Literature Review: Mechanisms and Hypotheses

Abstract

This chapter examines the causal links between welfare state policies and individual-level outcomes. To this end, common theoretical premises, mechanisms and hypotheses are summarised and systematised. This is done by linking specific functions of the welfare state to specific groups of dependent variables, leading to a proposal for deriving theoretically sound indicators.

The previous chapter highlighted a number of problems that result from a lack of standardisation when selecting operationalisations. There are two extremes when it comes to how we can possibly solve these problems. One would be to further improve the overall measurement of the welfare state in order to arrive at one single universally agreed-upon operationalisation. It seems unlikely that such an agreement can be reached in the near future and we might actually also question whether something like that could (or should) be realised at all (e.g. Spicker 2018). In fact, it seems doubtful that a catch-all approach is actually differentiated enough to capture all of the different relevant facets of welfare stateness. The opposed extreme would be to give up the search for a universal instrument and instead try to find fitting indicators for singled-out issues. Finally, a middle ground might be to propose clusters of conceptually related indicators that represent tailored independent variables for specific research questions, hypotheses, and dependent variables. This last solution appears to be the most promising, as it offers enough leeway for context-specific modelling without compromising comparability.

In the following part, I am going to develop these thoughts further. In order to explore which indicators are most suitable for specific research agendas, the causal relationships linking welfare state policies to individual-level outcomes are discussed in detail. For this purpose, theoretical premises, mechanisms and hypotheses are summarised and systematised. In order to capture the assumed causalities, I am proposing to approach the matter from two perspectives: on the one hand, I am taking up the functions of welfare states discussed previously (cf. section 2.2). On the other hand, I am systematising the individual-level outcomes that serve as popular dependent variables in the relevant literature. By linking specific functions of the welfare state to specific outcomes, I aim at pinpointing relevant mechanisms in more detail. This is intended to result in a proposal, which describes how to derive theoretically grounded indicators by shifting the way we look at the welfare state when treating it as an independent variable. Instead of attempting to best capture social policy arrangements as a whole, the focus should be on modelling the key assumptions about the relationships between policies and micro-level outcomes. Such proceeding—and this is important to emphasise—relates foremost to the type of cross-cultural comparative multilevel research addressed in this book.

4.1 The Mechanisms Matter: a Short Introduction

In the exemplary research objective chosen for illustration in the previous chapter, very basic theoretical assumptions about how welfare states influence individual attitudes towards social policies were briefly introduced. In this case, it has been hypothesised that generous and universal social policies following social-democratic principles generate political support and positive attitudes towards redistribution (Jaime-Castillo 2013; Roosma et al. 2014) while redistribution-based and targeted polices increase conflicts between beneficiaries and contributors, leading to disapproval of welfare policies (Jordan 2013). Since the aim of this chapter is to identify such hypotheses for a variety of popular research questions, a more comprehensive review of literature on welfare attitudes and many other research items is needed. Since the explicit measurement of cause and effect is the focus of this book, some general preliminary remarks on the nature and relevance of mechanism-based reasoning follow.

Research addressed in this book, usually fits into a framework shaped by theories of the middle range (Merton 1968). As the term suggests, such theories involve a perspective between very general, comprehensive approaches on the one hand and detailed, very specific ones on the other. As Merton puts it:

Middle-range theory involves abstractions, of course, but they are close enough to observed data to be incorporated in propositions that permit empirical testing” (Merton 1968: 39).

In this line of research, the identification of social mechanisms emerged as an important analytical task (Merton 1968: 45). Mechanisms have been defined in a variety of ways, which share a few important aspects. One of these shared features is that they reveal causal processes that show how and why a given phenomenon is observed (for an overview of definitions see Hedström & Ylikoski 2010: 51). Essentially:

[…] it is not enough to state that X leads to Y: a satisfactory explanation demands explication of the sequences and steps through which X and Y are causally linked, i.e. why and how X leads to Y (Tranow et al. 2016: 5).

In recent years, exploring mechanisms has become increasingly popular in the social sciences and the term is widely used—especially in research attributed to the field of analytical sociology (for an earlier overview see Manzo 2010). While it important to exame in more detail how and why “nuts and bolts, cogs and wheels” (Elster 1989: 3) exist and help explain social processes and phenomena, the term is sometimes used prematurely. As Kalter and Kroneberg (2014: 100–103) point out, the popularity of the term led to it being used even in cases where it is not appropriate. This includes (1) tautological (‘mechanism’ and explanandum are equal), (2) speculative (ad hoc explanations without empirical proof), (3) elliptical (‘mechanisms’ refer only to concepts) and (4) partial (incomplete) assumptions or evidences. Out of these four, the first two are especially problematic. Overall, the adequacy of a mechanism is determined by its ability to lead to specific and informative hypotheses (Kalter & Kroneberg 2014: 102).

A popular approach to conceptualising mechanisms is the macro-micro-macro scheme of sociological explanation. Since this approach is well compatible with the multilevel logic pursued in this book (cf. section 1.2), this analytical framework is chosen as conceptual representation of mechanisms (and other types of relationships) that are relevant for the research questions at hand. A very basic illustration of different kinds of mechanisms within the macro-micro-macro scheme is provided in Figure 4.1. Three types of mechanisms are usually explored when applying this analytical framework to research addressing how and why an outcome on the societal level (S2) results from an initial situation (S1). Following Hedström and Ylikoski (2010: 59), the link between macro- and micro-level has to be taken into account in the first step. Such situational mechanisms (1) address how and why societal contexts shape individual actors by affecting their position in the social structure, their desires and beliefs et cetera. The second type of mechanism grasps processes leading individuals from an initial situation (A1) to a specific outcome (A2). It is labelled formative mechanism (2) in the figure. This deviates slightly from popular versions of such models in the literature, which usually work in an action-based framework. Since, however, the relevant hypotheses regarding the impact of welfare states on individual-level outcomes refer to a variety of phenomena, going well beyond action-based approaches, this more general label is chosen to incorporate other processes and outcomes, such as attitude formation. Lastly, transformational mechanisms (3) refer to how and why individuals generate intended or unintended outcomes on the societal level.¹ As mentioned earlier, this last step is not typically part of the research covered in this book, as the outcome to be explained is usually at the micro- rather than the macro-level.

Figure 4.1

Types of mechanisms. (Figure based on Hedström & Ylikoski (2010: 59) and Tranow et al. (2016: 8))

Although research exploring how and why the characteristics of welfare states affect individuals seems to fit well within the analytical framework, detailed references to social mechanisms, causal processes, and the key issues involved are indeed rare in the relevant literature. This problem is sometimes acknowledged. For instance, O’Campo and colleagues (2015: 89) refer to it in the case of the impact of welfare policies on health and poverty.² In addition, many hypotheses in relevant contributions do not actually contain mechanisms or even causal relationships at all—even if some still refer to those terms (often in the problematic sense observed by Kalter and Kroneberg (2014) that was mentioned before). Instead, studies frequently focus on mere effects and statistical relationships, without much specification of causal links, thus often lacking clear-cut theoretical elaborations on how and why a correlation is expected. In a more recent publication Spicker (2018) takes a very critical perspective on the treatment of causality in the specific example of assumed effects of welfare policies on individuals:

[…] how can we say which is the cause, and which is effect? Wherever there is a complex, multi-faceted set of phenomena, it is almost impossible to distinguish generative mechanisms or to demonstrate genuinely effective processes (Spicker 2018: 225).

Even though there is some ground to this argument, I disagree with his conclusion that this ‘evidential’ issue—combined with methodological concerns about the reduction of complexity in statistical approaches and theoretical concerns about the equivalence of countries—should lead to questioning the worth of quantitative approaches in general (Spicker 2018: 226).³ If we want to understand at large why individuals differ not only within a country but also between countries, turning away from statistical approaches means abandoning entire research questions. A middle ground between giving up on mechanisms and giving up on statistical models, could be to put more effort into the theoretical specification of the assumed effects and the selection of measures. In addition, it is helpful to know the limits of the actual modeling of possible mechanisms. This includes acknowledging not all complex mechanisms can be modelled empirically. This can result from a lack of data on the policy-level or individual-level as well as from statistical limitations.

In contrast, some hypotheses simply do not entail complex mechanisms, as they describe in fact almost tautological relationships. This can be because they highlight one particular path in a potentially more complex model, or the outcome is a necessary consequence of the condition—sometimes even trivially so (e.g. income replacement decreases poverty). Still, regardless of whether we are dealing with complex mechanisms or singled-out relationships between welfare state and individuals: it it important to distinguish between the various epistemological interests motivating hypotheses and to ask how and why differences in social policy-making impact individuals differently.

To summarise the main arguments in this brief excursion into the field of mechanism-based explanations: its main advantage for the task at hand is its focus on breaking down assumed relationships between social contexts and individuals and exploring in more depth how and why such links exist. In the case of the specific problem at hand this means asking why an outcome is influenced by the welfare state—in the assumed way and how the underlying causal process works. If we can answer these questions—or at least approach possible answers –, it will considerably help to select appropriate and meaningful operationalisations of welfare stateness for the empirical test of theoretical models.

For this purpose, the following section summarises and systematises popular hypotheses. The term ‘mechanism’ will be used to refer to the explanation of a causal relationship between welfare stateness and specific outcomes. In some cases, it can be discussed very critically if using the term overstates the complexity of a given explanation in the sense described before. For the sake of consistency, I will use the term regardless, albeit cautiously. By linking the hypotheses to the functions of welfare states discussed before,⁴ I hope to further concretise the key causal assumptions—especially in cases where the relationship between welfare state and micro-level outcomes lacks strong theoretical foundation.

4.2 Summary of Popular Hypotheses

While it is easier to research relevant literature on a specific dependent variable, it is more complicated to obtain a comprehensive review of literature on a specific independent variable. There are various social phenomena, which are influenced by social policies and therefore receive much attention in the literature. Among the most popular dependent variables are attitudes towards the welfare state (e.g. Roosma et al. 2014; Steele 2015), health (e.g. Bambra et al. 2014; Foubert et al. 2014; Deeming & Jones 2015), poverty (e.g. van Lancker & van Mechelen 2015; Brady et al. 2017) and work-family balance (e.g. Beham et al. 2014; Lunau et al. 2014). However, it is important not to overlook other notable research that suggests a significant impact of the welfare state. For this purpose, the pool of examined literature is enhanced by a review of relevant journal publications in a five-year-period (mid-2013 to mid-2018) listed in the Social Science Citation Index (SSCI). Contributions are included in this review if they meet four criteria. First, the welfare state must be treated not just as a control variable, but as an important explanatory variables. For this purpose a reference to the welfare state, public policies or social policies has to be included in the abstract, title or keywords. Second, the examined dependent variable must be on the micro-level. Third, a cross-cultural comparison must be carried out. Thus, comparisons in general or multilevel analyses in particular have to be mentioned. Lastly, since this project focusses on western democracies and especially the European countries, a reference to this geographical region is required.⁵ These search parameters resulted in just above 530 contributions of which about 15 percent perfectly fit these criteria. This is by far not an exhaustive stocktaking of all relevant publications. Yet, since the aim is to identify popular topics in the body of literature, it is not necessary to include every study into the sample. The configuration of the search parameters should provide a sufficient picture of contemporary research in which the welfare state serves as an important independent variable, and thus key causal assumptions should be represented in the selection.

Even though the identified publications treat the welfare state as a central explanatory variable in their hypotheses and empirical models, many studies lack a comprehensive discussion of why and how features of welfare states are at work. Thus, the explicit mention, conceptualisation and operationalisation of social mechanisms is rare. This somewhat sobering finding makes it difficult to identify causalities beyond loose assumptions or statistical correlations. However, the mere fact that mechanisms are often not explicitly mentioned and modelled does not necessarily mean that the hypotheses posed are not based on causal explanations. Furthermore, while the majority of relevant studies relies only on very short theoretical explanations, there are also contributions, which set an example for well-grounded and differentiated conceptualisations. These studies will be referenced more extensively in the course of this chapter as they provide much-needed material to bridge some existing gaps in the argumentations.

The following sections provide a detailed overview of key objects of research and the hypotheses tying them to social policies. The majority of dependent variables can be subsumed under at least one out of four broader categories: (1) well-being, (2) risks and needs, (3) attitudes and (4) behaviours. Even though there are overlaps between them, hypotheses within each of these topics exhibit similarities. These similarities are discussed and systematised in the following four sections of this chapter, which all follow the same structure: first, main debates are summarised. Second, one especially characteristic topic is selected for an in-depth discussion within each of the four broader topics. This is followed by exploring how and why the welfare state is assumed to affect the specific outcome highlighted in this example. Lastly, hypotheses are summarised in an attempt to deduce general conceptualisations and patterns from the specific examples. While conceptual issues receive much attention in the following review, actual empirical findings are not going to be discussed in much detail. The reason is that this contribution is interested in the theoretical and conceptual arguments. Since the results are assumed to be biased, or at least not fully comparable, including them in the discussion could distort this view. Thus, I am interested in the main hypotheses and explanations, not in the question if they are confirmed (although—as they are the most popular ones—they usually are).

Before turning to the dependent variables, one final comment is required. Comprehensive bodies of work deal with each of the topics covered in the next part of this chapter. There are many complex contributions highlighting important theoretical and methodological issues in each of the four examples. My summaries of how and why welfare states affect each dependent variable cannot and do not do justice to each one of them. Instead, this chapter aims to summarise very general assumptions on a general level—in the sense of a minimal consensus. It will not permeate all facets of the subjects but substantiate the connection between them and the welfare state for the purpose of this contribution.

4.2.1 The Impact of the Welfare State on Well-Being

Well-being is an outcome of social policies, which is closely related to risks. However, there are several reasons why literature on the impact of welfare states on well-being deserve a separate inspection. One reason for this is the wide variety of manifestations of well-being, ranging from health outcomes to life satisfaction and other forms of subjective well-being, beyond the more 'classical' risks associated with work and social security. Another reason is that literature dealing with the most apparent expression of well-being—individual health—offers some elaborate debates about the theoretical conceptualisation and empirical operationalisation of welfare stateness as a determinant of health outcomes. As I will show in the course of this chapter, such debates can be applied to the relationship between welfare state and other outcomes as well.

When examining well-being, several different perspectives have to be taken into account. Two areas are especially prominent: subjective well-being (SWB) and health. While subjective well-being is based on individual perception, health outcomes are usually related to objective risks. Even though a broader focus on well-being covers a variety of topics and incorporates various conceptualisations of well-being (ranging from economic to philosophical aspects),⁶ the different dependent variables share many similarities when it comes to the assumed causalities tying the welfare state to a manifestation of well-being. In the following section, I focus on health outcomes. This dependent variable is selected for various reasons. First, health outcomes are—by their very nature—closely linked to the welfare state (particularly the healthcare system) and therefore receive much attention in the literature (for a review see among others Muntaner et al. 2011; Bergqvist et al. 2013). Second, literature on the relationship between welfare state arrangements and health outcomes is quite sensitised to the issue at hand—especially compared to other fields of research. Thus, conceptualisation and operationalisation of social policies as an independent variable and problems arising from inconsistent approaches are explicitly discussed in this line of research (Brennenstuhl et al. 2012; Bergqvist et al. 2013). While the problems of unstandardised proceedings were discussed from a more general perspective in the previous chapter of this book (cf. chapter 3), the different perspectives on social policies pinpointed in the critical literature on including social policies as determinants of health are a helpful addition. This includes theoretical approaches conceptualising mechanisms linking the welfare state to health outcomes (such as Beckfield et al. 2015), as well as methodological reviews (such as Bergqvist et al. 2013) and studies selecting their indicators as explicit representatives of distinct perspectives on the welfare state (e.g. Ferrarini et al. 2014a).

Empirically, health outcomes can be explored based on objective measures of health as well as self-rated ones. Since comparative research on health usually relies on survey data, self-perceived health is much easier to measure than objective health.⁷ Therefore, the following discussion mainly relates to this kind of health measurement. Furthermore, it is important to distinguish between two different perspectives on health in the literature. One focusses on health outcomes in general, the other on health inequalities. An increase in overall health in a population does not necessarily go hand in hand with a decline in socioeconomic health inequalities (e.g. Bambra 2013; Bergqvist et al. 2013). I will take up both perspectives when summarising relevant hypotheses later in this chapter as they are almost inseparable. The social gradient of health outcomes on the micro-level will be especially important when differentiating between direct effects of social policies on health and those that moderate the impact of social inequality (or are mediated by social inequality).⁸

I proceed in three steps to identify different conceptualisations of welfare stateness. First, I determine possible relationships between welfare state arrangements and health outcomes and the mechanisms responsible for such relationships. Second, I summarise conceptual views on the welfare state that emphasise different features. Third, I explore specific hypotheses in the literature to determine which mechanisms and perspectives are assumed in explanations. Overall, this process is used to identify shared aspects, which help to distil distinct perspectives on the welfare state.

Turning to the first step, the welfare state is linked to health outcomes in several ways. The relationship can take the form of a moderation of the severity of other determinants of poor health as well as indirect and direct effects. When looking at the first two, we have to take into account determinants of health on the micro-level, which are part of moderated or mediated effects. Key determinants of health outcomes on the micro-level are manifestations of social inequality: higher education, social status and income increase health (Beckfield et al. 2015: 228), while especially unemployment and poverty reduce health considerably (Lundberg et al. 2010; O’Campo et al. 2015). The welfare state is intimately linked to such inequalities. Therefore, it is clear that moderating effects of social policies on health outcomes exist because they shape the risk of ill health associated with individual vulnerability in different ways. The relationship between social inequality and health is sometimes also described as being mediated by social policies (Beckfield et al. 2015: 228; O’Campo et al. 2015: 88). It is my understanding that the described effects refer rather to a moderating function of the welfare state, instead of a mediating one.⁹ Instead of the welfare state, social determinants may be mediators, as they are shaped by social policies (stratification function) and in turn impact health outcomes. In addition to moderating and mediating effects, there can also be a direct impact of social policies in general and health policies in particular on individual health outcomes. However, such effects are not as prominent in the literature. Still, features such as access to medical services (e.g. Kim et al. 2017) can account for cross-national differences in health outcomes.

While links between features of welfare states and health outcomes appear in empirical correlations, they still do not reveal how and why a relationship exists. For the purpose of identifying such possible mechanisms behind social policies on health outcomes, a contribution by Beckfield and colleagues (2015) is noteworthy. The authors propose an institutional theory of welfare state effects on health because—as they point out—theories of health inequalities have focussed mainly on micro-level explanations of health outcomes,¹⁰ while an institutional perspective is missing. Overall, they highlight four “mechanisms” (Beckfield et al. 2015: 232), which have to be distinguished when examining the impact of welfare states on individual health and health inequalities: (1) redistribution, (2) compression, (3) mediation and (4) imbrication.¹¹

Redistribution refers to different modes of shifting resources to those in need. While the authors only discuss this in relation to the redistribution of resources between differently situated groups, this could also refer to the redistribution of resources over the life-course. This mechanism clearly corresponds to the redistributive function of welfare states described in section 2.2.

The mechanism of compression describes upper and lower bounds in the social gradient of health set by institutional efforts (such as a minimum wage). In addition, welfare states regulate the minimum for the provision of healthcare (or other) benefits, which is manifested in health-care access, benefit levels, and generosity, among other things. This process appears to be closely related to the responsiveness of a welfare state. In this sense, it seems to be largely linked to what was previously termed the security function of minimising risks and meeting needs. As Högberg and colleagues (2018: 314) add, benefit levels can even be beneficial if they are not used. They represent potential resources that are understood in the subjective perception as a protection against possible risks in the future, which reduces uncertainty. However, this should relate more strongly to subjective (psychosocial) well-being (Sjöberg 2010) than health outcomes. Still, as discussed in the following section, psychosocial stress is a predictor of morbidity and reducing such stress may therefore be a relevant moderating effect.

Mediation means a process in which institutional aspects shape a third variable, which in turn has an impact on health outcomes. More specifically, Beckfield and colleagues (2015) refer to the welfare state as a mediator in an indirect effect of socio-economic indicators on health. As noted before, it seems only partly convincing to refer to the welfare state as a mediator from a macro-micro perspective since it seems implausible that social policies are directly shaped by individual-level social inequality. It can of course be a moderator, shaping the impact of another variable (such as socio-economic status or unemployment) on the outcome. Such a moderating effect is often associated with the reduction of psychosocial stress as a morbidity-increasing factor (O'Campo et al. 2015: 89). Lowering stress by reducing uncertainty and compensating for (e.g. financial) deficits again corresponds to the security function. While such moderating effects of welfare states are more prominent in the literature, mediating effects can exist. Indicators of social inequality on the individual level might in fact be mediators in the sense that the impact of social policies on health is an indirect effect which is mediated by vulnerability of individuals. The link between welfare state and individual social status would thus represent the function of shaping patterns of social stratification and inequality, while the moderating role of social policies is a manifestation of the security function. Concluding, I suggest to add moderation to the proposal of Beckfield and colleagues, and in the following, I will discuss the welfare state as a moderator on the one hand and its impact on the outcome as a possible indirect effect, which is mediated by other features (especially related to social inequality) on the other hand.

Finally, the concept of imbrication refers to the simultaneous and overlapping effects of different institutions. While it seems evident that policies in the field of health care have an impact on health outcomes, other areas within the welfare state (and beyond) can be influential as well. In this sense, institutions can amplify each other’s impact; they can moderate it or undermine it. Taking this argument further, it seems prudent to consider other areas outside of the healthcare system as well, simply because they may be more relevant. To take examples from the literature: if social inequality is conceptualised as a mediator between welfare state and health outcomes, the health system may have a different and perhaps even less influential effect than—for instance—pension policies (Högberg et al. 2018), unemployment insurance (Renahy et al. 2018) or active labour market policies (Voßemer et al. 2018). Moreover, it may be worth discussing, whether imbrication does not only addresses overlapping policy areas, but also overlapping policy manifestations. An example of this could be empirical evidence finding an interplay of indicators such as insurance coverage and benefit generosity, suggesting that high replacement rates indicate well-functioning security only when combined with high insurance coverage (Ferrarini et al. 2014b). Imbrication thus raises many conceptual questions. As a mechanism, it seems most closely tied to the security function, as it addresses how closely knit a securing net is.

Figure 4.2

The impact of the welfare state on health outcomes

Figure 4.2 summarises possible impacts of the welfare state on health outcome and the interaction with social determinants of health. The first part of the graph summarises all possible effects (A). Here, a transformational process (final micro-macro link) is included. Since I focus on micro-level manifestations of health, it will not receive further attention. The second figure (B) illustrates direct effects of the welfare state on health outcomes. Such effects include what was described as welfare state’s primary function of providing security (cf. section 2.2.1). C and D in Figure 4.2 illustrate the moderating effect of social policies and the indirect effect in which social determinants serve as mediators.

Summarising potential mechanisms based on the proposal by Beckfield and colleagues as well as the functions of welfare states discussed in the second chapter of this book, several mechanisms can be at play when explaining the impact of welfare states on health outcomes and health inequalities. Health outcomes are shaped directly by the welfare state in its function of securing individuals against the risk of bad health by redistributing resources (e.g. in order to improve access to services of insurance coverage). In contrast, health inequality is shaped mainly in a non-direct way through mediation or moderation. Different outcomes can be explained by different patterns of social inequality (stratification) and different effectiveness of securing against the consequences of vulnerability (security). Regulations of benefits (compression) and the interplay between different areas of social policy-making (imbrication) are underlying those other mechanisms.

The second step in identifying different conceptualisations of the welfare state and how it works requires a closer look at the components of welfare states on which research is focussed. In the relevant literature, several perspectives are distinguishable. Following a proposal by Dahl and van der Wel (2013), three approaches to grasping welfare states can be identified that are especially relevant for empirical operationalisations. The first focusses on regime affiliation, the second on expenditure and the third on the institutional organisation (also Bergqvist et al. 2013; Ferrarini et al. 2014a). These three approaches already hint at specific operationalisations such as typologies, indicators of effort, generosity and more (as discussed in previous chapters). Apart from offering a guideline for the selection of indicators, these different approaches reveal different perspectives on the welfare state. While regime typologies highlight broad policy principles, expenditure addresses the size of the budget and institutional arrangements direct the focus on eligibility criteria, generosity and population coverage.¹² The expenditure and regime perspectives both suffer from the problems discussed in chapter 3. Both approaches lack clarity, as they are too broad in the case of regimes and too fuzzy in the case of expenditure. Depending on the conceptualisation, the expenditure approach is sometimes used to capture the “concept of ‘welfare resources’” (Dahl & van der Wel 2013: 61). Even though differentiating between areas of spending helps against the ambiguity of an overall spending indicator, it is still not entirely clear, whether those resources result from a generous welfare agenda or a high number of recipients. As a result, out of the three options the institutional approach is viewed as best suited to explore health outcomes and especially the social gradient of health inequality (Ferrarini et al. 2014a: 635–636; Lundberg et al. 2015: S. 32). Pinpointing these three fundamental perspectives on welfare states illustrates the importance of choosing an explicit conceptualisation of welfare stateness for a specific research question. The differentiation between expenditure, institutional and regime perspectives reveals distinct operationalisations, which are by no means identical in nature and therefore allow very different interpretations.

After discussing through which mechanisms the welfare state can influence health outcomes and how different perspectives on the welfare state may reveal distinct conceptualisations, the third and last step explores why outcomes are expected to differ based on particular policies. Even though the term mechanism has been adopted from the literature discussed, the actual causal processes that explain which social policies produce which outcomes and why this might be the case often remain somewhat elusive. In order to substantiate the processes, direct, moderating and indirect effects are now discussed with the help of key hypotheses on how and why the welfare state shapes health outcomes. Summarising prominent hypotheses is complicated by the fact that different studies highlight different aspects within the relationship between social policies and health. To illustrate the diversity of perspectives and arguments in the literature, Table 4.1 gives an overview of hypotheses in the papers sampled for this review. I mainly includes empirical studies and do not list hypotheses twice if several authors assumed the same exact effect. Evidently, this is only a selection and there are most likely many other hypotheses, which are not acknowledged. However, it serves to make an important point: even though the literature on health outcomes and health inequalities discusses the identification of relevant mechanisms and the standardisation of indicators more elaborately than other fields, the tested assumptions still cover a variety of different perspectives on the welfare state.

Table 4.1 Welfare states and health—summary of hypotheses

The selection of hypotheses presented in Table 4.1, illustrates a variety of issues. First, hypotheses often include an implicit (and sometimes explicit) reference to specific perspectives on the welfare state (e.g. expenditure or generosity). What is not represented in the table but visible in the publications is that the reasons for choosing a specific perspective are not always entirely clear. While some contributions elaborate in much detail why they choose a specific area of the welfare state (e.g. Ferrarini et al. 2014a), others do not. Second, hypotheses frequently go beyond a focus on health policies. Instead, all areas of social policy-making are addressed. Thus, health is assumed to be shaped by welfare state activities in the field of health care as well as unemployment and pensions—either (in the sense of imbrication) because policies are intertwined or because they are assumed to have their own distinct effects on the outcome. Third, while moderations are addressed frequently, indirect effects seem to play a subordinate role. Fourth, while what I call the security function receives the most attention in empirical studies, social stratification is only of secondary importance and redistribution is not addressed in the selected hypotheses. A reason for this may be that identifying a distinct effect of redistribution (as cause) and disentangling it from effects of income inequality (as outcome) in empirical analyses is complicated and so far rather unsuccessful (Pickett & Wilkinson 2015: 322).

Lastly, the list of hypotheses—no matter how anecdotal—highlights how diverse assumptions about the relationship between welfare state, health outcomes and health inequalities are. Considering the discussed contributions share a very comparable conceptualisation of the dependent variable (either self-reported good health or bad health), this diversity is remarkable. In order to narrow down the hypotheses and combine them with mechanisms identified before, we can summarise key explanations. Those explanations particularly highlight the institutional perspective and the expenditure approach. Furthermore, the first explanation highlights the direct impact of social policies on health, while the moderating function is covered by the second explanation. Lastly, for the sake of completeness, the third explanation addresses an effect that is mediated through social inequality, although this one is not as prominently pursued in the literature so far.

• Explanation 1.1: Comprehensive social rights (including benefit coverage, generosity and eligibility criteria) and redistributive budget increase health because these actions secure against risks by redistributing resources to those who require them.

• Explanation 1.2: Comprehensive social rights (including benefit coverage, generosity and eligibility criteria) and redistributive budget decrease health inequalities because these actions secure those who are especially vulnerable.

• Explanation 1.3: Comprehensive universal social rights decrease health inequalities because these actions shape patterns of social stratification in ways that increase equality of opportunity thereby lowering risks among those who are especially vulnerable.

In addition, two recommendations regarding the empirical operationalisations can be distilled from the literature: regime approaches are not well equipped to explore general health and health inequalities, since they are too crude to determine whether an empirical result actually captures an assumed causal effect or mechanism. In a similar manner, social expenditure is an ambiguous indicator. Its prevailing presence and importance is most likely a result of data availability. This should still inspire caution. Both recommendations relate to the critical discussion in the previous chapter (cf. chapter 3) and they should be transferable to other instances where welfare stateness is assumed to shape individual-level outcomes.

Although, literature on the link between welfare states and health outcomes and inequality is quite heterogeneous, several aspects can be systematically deduced. In terms of the mechanisms behind the relationship between social policy and health, there is significant overlap between the functions of welfare states and the mechanisms introduced in the relevant literature. A synthesis is quite possible by combining different attempts of grasping how and why the welfare state shapes health outcomes on the individual level. Based on the discussion in this section, the functions of welfare states introduced in the second chapter of this contribution, actually capture relevant mechanisms quite well. In the case of health inequality different ways of providing security (including compression and mediation), shaping social stratification (including moderation) and organising redistribution explain differences in the impact of welfare states on individual health. Overlapping effects of institutions (imbrication) can be subsumed under security as well. If the functions of welfare states actually do represent mechanisms explaining why welfare states shape individual outcomes, they should fit other dependent variables as well.

All of these mechanisms relate to a similar meta-perspective on the welfare state. Studies investigating well-being in general and health in particular mostly conceptualise the welfare state from a top-down perspective: it is assumed that the nature of social policies is responsible for an outcome or the moderation of another effect, instead of—for instance—individual perceptions of such policies. This common feature can be helpful in identifying which operationalisations of welfare stateness better fit the mechanisms discussed in this chapter.

Before turning to other dependent variables, it is important to note that while measures of health receive much attention in the literature, other manifestations of well-being are equally important—in particular subjective well-being. There is however some disagreement regarding their measurement.¹³ Regardless of the operationalisation, many assumptions about the relationship between social policies and well-being are similar to those addressing health outcomes and they are often examined following the same theoretical premises (Voßemer et al. 2018). Even though I focus on hypotheses and mechanisms referring to the link between welfare state and health outcomes, many considerations are considered transferable to the field of subjective well-being as well. Others however may not. If subjective well-being is more strongly linked to the individual perception of the welfare state—for example psychosocial stress is reduced if the welfare state is perceived as a good provider of security—this does not necessarily correspond to the actual responsiveness of welfare states. Because individual perceptions can differ from actual policy-making, a focus on subjective manifestations of well-being can conceptualise the welfare state not only from a top-down but also from a bottom-up perspective, in which it is not the main actor, but merely subject to an individual assessment. This may fall under mechanisms, labelled ‘psycho-social’ by Molnar and colleagues (2015: 4). Here, emphasis lies on psychological and cognitive processes, interaction with social environments and more. While these thoughts are not explored further in the case of well-being, similar arguments will be taken up again in the course of this chapter, when addressing the relationship between welfare states and attitude formation and behaviour.

4.2.2 The Impact of the Welfare State on Risks and Needs

The most obvious areas affected by the welfare state are related to risks and needs. Considering that lowering risks and responding to needs is one of the key functions of social policies, it does not surprise that a considerable amount of research is devoted to exploring the relationship between the welfare state and individual risk. Especially comparative studies on the matter usually ask in what way—and often also how well—different social policies lower risks and respond to needs.

As noted before (section 2.2.1), poverty represents the most elementary risk, the welfare state responds to (Saunders 2010). Therefore, it is one of the most common dependent variables in this line of literature. From a comparative perspective, the basic research interest is to explore the extent to which social policies account for cross-national variations in individual poverty (for a review focussing on unemployment policies see O’Campo et al. 2015). Albeit very popular and of course evident, poverty is by far not the only risk explored in the relevant literature. Other studies for instance investigate indebtedness (e.g. Angel & Heitzmann 2015) and (self-reported) economic deprivation¹⁴ (e.g. Visser et al. 2014). While the focus is often on risks and thus the likelihood of an emergency (such as poverty) occuring, deprivation is an example of a research topic in which needs are focussed. Still, both aspects are strongly intertwined as the need that the welfare state responds to originates from risks an individual is exposed to. As in the previous part of this chapter, I will focus on one manifestation of risk in particular. Since it is a major variable in this line of research, I highlight the relationship between social policies and poverty.

Empirical analyses conceptualise poverty in several ways. We can first distinguish between short-term and long-term poverty as well as absolute and relative poverty (Molnar et al. 2015: 4).¹⁵ In addition, poverty among individuals, which are not part of the labour force is often distinguished from in-work poverty (cf. Andress & Lohmann 2008). In most cases, poverty is conceptualised as relative poverty and measured through income. A person is considered poor if a household falls below a threshold of either 50 (e.g. Brady et al. 2017) or 60 percent (e.g. Polin & Raitano 2014) of a country’s median equivalised disposable income.

The previous section of this chapter already hinted at some overlap between explanations for morbidity and other risks. Thus, testing whether the mechanisms and conceptual perspectives identified in the case of health outcomes are transferrable to poverty as an explained outcome seems valuable and is further encouraged by a variety of contributions addressing the link between social policies and poverty and health simultaneously (Molnar et al. 2015; O’Campo et al. 2015). Therefore, the following discussion is structured analogously to the previous section of this chapter. I begin by describing the ways in which welfare states can affect individual poverty and discuss the underlying mechanisms. This is followed by a brief summary of the various features of social policies highlighted in the literature, before summarising related hypotheses.

Again, the welfare state is linked to poverty through direct as well as indirect and moderating effects. Similarly to the case of health outcomes, non-direct effects address how social policies interact with micro-level determinants of poverty either by moderating their influence or by reducing (or increasing) their occurrence in case of mediated effects. Overall, there is a number of relevant individual-level predictors of poverty, which are often referred to as penalties. Brady and colleagues (2017: 742) highlight four major socioeconomic determinants increasing the risk of poverty: low education, single motherhood, young household headship and unemployment. In case of mediated effects, welfare policies thus reduce the occurrence of such (and more) risk situations, while moderating effects shape how strongly they increase the risk of poverty. An example for a moderating effect of social policies is given by studies examining the risk of entering poverty as a consequence of childbirth (e.g. Barbieri & Bozzon 2016). In addition to such mediated or moderating effects, there can also be direct effects of welfare policies on poverty risk. Since poverty is closely related to income and most commonly measured with reference to income, such direct links seem almost tautological—especially when focussing on policies regulating income replacement (e.g. Lohmann 2009) and generosity of redistributive effort (e.g. Saltkjel & Malmberg-Heimonen 2017).

Figure 4.3

The impact of the welfare state on the risk of poverty

Turning to mechanisms behind such direct, mediated and moderating effects, again reveals many similarities to the previous discussion of welfare states’ impact on health (summarised in Figure  4.3). What Molnar and colleagues (2015: 3–4) summarise as ‘materialist’ mechanisms is of particular importance. The main explanation here is that individuals are secured against risk by replacing deficits through the redistribution of resources between individuals and across lifecycles. Furthermore, the provision of security is tied to determinants of poverty at the individual level, as it potentially moderates the impact of being at risk. Following the redistribution paradox (cf. section 2.2.2), targeted welfare states may respond more effectively to occurring risks and needs for the lower social classes, but universal welfare states generate more public support and thus a greater budget for redistribution. They are therefore more advantageous for middle and higher income classes but the increased redistributive budget leads to more effective overall poverty reduction (Korpi & Palme 1998).

Like in the case of health outcomes, compression can be seen as a subordinate mechanism underlying the security function. It is setting the upper and lower bounds in the provision of benefits and services that secure against poverty, while mediation (and moderation) relates to the relationship between welfare states and social stratification. Turning to the mechanism of imbrication, the relationship between welfare state and poverty again suggests that many fields of policy-making are relevant and amplifying and overlapping effects may occur. For instance, while unemployment policies are considered very important predictors of poverty (Molnar et al. 2015), they are still embedded in a context in which such effects are shaped by other policy measures as well. An example for this is the suggestion that generous unemployment benefits may even increase poverty as they can be a disincentive to find a job among unemployed individuals in some contexts (O'Campo et al. 2015: 92). Besides hinting at potential overlap and interaction between different policies, this reveals a potential additional mechanism to the ones described before. This additional mechanism relates to what was termed the function of activation in chapter 2. Especially active labour market policies (e.g. Clasen et al. 2016), which help and incentivise re-entering employment, aim at reducing risks. While their overall effectiveness is contested (cf. section 2.2.4), evidence suggests they are indeed able to reduce poverty levels (e.g. Cronert & Palme 2017). Thus, investment strategies of welfare states potentially bear a relevant explanation for different poverty outcomes and their relationship to unemployment and other situations of vulnerability—especially as a moderating institution. Differentiating between security and activation in such moderations, represents the difference between a protection strategy and an investment strategy (cf. Vandenbroucke & Vleminckx 2011).

After exploring mechanisms, which serve as explanations for the relationship between social policies and cross-national differences in prevalence of individual poverty on the one hand and penalties associated with risk situations on the other, the next step entails exploring which components of the welfare state are highlighted in relevant contributions on the matter. Here too, a distinction must be made between contributions that highlight welfare state effort (expenditure) and those focussing on the overarching nature of welfare states either by using regime typologies (e.g. Whelan & Maitre 2010) or by emphasising targeted versus universal premises (e.g. Brady & Bostic 2015). These perspectives overlap considerably, for example in studies using social expenditure as an indicator of universalism in welfare states (e.g. Zwiers & Koster 2015). While these two perspectives emphasise what was labelled expenditure and regime approach before, we also find institutional perspectives on the welfare state. Again, the latter include a focus on eligibility criteria (as discussed by O’Campo et al. 2015), benefit generosity (e.g. Brady et al. 2017), and logic of redistribution (e.g. Jacques & Noël 2018).

One last thing should be noted when it comes to perspectives on the welfare state and its impact on individual poverty. A focus on how welfare states meet old risks associated with the labour market and economy provides only a partial explanation for individual poverty. While it lowers penalties resulting from unemployment and low education, individuals affected by the ‘new risks’ (cf. section 2.1.3) can slip through the net (Brady et al. 2017: 771). Hence, the effectiveness of overall poverty reduction should be a function of how well welfare states adapt (Brady & Bostic 2015: 272). Therefore, examining the role of social policies in poverty related to new risks (e.g. single motherhood or youth unemployment) may require additional perspectives on the welfare state. This could include, for instance, a focus on social investment and active labour market policies.

To take a closer look at the application of the summarised mechanisms and perspectives, a brief summary of hypotheses helps to systematise how and why social policy and individual poverty are related. Analogously to the previous section on health outcomes, Table 4.2 provides a fragmentary listing of hypotheses. Again, this is not intended as an exhaustive review of all existing hypothesis, but as an anecdotic illustration of heterogeneity. Indeed, it shows that the assumptions regarding explanations for differing poverty risks are divers, as it is assumed that very different policy areas influence poverty risks, with particular attention being paid to unemployment-related policies (the link between unemployment insurance and poverty is covered in a comprehensive review by O’Campo et al. 2015).

Table 4.2 Welfare states and the risk of poverty—summary of hypotheses

We can summarise several main explanations for differing poverty levels. While the expectations regarding moderating effects and methodological issues are similar to those formulated in the case of health inequality (section 4.2.1), the first explanation—highlighting the direct impact of social policies on poverty—appears to be especially popular. This is to be expected considering the close link between income deficits and measures replacing income. Still, a moderating effect (2.2) and a mediated one (2.3) are present. In addition, a fourth explanation (2.4) captures the activating mechanisms expressed in those policies, which aim at enabling individuals to leave vulnerable situations such as unemployment.

• Explanation 2.1: Comprehensive social rights (including benefit coverage, generosity and eligibility criteria) and redistributive budget decrease poverty because these actions secure against risks by redistributing resources to those who require them.

• Explanation 2.2: Comprehensive social rights (including benefit coverage, generosity and eligibility criteria) and redistributive budget decrease poverty because these actions secure those who are especially vulnerable.

• Explanation 2.3: Comprehensive universal social rights decrease poverty because these actions shape patterns of social stratifications in a way that increases equal opportunities thereby lowering risks among those who are especially vulnerable.

• Explanation 2.4: Activating policies aiming at ending periods of vulnerability (such as unemployment) decrease poverty because these actions increase labour market participation through activation.

The main purpose of this chapter is to identify different perspectives on the welfare state, which underlie the central assumptions about its functioning and explain the different distribution of outcomes between countries. In the case of poverty, this perspective bears much resemblance to the one explaining different health outcomes and health inequality (cf. section 4.2.1). The welfare state is again conceptualised from a top-down perspective. As such, it is seen as an institution, which actively shapes individual outcomes through several intertwined mechanisms. A main mechanism is to secure individuals from the occurrence of poverty in the first place (security) through income replacement or other benefits and services. This is achieved through the redistribution of resources on the one hand, and through enabling individuals to overcome situations in which they are at risk (activation) on the other hand. The latter can obviously also be funded and organised by shifting resources towards active labour market policies and other activating measures. In many of these cases, the welfare state does not directly affect individual risks, but indirectly through its impact on patterns of social inequality (social stratification) and its potential to moderate the outcomes of social determinants of risk. Again, this shows that we cannot identify one singular mechanism responsible for the welfare state’s impact on individual risks. Instead, various mechanisms are at work simultaneously. Each of the explanations described can be singled out and examined in more detail—as can the combination of different mechanism.

4.2.3 The Impact of the Welfare State on Attitudes

Numerous publications deal with the effects of social policy on attitudes.¹⁶ Most of the contributions in this field deal with attitudes directly related to the welfare state, such as evaluations of social justice (e.g. Arts & Gelissen 2001), support for social policies (e.g. Brady & Finnigan 2014), general social policy preferences (e.g. Steele 2015; Breznau et al. 2019), welfare chauvinism (e.g. Cappelen & Peters 2018) or preferred spending in the field of social policy-making (e.g. van de Walle & Jilke 2014). I will focus on such welfare attitudes in this section, as they are not only the most extensively discussed, but also stimulate the more sophisticated hypotheses about how and why different welfare states are responsible for the different distribution of attitudes between countries.

The empirical measurement of welfare attitudes is diverse and allows for different foci besides general welfare support. They range from principles of justice to ideal redistribution principles to welfare chauvinism. Furthermore, studies of attitudes towards the overall welfare state have to be distinguished from those focussing on attitudes towards specific policy areas (such as attitudes towards public childcare provision, Chung & Meuleman 2017). This lack of agreement on how to measure welfare attitudes is attracting increasing attention (e.g. Svallfors 2012b: 9). In many cases, different aspects are emphasised in the dependent variable, although general ideas about social justice, for example, differ considerably from the assessment of how well the welfare state functions in one’s own country. There are several contributions that are sensitive to the issue of operationalising welfare attitudes, such as the ones bundled in Svallfors edited volume “Contested Welfare States” (Svallfors 2012a). Although my focus is on the macro-micro mechanisms linking the welfare state to attitude formation, this issue is at least noteworthy. Since this macro-micro-link produced mixed results (van Oorschot & Meuleman 2012: 27; Deeming 2018), the question arises whether this is related to different operationalisations of the dependent as well as of independent variables. The latter will receive more attention in this book, while keeping the former in mind. Since the aim of this section is the same as in the previous ones—to identify mechanisms, shared conceptualisations and common hypotheses—the discussion follows the accustomed structure.

In general, welfare attitudes are formed “reflecting a mix of ideology and interest” (Brady & Finnigan 2014: 21). They are shaped both by individual-level characteristics such as socioeconomic status and by contextual influences of the welfare state. The welfare state can logically be linked to attitudes directly, indirectly and as a moderator—although not all of these links are examined with equal prominence. At the micro-level, socio-demographic indicators (such as age, class position and gender) as well as political orientation and party preferences (e.g. Koster 2014) and value preferences (e.g. Kulin & Meuleman 2015) are considered relevant influences. Based on these characteristics, it is assumed that individuals have different perceptions of deservingness on the one hand and different self-interests on the other (as summarised by Deeming 2018: 1107). Here the welfare state serves as a potential moderator, shaping the effect of socioeconomic position on attitude formation. Alternatively, the effect of social policies can also be mediated—for example by individual perceptions of performance (van Oorschot & Meuleman 2012: 26). Another mediated effect could result from the fact that welfare states influence individual-level variations in social status, which in turn explains different levels of satisfaction with democracy (e.g. Sirovátka et al. 2018). Finally, direct links between welfare state and welfare attitudes can be found in the cultural imprint welfare states leave on citizens. Following the argument that welfare states represent institutionalised ideas about social justice (e.g. Sachweh 2016), they have the power to shape collective notions of norms and principles (Arts & Gelissen 2001). These processes have been described as the function of socialisation before (section 2.2.5).

The latter already leads to the question of mechanisms. In the case of attitude formation, the direct effect of welfare stateness can be explained, on the one hand, by the responsiveness (security) and the redistributive logic (redistribution) that generate political support for the welfare state. On the other hand, the function of socialisation plays an important part when explaining different attitudes. In the sense described above, the socialising function results from the welfare state as a cultural context in which individuals are embedded. Following the previously discussed processes (cf. section 2.2.5), welfare states cultivate dominant normative frames, which are adopted by individuals or at least shape them (Arts & Gelissen 2001; Pfau-Effinger 2005). The distinction between socialisation and the other mentioned mechanisms corresponds to what is differentiated by Sirovátka and colleagues as the “level of the protection and redistributive effect” and the “formative effect” (2018: 3). In addition to such direct effects, the welfare state and especially the socialising mechanism can also be seen as a moderator, shaping how social status is expressed in attitudes. Although plausible, contributions modelling such a relationship are rare. The same is true for the argument that welfare states shape socioeconomic status, which in turn serves as a mediator between social policies and attitudes. Although such effects, which highlight the mechanism of social stratification, are discussed (Sirovátka et al. 2018: 4), usually only one of the paths (welfare state—stratification or stratification—attitude) is considered further.

While these explanations for the relationship between welfare state and welfare attitudes follow the same top-down perspective encountered in the case of well-being and risks, this perspective is not sufficient in the case of attitudes. In addition to focussing on how welfare states impact individuals, an important question here is, how individuals perceive the welfare state. These perceptions are relevant for explaining indirect effects in which individuals’ views on social policies mediate an effect of actual policies as well as direct effects of social policies on the process of attitude formation. It is important to distinguish such a bottom-up perspective from the top-down perspective pursued so far. While the mechanisms explaining how and why welfare states shape individual outcomes correspond to the functions of the welfare state discussed before, the question how and why individuals perceive welfare state performance in a particular way requires additional insights. Especially self-interest offers a potential answer for the causal mechanism behind the relationship between welfare state and attitude. As Sachweh and Olafsdottir (2012: 151) point out, individuals evaluate, how likely and how strongly they profit personally from welfare state efforts and base their preferences for redistribution on such evaluations. In addition, evaluating how likely one could be at risk may influence such considerations as well (Iversen & Soskice 2001). However, self-interest is not the only driver of policy-driven attitude-formation. Instead, individual evaluations of welfare state performance can be inspired by other aspects such as ideology (Petersen et al. 2011; Svallfors et al. 2012; Chung & Meuleman 2017). If a welfare state perceived as guided by egalitarian principles meets egalitarian values at the individual level, this coherence may even amplify support (Calzada et al. 2014: 187). I will refer to such processes of perceiving and assessing the welfare state without necessarily doing so based on self-interest as evaluation.

The two different perspectives on the relationship between welfare state and welfare attitudes are illustrated in Figure 4.4. The top-down perspective (A) is characterised by similar pathways as in the previous sections: the welfare state is the main actor and several mechanisms are highlighted. The link between social policies and individual determinants is explained mainly by the stratifying effect and could to some extent also include securing against the occurrence of risk. Direct effects can be explained by socialisation and responsiveness (as indicated by how well welfare states provide security and how they redistribute). Lastly, the moderating effect could be characterised by similar mechanisms as the direct effect. Since it is not as prominently examined, it does not receive as much attention as the other paths (as indicated by the dotted lines). Evidently, the mechanisms overlap and this summary highlights relevant explanations (while others can be plausible as well). Most importantly, these mechanisms illustrate the described top-down influence of welfare states. Thus, impact of policies on attitudes does not necessarily have to be salient.

Figure 4.4

The impact of the welfare state on welfare attitudes

In contrast, where the focus rests on the individual perception and evaluation of social policies (B), the individual is assumed to have at least partial information about actual policy-making. Such information is used when evaluating social policies. Hence, the impact of the welfare state is filtered through individual perception. While both perspectives are equally plausible and coincide, by distinguishing between the two, I would like to emphasise that each corresponds to a specific theoretical perspective on the matter. Following the discussion in chapter 3, this distinction is highly relevant for the question, which areas and components of the welfare state should be highlighted conceptually as well as empirically.¹⁷

Following these arguments, identifying relevant areas of welfare stateness in this case requires considering both perspectives. Starting with the top-down perspective, relevant areas of welfare stateness can be found in literature on the formation of public preferences. As Breznau (2017: 584–585) summarises, one perspective (the thermostatic model) focusses on expenditure, while a second perspective (the increasing returns model) emphasises entitlements and generosity, and thus what falls under the institutional component of the welfare state mentioned earlier. Studies examining the extent to which welfare states have a socialising effect on individuals sometimes use a regime approach (Arts & Gelissen 2001). However, more recent contributions of this kind produce mixed findings (van Oorschot & Meuleman 2012: 27). Overall, using a selection of distinct characteristics, representing either spending or generosity is preferred (Jæger 2006). This is supported by my own analyses conducted in chapter 3. A further component highlighted in this context is the degree of egalitarianism in a welfare state. This can for instance manifest in egalitarian gender norms (Chung & Meuleman 2017: 54). Since gender ideologies and family policies are losely linked (Grunow et al. 2018), such cultural perspective on the welfare state are highly relevant.

While the above-summarised perspective on welfare stateness is familiar, as it corresponds to what has been discussed in the two previous parts of this chapter, capturing the bottom-up perspective adds an entirely new challenge. The literature agrees that individual perception of welfare state performance (Roosma et al. 2014) and the likeliness of ever being at risk and in need of welfare services are important aspects when explaining why and how individual attitude-formation takes place (Jordan 2013: 136). However, little is known about which components of social policies are relevant in this case. Instead, the above-discussed ways of approaching the welfare state (und ultimately measuring policy differences) are used. Following the discussion in chapter 3, too little seems to be known about which elements of welfare states are actually perceived—and if so correctly. After all, it is unlikely that individuals are fully and accurately informed about all aspects of social policy-making (Breznau 2013: 3). How problematic this can be, is illustrated in literature on welfare knowledge. This strand of research reveals that citizens perceive the reality of welfare provision in a distorted way. For the UK Taylor-Gooby and colleagues (2003) show that in particular spending for unemployed individuals is usually overestimated quite severely. Evidence from the US furthermore shows that the cost of replacement payments is overrated and that benefits are perceived to be more generous than they actually are (a good summary of main findings is given by Geiger 2018). Hence, it seems insufficient to rely on established approaches of capturing characteristics of welfare states if the bottom-up perspective is emphasised. Evaluation of policies in general and self-interest in particular, were identified as key mechanisms explaining how individuals look at the welfare state and why they form different opinions based on what they perceive. For this purpose, it seems highly relevant that the emphasised components within the welfare state represent salient features. As (Sachweh 2018: 50–51) points out, especially when confronted with crisis, it is likely to expect that indicators more closely related to security are more salient than those representing patterns of inequality. Furthermore, it seems plausible to assume that benefit generosity is related to perceptions guided by self-interest, as they represent potential benefits. Other indicators of eligibility such as benefit duration, waiting periods and overall insurance coverage may be of secondary importance, as they require very specific knowledge about benefit provision. These considerations, which go beyond the approaches highlighted in the literature, will be revisited in the next chapter when it comes to elaborating different conceptualisations and their empirical operationalisation. For now, suffice it to summarise that the link between welfare states and individual attitudes can and should be approached from two sides—a bottom-up perspective highlighting the individual process of attitude formation and a top-down perspective emphasising the welfare states impact on shaping the context in which attitudes are formed. Both perspectives can be combined well in one explanatory model, if both kinds of processes are examined.

Hypotheses in the literature on welfare attitudes are characterised by a variety of different approaches to modelling the relationship between welfare state and attitudes. Summarising hypotheses in this case is complicated because not only do perspectives on the welfare state vary, but so do operationalisation of welfare attitudes. Therefore, I refrain from presenting hypotheses anecdotally and instead turn directly to summarising the relevant assumptions. Overall, universalism is expected to be among the main determinants of public support for redistribution (cf. section 2.2.2 and the paradox of redistribution) as well as support for social policies in general (e.g. Jordan 2013). The same can be said for the generosity and size of welfare states, which one may call their performance. In both cases, the explanation for such relationships may lie in both the socialising effect and the active evaluation by the individual. Thus, these relationships can be approached from the top-down as well as the bottom-up perspective. Summarising the literature presented in this section of the chapter, the following explanations for how social policy influences attitudes can be derived from the discussion.

• Explanation 3.1: Egalitarianism, universalism and comprehensive social rights (including benefit coverage, generosity and eligibility criteria) lead to positive attitudes towards the welfare state because individuals are socialised corresponding to such principles of solidarity and justice (top-down perspective).

• Explanation 3.2: Comprehensive social rights (including benefit coverage, generosity and eligibility criteria) lead to positive attitudes towards the welfare state because they provide security and thus increase political support (top-down perspective).

• Explanation 3.3: Perceived fairness and good performance lead to more positive attitudes towards the welfare state because individuals evaluate these actions positively (bottom-up perspective).

• Explanation 3.4: Expected personal benefits from policy-making lead to positive attitudes towards the welfare state because this is line with self-interest (bottom-up perspective).

The impact of social policies on attitudes towards the welfare state raises more complex issues than those that arise in the case of poverty and well-being. While the latter are closely related to responsiveness of welfare states and mostly address top-down perspectives, the process of attitude formation is less straightforward in many ways. Not only does it conceptualise individuals as actors of their own, but it also highlights more intricate processes. Conceptualising und ultimately operationalising mechanisms related to socialisation or self-interest is much more challenging than in the case of responsiveness in situations of risk.

While attitudes towards welfare state policies are covered by the majority of literature assuming an impact of social policies on attitudes, there are notable further dependent variables addressing attitudinal phenomena. These include general political attitudes, such as satisfaction with democracy (e.g. Sirovátka et al. 2018) and support for Europe (e.g. Beaudonnet 2015). In addition, there are miscellaneous other attitudes explored, including attitudes towards migration (e.g. Rapp 2017) and self-employment (e.g. Rapp et al. 2017). Essentially, hypotheses tying welfare stateness to all three groups of dependent variables—welfare attitudes, political attitudes and miscellaneous other attitudes—are quite similar at their core. They all can focus on a bottom-up perspective or a top-down perspective (or both). Therefore, methodological issues arising from this distinction can be considered transferrable.

4.2.4 The Impact of the Welfare State on Behaviour

Various behaviours are shaped by welfare states. The most evident ones are those tied to the labour market and the balance between paid employment and household labour (including care responsibilities). This includes objects of research such as absenteeism (Sjöberg 2017), the likelihood of entering risky endeavours such as self-employment (Rapp et al. 2017) as well as gendered division of household labour (Hook 2010) and labour market participation of fathers (Bünning & Pollmann-Schult 2016) and mothers (Gangl & Ziefle 2015). In order to discuss mechanisms and perspectives on the welfare state, I choose the latter as an example for two reasons. First, mothers’ labour market participation is a popular and widely covered topic. Second, it allows for the discussion of very different areas of welfare stateness than those explored so far, as it is more closely linked to family policies and thus contains aspects that have not received much attention in the last three sections of this chapter. Following the same structure as before, I will first turn to paths and mechanisms, before elaborating on relevant facets of welfare stateness and summarising main hypotheses and explanations in the literature.

Mothers’ labour market participation can be categorised in the field of new risks and risk groups (cf. section 2.1.3). As Chung & Meuleman (2017: 52) point out, responses to such new risks and new risk groups may not fit mechanisms developed with reference to the old risks. So far, I have indeed focussed on classical risks tied to benefits and services that are provided if unemployment, sickness or old age threaten income. In explaining behaviour related to the labour market and gender issues in particular, it is necessary to consider not only the classical functions of the welfare state but also more recent policy agendas—such as the increasing promotion of female employment in general and maternal employment in particular. When researching labour market participation of mothers empirically, the existence of employment and its scope in terms of working hours are often differentiated (Erhel & Guergoat-Larivière 2013: 78). Welfare states are assumed to shape both outcomes in a similar manner.

The links between social policies and maternal labour market participation partly overlap with those discussed in the previous sections. Again, we are confronted with direct effects, as well as indirect and moderated ones. Furthermore, this type of behaviour as an outcome is tied to both essential perspectives on the welfare state as an explanatory factor: top-down and bottom-up. When the research question is why and how individuals respond in a particular way to social policies, a bottom-up perspective is taken. In contrast, studies exploring the impact of welfare policies on behaviour follow a top-down perspective.

Within the top-down perspective, direct effects can especially be found in policies that target a specific behaviour—in this case labour market participation of mothers. This is especially relevant in the case of activation. The design of work-family reconciliation policies enables equal participation in the labour market to varying degrees (Grunow et al. 2018: 42–43). Such activation can be directly tied to labour market participation—for instance in case of a general tendency to enable all women. In addition, it can also have a moderating effect if activation especially affects women in a particular situation (e.g. unemployment or single motherhood) or shapes the impact of other determinants such as the gendered division of household labour. It can also be argued that there is a link to the provision of security because increasing labour market participation of parents in general and mothers in particular also leads to economic benefits (Morrissey 2017: 3). Since this happens through activation, provision of security seems to be of secondary importance in this case. However, family-policies can also have an adverse effect. Expanding policies related to the length and generosity of parental leave may reduce maternal labour market participation (Gangl & Ziefle 2015: 520). For instance, generosity of financial support is therefore assumed to have a negative impact on mothers’ employment (Nieuwenhuis et al. 2012: 615). In sharp contrast to an activating impact, this can be referred to as a deactivation of mothers (cf. section 2.2.4).

Besides an activating component, the welfare state is tied to gendered outcomes through its impact on role models and gender ideologies (Grunow et al. 2018). In particular, work-family policies are shaped by both welfare culture and gender culture (Pfau-Effinger 2018: 170) and provide legitimised reference points for preference formation (cf. Pollmann-Schult 2016) as well as incentives for the dissemination of specific gender roles. On the one hand, they set certain norms about the relationship between employment and motherhood and, on the other hand, they determine how widespread caregiving is in a society. While the first mechanism is referred to as norm setting, the second is related to role exposure (Gangl & Ziefle 2015). It is my understanding that both explanations can be subsumed under socialisation because shaping what is perceived as a legitimised cultural frame is the driving mechanisms behind both. In this case, socialisation exhibits direct effects, as the dominant cultural norms regarding gender equality influence how natural, supported and even socially desirable female labour is. However, it can also be incorporated as moderating influence, which shapes the severity of individual-level determinants of female employment such as education, partnership status, age, but also preferences and more (Gangl & Ziefle 2015). Since at least the determinants related to the individual-level manifestations of social inequality are potentially shaped by social policies, indirect effects are conceivable as well. Such effects, where the impact of family policies is mediated by mothers’ characteristics, are not examined in the sampled studies (a similar conclusion is reached by Pollmann-Schult 2016: 24).

Turning to the bottom-up perspective, individual perception again represents an important link between policy-making and behaviour. In the case of behaviour, self-interest is a commonly assumed explanation for observed outcomes. Like in the case of attitudes, individual perceptions of social policy-making are assessed in terms of costs and benefits, resulting in a rational behavioural choice. Regarding labour market participation, this is based on the premise that having children negatively affects female employment. From an economic point of view, a major reason for this is higher relative income among men (Erhel & Guergoat-Larivière 2013: 77–78), which promotes the male breadwinner model. Thus, the more welfare states focus on compensating for income differences or incentives for parental leave among fathers (cf. Geisler & Kreyenfeld 2019), the higher maternal labour market participations is assumed to be.

Another reason for employment can simply be individual preferences although literature confirms a prevailing mismatch between preferred and actual working hours—especially among women and parents of children below school age (Steiber & Haas 2018). Welfare cultures may be responsible for the formation of such preferences, which are then weighted against the costs and benefits of labour market participation or are evaluated according to other principles (as in the case of attitude-formation). This is related to the effects of socialisation mentioned. In this bottom-up perspective, however, the difference is that normative reference frames not only leave an imprint on individuals, but are perceived and actively incorporated in a behavioural choice. Still, it is difficult to attribute findings like the fact that women in countries that are closer to the conservative ideal appear to prefer less working hours (Pollmann-Schult 2016) to either socialisation or an active assessment. It seems advisable for further research to explore which of the perspectives on the interplay between policies and behaviour is more accurate or if both are similarly at work.

A summary of the paths and mechanisms introduced briefly in this section is provided in Figure 4.5. From a top-down perspective (A), social policies in general and particularly family policies can serve as direct and as moderating effects. Here, activating mothers by incentivising them to work and relieving them of care responsibilities can increase labour market participation and it can shape the impact of social determinants. Similarly, the prevailing welfare culture may foster a social climate that can account for variations in maternal labour market participation across countries. In contrast, the bottom-up perspective includes individual perception as an intermediary premise (B). Here, behaviour is based on an assessment of social policies in a similar manner as in the case of attitude-formation. However, especially in the example of mothers’ employment, cost-benefit considerations play a more important role.

Figure 4.5

The impact of the welfare state on employment of mothers

Similarly to the three previous exemplary dependent variables, different areas of welfare stateness are highlighted in literature on the impact of policy-making on maternal employment. The main difference is that in the previous examples the focus was on managing old risks and the classical policy areas (mainly unemployment, sickness, disability and old age), whereas here the focus is on family policies. Accordingly, in sharp contrast to the previous examples, maternal labour market participation is a phenomenon that is assumed to be shaped mainly by very specific components of welfare states. Evidently, work-family policies are highly relevant and an impact of other policy areas (such as unemployment or health) is unlikely or marginal. While this is a very different emphasis, the way in which these policies are operationalised is quite similar to the other examples discussed in this chapter. Here as well, expenditure is among the approaches capturing differences between countries. In this case, expenses on childcare are highlighted (e.g. Andringa et al. 2015). Moreover, literature frequently focusses on institutional characteristics, which—again analogously to the previous examples—include work-family reconciliation policies; in particular childcare provision (Erhel & Guergoat-Larivière 2013; Morrissey 2017). In addition, eligibility criteria and other aspects that emphasise the level of defamilisation represent institutional characteristics (Keck & Saraceno 2012). Lastly, regime approaches do exist in this field of research. Usually, they take a more general perspective and examine patterns within policy-making (Saraceno & Keck 2011). As independent variables, they receive criticism in line with previously stated arguments in this book such as that they are too broad to determine, which specific aspects are at play (Nieuwenhuis et al. 2012: 615–616).

In summary, hypotheses about how and why family-policies shape female employment are strongly related to incentives provided top-down and cost-benefit considerations by the individual (bottom-up). An additional perspective is provided by contributions that emphasise cultural explanations and focus on the potential to foster egalitarian behaviour through policy-making that embodies an egalitarian principles. The following three explanations summarise the most popular hypotheses. Again, this list does not claim to be exhaustive and focusses on the most basic assumptions about the impact of policy-making on labour market participation of mothers.

• Explanation 4.1: Strong work-family reconciliation policies increase maternal labour market participation because they activate mothers by providing incentives, removing obstacles and compensating gender differences (top-down perspective).

• Explanation 4.2: Strong work-family reconciliation policies increase maternal labour market participation because they shape cultural reference frames in which individuals are socialised (top-down perspective).

• Explanation 4.3: Strong work-family reconciliation policies increase maternal labour market participation because employment is perceived and evaluated as more beneficial (bottom-up perspective).

The first two explanations could also be reformulated to include a moderating effect in which certain groups of mother (e.g. based on education or number of children) are particularly influenced by family-policies. Since such analyses are not prominent in the sampled studies, they are not listed here. Furthermore, the discussed deactivating effect that may result from policies that support women’s responsibility for childcare is not explicitly mentioned, as it is the counterpart to Explanation 4.1.

While labour market participation of mothers in particular and of women in general is a popular issue, other behaviours are shaped by welfare states as well. Several examples have been mentioned in the beginning of this section. In the examples chosen so far in this chapter—well-being, risks and attitudes—it was assumed that the summary of mechanisms and hypotheses for the exemplary dependent variables is at least partially transferable to many other research topics. As far as the employment of mothers is concerned, this is not so obvious, since in this case a very specific part of the welfare state is emphasised. Nevertheless, there is reason to believe that explanation such as activation and self-interest are also relevant to other research questions that examine the impact of policy-making on behaviour.

4.3 Summary: Analytical Perspectives and Mechanisms

The summary of dependent variables and hypotheses shows that there is a small number of distinct mechanisms and perspectives on the welfare state embedded in literature on the impact of social policies on individual-level outcome. However, these different perspectives are not necessarily tied to a specific dependent variable—we find similar hypotheses about the way in which the welfare state works across different outcomes.

The short review in this chapter reveals another noteworthy issue. Although theoretical assumptions are formulated in many fields about why and how the welfare state influences individual-level outcomes, this does not usually lead to a recommendation for specific operationalisations. For example, when responsiveness is identified as a key feature in explaining differences in poverty risk across countries, the empirical operationalisation of responsiveness still varies. This is true even in areas that seem to be sensitive to the issue, such as in the literature on health outcomes. While processes and mechanisms are laid out in much detail, they are still not operationalised in a standardised manner—even though proposals for such standardisation exist. This seems rather unsatisfactory and raises the question why it seems to be so difficult to agree on a set of operationalisations that fit certain hypotheses better than others. Following this thought, I would like to concretise this missing link.

Essentially, we can take away three things from this literature review. The first is a set of mechanisms on which the hypotheses discussed are based. These mechanisms are consistent with the main features and functions of welfare states (cf. chapter 2) and emphasise different modes of functioning. The second point we can take away is that there are two overarching perspectives on how the welfare state is addressed. One of them sees it as a top-down institution that actively shapes processes and outcomes at the individual level. The other approach is based on a bottom-up perspective, where the welfare state is a passive element perceived and judged by the individual. Lastly, these two elements—the general perspective on the welfare state and the mechanisms emphasised—are linked by theoretical and empirical conceptualisations that highlight specific areas that are assumed to be at work. This includes regime approaches as well as those that focus on expenditure (welfare effort) or institutional aspects (social rights in general and/or benefit generosity in particular). Benefit receipt as an additional perspective, does not appear to be a popular choice as an independent variable.

Depending on how one approaches these concepts, this can involve a variety of different indicators and operationalisations, which brings us back to the main objective and the second research question of this study: How can we derive a more standardised, transparent and comparable approach to operationalising the welfare state as an independent variable? I believe that the identified mechanisms and the two meta-perspectives on the welfare state can contribute very much to answering this question. The first step in finding the right indicators that capture those features of the welfare state that are essential for particular research questions and hypotheses should be to state clearly which mechanisms are assumed to be at work and which perspective on the welfare state is chosen. Before however we can approach explicit measurements, the welfare state has to be conceptualised as an explanans. This intermediate step could be the missing link between the existing hypotheses and measurements and has the potential to standardise the approach considerably, as explicitly modelling different perspectives on the welfare state should narrow down the choice of indicators in a meaningful way.

Table 4.3 summarises the main perspectives, mechanisms and associated processes identified in the literature review. This includes the distinction between top-down and bottom-up conceptualisations. Underlying these two higher order perspectives are at least¹⁸ seven mechanisms. The mechanisms framing the welfare state from a top-down perspective correspond to the functions of the welfare state discussed in the beginning of this book. Here hypotheses posit that social policies have a distinct influence on the examined outcomes. These effects can be intentional (as in the case of securing against risks), unintentional (as often in the case of socialisation of individuals), or both (in the case of activation). Empirically, this presupposes that the features of the welfare state are measured against objectively relevant criteria and that its effects can be invisible to the individual. Furthermore, two mechanisms—stratification and redistribution—are theoretically relevant, but it proved difficult to include them in the discussion. This is most likely due to the exemplary dependent variables chosen, where the influence of stratification and redistribution is only relevant as part of the provision of security. At this point, it is not possible to determine whether this is always the case or if the two provide important explanations for other dependent variables. They therefore remain in the concluding overview (in italics in Table 4.3), but should receive more attention in future research.

In contrast, mechanisms falling under the bottom-up perspective assume a more passive role of the welfare state. What is most important here is not the objective features of a welfare state, but the individual perception thereof. Such subjective perspectives on social policies, which are mostly found in studies exploring individual evaluations in general and the role of self-interest in particular (rational choice), can vary considerably from objective characteristics and performance.

Table 4.3 Summary of key perspectives, mechanisms and processes

Three processes that were also introduced as mechanisms are missing from this table: compression, imbrication and mediation (all inspired by Beckfield et al. 2015). Throughout this chapter, compression (regulating upper and lower bounds of health inequality) and mediation (or moderation) were always subsumed under either security or stratification. I believe that they relate to important processes within these two mechanisms, but should be seen as components rather than independent mechanisms. Since I am concerned with minimising the complexity of the theoretical premises, this seems to be a viable option for my purposes. However, if these specific elements are to be studied separately, there is no reason not to consider them as isolated mechanisms. In contrast, imbrication seems to be situated on an even higher level than the mechanisms in Table 4.3. Overlap and interaction between policy areas (and perhaps also mechanisms themselves) should always be taken into account, but as a downstream step.

This chapter has summarised and systematised the main hypotheses on the impact of social policies on various individual-level outcomes. The literature review has revealed distinct perspectives on the nature of welfare states. It now remains to specify which concepts of welfare stateness emerge and how empirical operationalisations can best capture them.