Four Faces of Marginalisation: Variations in Institutional Frameworks of Welfare State Provisions and Social Trust in Europe

Abstract

By assuming that marginalisation threatens social trust formation, this study introduces a new analytical framework to explain the relationship between a welfare state’s institutional design and trust levels in European societies. A good’s lifecycle view, consisting of production and consumption, is applied to the provision of social benefits to discern four forms of marginalisation in an individual’s experience with the welfare state: (a) marginalisation through attitude, (b) marginalisation through context, (c) marginalisation through poverty and (d) marginalisation through opportunity lack. We argue that universalism in benefit provisions minimises each of the four marginalisation forms whereas selectivity is characterised by higher odds of marginalisation. We further demonstrate that this especially holds true when universal social programmes are generous, and the State dedicates substantial resources to their funding. When the State’s resources are scarce, selectivity becomes a good alternative to universalism and may enhance social trust formation among individuals. We tested our hypotheses using data from the European Social Survey (2010).

JEL: D02, D03, Z10, Z1

Keywords

Welfare state social trust institutional design universal social benefits selective social benefits

I. Introduction

People living in universal welfare states have substantially higher trust levels than people in selective welfare states (Kumlin & Rothstein, 2007; Larsen, 2007; Rothstein & Stolle, 2003; Rothstein & Uslaner, 2005; Rothstein, 2001a, 2001b). Controlling for conventional country-level or individual-level determinants of trust does not fully eliminate this difference. This creates space for asking whether the institutional design of benefit provisions can be held responsible for variations in trust scores across societies (Kumlin & Rothstein, 2007; Rothstein & Uslaner, 2005).

It is well-established that formal institutional frameworks are essential for trust formation processes (Farrell, 2005; Herreros & Criado, 2008; Rothstein, 2003; Tillmar & Lindkvist, 2007). By defining the welfare state’s institutional design as the structure of the direct encounter between citizens and welfare state institutions (Kumlin, 2004, p. 51), literature suggests that it influences social trust in a twofold manner. On the one hand, the analysis focuses on an individual’s emotional experience with welfare state institutions and their officials. On the other hand, studies draw on the concept of inequality and how it potentially impacts trust levels among individuals. Despite strong empirical evidence supporting both perspectives, these streams have never been addressed jointly or juxtaposed in terms of their relative importance in trust emergence.

This study offers a single analytical framework to explain how the selectivity or universalism of welfare state programmes influences trust formation. By assuming that any form of marginalisation is a primary threat to trust emergence, we distinguish between four mechanisms through which marginalisation may arise as a result of contact with the welfare state: (a) attitude-related marginalisation, (b) context-related marginalisation, (c) poverty-related marginalisation and (4) opportunity-lack-related marginalisation. We further demonstrate that welfare states with distinct institutional designs have different levels of marginalisation, resulting in widely varying trust levels among individuals. The European Social Survey (ESS) data is used to test the validity of our analytical framework.

II. Literature Overview

In his study from 1998, Rothstein distinguished between universal and selective welfare states. He argues that universalism in welfare state provisions is more effective than selectivity in terms of the substantive justice, procedural justice and distribution burden. Further research linked the universal versus selective welfare state taxonomy to social trust by offering two key explanations for how the welfare state’s institutional design can impact trust levels in society: (a) an institutional approach and (b) an inequality approach.

The institutional approach assumes that social trust is a function of one’s perception of impartiality and fairness regarding the functioning of public institutions, formed from one’s experience with these institutions or their officials (Rothstein & Stolle, 2001, 2002, 2003). A positive experience leads individuals to perceive public institutions as fair and unbiased, which may spur feelings of safety and trust that can later be extrapolated to interactions with others (Rothstein & Stolle, 2008). Also, public officials are viewed as exemplars of society’s moral code (Kumlin, 2004; Kumlin & Rothstein, 2007; Rothstein & Stolle, 2001; Rothstein & Stolle, 2008), whereas a contact with them provides exposure to general social norms. The institutional argument further asserts that bureaucrats in selective welfare states have more discretionary power and are more likely to create suspicions or engage in actual cheating and arbitrary treatment, compared to bureaucrats in universal welfare states, which reduces social trust among individuals (Kumlin & Rothstein, 2007; Rothstein & Stolle, 2007).

The institutional argument also expands the notion of fairness and impartiality by discussing how individuals are stigmatised when they apply for social benefits (Larsen, 2007; Lee, 2012; Rothstein & Stolle, 2001). Selectivity or means-tested programmes require a qualification procedure and hence pit population groups against each other (Rothstein & Stolle, 2001). By dividing the population into a majority and ‘the bottom’ (Larsen, 2007), the welfare state cultivates a dualism between ‘the good independent citizens’ and ‘bad welfare dependents’ (Lee, 2012), which increases distance between individuals in society (Jensen & Svendsen, 2011; Larsen, 2007). Singling out citizens as special problems violates the principle of fairness and divides the population economically and culturally, fomenting distrust not only among the applicants but the entire society (Rothstein & Stolle, 2001). By contrast, universal welfare states or welfare programmes guarantee their recipients general inclusiveness, preventing discrimination and promoting norms of impartiality, fairness and respect, which generates high trust levels in society (Rothstein & Stolle, 2001).

The inequality approach draws on the premise that trust is unlikely to appear between unequal individuals. Unequal individuals do not feel the common bonds that are necessary to develop trust (Kawachi et al., 1997; Uslaner, 2000). Studies on trust often limit their inequality analysis to income inequality by considering that a key threat to trust emergence (Jordahl, 2007; Uslaner, 2000). People with fewer resources tend to feel less optimistic about their future, which results in lower trust levels towards others (Rothstein & Uslaner, 2006). There may also be a perceived injustice on the part of the poor. ‘Haves’ may be seen as having exploited those who ‘have not’ (Bjørnskov, 2005). The poor may also be dissatisfied with the existing distribution of wealth and seek to deviate from cooperative agreements that perpetuate the existing income distribution (Boix & Posner, 1998). The inequality approach further suggests that welfare states with different institutional designs support different levels of income inequality, resulting in contrasting levels of social trust among citizens. Universal programmes are deemed more effective at generating trust because they are based on proportional or progressive taxes and nominal service/benefits, ensuring better redistribution from the rich to the poor (Bergh, 2011). By contrast, selective welfare states often link the benefit amount to previously made payments, perpetuating existing inequalities and reducing solidarity between various population groups (Arts & Gelissen, 2002; Jensen & Svendsen, 2011).

Finally, the inequality approach considers the inequality of opportunities by looking at people’s access to education, labour markets, gender equality, and so on (Brehm & Rahn, 1997; Knack, 2002; Uslaner, 2000). The degree of equal opportunities among individuals is believed to shape an individual’s sense of optimism about the future and thereby influence their sense of social cohesion (Rothstein & Uslaner, 2006). Again, universal welfare states are viewed as more effective in enhancing trust levels, since benefits are provided to everyone, which equalises the status and responsibilities of citizenship, unlike benefit distribution in selective welfare states (Rothstein & Uslaner, 2006).

Research has addressed the impact of welfare states’ institutional design on social trust, but certain drawbacks require further analysis. First, most of the empirical studies are limited to establishing a negative relationship between an individual’s trust level and contacts with means-tested social programmes without modelling the logic of causal mechanisms promoted by their line of reasoning. This drawback can primarily be attributed to the lack of an adequate operationalisation in assessing the extent of universalism and selectivity in welfare state provisions. Second, no one has juxtaposed the existing explanations’ relative importance in trust formation processes. Rather, the existing arguments are viewed as independent, thereby preventing a simultaneous analysis of multiple channels through which the welfare state can impact trust levels in society. Third, the division of welfare states into a universal versus a selective group is closely associated with the level of social support available to citizens. Universalism presupposes that social spending is substantial to ensure that everyone receives the same set of social benefits or services (Rothstein, 2001a, 2001b). By contrast, selectivity is associated with lower social spending, since social support is limited to the needy. This raises the question of whether it is the welfare state’s institutional design or the level of social spending that better explains variations in social trust across societies. Sufficient empirical evidence supports the existence of a positive relationship between a welfare state’s social spending and social trust levels (Rothstein, 2005; Van Oorschot & Arts, 2005).

We seek to eliminate these drawbacks by integrating the existing theories on the welfare state–trust nexus into a single analytical framework. We further test this analytical framework with the ESS data by using the ratio of non-means-tested to means-tested social spending to measure the level of universalism and selectivity in organising a country’s social benefit provisions.

III. Analytical Framework

For our study, we adopt a psychological approach to explain trust formation. We see ‘trust’ as a trustor’s positive expectations concerning a trustee’s course of action (Barber, 1983). As such, ‘trust level’ refers to the extent to which a trustor is willing to be open to the trustee and does not fear emotional harm from him or her (Lewis & Weigert, 1985). We further use social psychology’s premise that an individual’s marginalisation is a major factor in how individuals see themselves and others. Marginalisation is defined as the process of pushing a particular group or groups of people to the edge of society by not allowing them an active voice, identity or place in it (Nelson & Prilleltensky, 2005). Since marginalisation relates to negative psychological and emotional responses, such as anger, depression, sadness, frustration, hopelessness, resentment, isolation, stress, confusion, and so on (Nelson & Prilleltensky, 2005), we further assume that marginalisation is one of the key factors in turning social trust into distrust or in preventing trust formation in the first place.

To locate all of the possible marginalisation instances that can occur through contact with the welfare state, we view social benefits as a good and utilise the economics’ simplistic understanding of a good’s life cycle as consisting of two stages (a) production and (b) consumption. As such, we consider two stages in the individual’s experience with the welfare state (a) application/qualification for social benefits or services and (b) receipt/consumption of social benefits (see Figure 1). The application stage is a ‘production stage’ and refers to the process of going through procedures that are required to obtain social benefits. Based on direct or indirect interactions with welfare state institutions and their personnel, this stage produces an experience with the welfare state where the individual forms knowledge regarding the quality of welfare state institutions and their officials, which is used to evaluate their performance. The second stage refers to receiving and consuming such benefits and involves evaluating how that impacts individuals in terms of the relative shift in their current or future well-being. We argue that experiences with the welfare state from both of these stages can be linked to marginalisation. We also argue that welfare states with different institutional designs produce different marginalisation levels in any of the stages and may therefore have varying results regarding the intensity of trust building processes.

Figure 1.

Source: The authors.

In analysing sources of marginalisation in the application stage, we use the common understanding that social benefit provision is embedded within a certain legal framework (postulated in formal rules and regulations) and is handled by public bureaucrats. Public bureaucrats must interpret the general body of regulations and apply them to each individual seeking to qualify for public provisions. These formal rules and regulations are rarely exact enough to provide unambiguous direction regarding the right decision in each case (Kumlin & Rothstein, 2007). And even if the formal legal framework is sufficiently exact, an official’s belief system, or so called mind-set, may impact how public bureaucrats make decisions or predictions and what meanings they draw from social contexts. These implicit belief systems shape how officials find facts, the inferences they draw and how they reason and interpret the law (Quintanilla, 2012). The application procedure can therefore create two sources of marginalisation: (a) marginalisation through attitude and (b) marginalisation through (legal) context.

Marginalisation through attitude occurs when public officials receive or process an individual’s application for social benefits. This can manifest itself either through their actual behaviour towards the applicant or through the ultimate decision that public servants make regarding granting or not granting social benefits/services. Both aspects are a function of the public officials’ understanding of the formal rules and their perception of the applicant’s eligibility or deservedness of social support. A negative attitude towards immigrants may, for instance, lead to the application procedure for an individual with an immigration background to be more time-consuming and stressful compared to an individual with similar needs but of the same nationality as the public official. Such experiences can shape an applicant’s perception of how fair and impartial the entire welfare system is, influencing the applicants’ institutional trust and, as a result, trust towards other individuals.

Marginalisation through context refers to formal rules and regulations regarding welfare eligibility that public officials must interpret during the benefit production stage. As such, the applicant’s perception of fairness and impartiality regarding decisions made by welfare state officials is a function of (a) the actual quality of rules and regulations that are to be interpreted and (b) the extent to which existing rules and regulations are enforced. Numerous regulations or ambiguous stipulations associated with the application procedure or eligibility criteria increase chances of arbitrariness and cheating among public bureaucrats in their decision-making about granting social support to applicants. Low levels of law enforcement and widespread corruption in a country’s social benefit provisions may cause more public officials to deviate from existing regulations. Widespread corruption in a country may also motivate applicants to manipulate eligibility criteria to qualify for social support or contribute to the assumption that corruption constitutes the only way of obtaining even justified benefits. Inadequate regulations and high levels of corruption both undermine the concept of impartiality and fairness, increasing instances of marginalisation and reducing trust in both of the two forms: institutional and social.

The consumption-related phase refers to evaluating current or future changes in the individual’s quality of life arising from either receiving or not receiving the requested social benefits. Based on this premise, we derive two more sources of marginalisation: (a) marginalisation through poverty and (b) marginalisation through opportunity lack. The marginalisation through poverty relates to a change in the benefit recipient’s present economic well-being where receiving the requested social benefits lifts him or her above the poverty line. Poverty reduction nurtures feeling optimistic and secure about the future, which should lead to higher trust levels not only towards the welfare state system but also towards other individuals, in general (Uslaner, 2000). The marginalisation through opportunity lack is derived from the equality of opportunity argument and relates to the government providing insufficient access to sectors that determine an individual’s opportunity set, thereby constraining this individual’s prospects for improving their situation in the future. Here, we consider conventional opportunities concerning employability, education and the health system. A wider set of available opportunities is expected to lead to higher trust levels.

We use these four forms of marginalisation to compare the impact of universal and selective welfare states on social trust. In doing this juxtaposition, we adopt the conventional understanding that a universal welfare state offers all or most of the social benefits on a universal or non-means-tested basis without linking the benefit amount to a person’s need or income level. By contrast, the selective welfare state provides all or most of the social benefits on a means-tested basis, with public officials having essential discretionary power in deciding whether to grant an applicant’s requested social benefit. Here, the benefit coverage is limited to those who need social support. Drawing from these key differences, we suggest the following:

Proposition 1. By minimising each of the four forms of marginalisation, universalism in welfare state provisions should lead to higher levels of social trust than selectivity. By eliminating the application procedure, universal social programmes require less interaction with social benefit applicants and hence minimise the odds of marginalisation through attitude or behaviour compared to selective welfare states. Eliminating application procedures also suggests that poorly formulated social regulations or widespread corruption may only marginally impact an individual’s experience with universal welfare states. Finally, universalism in social provisions guarantees better redistribution from the rich to the poor and minimises instances of poverty-related or opportunity-lack-related marginalisation in society. As such, we expect that

H₁: Compared to selectivity, universalism in welfare state provisions leads to lower levels of marginalisation in each of the four forms, resulting in higher levels of social trust.

Proposition 2. Compared to selectivity, the universal coverage of social benefits leads to a greater increase in social trust levels as a response to a reduction in each of the four marginalisation forms. With universal social programmes, most of the population is exposed to the impact that a reduction in the four forms of marginalisation has on their life, which can directly influence their perception of other people and thereby their trust level towards others. In selective welfare states, the impact that receiving social benefits has on recipients is limited to them and can only indirectly be analysed with regard to the rest of the population. This suggests that the social trust levels of benefit recipients in selective welfare states are subject to direct amelioration whereas the rest of the population can reconsider their trust levels only marginally through social programmes’ indirect economic, social or political consequences. Therefore, we anticipate that

H₂: An equal reduction in marginalisation of any form leads to a greater increase in social trust levels in universal welfare states than in selective welfare states.

Proposition 3. Both universalism and selectivity yield more social trust in society when social spending is higher. Higher social spending may reduce marginalisation instances in both universal and selective welfare states by offering higher social benefits or more generous social coverage to recipients. More generous social programmes set fewer constraints on welfare state institutions or their officials thereby limiting the attitude- and context-related marginalisation forms. Generous social programme funding (whether universal or selective) also reduces marginalisation through poverty to a greater extent and enables the provision of a wider set of opportunities in key economic sectors, thereby leading to higher trust levels.

H₃: The impact of universalism and selectivity on social trust is greater when social spending is higher.

IV. Data and Methods

For testing our hypotheses, we use ESS. We select the data for the year 2010 because this dataset offers operationalisation choices for all of the four proposed mechanisms (marginalisation through attitude, marginalisation through context, marginalisation through poverty and marginalisation through opportunity lack mechanisms), while also providing a measure of social trust. The ESS public use dataset has been downloaded from its website¹. The sample is restricted to countries for which the data on the key variables are available and includes Belgium, Bulgaria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Lithuania, Norway, Poland, Slovakia, Slovenia, Spain, Sweden, Switzerland, the Netherlands and the United Kingdom. The total number of cases amounts to 23,631. The following set of variables is utilised in the analysis (see Table 1 for the descriptive statistics).

Table 1.

Descriptive Statistics for the Key Variables Used in the Analysis

Variables	N	Minimum	Maximum	Mean	Std. Deviation
Social trust	23,631	1.000	10.000	5.189	2.368
Treatment with respect	23,631	0.000	1.000	0.297	0.172
Attitude towards income inequality	23,631	0.000	1.000	0.424	0.210
Attitude towards poverty reduction	23,631	0.000	1.000	0.342	0.168
Wrong morals	23,631	0.000	1.000	0.827	0.173
Impartiality and fairness	23,631	0.000	1.000	0.341	0.149
Corruption	23,631	0.000	1.000	0.377	0.248
Experience with poverty	23,631	0.000	1.000	0.213	0.410
Distribution around poverty line	23,631	0.000	1.000	0.196	0.130
Income insufficiency	23,631	0.000	1.000	0.515	0.224
Opportunity lack in the economy	23,631	0.000	1.000	0.569	0.252
Opportunity lack in the education system	23,631	0.000	1.000	0.441	0.234
Opportunity lack in the health system	23,631	0.000	1.000	0.465	0.253
Frequency of TV watching	23,631	0.000	7.000	4.315	2.044
Interest in politics	23,631	1.000	4.000	2.555	0.895
Frequency of socialisation with friends or colleagues	23,631	1.000	7.000	4.865	1.542
Subjective health condition	23,631	1.000	5.000	2.189	0.916
Frequency of attending religious services	23,631	1.000	7.000	5.510	1.461
Born in the country	23,631	0.000	1.000	0.919	0.272
Age	23,631	14.000	94.000	47.264	17.502
Years of education	23,631	0.000	25.000	12.874	3.693
Gender	23,631	0.000	1.000	0.488	0.499
Household income	23,631	1.000	10.000	5.381	2.771

Source: The authors.

Note: The values of the four mechanisms’ items are rescaled to change between 0 and 1.

Dependent Variable

Social trust is defined on the basis of the following question: ‘Generally speaking, would you say that most people can be trusted or that you cannot be too careful in dealing with people?’ The responses vary from 0 (no trust at all) to 10 (complete trust).

Independent Variables

We operationalise the welfare state’s institutional design through non-means-tested and means-tested social spending expressed as a percentage of countries’ GDP. We also calculate the ratio of means-tested to non-means-tested social spending in order to deviate from the extent of welfare state development in a country and focus on the level of universalism or selectivity in the overall volume of welfare state provisions. The data are sourced from the European System of Integrated Social Protection Statistics (EUROSTAT website).

The marginalisation through attitude mechanism refers to one’s perception of treatment by public bureaucrats and the perception of the needy that is prevalent in society. Since the ESS does not contain any variables measuring one’s perception of welfare state institutions, we use the question asking the respondents to specify the frequency with which the police treat people in their country with respect. This question is expected to approximate the level of respect with which public institutions treat applicants, in general. The responses vary from 1 (very often) to 4 (not at all often). The perception of the needy is operationalised through two questions. The first question measures the individual’s attitude towards poverty by asking the respondent whether the government should do more to prevent people falling into poverty. The second question captures one’s attitude towards income inequality by asking whether the government should take measures to reduce differences in income levels. Both questions have a response scale varying between 1 (agree strongly) and 4 (disagree strongly).

The marginalisation through context mechanism is operationalised through the quality of a country’s informal and formal institutional contexts. The informal institutional context is measured through the average of two questions regarding how wrong the respondent considers (a) making an exaggerated or false insurance claim and (b) buying something one thought might be stolen. Each of the questions has responses with values varying between 1 (seriously wrong) and 4 (not wrong at all). The quality of formal environment is approximated through (a) impartiality and fairness and (b) corruption levels in public institutions. The impartiality and fairness variables are measured through an average of responses provided to two questions regarding how the respondent would say that (a) the police or (b) courts make fair, impartial decisions in the cases they deal with. The response scale varies from 1 (very often) to 4 (not at all often). The corruption levels are captured by an average of responses to two questions regarding the frequency with which the respondent would say (a) the police and (b) the courts take bribes, with responses ranging between 0 (never) and 10 (always).

The marginalisation through poverty mechanism is operationalised through an individual’s experience with poverty, income dispersion around a country’s poverty line and income insufficiency. The individual’s experience with poverty is a dummy variable that takes the value of one when the individual’s income is above the national poverty line. A country’s poverty line is calculated for each country individually by taking 60 per cent of median household income provided by the ESS. The dispersion around poverty line is constructed as follows: [(individual income – a country’s poverty line)/ a country's poverty line]. We then rescale the values to change between 0 and 1, with higher values corresponding to greater negative distance to the poverty line. Income insufficiency is operationalised through the question about the respondents’ feeling regarding their present income. The initial responses vary from 1 (living comfortably on present income) to 4 (finding it very difficult on present income).

The marginalisation through opportunity lack mechanism is operationalised through three questions regarding one’s perception about (a) the present state of economy, (b) education and (c) the health system. Each of the questions has responses varying between 0 (extremely satisfied) and 10 (extremely dissatisfied). The quality of three above domains is expected to approximate opportunity sets of individuals through the economy, education and health systems.

We rescale the values of the four mechanisms’ operationalisations to change between 0 and 1, with higher values corresponding to greater marginalisation in each of the four forms.

Control Variables

We control for the conventional determinants of trust such as frequency of TV watching, interest in politics, frequency of meeting friends, health condition and religiosity levels (Christoforou, 2004; Hall, 1999; Van Oorschot & Arts, 2005). The frequency of watching TV variable has values varying from 1 (no time at all) to 7 (more than three hours a day). Respondents’ level of interest in politics ranges between 1 (very interested) and 4 (not interested at all). The frequency of meeting friends or colleagues is measured through a 7-point scale with values changing between 1 (never) and 7 (every day). The respondents’ health condition is operationalised through the question asking respondents to self-rate their health status on a scale between 1 (very good) and 5 (very bad). One’s religiosity is approximated through the frequency with which one attends religious services, with responses varying between 1 (every day) and 7 (never). Additionally, we control for the respondents’ immigration status, age, years of completed education and gender (Bjørnskov, 2007; Hall, 1999; Van Oorschot & Arts, 2005). Finally, we include the household income variable measured on a 10-point scale.

Methods Used in the Analysis

We test our hypotheses by running a multilevel analysis, which allows us to control for hierarchical structure of our data. This is necessary to prevent the un-modelled country information from ending up all being pooled into the single individual error (Luke, 2004). The empirical model takes the following form:

\begin{array}{l} {Social_Trust}_{i j} = γ_{00} + γ_{10} Attitude_M \arg {inalisation}_{i j} + γ_{20} Context_M \arg {inalisation}_{i j} \\ + γ_{30} Poverty_M \arg {inalisation}_{i j} + γ_{40} Opportunity_Lack_M \arg {inalisation}_{i j} + γ_{50} X_{i j} \\ + m_{0 j} {+ ε}_{i j} \end{array}

(1)

Here, Attitude_Marginalisationij, Context_Marginalisation_ij, Poverty_Marginalisation_ij and Opportunity_Lack_Marginalisation_ij are the four marginalisation mechanisms, X_ij is a set of control variables, m_0j is variance at the country level and ε_ij is variance at the individual level. We use STATA command GLLAMM for calculating the model parameters.

V. Empirical Results and Discussion

We begin our analysis by calculating a ratio of means-tested to non-means-tested spending and determining the median for the selected countries. The median (0.070) is used as a benchmark for dividing the sample into two sub-samples. The first comprises countries with a ratio of less than 0.070 and includes Belgium, Bulgaria, the Czech Republic, Estonia, Finland, Hungary, Lithuania, Norway, Poland, Slovakia and Sweden. This group is characterised by more universal welfare state designs. The second group includes the remaining countries, which have a ratio above 0.070 and are hence characterised by a rather selective welfare state system.

We further calculate the marginalisation mechanisms’ mean values for the universal and selective groups. Juxtaposing these values reveals substantial differences (see Table 2) in marginalisation levels stemming from both the production and consumption stages of the social benefit life cycle. Universalism guarantees a narrower distribution of income around the poverty line and is supported by a greater concern for poverty reduction. Universalism also enables more impartiality and fairness in interactions with public institutions and their officials. The greatest differences, however, lies in the respondents’ perception of available opportunities as measured through their satisfaction with the economy and its key sectors, such as education and health care.

Table 2.

The Four Marginalisation Forms’ Mean Values, by Institutional Design

Mechanism Items	Universal	Selective	t-Test for Equality of Means
Mechanism Items	Universal	Selective	Mean Difference	Sig. (2-Tailed)
Social trust	5.400	5.035	0.365***	(0.000)
Marginalisation through attitude
Treatment with respect	0.296	0.297	–0.001	(0.490)
Attitude towards income inequality	0.421	0.425	–0.004	(0.124)
Attitude towards poverty reduction	0.338	0.346	–0.008***	(0.000)
Marginalisation through context
Wrong morals	0.828	0.827	0.001	(0.580)
Impartiality and fairness	0.334	0.345	–0.011***	(0.000)
Corruption	0.384	0.372	0.012***	(0.000)
Marginalisation through poverty
Experience with poverty	0.210	0.210	0.000	(0.733)
Distribution around poverty line	0.189	0.203	–0.014***	(0.000)
Income insufficiency	0.529	0.506	0.023***	(0.000)
Marginalisation through opportunity lack
in the economy	0.513	0.610	–0.097***	(0.000)
in the education system	0.391	0.478	–0.087***	(0.000)
in the health system	0.451	0.475	–0.024***	(0.000)

Source: The authors.

Notes: Standard errors in parentheses.

*** p < 0.01, ** p < 0.05, * p < 0.1 (2-tailed).

We use the universalism versus selectivity division to construct a dummy that takes the value of one if a country belongs to the group with a more universal institutional design and the value of zero if a country has a welfare state with a more selective institutional design. Since the welfare state is a country-level category, we first resort to a country-level analysis. We link the universalism dummy to social trust levels for the aggregated dataset obtained by calculating countries’ mean values for the selected variables (see Table 3). Two interesting nuances emerge from this analysis. First, we establish that universalism in social provisions has a strong positive effect on trust scores. This positive effect remains robust after controlling for the conventional trust determinants, such as criminality levels, being a post-communist country and altruism levels. The perception of criminality variable reflects the quality of the institutional environment in a country. It is derived from the institutional theories of social trust formation, according to which keeping an individual safe and protected through well-functioning public institutions may form the necessary grounds for higher institutional trust and, as a result, higher levels of interpersonal trust (Rothstein & Stolle, 2008). The post-communism dummy controls for the heritage of the socialist regime that emphasised collectivism. Actors in collectivist societies prove to give greater priority to group goals leading to higher levels of trust among individuals (Simpson, 2006; Yamagishi et al., 1998). Volunteering is believed to offer regular and close contacts with others, helping individuals develop reciprocity, cooperation, empathy for others and yielding more social trust (Brehm & Rahn, 1997; Newton, 1999a, 1999b; Paxton, 2002; Putnam, 1993, 2000). Second, universalism’s impact on trust changes most dramatically when including the overall level of social spending, measured as a percentage of GDP.

Table 3.

The Impact of Welfare State’s Institutional Design on Social Trust Levels, An Ordinary Least Squares Regression

Variables	(1)	(2)	(3)	(4)	(5)
Universalism dummy	0.277	0.693*	0.771**	0.808***	0.587**
	(0.374)	(0.380)	(0.309)	(0.279)	(0.296)
Total social spending		0.127***	0.071*	–0.022	-0.026
		(0.041)	(0.036)	(0.052)	(0.051)
Perceived criminality			–2.049***	–1.978***	–2.069***
levels			(0.606)	(0.547)	(0.545)
Post-communism				–1.086**	–1.117**
dummy				(0.470)	(0.464)
Altruism levels					0.770
					(0.625)
Constant	4.931***	1.829*	7.164***	9.654***	8.425***
	(0.243)	(1.055)	(1.794)	(1.946)	(2.162)

Observations	23	23	23	23	23
R-squared	0.022	0.332	0.583	0.678	0.705

Source: The authors.

Notes: Standard errors in parentheses. The perceived criminality variable is operationalised through the question in which respondents should specify how safe they feel walking alone in their area after dark. The responses vary between 1 (very safe) and 4 (very unsafe). The post-communism dummy takes the value of one when a country has experienced the post-communist transition. Altruism levels are measured through the degree to which respondents agree with the statement that it is important to help people and care about others’ well-being. The response values vary from 1 ‘not like me at all’ to 6 ‘very much like me’. *** p < .01, ** p < .05, * p < .1 (2-tailed).

To provide a better insight into the first nuance, we run a simultaneous equation model to check whether the impact of universalism versus selectivity goes through the four marginalisation channels proposed by our analytical framework. More specifically, the system of equations includes a cross-country trust equation and a channel equation that models one of the four marginalisation mechanisms. Due to a limited number of countries in our sample, we include the marginalisation mechanisms sequentially. The universalism dummy enters both equations. The four marginalisation mechanisms appear endogenous in the system. Therefore, we consider it necessary to use instrumental variables estimation to ensure that our structural parameters are identified. The choice of the instruments is done based on their strong correlation with the corresponding variables, while ensuring that they are theoretically or statistically unrelated to the main dependent variable (social trust). In addition, we include other control variables in the channel equations, with the number of inclusions being sufficient for the order condition for identification to be satisfied. We estimate the full set of equations jointly using three-stage least square by applying STATA command reg3 to the aggregated data-set. The base model is

{Trust}_{j} = α_{0} {+ α}_{1} {Marginalisation_mechanism}_{j} {+ α}_{2} {Universalism_dummy}_{j} + ε_{j}

(2)

\begin{array}{l} {Marginalisation_mechanism}_{j} = β_{0} + β_{1} {Universalism_dummy}_{j} + β_{2} Interested_in_ \\ {politics}_{j} + β_{3} {Post_Communism_dummy}_{j} + β_{4} {Instrument}_{j} + ε_{j} \end{array}

(3)

Where, ‘Trust’ is social trust scores. Marginalisation_mechanism is a single measure of one of the four marginalisation mechanisms which is constructed by applying the STATA predict option for factor analysis. The single measure variables have values rescaled to vary between 0 and 1, with higher values corresponding to greater marginalisation in each of the four forms. Universalism_dummy is a dummy variable that takes the value of one when a country’s welfare state can be characterised as having a rather universal institutional design and the value of zero when otherwise. Interested_in_politics is a country-level mean value for a question in which the respondents should self-rate their interest in politics by using a 4-point scale, with response values varying from 1 (very interested) to 4 (not interested at all). Post_Communism_dummy is a dummy that takes the value of one when a country has experienced a post-communist transition. ‘Instrument’ stands for the variables used to instrument one of the marginalisation mechanisms. The marginalisation through attitude variable is instrumented with the question asking the respondents to specify how important it is that people are treated equally and have equal opportunities. The responses vary between 1 (not like me at all) and 6 (very much like me). The marginalisation through context mechanism is instrumented with the question about the degree to which respondents agree that all laws should be strictly obeyed. The response values range between 1 (disagree strongly) and 5 (agree strongly). The marginalisation through poverty line is instrumented with a country’s actual poverty line calculated as 60 per cent of a country’s median household income. The marginalisation through opportunity lack mechanism is instrumented with the question about the respondents’ overall satisfaction with life. The responses vary from 0 (extremely dissatisfied) to 10 (extremely satisfied).

Table 4 provides the results for our simultaneous equation analysis. It reveals that the behavioural and contextual marginalisation mechanisms are weak channels for the impact that the welfare state institutional design has on social trust. By contrast, the poverty and opportunity lack marginalisation mechanisms were found to channel the impact of the universalism dummy on social trust. More universalism in welfare state provisions leads to less marginalisation through poverty and opportunity lack thereby enhancing social trust formation. Note that in the case of the consumption-stage marginalisation forms, the universalism dummy did not register a direct impact on social trust. Given the operationalisation strategies for the marginalisation mechanisms used in this analysis, we conclude that one’s experience with the welfare state and its institutional design in the consumption dimension plays a more significant role in the individuals’ social trust formation process than the negative perceptions that result from one’s experience during the application stage. We hence received only partial support for Hypothesis 1.

Table 4.

The Four Marginalisation Mechanisms’ Impact on Social Trust, Three Stage Least Squares Models

Variables	Marginalisation
Variables	Through Attitude	Through Context	Through Poverty	Through Opportunity Lack
Trust equations
Universalism dummy	0.472**	0.616***	0.385	0.039
	(0.222)	(0.231)	(0.274)	(0.237)
Marginalisation forms	–3.587***	–4.554***	–3.271***	–3.301***
	(0.535)	(0.641)	(0.647)	(0.535)
Constant	3.552***	2.100***	2.648***	3.214***
	(0.266)	(0.438)	(0.505)	(0.335)
Rsq	0.692	0.678	0.523	0.639
Observations	23	23	23	23
Marginalisation mechanism equations
Universalism dummy	0.061	0.035	–0.115**	–0.143***
	(0.067)	(0.051)	(0.056)	(0.046)
Interest levels in politics	0.399***	0.340***	0.085	0.042
	(0.123)	(0.094)	(0.110)	(0.105)
Post-communism dummy	0.229***	0.048	0.101*	0.054
	(0.064)	(0.062)	(0.056)	(0.053)
Instrumental variables
Important to be treated equally	–0.045***(0.144)
Laws should be strictly obeyed		–0.452***(0.111)
Poverty line levels			–0.276***(0.042)
Overall satisfaction levels				–2.176***(0.299)
Constant	–0.641(0.451)	–0.574*(0.330)	–0.208(0.334)	0.867***(0.412)
Rsq	0.703	0.707	0.761	0.823
Observations	23	23	23	23

Source: The authors.

Notes: Standard errors in parentheses.

*** p < .01, ** p < .05, * p < .1 (2-tailed).

We now juxtapose the universal versus selective country categories to consider how an equal change in each of the marginalisation forms impacts social trust in the two subsamples. In doing so, we shift to an individual-level analysis and use the dataset consisting of 23,631 cases. Table 5 shows the results for the key control variables that are utilised as a base model for our trust equation.

Table 5.

Social Trust Base Regression, Multilevel Modelling

Variables
Frequency of TV watching	–0.029***
	(0.007)
Interest in politics	–0.160***
	(0.017)
Frequency of socialisation with friends or colleagues	0.102***
	(0.009)
Subjective health condition	–0.261***
	(0.017)
Frequency of attending religious services	–0.043***
	(0.011)
Born in the country	0.237***
	(0.053)
Age	0.009***
	(0.001)
Years of education	0.061***
	(0.004)
Gender	–0.027
	(0.028)
Household income	0.048***
	(0.006)
Constant	4.255***
	(0.228)
Variance at Level 1	4.378
	(0.040)
Variance at Level 2	0.710
	(0.211)
Log likelihood	–51,036.899
Number of Level 1 units	23,631
Number of Level 2 units	23

Source: The authors.

Notes: Standard errors in parentheses.

*** p < .01, ** p < .05, * p < .1 (2-tailed).

Table 6 contains results for the augmented models in which we sequentially include the four marginalisation mechanisms’ items. Our individual-level analysis suggests that, in most cases, an equal reduction in marginalisation levels of any form produces more social trust among individuals in universal welfare states than in selective welfare states. This is in line with Hypothesis 2. Since universalism in welfare states aims to provide support to everyone, social trust levels are more responsive to social programmes than in selective welfare states where social support is only given to those in need. Again, the greatest differences are found in the consumption dimension that describes changes in one’s current or future well-being. Both findings suggest that the inequality argument better explains how a welfare state’s institutional design impacts social trust than the behavioural argument.

Table 6.

The Four Marginalisation Mechanisms’ Impact on Social Trust Scores for More Universal versus More Selective Welfare States, Multilevel Modelling

Mechanism Items	Universal	Selective	Test for Differences in Coefficients
Marginalisation through attitude
Treatment with respect	–1.635***	–1.209***	22.860***
	(0.134)	(0.118)	(0.000)
Attitude towards income inequality	0.094	-0.009	3.160*
	(0.107)	(0.091)	(0.075)
Attitude towards poverty reduction	0.497***	0.434***	1.230
	(0.136)	(0.115)	(0.267)
Marginalisation through context
Wrong morals	–0.494***	–0.054	29.700***
	(0.129)	(0.121)	(0.000)
Impartiality and fairness	–2.679***	–2.220***	49.030***
	(0.154)	(0.140)	(0.000)
Corruption	–1.296***	–1.219***	47.570***
	(0.108)	(0.097)	(0.000)
Marginalisation through poverty
Experience with poverty	–0.062	–0.008	0.680
	(0.072)	(0.062)	(0.408)
Distribution around poverty line	–1.968***	–0.764*	97.820***
	(0.700)	(0.408)	(0.000)
Income insufficiency	–0.677***	–0.978***	22.920***
	(0.120)	(0.109)	(0.000)
Marginalisation through opportunity lack
in the economy	–2.521***	–2.324***	96.610***
	(0.104)	(0.094)	(0.000)
in the education system	–1.905***	–1.527***	23.370***
	(0.103)	(0.090)	(0.000)
in the health system	–1.843***	–1.641***	40.130***
	(0.098)	(0.085)	(0.000)
Number of Level 1 units	23,631	23,631	23,631
Number of Level 2 units	23	23	23

Source: The authors.

Notes: Standard errors in parentheses. The test for differences in coefficients was conducted by using STATA commands ‘suest’ and ‘test’. We report the value of ch2 and its p-value (in brackets).

*** p < .01, ** p < .05, * p < .1 (2-tailed).

Our results suggest that taking into account the universal versus selective nature of welfare states is necessary when analysing social trust formation. But, there is more to understanding the overall relationship between the welfare state and trust levels. The second nuance derived from Table 3 indicates that social trust in society is influenced not only by the level of universalism in welfare state provisions but also by the extent to which the welfare state is developed (viewed as the level of social spending). To account for both the level of universalism and the extent of welfare state development, we calculate the median for the overall social spending variable and use it along with the universalism versus selectivity ratio’s median to divide countries into four groups (see Figure 2). The first comprises Belgium, Finland, Norway and Sweden. These countries have a ratio of means-tested to non-means-tested social spending below 0.070, whereas the overall social spending is above the sample’s median of 23 per cent. We call this group ‘Generous Universalism’. The second group still includes countries with a universal institutional design (with a ratio below 0.070) but with lower levels of overall social support (social spending below 23% of GDP) and is called ‘Meagre Universalism’. This group includes post-communist countries (Bulgaria, the Czech Republic, Estonia, Hungary, Lithuania, Poland and Slovakia) that retained the Socialism’s universal approach to social provisions but, due to transition and relative economic weaknesses, are unable to direct substantial public resources to social programmes.

Figure 2.

Source: The authors.

Similarly, we define the group of Generous Selectivity as countries having a ratio value above 0.070 with social spending above 23 per cent of GDP. This group is represented by Germany, Denmark, Spain, France, the United Kingdom, Greece and the Netherlands. Finally, the group of Meagre Selectivity is comprised of Croatia, Cyprus, Ireland, Slovenia and Switzerland and includes countries with a ratio value above the benchmark and overall social spending below 23 per cent of GDP. Figure 2 illustrates the distribution of the selected countries across the four groups. Table 7 reports mean values for the marginalisation mechanisms by country group.

Table 7.

Four Marginalisation Mechanisms’ Mean Values, by Typology Group

Mechanism Items	Generous Universalism	Meagre Universalism	Generous Selectivity	Meagre Selectivity
Marginalisation through attitude
Treatment with respect	0.250(Weak)	0.348(Strong)	0.271(Relatively strong)	0.261(Relatively strong)
Attitude towards income inequality	0.404(Weak)	0.401(Weak)	0.454(Strong)	0.426(Strong)
Attitude towards poverty reduction	0.381(Strong)	0.297(Weak)	0.358(Strong)	0.345(Strong)
Marginalisation through context
Wrong morals	0.764(Weak)	0.795(Relatively weak)	0.846(Strong)	0.827(Relatively strong)
Impartiality and fairness	0.280(Weak)	0.304(Relatively weak)	0.320(Relatively strong)	0.335(Strong)
Corruption	0.241(Weak)	0.513(Strong)	0.328(Relatively strong)	0.348(Relatively strong)
Marginalisation through poverty
Experience with poverty	0.213(Average)	0.212(Average)	0.214(Average)	0.215(Average)
Distribution around poverty line	0.174(Weak)	0.202(Strong)	0.190(Weak)	0.233(Strong)
Income insufficiency	0.427(Weak)	0.621(Strong)	0.466(Relatively weak)	0.484(Relatively strong)
Marginalisation through opportunity lack
in the economy	0.372(Weak)	0.642(Strong)	0.602(Average)	0.600(Average)
in the education system	0.315(Weak)	0.640(Strong)	0.482(Relatively weak)	0.609(Strong)
in the health system	0.321(Weak)	0.569(Strong)	0.447(Relatively weak)	0.470(Relatively strong)

Source: The authors.

We use the above typology to construct dummies for the four country groups. We further include these dummies in the trust base model by using Generous Universalism as our reference category (see Table 8). Our results indicate that social trust is more easily formed in Generous Universalism and most difficult to form in Meagre Selectivity or Meagre Universalism. Additionally, our results indicate that Generous Selectivity is more effective in promoting social trust than Meagre Universalism, thereby suggesting that when resources are scarce they should be directed only to those in need. Still, our results emphasise that social trust is higher in Generous Universalism than in Meagre Universalism, whereas Generous Selectivity promotes trust formation more effectively than Meagre Selectivity. This is in line with Hypothesis 3. The institutional design of welfare states should be analysed along with the actual amount of social spending that the welfare state dedicates to funding social benefits through universalism or selectivity. Our results remain robust after including additional control variables and only slightly change when we control for a country’s experience with post-communist transition (see Table 3).

Table 8.

Variations in Social Trust Levels Across the Four Typology Groups, Multilevel Modelling with Dummies

Variables	(1)	(2)	(3)	(4)
Frequency of TV watching	–0.029***	–0.040**	–0.041**	–0.038***
	(0.007)	(0.015)	(0.016)	(0.011)
Interest in politics	–0.160***	–0.204***	–0.203***	–0.189***
	(0.017)	(0.044)	(0.044)	(0.045)
Frequency of socialisation with	0.102***	0.120***	0.120***	0.121***
friends or colleagues	(0.009)	(0.026)	(0.026)	(0.027)
Subjective health condition	–0.260***	–0.221***	–0.219***	–0.201***
	(0.017)	(0.039)	(0.039)	(0.037)
Frequency of attending religious	–0.044***	0.006	0.004	–0.003
services	(0.011)	(0.032)	(0.032)	(0.031)
Born in the country	0.237***	0.114	0.113	0.165
	(0.053)	(0.117)	(0.117)	(0.118)
Age	0.010***	0.012***	0.012***	0.012***
	(0.001)	(0.002)	(0.001)	(0.001)
Years of education	0.061***	0.063***	0.064***	0.062***
	(0.004)	(0.008)	(0.008)	(0.007)
Gender	–0.027	–0.234***	–0.235***	–0.233***
	(0.027)	(0.059)	(0.058)	(0.056)
Household income	0.048***	0.049***	0.049***	0.054***
	(0.006)	(0.013)	(0.013)	(0.014)
Perceived criminality levels		–0.475***	–0.476***	–0.485***
		(0.047)	(0.047)	(0.046)
Altruism levels			–0.015	–0.023
			(0.029)	(0.030)
Post-communism dummy				–0.954***
				(0.278)
Generous Universalism	Reference	Reference	Reference	Reference
	Category	Category	Category	Category
Meagre universalism	–1.320***	–1.031**	–1.033**	–0.088
	(0.398)	(0.391)	(0.400)	(0.486)
Generous selectivity	–0.953***	–0.763*	–0.762*	–0.763*
	(0.360)	(0.390)	(0.400)	(0.400)
Meagre selectivity	–1.402***	–1.102***	–1.106***	–0.682***
	(0.428)	(0.373)	(0.384)	(0.226)
Variance at Level 1	4.378	4.008	3.600	3.200
	(0.040)	(0.023)	(0.022)	(0.019)
Variance at Level 2	0.417	0.300	0.280	0.269
	(0.100)	(0.080)	(0.058)	(0.026)
Number of Level 1 units	23,631	23,631	23,631	23,631
Number of Level 2 units	23	23	23	23

Source: The authors.

Notes: Standard errors in parentheses.

*** p < .01, ** p < .05, * p < .1 (2-tailed).

To further explore the relationship between a welfare state’s institutional design and social spending levels, we examine the impact of the marginalisation mechanism items on social trust in each of the four country groups (see Table 9). Our interaction analysis suggests that there are substantial variations in how the four marginalisation mechanisms affect trust scores across the four country groups. An equal reduction in marginalisation levels tends to produce more social trust in Generous Universalism than in Meagre Universalism. A similar observation was found between Generous Selectivity and Meagre Selectivity: When selectivity is more generous, social trust can be more easily formed by improving marginalisation variables. There tends to be a smaller but still positive impact with Generous Selectivity compared to Generous Universalism.

Table 9.

Interactions Between the Four Groups of Welfare State’s Institutional Design and the Marginalisation Mechanisms, Multilevel Modelling

Mechanism Items	Interaction Terms with
Mechanism Items	Generous Universalism	Meagre Universalism	Generous Selectivity	Meagre Selectivity
Marginalisation through attitude
Treatment with respect	Ref. category	0.244	0.699***	0.107
		(0.278)	(0.261)	(0.309)
Attitude towards income inequality	Ref. category	1.023***(0.210)	0.466**(0.185)	0.439*(0.235)
Attitude towards poverty reduction	Ref. category	1.635***(0.274)	0.669***(0.226)	0.656**(0.290)
Marginalisation through context
Wrong morals	Ref. category	–0.212	–0.701***	–0.079
		(0.265)	(0.262)	(0.297)
Impartiality and fairness	Ref. category	0.555*	0.522*	0.665**
		(0.304)	(0.304)	(0.346)
Corruption	Ref. category	0.404**	0.143	–0.016
		(0.198)	(0.206)	(0.242)
Marginalisation through poverty
Experience with poverty	Ref. category	0.022	–0.244***	–0.342***
		(0.101)	(0.091)	(0.113)
Distribution around poverty line	Ref. category	–0.608*(0.365)	0.984***(0.360)	0.513(0.369)
Income insufficiency	Ref. category	0.711***	–0.399**	–0.106
		(0.215)	(0.200)	(0.237)
Marginalisation through opportunity lack
in the economy	Ref. category	–0.509***	–0.004	–0.285
		(0.190)	(0.198)	(0.238)
in the education system	Ref. category	0.114	0.503***	0.251
		(0.217)	(0.180)	(0.238)
in the health system	Ref. category	–0.309*	0.020	0.157
		(0.160)	(0.196)	(0.225)

Source: The authors.

Notes: Standard errors in parentheses.

*** p < .01, ** p < .05, * p < .1 (2-tailed).

We conclude our analysis by illustrating the relative positioning of the four typology groups regarding their values on the four marginalisation mechanisms (see Figure 3). Generous Universalism ensures weak marginalisation in each of the four mechanisms and is characterised by the sample’s highest trust levels. By contrast, Meagre Selectivity produces values suggesting that strong marginalisation is possible through each of the four mechanisms, resulting in the sample’s lowest trust levels. Generous Selectivity focuses on minimising marginalisation through poverty and opportunity lack but still allows some marginalisation through behaviour and context, with social trust exhibiting average values. In contrast, Meagre Universalism has slightly worse values on the poverty and opportunity marginalisation mechanisms; while the values for behavioural and context marginalisation are better, it still has the sample’s lowest trust score. This again supports our claim that the relationship between the welfare state and social trust is not only influenced by the social programmes’ institutional design but also by the actual level of social spending dedicated by the state to funding social programmes.

Figure 3.

Source: The authors.

VI. Conclusion

The welfare state’s institutional design is an essential element in defining social trust levels not only among benefit recipients but among society as a whole. Our analysis suggests that the welfare state’s institutional framework is better understood when analysed in combination with the overall level of social spending. When universalism is generous and hence the welfare state offers a wide range of social programmes that cover the entire society, universalism can promote social trust formation. When public resources are scarce but the welfare state provisions are still organised on a universal basis, social trust levels may remain low. In this case, selectivity becomes a good alternative to universalism, by allowing the welfare state to direct its limited resources to those in need. Nonetheless, selectivity in social programmes cannot produce as much trust in a society, even with the same level of generosity in social programmes, as universal welfare states.

Our results also suggest that the outcome dimension is more important than the behavioural dimension in one’s experience with the welfare state regarding trust formation. Since reducing current levels of poverty and minimising the odds of experiencing poverty in the future by expanding one’s current set of opportunities can best be achieved through generous universalism or relatively generous selectivity, these modes of welfare state institutional design prove to be the most effective in advancing trust in society. Our findings can lay the foundation for designing welfare policies not only in Europe but also outside it, where current debates actively address the issue of the degree to which the state should intervene in the economy in general and in reducing inequality in particular (Aneja et al., 2020; Balasubramaniam et al., 2020; Chowdhury & Chowdhury, 2018; Datt et al., 2016; Paul & Lal, 2020; Wang, 2020).

Further analysis is still needed to validate the analytical model introduced by this study. First, it is necessary to test the dynamic nature of the relationship between the institutional design and trust levels by using longitudinal data. Second, our empirical results suggest that the impact of selectivity and universalism in social provisions may go through channels other than those proposed by this study thereby implying that we might not know enough about the overall set of mechanisms through which the welfare state impacts social trust. This can in turn be attributed to a more general problem concerning the lack of a clear theoretical framework that explains the trust building process in relation to social policy.

Footnotes

Acknowledgements

We thank reviewers for their insightful comments that essentially contributed to improving the manuscript's quality. We are also grateful to Stephanie Vogel for proof-reading the manuscript.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author received no financial support for the research, authorship and/or publication of this article.

Note

References

Aneja

Barkha & Banday

U. J.

(2020). Regional economic growth and inequality in India: A sector-wise decomposition analysis. Arthaniti: Journal of Economic Theory and Practice . https://doi.org/10.1177/0976747920910825

Arts

Gelissen

(2002). Three worlds of welfare capitalism or more? A state-of-the-art report. Journal of European Social Policy , 12(2), 137–158.

Balasubramaniam

Chatterjee

Mustard

D. B.

(2020). Public versus private investment in determining child health outcomes: Evidence from India. Arthaniti: Journal of Economic Theory and Practice , 19(1), 28–60.

Barber

(1983). The logic and limits of trust . Rutgers University Press.

Bergh

(2011). The rise, fall and revival of the Swedish welfare state: What are the policy lessons from Sweden? (IFN Working paper No. 873). Research Institute of Industrial Economics, Stockholm, Sweden.

Bjørnskov

(2005). The determinants of trust (Working paper). Retrieved from https://pure.au.dk/ws/files/39599381/cb_trust.pdf

Bjørnskov

(2007). Determinants of generalized trust: A cross-country comparison. Public Choice , 130(1–2), 1–21.

Boix

Posner

D. N.

(1998). Social capital: Explaining its origins and effects on government performance. British Journal of Political Science , 28(4), 686–693.

Brehm

Rahn

(1997). Individual level evidence for the causes and consequences of social capital. American Journal of Political Science , 41(3), 999–1023.

10.

Chowdhury

I. R.

Chowdhury

P. R.

(2018). Public–private partnerships, commitment and X-inefficiency. Arthaniti: Journal of Economic Theory and Practice , 17(2), 157–167.

11.

Christoforou

(2004, September 9–11). On the determinants of social capital in countries of the European Union . Paper presented at ESPAnet Conference ‘European social policy: Meeting the needs of a New Europe’, Oxford, England.

12.

Datt

Ravallion

Murgai

(2016). Four questions and answers on growth and poverty reduction in India. Arthaniti: Journal of Economic Theory and Practice , 15(2), 1–10.

13.

Farrell

(2005). Trust and political economy. Comparative Political Studies , 38(5), 459–483.

14.

Hall

P. A.

(1999). Social capital in Britain. British Journal of Political Science , 29(3), 417–461.

15.

Herreros

Criado

(2008). The state and the development of social trust. International Political Science Review , 29(1), 53–71.

16.

Jensen

Svendsen

G. T.

(2011). Giving money to strangers: European welfare states and social trust. International Journal of Social Welfare , 20(1), 3–9.

17.

Jordahl

(2007). Inequality and trust (IFN Working paper No. 715). Research Institute of Industrial Economics, Stockholm, Sweden.

18.

Kawachi

Kennedy

B. P.

Lochner

Prothrow-Stith

(1997). Social capital, income inequality and mortality. American Journal of Public Health , 87(9), 1491–1498.

19.

Knack

(2002). Social capital and the quality of government: Evidence from the states. American Journal of Political Science , 46(4), 772–785.

20.

Kumlin

(2004). The personal and the political: How personal welfare state experiences affect political trust and ideology . Palgrave Macmillan.

21.

Kumlin

Rothstein

(2007). Making and breaking social capital: The impact of welfare- state institutions. Comparative Political Studies , 38(4), 339–365.

22.

Larsen

C. A.

(2007). How welfare regimes generate and erode social capital: The impact of underclass phenomena. Comparative Politics , 40(1), 83–101.

23.

Lee

(2012). Welfare states and social trust. Comparative Political Studies , 46(5), 603–630.

24.

Lewis

D. J.

Weigert

(1985). Trust as a social reality. Social Forces , 63(4), 967–985.

25.

Luke

D. A.

(2004). Multilevel modeling . SAGE Publications.

26.

Nelson

Prilleltensky

(Eds.). (2005). Community psychology: In pursuit of liberation and well-being . Palgrave Macmillan.

27.

Newton

(1999a). Social trust and democracy in modern Europe. In van Deth

Maraffi

Newton

Whiteley

(Eds.), Social Capital and European Democracy (pp. 3–24). Routledge.

28.

Newton

(1999b). Social and political trust in established democracies. In Norris

(Ed.), Critical citizens: Global support for democratic government (pp. 169–187). Oxford University Press.

29.

Paul

Lal

(2020). Technology intensity and employment in the Indian economy. Arthaniti: Journal of Economic Theory and Practice . https://doi.org/10.1177/0976747919895326

30.

Paxton

(2002). Social capital and democracy: An interdependent relationship. American Sociological Review , 67(2), 254–277.

31.

Putnam

(1993). Making democracy work: Civic traditions in modern Italy . Princeton University Press.

32.

Putnam

(2000). Bowling alone . Simon & Schuster.

33.

Quintanilla

V. D.

(2012). Judicial mindsets: The social psychology of implicit theories and the law. Nebraska Law Review , 90(3), 611–646.

34.

Rothstein

(1998). Just institutions matter: The moral and political logic of the universal welfare state . Cambridge University Press.

35.

Rothstein

(2001a). Social capital in the social democratic welfare state. Politics and Society , 29(2), 207–241.

36.

Rothstein

(2001b). The universal welfare state as a social dilemma. Rationality and Society , 13(2), 213–233.

37.

Rothstein

(2003). Social capital, economic growth and quality of government: The causal mechanism. New Political Economy , 8(1), 49–71.

38.

Rothstein

(2005). Social traps and the problem of trust . Cambridge University Press.

39.

Rothstein

Stolle

(2001, November). Social capital and street-level bureaucracy: An institutional theory of generalized trust . Paper presented at ‘Trust in Government Conference’, Centre for the Study of Democratic Politics, Princeton University.

40.

Rothstein

Stolle

(2002, August–September). How political institutions create and destroy social capital: An institutional theory of generalized trust . Paper presented at 98th meeting of the American Political Science Association in Boston.

41.

Rothstein

Stolle

(2003). Introduction: Social capital in Scandinavia. Scandinavian Political Studies , 26(1), 1–26.

42.

Rothstein

Stolle

(2007). The quality of government and social capital: A theory of political institutions and generalized trust (QoG Working paper series 2). The Quality of Government Institute, Göteborg.

43.

Rothstein

Stolle

(2008). The state and social capital: An institutional theory of generalized trust. Comparative Politics , 40(4), 441–459.

44.

Rothstein

Uslaner

E. M.

(2005). All for all: Equality, corruption, and social trust. World Politics , 58(1), 41–72.

45.

Rothstein

Uslaner

E. M.

(2006). All for all: Equality and social trust (Working paper series no. 117). Center for European Studies.

46.

Simpson

(2006). The poverty of trust in the Southern United States. Social Forces , 84(3), 1625–1638.

47.

Tillmar

Lindkvist

(2007). Cooperation against all odds: Finding reasons for trust where formal institutions fail. International Sociology , 22(3), 343–366.

48.

Uslaner

E. M.

(2000, April). Trust, democracy and governance . Paper presented at the European Consortium for Political Research (ECPR) Workshop 13 on Voluntary Associations, Social Capital, and Interest Mediation: Forging the Link, Copenhagen, Denmark.

49.

Van Oorschot

Arts

(2005). The social capital of European welfare states: The crowding out hypothesis revisited. Journal of European Social Policy , 15(1), 5–26.

50.

Wang

L. F. S.

(2020). Relative performance versus market share delegation in a vertically related market. Arthaniti: Journal of Economic Theory and Practice , 19(1), 16–27.

51.

Yamagishi

Cook

Watabe

(1998). Uncertainty, trust and commitment formation in the United States and Japan. The American Journal of Sociology , 104(1), 165–194.