Redistribution policies towards poor families in Europe

Abstract

Poverty prevention is a central concern of welfare states, and the redistribution of financial resources has been a major strategy to realise it. The differences in addressees, extent, and conditions of this redistribution have been intensively studied. The relevance of family in poverty prevention policies, though, has hardly been analysed, although all forms of welfare redistribution “factor in” family in one way or another, and particularly so in poverty prevention. We analyse how family membership impacts welfare state redistribution to the poor to identify redistributive logics in terms of family, that is the unequal redistribution of public resources to particular family types. We systematically analyse and present the similarities and differences in these redistributive logics, using the micro-simulation model EUROMOD for the countries of the EU. The results show that poor families benefit from anti-poverty measures in form of additional benefits, but family-related financial obligations often exceed these.

Introduction

Combating poverty is one of the main goals of post-industrial societies, and the redistribution of financial resources has been one of its most important strategies (Townsend, 1979; Barr, 1993). Research has been particularly keen to understand the criteria for successful poverty prevention and to identify the underlying logics of granting resources to the poor. Comparative research has focused on the country variation in addressees and levels of redistribution (Jacques and Noël, 2021; Korpi and Palme, 1998), and on the concrete conditions for receiving poverty-reducing benefits (Gough, 2001; Knotz, 2018). Current welfare states often link benefits to the labour market by providing “carrots” to the poor, that is economic incentives for those who show work effort (thus aiming to “make work pay”) (Immervoll and Pearson, 2009; Marchal et al., 2018), and “sticks” to enforce activity on the labour market by otherwise restricting access to benefits (Clasen and Clegg, 2007; Knotz, 2018; Marx and Nelson, 2013). And while poverty remains a challenging issue in all countries, these, in part, strongly differ in how they address the poor (Esping-Andersen, 1990; Korpi and Palme, 1998; Gough, 2001; Lohmann, 2009; Nelson, 2013).

In poverty prevention, however, there is a second category, besides labour, that strongly impacts welfare state redistribution: the family. As people often live in economic units of family members, depending on each other’s income, also poor relief and poverty prevention policies have ever since been linked to the family, with benefits related to the size and economic situation of the family (Marshall, 1964; Obinger and Schmitt, 2018). The family, though, is not only considered in welfare state redistribution as potentially entitling to additional benefits; there are also two forms of financial obligations that welfare states impose on families: higher taxes and social insurance contributions for one, and for another, financial support within the family before members can be entitled to public means (resulting in lower benefits). They represent a financial disadvantage in comparison with individuals without family. Family-related obligations are hardly systematically considered in the literature though. Redistribution, therefore, can only be analysed properly when the relevance of the family – both in terms of benefits and financial obligations – is systematically considered. In addition, redistribution addresses social groups differently and, in terms of family, prioritises certain family types over others (Van Lancker and Van Mechelen, 2015; Christl et al., 2021). Since the impact of family membership on poverty prevention has not yet been systematically analysed though, our knowledge of family-related redistribution is, at best, incomplete. And this holds true for cross-country and within-country variation.

We aim to help fill this research gap by answering the question in how far countries differ in their redistributive logics towards poor families, defining redistributive logics as the lines of inequality as set down in welfare states’ redistributive regulations (Frericks 2023). For doing so, we examine the interplay of financial resources granted to and demanded from families. We compare financial redistribution to 76 hypothetical family types that are at or below the poverty threshold, with redistribution to individuals without (officially recognised) family members. Among the family types are those with children of different ages, adults differing in marital status, and the distribution of paid work between the partners. The analysis uses the micro-simulation model EUROMOD for the countries of the European Union and processes data on means-tested and non-means-tested benefits, taxes, and social insurance contributions.

In the following, we first review the theoretical background on the issue at stake, from which we then draw our assumptions and methodological decisions, after which we present the empirical findings and draw major conclusions.

Theoretical background

Poverty prevention depends on three factors that are differently substantiated in the welfare state’s poor policies: deservingness, conditionality, and differentiation (Van Oorschot, 2006; Clasen and Clegg, 2007). These factors are conceptually interwoven but emphasised differently. Deservingness, dating back to Aristotle, was fundamental to early “poor laws,” which granted resources to the so-called “deserving poor,” that is poor men who were unable to work because of illness, disability, or age, and their dependent family members. They were deemed deserving of societal resources as their inability to work was recognised as no fault of their own (Polanyi, 2001). And even later, welfare states considered the “non-able-bodied,” including wounded soldiers and disabled workers with their families, or in case of death, their widows and orphans, as the most deserving (Titmuss, 1965; Skocpol, 1995; Obinger and Schmitt, 2018). In current European societies, it is mainly a particular income threshold, as a percentage of a society’s average income, which is considered an income below which individuals or families are defined as poor (or at risk of poverty) and in need of (financial) help to be able to adequately participate in society (Callan and Nolan, 1991; Whelan, 2021).

Social benefits to prevent poverty, like all benefits, are linked to particular conditions. If claimants do not meet these conditions, benefits can be reduced, suspended, or denied altogether (Gough et al., 1997; Venn, 2012). These “sticks” that welfare states wield are concretised in different ways. First, in means-tests, income and/or assets of the claimant and often of his/her family are made relevant to the access to and level of benefits. This has been interpreted as a means to “single out the more deserving of the poor” (Marchal et al., 2020). Second, conditions are concretised in instruments to enforce labour-market participation as the main way to prevent poverty. Here, welfare states link benefits (or restrictions on them) to specific behaviours of “able-bodied” claimants, in particular, the active search for full-time employment (Clasen and Clegg, 2007; Knotz, 2018).

But poverty prevention also uses “carrots” to reward those who make an effort to find work. In line with the shift from the more “passive” income compensation towards economic self-sufficiency, subsidised commodification, and “making work pay” (Eichhorst et al., 2008; Immervoll and Pearson, 2009), welfare states have introduced additional employment-dependent benefits, the so-called “in-work” benefits that make it financially more attractive to work (Immervoll and Pearson, 2009; Marchal et al., 2018). With that, they address not only the non-working poor but also the working poor (Marx et al., 2016).

Welfare states also treat the various groups of poor differently, granting them higher or lower benefits and easing or toughening the conditions on benefit receipt (Marchal and Van Lancker, 2019; Jacques and Noël, 2021). Distinct programmes and targeted benefits address particular risk groups more directly (Bahle et al., 2010). In general, it is the long-term unemployed, low-wage earners or older persons close to pension age or without adequate pensions who are considered particularly vulnerable or disadvantaged, and thereby more deserving than others (van Oorschot, 2006; Van Oorschot and Roosma, 2017).

These differentiations between social groups in poverty prevention manifest themselves in the welfare states’ regulations themselves. Consequently, it is not surprising that countries differ in how they redistribute. One major distinction in poverty prevention that still dominates the discourse was made by Esping-Andersen (1990). He sees selectivity through targeting as characteristic of the so-called liberal regime type, high-level universalism as characteristic of social-democratic countries, and status maintenance as the major concern of so-called conservative countries. Further studies have shown, though, that the picture is much more complex. While they do observe substantial cross-country variation in Europe, they also show that a country’s allocation to a specific cluster varies with the facets of poverty prevention it addresses (e.g. Marchal et al., 2020; Natili, 2020).

Because people often live in economic units with family members and depend on each other’s income, also poor relief and poverty prevention policies have usually been linked to the family (Marshall, 1964; Obinger and Schmitt, 2018). But in fact, families are supported by welfare states in various ways and not only when they are poor. Welfare states may compensate families for certain income losses (e.g. parental leave payments), support them by means of direct cash allowances (e.g. for children), or with indirect allowances in the form of family-based social insurances or tax deductions (Bradshaw and Finch, 2002; Daly and Ferragina, 2018). For poor families, there are, in general, several additional benefits. For those who work but do not benefit sufficiently from the general redistributive measures, there are additional programmes and benefits that Marchal and Van Lancker (2019) refer to as “low-income targeting.”

Interpreting family as an economic unit though also results in imposing financial obligations on it. First, welfare states oblige families to support financially their members before being eligible to receive public means. These family-related obligations are put into practice by means-tests that generally assess the income and/or assets of the whole family rather than just those of the individual claimant (Saraceno, 2006; Daly and Scheiwe, 2010). The assessment of total family resources may result in benefit reduction in comparison with benefit levels to individuals without family or even benefit denial (Millar and Warman, 1996). These financial obligations are imposed on poor families in particular because means-testing on a household or family basis is most widespread in anti-poverty measures (Saraceno, 2006). As Saraceno (2006), p. 78) observes, “the nearer to social assistance a benefit, the stronger the reference to the family/household as a solidaristic community where sharing and saving of resources is expected to occur.” Second – and hardly addressed in the literature – some welfare states oblige families to pay higher mandatory contributions to social security schemes and/or higher taxes (Daly and Scheiwe, 2010).

What connects both strands of research is that not only the labour market but also family strongly impact welfare state redistribution. Importantly, we do not yet know much about the extent of this impact, or about the differences between countries. This is because previous research on welfare state redistribution in terms of family covered only a rather small number of countries (with others understudied) and on a reduced range of features, for example only welfare benefits, single welfare regulations, or a smaller number of family types.

Furthermore, it seems particularly difficult to make assumptions about differences between countries. Considering the family an economic unit with related obligations has been identified as a central feature of the male breadwinner model (Millar and Warman, 1996). This is logically sound. However, matching this model with concrete welfare states is highly challenging and controversial, particularly after major changes in those countries where this model previously prevailed (Pfau-Effinger, 2004). Consequently, though we expect international variation in redistribution to poor families, we are unable to develop concrete assumptions from the literature on how countries vary.

Finally, we also know that different types of families are addressed differently by welfare state redistribution, be it in terms of benefits and/or financial obligations (Author A). This differentiation in terms of family has been shown to have both vertical and horizontal dimensions: poor families are addressed differently than families that are not poor, and within middle-income groups, some families are defined categorically (e.g. married and single-earner couple families), or some on the basis of need (e.g. families with dependent children versus families without children or single-parent versus two-parent families) are provided with higher benefits and/or obliged less (Bahle et al., 2010; Van Oorschot and Roosma, 2017; Marchal and Van Lancker, 2019). Some of these redistributive logics with regard to distinct family types are argued to be common to all welfare states. Yet, how this differentiation manifests itself for poor families has never been systematically analysed. This is not only relevant for identifying institutionalised forms of inequalities in poverty prevention systems but also for identifying which families are successfully raised above the poverty line through welfare state redistribution and which are not.

In short, the impact of family membership on welfare state redistribution policies towards the poor is still little understood, and systematic analysis and a comparison of European welfare states in this regard has yet to be delivered. We aim to help fill this research gap by answering the question of how far countries differ in their redistributive logics towards poor families. The concept of “redistributive logics” towards particular social groups – here poor families – refers to the unequal redistribution of public resources to particular subgroups (Author B), here, for example, to poor single mothers who might profit much less from redistribution than poor single-income families. We analyse these logics in an international comparison.

Assumptions

Using current knowledge of how welfare states address poor families, we now develop assumptions (broad generalisations) that we then empirically test (see Table 1). The first three assumptions address commonalities between welfare states, that is shared redistributive logics with regard to distinct family types. The latter three assumptions address country variation.

Table 1. Assumptions about poverty prevention; expected cross-country differences (yes/no)

Source: Authors’ compilation.

First, it has been argued that poor children are a pivotal focus of anti-poverty policies and that welfare state redistribution intends to reduce or abolish poverty in poor families with dependent children (Bradshaw, 2013; Van Mechelen and Bradshaw, 2013). Consequently, all European welfare states, independent of their generosity, have income support packages for poor families of different types and size (Bradshaw and Finch, 2002). This lets us assume that all welfare states redistribute to poor families with dependent children by granting resources to them that exceed financial obligations – if any apply to these families at all (A1).

Second, it has been argued that all welfare states consider single parents to be particularly vulnerable. Research shows that single parents are often targeted by specific programmes tailor-made to their needs (Chzhen and Bradshaw, 2012; Van Lancker and Van Mechelen, 2015). Our assumption is thus that all welfare states treat poor single parents differently from poor couples with dependent children by granting them more resources and, if at all, by demanding from them less (A2).

Third, “making-work-pay” policies and related in-work benefits have been introduced in European welfare states with the aim to “activate” the non-employed and overcome the “culture of dependency” by “accentuating the gap between incomes in and out of work” (Immervoll and Pearson, 2009, p. 5). Consequently, poor families who show work-seeking efforts are given employment-related and family- or child-conditioned income supplements or tax credits (Kenworthy, 2015; Vandelannoote and Verbist, 2020). We assume therefore that in all European countries, income after welfare state redistribution is higher for working families than for families who are fully unattached to the labour market, since income from work is not (entirely) factored into the level of social benefits (A3).

As for our assumptions about other redistributive logics with regard to distinct family types, we can expect European welfare states to significantly differ. The first of these assumptions draws on the literature on differences in means-tests and family-related obligations and differences in “factoring in” income from employment (Millar, 2004; Nelson, 2010). We assume that countries differ in their imposition (or not) of financial obligations on those who are fully dependent on welfare state support (A4).

The literature prominently features arguments that welfare state redistribution differentiates support on the basis of the families’ distribution of paid work and that countries differ in the type of family that benefits from welfare state redistribution the most (Korpi, 2000; Lewis et al., 2008). We therefore expect to find country variation in the redistributive logics applied to poor families, depending on the families’ employment income distribution (A5).

Lastly, welfare states distinguish differently the family types that are entitled to benefits or have financial obligations (Daly and Scheiwe, 2010; Saraceno, 2018). Consequently, it has been argued that married and cohabiting couples (whether or not with dependent children) are treated differently, with financial advantages accorded to married families in some countries and to cohabiting couples in others (Christl et al., 2021). Therefore we assume that marital status matters in some countries’ redistribution, but not in others’ (A6).

As anticipated, clear cross-country or cross-regional variation in the latter assumptions is hardly possible. The first reason is that there is a lack of up-to-date classifications of European welfare states as to their relevant characteristics after the countries have considerably changed policies (Jacques and Noël, 2018; Kuivalainen and Niemelä, 2010; Pfau-Effinger, 2004), so that basing our assumptions on older classification studies would run the risk of their being severely obsolete for purposes of analysis. Second, previous research on welfare state redistribution in terms of family is, as shown above, limited, covering a rather small number of countries and a reduced range of study. And because family-related redistributive logics have never been systematically analysed and completely understood even for single countries, it is extremely difficult, if not impossible, to develop sound assumptions about country variation.

Methodology

Verifying the above assumptions depends on a sophisticated methodology. First, it must allow us to consider a large number of family types to gain insights into how these are addressed by poverty prevention. With this, differentiation – as redistributive logics – can be identified. Next, a suitable income indicator must be chosen that, on the one hand, enables a comparison of family types and, on the other hand, shows only the intended effects of the welfare state redistribution (Van Lancker and Van Mechelen, 2015), that is the income indicator is unaffected, for example, by factors such as individual preferences or social norms. Furthermore, we need to establish a reference point against which the extent of the redistribution in terms of family is measured. Every family type has its own reference point. And finally, we must choose a measurement instrument that allows us to simulate welfare state redistribution for a large number of family types and reference points and at the same time simulate the effects of a large range of regulations by taking into account their interactions.

Drawing on the rich literature on child support packages, we applied the so-called model family method (fundamentally developed by Bradshaw et al., 1993; Bradshaw and Finch, 2002; see also Penne et al., 2020). The essence of this method is to define a set of hypothetical family types based on characteristics that correspond to the particular research focus. The well-defined but necessarily limited family types “cannot be considered representative for the population as a whole” (Penne et al., 2020, p. 84). They are helpful though for an understanding of welfare state logics in terms of family when we are including a large number of hypothetical family types (and comparing them to a reference point; see below). National laws and social policies give rise to different notions of the family (Naldini and Long, 2017). We use its broad definition, according to which “the family is increasingly defined as a community of responsibility and care that extends beyond the boundaries of household, marriage, kinship, and even parenthood” (Schneider and Kreyenfeld, 2021, p. 5). This means that in addition to nuclear family types with children and two parents, we also take single parents and childless couples into account. To be addressed by welfare state redistribution, couples, resp. children, need to be officially recognised. Whether this is restricted – as in some countries – to heterosexual couples (and parents) is not addressed in this study. For experts in countries’ recognition of family types, the provided results can easily be translated into probable (dis-)advantages for non-recognised families in contrast to recognised ones. Finally, the family types we analysed all live in one household. This is because, first, this focus helps reduce complexity since regulations on redistribution differ in their details for families in different households, and second, the database and tool we use are limited to these family types.

Acknowledging discussions in the relevant research (e.g. Lewis et al., 2008; Bradshaw, 2013; O’Connor, 2013), we distinguish our family types along four dimensions: the distribution of paid work between the partners (in the case of two adults), the presence of dependent children, the family’s income level, and marital status. We operationalise the distribution of paid work as the adults’ different job-income combinations, and we adjust the established models (in brackets) that normally refer to households that are not poor. We distinguish the equal-earner model (dual-earner model) of a family with two adults who are both employed and, in our operationalisation, earn exactly the same employment income; the unequal-earner model (1.5-earner model) where both adults are in paid work but here asymmetrically; the single-earner model (same), when one adult alone earns the family income; and finally the absent-earner model where none of the adults earn employment income but rely on benefits. All adults in our hypothetical family types are of working age. In addition, we analysed single parents, that is single-adult family types.

We included dependent children – in line with the established literature on child-related redistribution (see Bradshaw et al., 1993). The young children we constructed as a preschool child (2 years) or children (2 and 3 years), the older ones as a secondary school-aged child (13 years) or children (13 and 15 years).

Our notion of poverty in this article refers to income poverty, defined as 60% of the national median equivalised disposable income (i.e. the at-risk-of-poverty threshold). To ensure that the hypothetical family types were comparable (also across countries) and that we could apply our simulation tool, we needed to start from gross income to simulate the “net disposable incomes” (NDIs). Such an income is appropriate for our purpose because we are interested in the welfare state redistribution as intended, and it allows us to compare this redistribution across family types (such as single- and two-parent families). We decided to analyse family types with very low (30%, 50%, and 60%) or no average employment income (EU-SILC). By simulating the data, we could then check whether these starting gross incomes resulted in incomes below the at-risk-of-poverty threshold after family-related redistribution. With this methodological decision, we were able to identify whether the various very low employment incomes translated into disposable incomes above the at-risk-of-poverty threshold, and if so, for which family types.

As a final dimension, we considered whether the couples are married or only cohabit. Thus we analysed in total 76 hypothetical family types (see Supplementary Appendix for an overview).

To identify redistribution in terms of family, we needed to develop a referential “zero point” where family-related redistribution cannot occur. Our reference point is an individual (resp. two individuals) with no (legally identified) partner or child. The difference in the simulated NDI of the reference point that is, in terms of gross income, constructed in parallel to the adult(s) of our family type, and the family type, shows the welfare state redistribution in terms of family, since only family-related regulations can explain this difference (see Figure 1). For example, if we are interested in a single-earner family in which the adults earn zero and half the average income, we compared their aggregated NDI to the aggregated NDI of two single individuals, one earning zero and the other half the average income. An equal NDI of the family and of the two individuals would indicate that there is no family-related redistribution.

Figure 1. Calculation formula.

Source: Authors’ compilation.

We simulated the incomes by means of the tax-benefit microsimulation model for the European Union, EUROMOD (version i4.0+), and its Hypothetical Household Tool (HHoT). The model includes a large set of redistributive measures among which are social insurance contributions (with health insurance particularly family related), means-tested benefits, child benefits, leave payments, and fiscal welfare, and it accounts also for the interplay of relevant parts of the tax system (Sutherland and Figari, 2013). All these components and their particular linkage play an essential role in poverty prevention – in the way the resources of a family increase, decrease, or remain unchanged compared to those of individuals without family (Ferrarini et al., 2012). The simulation, for the year 2020, used the latest data available at the time of analysis.

To test the first assumption – that European welfare states do not put financial obligations on poor families with dependent children (and if they do, they are lower than what families are granted; see A1) – we simulated and compared welfare state redistribution for 64 family types (with a dependent child/children) with that of the respective reference points. A decrease in benefits or an increase in taxes and/or social insurance contributions can be identified as a “financial obligation.” Whether the obligations exceed the benefits can be seen from the difference in NDI between these 64 family types and the respective reference points. To test the second assumption – that welfare states treat poor single parents better than poor couples with dependent children by obliging them less and granting them greater resources (A2) – a two-step analysis was required. First, to observe differences in benefits, we compared the NDI of couples and single parents of the same employment income, which is 0%, 30%, 50%, and 60% of the average employment income. As with couples, we analysed only single-earner families. As this does not show us how welfare states impose obligations, we repeated the analytical step for the first assumption; that is, we compared the simulated welfare state redistribution to “couple families” (with a dependent child/children) with their respective reference points, and we compared the simulated redistribution to single-parent families (with a dependent child/children) with their respective reference points. To test our fourth assumption – that financial obligations are not imposed on those who are fully dependent on welfare state support (A4) – we compared the NDI of 10 family types unattached to the labour market (with or without a dependent child/children) with those of the respective reference points.

The remaining three assumptions are addressed by subtracting the NDI of one family from that of another one. For the third assumption – that the NDI of working families is higher than that of families fully unattached to the labour market (A3) – we simulated the NDI of families that are out of the labour market, and we simulated the NDI again for those employed. To investigate the fifth assumption – that poverty prevention differs with the distribution of paid work (A5) – we compared the NDI of three family types with a gross income of 60% of the average income: single-earner families (60–0 income), unequal-earner families (40–20 income), and equal-earner families (30–30 income). Lastly, to test the assumption that poverty prevention differs with marital status (A6), we compared the NDI of families in which the adults are married with that of those in which the adults cohabit (as always, same gross income).

We examined the role of specific regulations to understand which regulation(s) led to financial (dis)advantages for one family type over another. The countries included are those of the European Union.

Empirical findings

Our first assumption is that welfare state redistribution generally does not financially oblige poor families with dependent children, but if it does, obligations do not exceed the benefits. Our data show that all EU welfare states, although to different degrees, impose financial obligations on poor families with children (see Table 2). These obligations are most pronounced in six countries (Finland, Germany, Italy, Malta, the Netherlands, and Spain). Here, benefits are reduced or denied to more than half of our family types, and obligations exceed benefits for more than half of them. The reductions affect families with young and older, one or two dependent children. In only four countries do benefits granted to families exceed the obligations put on them (Denmark, Lithuania, Poland, and Romania). Here, the NDI of all family types with children is higher than that of the respective reference points. Table 2 shows us the differences between the families’ and the individuals’ NDI; in Cyprus, Italy, the Netherlands and Spain, the NDI of some family types is more than 70% below that of single individuals there.

Table 2. Overview of family-related obligations on families with dependent children (values in %, light grey – leaders and dark grey – laggards)

Source: Authors, based on EUROMOD.

As to the second assumption, Table 2 shows that, indeed, welfare states treat poor single parents differently from poor couples with dependent children. As expected, obligations are imposed on single parents only exceptionally, in seven countries: Denmark, France, Italy, Ireland, Lithuania, Luxembourg, and Slovakia. These are cases based on unfavourable taxation (Denmark, France, and Lithuania), reduced means-tested benefits (Italy), higher social insurance contributions (Luxembourg), and simultaneously reduced means-tested benefits and higher social insurance contributions (Slovakia). Obligations are thus mainly imposed on couples across all the study countries. And while for many couples with dependent children the obligations exceed benefits, this is not the case for any single-parent family.

While this result seems to confirm the concept of deservingness, comparing the NDI of couples and single parents, however, blurs the idea that single parents are necessarily more targeted than couples. On the contrary, if we look at families without labour income or with earnings of 30% average income, almost all European countries provide more financial advantage to couple families than to single parents, as their NDI is higher. The redistributive instruments that cause this are higher means-tested benefits. The picture is more diverse for the higher incomes (that is 50% and 60% of the average income). In several countries (Cyprus, Germany, Denmark, Finland, Luxembourg, and Slovenia), couple families still benefit more, in spite of higher taxation. There are five countries (Bulgaria, the Czech Republic, Estonia, Greece, and Latvia) where redistribution does not differ between couples and single parents. The rest of the countries financially advantage single over couple parents or show a mixed picture that depends on the type of family – often a result of higher means-tested and non-means-tested benefits and lower taxation. Interestingly, in several countries single parents with 50–60% average employment income find themselves above the at-risk-of-poverty threshold of 60% median income after redistribution, while couples with the same employment income find themselves below it. This might be interpreted as “making-work-pay” regulations, since a single-earner, single-adult family is better off than a single-earner, dual-adult family.

The third assumption is that welfare state redistribution differentiates between claimants with and without attachment to the labour market by creating substantial income gaps between the two. This assumption can be confirmed for most of our countries, with the most substantial differences to be observed for Lithuania, followed by Romania, Bulgaria, Poland, and Hungary. But as Figure 2 shows, there are also countries where the income of some family types is fully equal to (in Estonia, Latvia, and Spain) or even lower (in Austria, Cyprus, Denmark, Ireland, Malta, and the Netherlands) than that of families out of work. Here, Cyprus and Denmark are particularly interesting cases. In Denmark, the NDI of families without labour market attachment is above the at-risk-of-poverty threshold, while that of dual-earner (30–30 income and 40–20 income) and single-earner (30–0) families is below it. In Cyprus, this is also related to marital status (see also below), where cohabiting out-of-work families are above this threshold, while all working married families that we studied are below it.

Figure 2. Comparison of families in work and out of work.

Note: The value 100 means the working family’s income equals that of the inactive family. All family types were included in the measurement, and the bar indicates the range from the lowest to the highest share of income detected. Countries coloured light grey also have values that are lower than those of the inactive family.

Source: Authors’ compilation based on EUROMOD.

While knowing that all welfare states impose financial obligations also on poor families, we assumed, fourth, that they are not imposed on non-working families who fully rely on welfare state support. The data, though, are counterintuitive, and we find that (with the exception of Lithuania) all welfare states do impose obligations on these family types (see Table 3). They do so primarily through reductions in means-tested benefits, but also, for some family types, through higher income tax (Denmark) or higher income tax and social insurance contributions (Luxembourg). This is because the means-tested benefit is taxed before it is paid to families, and in both countries, the tax deduction for families is higher than for the reference points; in Denmark, because the means-tested benefit (before tax deduction) is higher for families with dependent children than for individuals without family; in Luxembourg, the means-tested benefit (before tax deduction) is lower for families and is even taxed more. Thus, while all European countries impose obligations on the various family types, they differ in the levels at which they do so. And we observe a group of countries – Finland, Germany, Italy, Malta, the Netherlands, and Spain – where obligations were not only found for all ten family types we simulated but for all of them, they also translated into lower NDI in comparison with individuals without family. Importantly, while there are also many family types with dependent children for whom obligations exceed benefits, it is the childless families (and in many countries only this family type) whose income is particularly reduced in comparison with the reference point – most extremely in Malta where the NDI of these family types is only 52 per cent of the NDI of individuals without family. Obligations are most pronounced in Finland, Germany, Malta, and Italy, where poor families’ NDI is, on average, less than 77 per cent of the reference points.

Table 3. Obligations on out-of-work families (light grey – leaders and dark grey – laggards)

Note: The value of 100 equals to a reference point’s NDI.

Abbreviation: MTB, means-tested benefits.

Source: Authors, based on EUROMOD.

Our fifth assumption – that poverty prevention differs with the distribution of paid work – is confirmed by our data for most countries. As Figure 3 shows, redistribution is often more favourable for families in which both adults work. Equal-earner families’ NDI is higher than those of other family types with the same employment income in the Czech Republic, Estonia, Finland, Germany, Hungary, Ireland, Lithuania, and Sweden. These differences result from different redistributive instruments, mostly higher means-tested benefits and lower taxes. As a result, the NDI of single earners in Finland and Lithuania, for example, is below the at-risk-of-poverty threshold, while that of families with two earners, both equally and unequally distributed, are much closer to or even above it. And Luxembourg, for instance, redistributes most towards unequal-earner families. However, in other countries the picture is more complex and depends on the type of family. Family types with younger children, for instance, are in some countries more generously supported when they are equal-earner families, as are family types with older children on the other hand, when the family is an unequal-earner family. There are only three countries – Denmark, France, and Croatia – where single parents are financially the most generously supported. In France, this results from higher means-tested benefits granted to them, in Croatia, from lower social insurance contributions, and in Denmark, from the combination of both. Moreover, in Denmark, the disposable income of single-earner families is above the at-risk-of-poverty threshold, while that of dual-earner families is not. In some countries, differences in poverty prevention based on the distribution of paid work are much more difficult to identify since they redistribute mostly to single-earner and equal-earner family types (e.g. Italy and Slovakia), but also to unequal-earner models (e.g. Latvia and Spain) depending on the concrete family type; that is, there is no clear redistributive logic in these countries favouring a particular family type in terms of the distribution of work. And finally, Bulgaria, Greece, and Portugal show no difference at all among the three ways of distributing paid work in the family.

Figure 3. Differentiation of family types based on distribution of paid work.

Note: National redistributive logics favouring equal earners (the Czech Republic, Estonia, Finland, Germany, Ireland, Lithuania, and Sweden); single earners (Croatia, France, and Croatia); unequal earners (Luxemburg); mix of equal earners and unequal earners (Austria, Belgium, Cyprus, Hungary, Malta, the Netherlands, Poland, Romania, and Slovenia); mix of equal earners with single earners (Italy and Slovakia) and mix of all three (Latvia and Spain). No difference in Bulgaria, Greece, and Portugal.

Source: Authors’ compilation based on EUROMOD.

Finally, we tested the sixth assumption that poverty prevention differs with marital status. The findings show that the majority of countries do not differ in their redistribution based on marital status. There are, however, countries where marital status is relevant: differences in redistribution in favour of married couples were found in the Czech Republic, Denmark, Estonia, Ireland, Malta, and Poland. They are caused by lower taxes and/or higher means-tested benefits. Differences in favour of cohabiting couples were found in Bulgaria, Cyprus, and Romania, caused by higher means-tested benefits. In Cyprus, this redistributive logic results in cohabiting families’ NDI being above the at-risk-of-poverty threshold, while that of married families is found to be below it. Lastly, Belgium, Germany, Italy, and Luxembourg favour some family types when the partners are married, and some others when they are cohabiting, this again due to a number of redistributive instruments, including social insurance contributions (Belgium and Germany).

Conclusion

This research has striven to understand the criteria for successful poverty prevention and identify the underlying logics of granting resources to the poor. Major concerns of previous comparative analyses have been to identify the differences among the addressees and in the extent of welfare state redistribution, as well as the conditions on the receipt of poverty-reducing benefits that impact the access to and levels of benefits. With earlier research focusing heavily on poverty prevention in the context of employment though, it has hardly taken into consideration the relevance of family in poverty prevention. But since people often live in economic units of family members, depending on each other’s income, also poor relief and poverty prevention policies have ever since been linked to the family. Family, thus, might increase or decrease redistribution. There has been little analysis though, let alone systematic comparison, of how welfare states vary in their definition of deservingness, corresponding conditions, and differentiations in terms of family. Our study has contributed to filling this research gap by comparing redistribution policies to 76 hypothetical family types that are identified as being poor or at risk of poverty. The family types differ in terms of the number and age of children, the adults’ marital status, their distribution of paid work, and the level of income. By distinguishing numerous family types, we were able to show how and in how far welfare states treat families differently. Our analysis of the countries of the EU, based on the micro-simulation model EUROMOD, considered the interplay of different redistributive instruments such as means- and non-means-tested benefits, taxes, and social insurance contributions. We draw on regulation data – and explicitly not on outcome data – to understand the redistributive logics of the EU welfare states’ policies towards different family types.

In many ways, the results of our study are at odds with the general understanding of poverty prevention. We tested six assumptions on welfare state redistribution to poor families that we deduced from the literature, with half of them positing common redistributive logics, and half of them, international variation in them. None of the assumptions were fully verified. In the literature on child poverty, for instance, welfare state redistribution is assumed to be set up in a way to reduce or abolish poverty in families with dependent children. And indeed, our study confirms the results of previous studies: that poor families with children benefit financially from welfare state redistribution via entitlement to some additional benefits, lower or negative taxes, and lower social insurance contributions. However, we show that this is only half of the story, since all welfare states also impose family-related financial obligations on poor families with children by reducing or stopping means-tested benefits or increasing social insurance contributions and/or taxes. In many cases, these obligations significantly exceed the additional benefits granted. Thus the combined analysis of benefits and obligations raises doubts about welfare states’ intention to – or understanding about how to – indeed reduce child poverty. The current efforts of some welfare states, for instance, Germany, to reduce poverty in families with children by reducing excessive redistributive regulations probably show that they recognise their welfare state redistribution setups to be unfavourable and inadequate.

Another assumption drawn from the literature was that welfare states do not financially oblige families without labour market income who fully rely on welfare state support. However, apart from Lithuania, all countries impose obligations even on families who are not economically active. This affects families without children to a greater extent, but the NDI of many families with dependent children is also reduced, in comparison with persons without family, by these financial obligations.

Family types are addressed differently by welfare state redistribution in various ways, such as in relation to poor families’ attachment to the labour market, resulting (in some countries) in higher NDIs of those working, or making it (in other countries) financially more advantageous to be economically inactive. Also, countries treat single parents differently from couples with children, for example by making it financially more advantageous for couples than for single parents to be economically inactive – though, when in employment, it is often the single parents who are granted comparatively higher benefits. In addition, poverty prevention varies with the distribution of paid work between a family’s adults, and in some countries also with their marital status. The in-depth analysis of the role of family-related regulations in redistribution shows that differences across family types are mostly caused by differently set up taxation and means-tested benefit schemes. In terms of poverty, this means that redistribution by the welfare state reduces poverty for some family types to a greater extent than for other family types and, as our study shows, brings some of them over the at-risk-of-poverty threshold, but others not.

The findings unveiled some other unforeseen results. First, in none of the aspects examined did the country differences correspond to any established country clustering (Esping-Andersen, 1990; Daly, 2020; Natili, 2020). This suggests that different perspectives on redistribution, such as ours on the role of family, are needed to better comprehend international variation in welfare state redistribution. Second, one might identify “leaders” in our study that correspond to those of earlier studies; the “laggards,” instead, cannot be fully confirmed, particularly not in the case of Finland and France (see Bradshaw, 2013), because Finland is known for its strong focus on income security and equality (Kvist et al., 2012), and France, for being very supportive of families (Bradshaw et al., 1993; Saraceno, 2016).

The major findings of our analysis are, therewith, counterintuitive or at least surprising, and they reveal that comparative welfare state research needs to further develop concepts and measurements useful for the comprehension of welfare state logics. All forms of welfare redistribution “factor in” family in one way or another, and particularly so in poverty prevention. Consequently, its marginal role in studies on welfare states, with their differences and developments, is astonishing. Our study contributes to a better understanding of the complexity of poverty prevention and to uncovering previously unseen social inequalities inherent in redistributive regulations.

There are, of course, also limitations to our approach. While we take into account all relevant redistributive regulations and therewith our study has a much broader scope than most on welfare state redistribution, we were nonetheless unable to include in-kind benefits, which are another major source of societal support for poor families and therefore perhaps a reason for some additional or reduced redistributive variation (e.g. due to the availability or not of such benefits for non-employed parents/mothers). Simulating them, however, requires different modelling. In addition, one could study additional risks such as disability, early retirement, or migration, to gain further insights into the logics of welfare state poverty prevention and the relevance of family in it. To keep complexity low, we did not include these here, but further research focusing additionally on them could reveal other important “blind spots” and inequalities in redistributive systems.