening the theorization of the welfare state, however, Esping-Andersen's empirical data was limited to differences in social protection programmes, which led Kasza (2002) to question the validity of the clusters on the grounds that welfare systems comprise multiple programmes, including housing, health and education, which developed at different times and, depending on the constellation of actors in various fields, in different ways.Jim Kemeny's account of housing system differences was based on comparison of eight countries and he argued that the central differences between housing systems originated from the nature of their rental markets -between what he labelled dual and unitary rental systems.In the former, the government's attempt to promote the profitorientation of the rental market leads them to adopt a minimalist cost rental sector reserved for those unable to meet their needs in the market (Kemeny 1995: 16).But since the private market is unlikely to be able to meet the housing needs of ordinary households satisfactorily, this type of rental system shapes preferences in favour of owner-occupation.The result is high levels of owner-occupation and a "dualist" rental system, with clear divisions between market and cost rental sectors.In the latter, by contrast, the government invests in and provides extensive cost rental provision.Once the initial investments in this provision mature and generate surpluses, government subsidies and regulations can be removed and the cost rental sector can compete with the private rented sector and owner-occupation, leading to a unitary rental system (Kemeny 1995: 16-19).Importantly, Kemeny is clear that these different types of system will have predictable consequences in terms of housing outcomes: extensive cost rental housing is understood as having the potential to lower rents charged in the market rental sector and to set a floor on housing standards (e.g.Kemeny 1995: 132, 18).
Two analytic challenges that have emerged in relation to these typologies are significant here.First, a key challenge in both the comparative welfare state tradition and in housing studies is how to extend the typologies to include countries beyond those included in the original studies.An early challenge came in terms of including countries in Southern Europe, where it was claimed that these nations represented a separate grouping in both the worlds of welfare (Ferrera 1996) and housing literatures (Allan et al. 2004), in both cases stressing the role of the family as a central welfare actor.A greater dilemma came with EU enlargement to Central and Eastern Europe (CEE) from 2004.Within the field of housing studies, the largescale transfer of formerly state-owned housing into private hands after the fall of Communism has been argued to have led to "housing welfare regimes by default" (Stephens, Lux, and Sunega 2015), with very high rates of outright homeownership.In the comparative welfare state literature Hemerijck (2013: 160) includes the nations of CEE in one welfare cluster noting that they are distinguished from other regimes by the shared experience of major discontinuitiesnamely, the common experience of the shift from capitalism to state-socialism in the 1940s followed by the reversion back to capitalism from 1989.Occasionally, in both housing studies and the welfare state literature, the Central and Eastern European nations are sub-divided into multiple regimes (e.g.Nolan and Whelan 2011: 104-105); however, it is not clear that these would meet the criteria of having distinctive policy logics that was the focus of Esping-Andersen in making the case for welfare regimes as a key organizing concept, yet neither does their amalgamation into one regime seem entirely satisfactory.
Second, both sets of typologies faced challenges posed by subsequent change though these challenges have played out differently in the welfare regime and housing systems literatures.Paul Pierson's (1994) "new politics" thesis claimed that attempts to roll back the welfare state were not simply a mirror image of the dynamics of welfare state expansion and offered a distinct theoretical account and new conceptual language to understand the mechanisms by which austerity might be achieved or (in many cases) frustrated.While attention shifted to new dynamics underpinning welfare state change, it has become commonplace to examine these different pressures through the prism of how they play out in different welfare regimes (for example, Pierson 2001).This has proved possible because the differences between welfare regimes had largely been accepted.By contrast, Kemeny's rental systems have faced greater challenge from subsequent political and economic developments.In observing changes in Sweden since the development of these typologies, Stephens (2020a: 533) notes that "the relationship between the cost and for-profit rental sector is no longer the key dynamic in the Swedish housing system, and in this crucial sense the theory has lost its explanatory power".Many housing scholars have turned their attention to the processes of commodification and financialization, and these are typically examined independently rather than, as in the welfare state literature, through the prism of the older typologies.
Thus, the clustering of countries into welfare regimes in the comparative welfare state literature continues, though the policy logics of the different regimes are arguably given less weight than is evident in Esping-Andersen's original work.The limitations of the regime-based approach in explaining change is well-recognized, but they are nonetheless often used to "frame" other developments and pressures on the welfare state -for example, to explore how social policy responses to COVID-19 vary by welfare regime (Béland et al. 2021).This, in turn, requires exogenous (or, supplementary) accounts of how welfare state institutions are changing.By contrast, Kemeny's housing systems have come under greater challenge, because differences between rental systems do not reflect the housing systems of a wider group of nations (e.g. the nations of Central and Eastern Europe) and because the focus on rental systems as being the core dynamic explaining contemporary housing systems in the countries in the original typology has been questioned (Stephens 2020a).This suggests that, of the available typologies, clustering countries by welfare regimes may prove more fruitful for empirical researchers working with large numbers of cases than attempting to classify nations by housing systems (see Hick, Pomati and Stephens, 2022a), though the practice of relying on welfare regime classifications to study differences in housing is not uncontroversial (see Stephens and Hick, forthcoming for a discussion).Within the field of housing studies, there has been a substantial growth in scholarship on the financialization of housing, which is argued to be a newer process shaping housing institutions.While some accounts of this process proceed via the typological approach (notably, Schwartz and Seabrooke 2008), most do not.Significantly, this process has the potential to be of relevance not only for housing scholars but also for scholars of the welfare state and of poverty.It is this to which we now turn.

Financialization, Commodification and Social Outcomes
In departing from the relationship between cost rental and for-profit housing as being the central dynamic explaining developments in housing systems, contemporary accounts often focus attention on the related processes of commodification and financialization.These accounts are arguably of interest to scholars of poverty because, while they vary in form, many emphasize the centrality of housing in new forms of stratification as these processes lead to a deterioration in housing affordability.In this section we discuss how commodification and financialization have been understood and why they are potentially significant in distributional terms.
The concept of commodification bears at least two distinct, but related, interpretations.The first focuses on circumstances where housing is increasingly treated as a commodity.Madden and Marcuse (2016: 17) suggest that commodification: "is the name for the general process by which the economic value of a thing comes to dominate its other uses . . .The commodification of housing means that a structure's function as real estate takes precedence over its usefulness as a place to live".Commodification has in this sense been associated with an emphasis on homeownership, state retrenchment from housing provision, embedding market mechanisms in determining rent levels, and a rebalancing from a focus on use to exchange values (Drudy and Punch 2002;Rogers, Nelson, and Wong 2018, inter alia).These understandings span objective developments in tenure balances, the extensiveness of state provision and regulation of housing as well as shifting subjective meanings given to homeownership by governments and people alike -for instance, the aspiration to homeownership as a means of achieving financial security in later life.At the macro-level, the emphasis on homeownership was said to lead to a "really big trade-off", where the need of newly-established households to secure homeownership would undermine support for more encompassing and expensive welfare states (Kemeny 2001).
For quantitative comparative studies, a measure of commodification based on the homeownership rate (Schwartz and Seabrooke 2008) or the extensiveness of the below market rent sector (as a measure of decommodification; see Angel 2021) is common, though it captures only some of these elements.This conceptualizaton of commodification has, however, been argued to struggle to deal satisfactorily with outright ownership which predominates in parts of Eastern Europe and forms part of national housing systems everywhere (see Hick, Pomati and Stephens, 2022a).
A second interpretation is closer to Esping-Andersen's concept of decommodificationi.e. the extent to which the housing system weakens the link between labour market incomes and housing consumption.These two perspectives have parallels -for instance, socially rented housing would be considered decommodifying on both interpretations, but outright ownership would be counted as commodified in relation to the first interpretation but not the second.Moreover, they are complicated by their potential interaction through time, as we observe when distinguishing between consumption and exchange.In this second interpretation, outright ownership may be decommodifying in terms of consumption, even as it can be commodified in terms of exchange -for example, if a person downsized, or a dwelling was sold after their death (see also Stephens, 2020b).
The de-commodification of housing might be expected to be associated with a clear set of outcomes in that state provision of social housing de-commodifies housing especially, though not exclusively, for families on low incomes.We might expect that more highly commodified systems will have higher housing cost burdens for lower-income families and, for a given income distribution pre-housing costs, higher after housing costs (AHC) poverty rates.In more highly commodified systems, we may also expect housing outcomes to be more highly stratified, with housing inequality mirroring the inequality of personal incomes to a greater extent than in systems with greater decommodification (Zhou and Logan 1996: 402).To that extent, while neglected, commodification is clearly significant for welfare state scholars seeking to make sense of policy reforms and the evolution of poverty and living standards in Europe.
While commodification is understood as a process linked with the wider rollback of the welfare state since the late 1970s, the financialization of housing has received growing attention as a process explaining changes within housing systems, especially since the turn of the century.In earlier accounts of financialization, developments in mortgage finance and homeownership were emphasized.A significant development was the deregulation of mortgage finance in the 1980s in the USA, UK, Australia, Nordic nations and others.Indeed, one interpretation of the financialization of housing is that its essence relates to the liberalization of credit and that subsequent developments are variations on this theme.One consequence of the liberalization of credit markets is that it has made housing wealth more liquid, through for instance the use of equity withdrawal and release products, which strengthen the link between housing wealth and current income.
From the turn of the century, the huge growth in mortgage securitization was seen as a significant development (e.g.Schwartz and Seabrooke 2008), facilitating an extension of homeownership, but also contributing to house price inflation (Aalbers 2017).The fundamentally unsound nature of many sub-prime mortgages extended in the early 2000s in the United States, facilitated by securitization, was the trigger that led to the Global Financial Crisis (Roubini and Mihm 2011).In the period since the Global Financial Crisis, central banks have slashed interest rates and have engaged in largescale quantitative easing, and there have been concerns that these policies have served to inflate asset prices further (e.g.Allen 2015).
In a second, more recent phase of literature on housing market financialization, developments in the private rented sector are also emphasized (e.g.Byrne and Norris 2019).In the UK, the introduction of Buy-to-Let mortgages facilitated a rise in petty landlordism which spurred a revival of the private rented sector, with most landlords owning only one property (Kemp 2015;Ronald, Kadi, and Lennartz 2015).Other accounts emphasize the emergence of new actors -namely, the growth in foreign institutional investors (Hearne 2020;Aalbers 2017) who have entered the buoyant property market in a "search for yield" as ultra-low interest rates make rental returns attractive (Kemp 2015: 616).While the welfare and housing systems approaches stressed divergence (Kemeny) or difference (Esping-Andersen), Stephens (2020a: 523) describes the work of Aalbers as presenting financialization as "a form of 'soft' convergence" in housing systemsa common pressure which is mediated through distinctive national institutional structures resulting in diverse consequences (see, for example, Wijburg and Aalbers 2017 on foreign institutional investment in the German rental sector).
Even this brief overview of what is a burgeoning literature shows that that multiple aspects of this process are emphasized.Mader, Mertens, and Zwan (2020: 2, 7) suggest that these multiple accounts "are not mutually exclusive, and only together they give the whole picture", pointing to the risk of "solidification", where overly rigid definitions of this process may lead to neglect of its evolving and emerging forms.This risk may be real but if financialization is a process that can be examined empirically and if we are to evaluate whether, like previous frameworks, it has predictive power, then clear definitions are needed.
While the causal pathways differ in these accounts, there is greater agreement that financialization leads to an increase in asset prices -for instance, by extending homeownership to sub-prime borrowers or by increasing multiple homeownership, thus increasing demand.Understood in this way, financialization appears as an unstable mechanism of promoting prosperity (Streeck 2017), yet is, in this sense, ambivalent: it may enable some people to purchase homes who otherwise might not be able to, but at the price of an increase in both individual and systemic risk.This makes the potential outcomes that might arise from financialization difficult to specify with precision, not least since they may change over time.
Moreover, a recent theme in the financialization of housing literature concerns the extent to which this is "variegated", which reflects the fact that this is not a broad one-size -fits-all process.One way that the idea of variegation has been used is to refer to the variable penetration of housing market financialization across nations.Working within the Varieties of Capitalism tradition, a recent specification of European "growth regimes" brings the financialization of housing to the political centre-stage (Hassel and Palier 2021).This account is founded on the core difference between export-oriented and import-oriented "growth regimes", noting that financialized housing markets are restrained in the former, such as in Germany, due to concerns over costcompetitiveness within the core export sectors, whereas they are encouraged in demandled regimes, such as the UK, where housing demand is more directly central to national growth strategies (Reisenbichler 2021: 325).Variegation, understood in this way, does not pose obstacles to the kind of comparative analysis we are concerned with here (that is, concerned with social outcomes).Aalbers (2017) offers a second meaning of variegation in relation to the financialization of housing -namely, its potential to involve different constellations of actors in different places -this raises potentially more significant issues, as we discuss in Section 3. A third interpretation might be that the impacts of financialization are variegated depending on the underlying structure of housing institutions.This is plausible -likely, even -but this is, fundamentally, an empirical question, and requires empirical studies to examine how impacts of housing market financialization are variable in different places.
Tracing the welfare and housing consequences of financialization In their review of the literature on financialization across multiple disciplines and involving multiple processes, Mader, Mertens, and Zwan (2020) argue "financialization scholars generally agree that the[se] processes . . .are unstable and have harmful distributional consequences".The consequences of housing market financialization are likely to be significant not only for housing studies scholars, but for scholars of the welfare state and poverty, too.
Tranøy, Stamsø, and Hjertaker (2020) for instance express concerns about financialization leading to "increased stratification and re-familialisation", as rising property prices prompt a greater reliance amongst young people on the "Bank of Mum and Dad" to secure homeownership.This, they suggest, generates new cleavages between "winners" and "losers" within generations based on access to inheritance, threatening commitments to social citizenship, even in the more egalitarian social democratic nations such as Sweden, Denmark and Norway.There may be changes between generations, too, with later homeownership for those who do become homeowners, the emergence of "Generation Rent" for whom homeownership is, at least, in question, and the prospect of a growth in numbers of retired private renters in primarily homeownership societies for those who do not become homeowners (Forrest and Hirayama 2015;Ronald, Kadi, and Lennartz 2015: 53).Debates about the relevant merits of "generations" vs "disadvantage" as being core analytic lenses often miss the point that, in a period of rising asset prices, stratification may increasingly come to be determined by their interaction, with intragenerational inequalities being determined by inheritances and other financial transfers between generations (Bangham et al. 2019).
Tracing the outcomes of these developments in empirical terms using household surveys is not straight-forward.Measures relating to house prices may be sensitive to such changes, but those relating to housing costs (or expenditures) will not, since recent homeowners represent only a small share of the total (Meen and Whitehead 2020).This may mean that change at the margins is hard to detect in, for example, studies of poverty, which focus on current incomes and expenditures, and typically on tenures in their entirety (i.e.all mortgaged homeowners), and with data that is typically blind to inheritance and other forms of wealth.But rising house prices can also be expected to lead to rising rents for those in the for-profit sector (Tranøy, Stamsø, and Hjertaker 2020: 396-7), and it is perhaps here where the effects might be expected to first be registered since rents on the whole rental stock are, differences in regulation notwithstanding, ultimately subject to revision while rising house prices impact only those who have recently purchased.This is also where the impact on poverty might be most acutely felt, given that a disproportionate number of those in or near poverty live in the private rented sector in many countries.Thus, we may expect financialization to be associated with worsening affordability for families both in the market rental sector and for (recent) mortgaged homeowners, but the former may be more immediately identifiable than the latter in empirical terms, and can be expected to be a particular problem for lowincome families.These propositions receive some support in Dewilde (2021), who finds that more highly financialized housing systems are associated with increased higher living conditions deprivation (a housing-sensitive measure of material deprivation) for renters and low-income households.
All of which is to say that the process of financialization has at least the potential to predict change in socio-economic outcomes.However, we need further studies that examine whether anticipated outcomes do, in fact, materialize.One reason for this is that housing outcomes are complex and multidimensional; rather than triggering affordability burdens or poverty entries, rising prices may be offset in some other way -e.g.falling homeownership rates, delayed independent living amongst young people, and so forth.This shows us both why this process is important for scholars of the welfare stateacting as an evolving set of intermediate institutions between the redistribution of incomes in social protection systems and welfarist outcomes such as poverty -and also -since the potential outcomes are multiple -why further empirical work on these questions is important.

Independent and Dependent Variable Problems and Impediments to Comparative Research on Housing
A second set of questions relate to how these processes might be measured and, again, these questions relate to both commodification and financialization.We approach these questions in light of what has been labelled the "dependent variable problem" in the literature on welfare state change and we believe that greater appreciation of this problem can contribute to advancing comparative housing scholarship.
The "dependent variable problem" refers to what Clasen and Siegel (2007: 4) describe as "a noticeable absence of reflection on how to conceptualize, operationalize and measure change within welfare states" and centres on the issue that alternative operationalizations of welfare state performance lead to substantially different perspectives about the extent of change.This matters both because it led to disagreement about the extent to which welfare states had, in fact, changed, but also because, for political economists, it frustrated the analysis of theoretical processes.As Paul Pierson has argued: 'it is difficult to exaggerate the obstacle this dissensus creates for comparative research . .., it is impossible to seriously evaluate competing explanations [of change] when there is no agreement about the pattern of outcomes to be explained' (Pierson 2001cited in Green-Pedersen 2007: 15).
Our aim here is to consider what the dependent variable problems might be for housing studies -a question which, we believe, has not yet been seriously considered.The point here is not simply that attention to conceptualization and measurement is necessary, but that we must look to understand the ways in which particular concepts and measures of, say, the financialization of housing influence our understanding of substantive questions, such as which countries have the most financialized housing systems, or whether the financialization of housing leads to deteriorating housing affordability.
The central approach to measuring commodification focuses on tenure balances between homeownership (or market renting), on the one hand, and reduced-cost renting, on the other, as measures of commodification and de-commodification, respectively, though we have noted above how outright ownership can be considered decommodifying, too.Studies adopting this approach to measurement focus on tenure balances, albeit in different ways, and there is little sense of the way that the coding of these variables may influence results.There are parallels here with the comparative welfare state literature, where it has been shown that the effects of political partisanship on welfare retrenchment vary depending on how partisanshipin particular, the balance between left-, centre and right-wing parties -is coded in binary terms (Allan and Scruggs 2004).
But there are important limitations to focussing on tenure balances, however coded: governments may seek to decommodify -weakening the link between household income and housing consumption -in multiple ways, including the provision of housing allowances or rent regulation.These multiple means of achieving similar ends "makes it difficult to arrive at overall judgements on the extent of social provision in this field" (Fahey and Norris 2011: 441).Nonetheless, focussing on the size of the social rented sector or on the generosity of housing allowances may amount to a feasible empirical strategy.What is lost here, though, is the supposed shift in the meaning of tenure (i.e. from use to exchange values) as well as other "objective" aspects of commodification -e.g. the strengthening of legal rights of landlords over their tenants, for example in relation to eviction, which is the kind of "micro-theory" that is hard to get to grips with in large-N comparative studies that may rely on household surveys.
But as we have suggested, the objective fact of homeownership, even when it is high and/or rising, is not necessarily taken in itself as an indication of a highly commodified housing system.Olt and Csizmady (2020) for instance claim that the privatization of housing in Budapest was not a clear example of commodification on the grounds that use value, as opposed to exchange value, was the main motivation of (new) homeowners.This echoes the earlier account by Allan and Scruggs (2004) in relation to housing systems in Southern Europe where they emphasize norms and institutions incentivizing property retention rather than exchange.Such meanings attributed to homeownership may be real, but they are difficult, if not impossible, to detect using the major social surveys that are available for comparative study.
Attempts to operationalize the financialization of housing also face significant challenges.Indeed, some accounts of financialization rely on vaguer claims -for example, of "the increasing power and prominence of actors and firms that engage in profit accumulation through the servicing and exchanging of money and financial instruments" (Madden and Marcuse 2016: 31) -that may not be amenable to direct observation and measurement and thus empirical scrutiny.Schwartz and Seabrooke's (2008: 243) influential account takes mortgage debt as a proportion of GDP as a measure of 'the degree to which housing finance is "liberal" or "controlled" and owner-occupation rates as a proxy for commodification. 1 On the basis of this, they create a 2 × 2 matrix and arrive at a fourway typology of varieties of what they label "residential capitalism".Schwartz (2020: 503) observes that "while the United States is often held up as the epicentre of financialization" it has a lower mortgage-debt-to-GDP ratio than many European nations.Do we take this as evidence that the housing markets of some European nations are more financialized than those of the United States, or is this conclusion dependent on this specific operationalization?These questions relate, too, to attempts to trace the consequences of financialisation.Angel (2021) finds that housing market financialization, using this proxy, is not associated with the residualisation of below-market rental housing.Dewilde (2018Dewilde ( : 2633Dewilde ( -2634)), also using this proxy, finds a marginally significant association between changes in financialization and changes in housing affordability in rural, but not urban, regions in 12 European countries.Moreover, in relation to measures of housing quality, Hick, Pomati and Stephens (2022b) find that while housing differences (especially, housing tenure) are important in terms of explaining the within-country incidence of severe housing deprivation -a composite measure of overcrowding and the deprivation of housing conditions -the extent of financialization, as measured by the outstanding mortgage debt to GDP rate or the degree of mortgaged homeownership appears to matter much less in explaining differences between countries.Do these studies show us the limits of the financialization thesis, or would other measures have led to different conclusions?
In conceptual terms, Schwartz and Seabrooke's (2008) proxy offers a pragmatic way forward, but is not without its limitations.It does not capture, for instance, potentially significant but hard to measure changes, such as the rise of AirBnB crowding out rental accommodation and thus prioritizing property-owners and tourists over prospective tenants, which forms part of Hearne's (2020: 9) account of the housing crisis in Ireland.Moreover, since it is mortgage debt and not, say, the rules governing housing market liberalization or the degree of securitization that is measured, there is an assumption that the process of financialization is leading to increased mortgage debt, as opposed to, say, simple supply-and-demand dynamics (Meen and Whitehead 2020).Disentangling financialization from alternative accounts may, on this measure, not be possible.Few alternate operationalizations exist, though Hulse and Reynolds (2018) propose a measure of investor housing finance as a proportion of the total.
At present, we have an insufficient understanding of whether the operationalization of these processes has an important bearing on their predictive power.We know from the comparative welfare state literature that shifting between spending ("welfare effort") to social rights measures of the welfare state can result in very different understandings about the extent of change in welfare states over time (e.g.Allan and Scruggs 2004), which has led to greater awareness of the "dependent variable problem" in comparative welfare state studies.The distinction between independent and dependent variables depends on one's analytic purpose -what are dependent variables for political economy scholars seeking to account for the development of welfare state institutions can be independent variables for poverty scholars looking to understand changes in rates of poverty and deprivation.This categorization question aside, the lessons from this debate add gravity to the difficulties in finding a sufficient measurement of commodification and financialization.We do know that one measure of financialization (the proportion of mortgage debt as a % of GDP) is strongly correlated at the country-level with another that has been used in the literature (the proportion of mortgaged homeownership; the country-level correlation across the EU-28 is 0.93 in 2018), 2 which provides some degree of support for the former, Schwartz and (2008) recommended proxy.Nonetheless, we do not regard the question of whether current measures are sufficient as settled.
In short, the theoretical literature on both commodification and financialization, while promising not only for housing studies but for our understanding of the welfare state and distributional studies more generally, needs greater attention to questions of conceptualization and measurement.This will require consideration of whether and how important developments -for example, the role played by institutional landlords -can be aggregated-up into country-level variables.There is a parallel here with the comparative welfare state literature, which has been significantly advanced by the construction of cross-national databases of policy measures such as the Comparative Welfare Entitlements Dataset (CWED2, see Scruggs and Ramalho Tafoya 2022), the Social Policy Indicators database (SPIN, see Nelson et al. 2020) and others, providing cross-national data on policy instruments that have enriched numerous empirical studies.In relation to housing, data from the European Mortgage Federation ( 2021) is often drawn upon for the mortgage debt to GDP proxy and, indeed, they provide a wider range of comparative indicators that may warrant further examination, including flow, as well as stock measures, in relation to mortgage debt, and annual measures of the number of dwelling completions and housing transactions.An alternative to relying on existing databases would be to scale-up survey micro-data into country-level variables.The ad hoc module of the 2020 wave of the Statistics on Income and Living Conditions survey, for example, asks respondents whether they own any real estate other than their main residence.When combined with tenure information, this variable could be used to create an indicator capturing inequality in housing ownership.Clearly, not all of these variables relate to commodification or financialization, but they are the kinds of indicators that may prove fruitful for empirical researchers seeking to disentangle competing theoretical propositions that attempt to explain housing outcomes.
Finally, in addition to ambiguities in relation to (in our terms) the independent variable -institutional characteristics and processes -there are also dependent variable problems in both the comparative welfare state change literature and in housing studies.The comparative welfare state literature has often relied on poverty rates or the povertyreducing capacity of welfare states to test the validity of welfare state classifications.Esping-Andersen and Myles (2011) tell us that: "poverty reduction is arguably the single most relevant measure of welfare state redistribution and, unsurprisingly, it has become the favored approach in empirical research".But policy-makers are often concerned not only with poverty reduction but, simultaneously, with the extent to which generous social protection may disincentivise work (Collado et al, 2019).
In housing studies there are perhaps a wider range of potentially relevant outcome variables still, reflecting the inherently multiple way that housing outcomes can be understood -for example, in relation to housing affordability, overcrowding, dwelling conditions, subjective evaluation of housing circumstances, and others.While much of the comparative welfare state literature treats the outcome of poverty as an unambiguous "bad", dependent variables in the housing studies literature can often seem highly qualified: a growing number of young people are struggling to get on the housing ladder -well, what is so special about owner-occupation?Housing affordability is an issue in the private rented sector? -well, some people choose to over-occupy housing and to prioritize this as a consumption good.Overcrowding rates are very high in some parts of Central and Eastern Europe?-perhaps, but there are questions about whether objective determinations of overcrowding are understood differently by people themselves, perhaps reflecting different preferences for living in multigenerational households.
There is, as we have argued, a need for a greater number of empirical studies examining the effects of commodification and financialization of housing at the micro level.Indeed, analysis of these housing and welfarist outcomes is central, we argue, to understanding the significance of the processes of commodification and financialization.This requires getting to grips with some of the conceptual and methodological challenges of operationalizing these processes and their anticipated effects, and we believe that the literature on the "dependent variable problem" can be a useful guide to comparative housing scholars in this task.

Conclusions
Debates in housing studies, in the comparative welfare state literature and in relation to poverty, have taken place in disjointed literatures for too long.In this paper, we have sought to demonstrate areas where productive integration of the literatures on housing studies, the welfare state and poverty is possible.Housing institutions act as an important mediator of the relationship between disposable household income and living standardsbetween the welfare state, the satisfaction of shelter needs and the experience of poverty at the micro-level.In particular, we see the commodification and financialization of housing as representing important processes that threaten living standards, and that the literature on these developments is of importance for scholars of the welfare state and poverty.On the other hand, we argue that scholarship on the welfare state, with its attention to the "dependent variable problem", which has led to greater awareness of the impact that decisions made in relation to conceptualization and measurement can have on substantive empirical findings, can enrich the comparative study of housing.
Though accounts of financialization (and commodification) vary in form, they typically suggest that this process is one that poses a threat to the housing outcomes of, especially, the young and those on lower incomes.However, while the financialization of housing has generated a substantial literature, much of this focuses on macro-level differences that are the focus of political economy and there is scope for a greater number of microlevel studies exploring whether housing market commodification and financialization do, in fact, generate the anticipated outcomes at the household level.
In order to contribute to this micro-level literature, we have sought to trace the potential social outcomes associated with the financialization of housing.If financialization is associated with a growth in mortgage debt because of an initial extension of mortgage finance and rising asset prices, then we can expect this to result either in a growth of "leveraged homeowners" or to precipitate a decline in homeownership amongst younger and poorer citizens (or both).This in turn, and especially in societies that privilege homeownership, may result in greater reliance on intra-family transfers to keep dreams of homeownership alive, leading to new forms of stratification both within and between generations.Rising property prices can also be expected to translate into rising rent levels.When combined with retrenchment of state-provided housing, this risks making housing increasingly unaffordable for market renters.In this way, financialization can lead to deteriorating housing affordability for market renters, delayed and frustrated homeownership, higher rates of leverage for those who become homeowners, and rising levels of inequality between and within generations.The financialization of housing thus has the potential to become a significant influence on stratification in contemporary society.
There are, however, limits to the financialization of housing becoming more widelyadopted in comparative study.First, commodification and financialization are subject to multiple, sometimes inconsistent, interpretations.Second, not all of these interpretations are capable of direct measurement, frustrating empirical analysis, especially of a quantitative kind.And, third, empirical understandings of financialization as a cause of change in housing systems do not tap into this "upstream", at a level where it might successfully be distinguished from other processes, but typically focus on a growth in mortgage debt, which might occur via alternative theoretical pathways (such as supply-and-demand imbalances).These issues impede the capacity for empirical analyses to test the claims made in relation to the distributional consequences of housing market financialization (e.g.Mader, Mertens, and Zwan 2020).Importantly, at least some of these limits reflect current impasses rather than inherent barriers and the hope is that by seeing them more clearly, we can move beyond them and have argued that greater attention to conceptualization and measurement questions (in line with attention to the "dependent variable problem") has the capacity to advance understanding of the variety of forms of financialization and their consequences.
Claims about deteriorating housing affordability have forced welfare state and poverty scholars to once again turn their attention to developments in housing systems (e.g.Hick and Lanau, 2017) and render the "disjointed literatures" of housing, poverty and the welfare state increasingly problematic.In trying to make sense of the impacts of housing systems on poverty, we have argued not only that scholarship on the welfare state and that of housing studies could benefit from a deeper integration but have suggested areas where further work is needed.After a long period where scholarship on the welfare state and poverty operated at significant remove from that of housing studies, we are, perhaps, moving on from the disjointed literatures of the past.

Notes
1. Curiously, the authors suggest that the owner-occupation rate "reflects size of social rental sector and thus commodification" (Schwartz and Seabrooke 2008: 256).If the size of the social rented sector is their real variable of interest, it is not clear why this was not examined directly.2. A far higher figure than is suggested by the scatter plot presented in Schwartz and Seabrooke (2008).