Who supports redistributive policies in contemporary Japan? An integrative approach to self-interest and trust models

Abstract

Spanish

French

This article investigates determinants of support for redistributive policies by taking an integrative approach with two main models that explain welfare attitudes: a self-interest model and a trust model. While the former focuses on effects of social positions and the latter focuses on effects of trust, there might be interactive effects between these two factors. The authors focus on three types of trust – generalized trust, primordial trust, and trust in institutions – and analyze how trust intervenes in effects of self-interest. They analyze data from a public opinion survey conducted in Japan, and present the results. Whereas generalized trust and trust in institutions alleviate effects of self-interest, primordial trust strengthens them. Furthermore, generalized trust increases support for selective redistributive policies regardless of one’s socio-economic status. These results prove that trust works differently according to one’s socio-economic position and that the three types of trust play different roles in support for redistribution policies.

Keywords

Japan public opinion redistributive policies trust welfare state

This article investigates the determinants of support for redistributive policies by taking an integrative approach to two major models that explain welfare attitudes: a self-interest model and a trust model.¹ The self-interest model assumes that socio-economic status affects support for redistributive policies (Iversen and Soskice, 2001; Rehm, 2009; Svallfors, 1997). The trust model, in contrast, regards generalized trust and trust in institutions as important determinants of welfare attitudes (Crepaz, 2008; Edlund, 1999; Rothstein, 1998; Svallfors, 2002; Van Oorschot, 2006). Although these two models have made great contributions to the study of support for redistributive policies, we argue that the supporting mechanism is more complicated than these models assume; we assume that there is an interaction between self-interest and trust, as explained later.

Our research contributes to existing literature by integrating the two models into one model on the basis of the Model of Frame Selection (hereafter MFS; Kroneberg, 2006). In their analysis of support for family transfer, Mehlkop and Neumann (2012) utilize the MFS in order to integrate the rational choice approach and the institutionalism approach which focuses on the impact of normative orientation. We concur with them and utilize the MFS in order to integrate the self-interest model and the trust model. By taking this approach, our research contributes to existing literature by clarifying the impact of trust on welfare attitudes as well.

Furthermore, taking the integrative approach is crucial when we consider the Japanese case. Despite public awareness of increasing economic disparity within society, Japanese citizens are reluctant to support redistribution policies. One study shows that the proportion of people who think that the government should prioritize ‘improvement of citizens’ welfare’ tends to decline during economic recessions (Kohno et al., 2009: 5). Furthermore, even though economic disparity has broadened in contemporary Japan, citizens have tended to withdraw from the comprehensive welfare state – a situation that has made the implementation of redistributive policies a controversial issue. Micro-level analysis found few effects of socio-economic status on welfare support (e.g. Kamimura, 2000), which means that even poor people do not show stronger support for social welfare than rich people. These studies imply that economic self-interest might intersect with another factor, i.e. trust.

Japan as a case

We focus on Japan as a case to test the empirical validity of the integrative approach. How the Japanese welfare state is positioned in a welfare regime typology has long been debated. Esping-Andersen (1997) tried to classify Japan in his ‘Three Worlds’ of welfare regime. The social democratic welfare regime is characterized by high levels of decommodification and defamilialization as well as a universalistic social security system. In contrast, the liberal welfare regime is characterized by low levels of decommodification. Defamilialization in a liberal welfare regime is high to some extent because it stresses market principles rather than familialism. The conservative welfare regime is characterized by corporatism and familialism. Social rights in this regime are based on employment and contribution, and families are expected to service their own welfare needs. Among these three welfare regime, Esping-Andersen characterized the Japanese welfare state as a conservative-corporatist model with characteristics of the liberal model.

As a conservative-corporatist welfare regime, the Japanese social insurance system is status-segregated. One difference between Japan and the conservative-corporatist regime is that the social insurance system is segregated not only according to occupations but also according to labor force ‘insiders’ (workers in a large firm) and ‘outsiders’ (workers in a small firms and non-regular workers) (Esping-Andersen, 1997: 183). Furthermore, workers in a large-sized firm can enjoy lifelong employment, fringe benefits, and life-stage adjustment wages (Imai, 2011). On the other hand, these benefits are not available to workers in a small-sized firm nor to non-regular workers. However, a large proportion of non-regular workers are married women whose husband is a regular worker. Their lives are supported by their husband in exchange for taking care of their children and parents. Strong familialism is another characteristic that Japan shares with the corporatist-conservative regime.

While firms and families support citizens’ lives, Japan’s welfare state is only residual; a feature that it shares with the liberal welfare regime type. Social assistance in Japan is means-tested and highly stigmatized (Fukuda, 1995; Goodman and Peng, 1996). As a result, according to estimations summarized by Fujisawa (2008), only around 20% of those who are eligible are supported by social assistance. This is much smaller than the take-up rate in other countries. For example, in Germany and the United Kingdom, it is around 70% (Eardley et al., 1996).

Although state welfare is residual, Japanese society has maintained a low poverty rate because of corporate welfare and family welfare. However, because of the pressure of globalization and economic recessions, firms cannot maintain lifelong employment. Under the influence of deregulation of labor market policy, they have been increasing their proportion of non-regular workers. As a result, a growing number of young male workers are employed as a non-regular worker. Despite this social change, state welfare has remained residual, which leads to increasing poverty and economic disparity. Figure 1 compares Gini coefficients before and after tax and transfers in 2009 among four countries: Japan, Germany, Sweden, and the UK. The gap between Gini coefficients before and after tax and transfers is the smallest in Japan (.15) among the four countries. A similar result can be seen in Figure 2, which compares the poverty rate (50% poverty line). The gap between before and after tax and transfer is .16 in Japan while it is higher in the other three countries (.23 in Germany, .19 in Sweden and .21 in the UK).

Figure 1.

Gini coefficient gap between before and after tax and transfer among four countries.

Figure 2.

Poverty rate gap between before and after tax and transfer among four countries (50% poverty line).

Self-interest versus trust

Welfare attitudes research has examined the effects of self-interest and trust. The self-interest model assumes that people judge policies based on the effects on their material interests. Thus, with regard to redistributive policies, it is assumed that less wealthy people or those who anticipate more risk of poverty support the policies because they would receive benefits from the policies.² In contrast, wealthier people or those who face little risk of poverty are against them because the policies would damage their interests.

Although this model is empirically supported in Western societies (Edlund, 1999; Iversen and Soskice, 2001; Rehm, 2009; Svallfors, 1997), its validity is controversial in Japan. Kamimura (2000) analyzes International Social Survey Program 1996 data and shows that economic conditions, as well as occupational status, do not affect support for redistributive policies; they are only affected by educational level. More recent data show the effects of socio-economic status on support for redistributive policies (Maruyama, 2011): those whose equivalent income is relatively low, the unemployed, and non-regular workers are likely to support redistributive policies. Furthermore, those who perceive a low risk of economic difficulty are more likely to support policies of higher social welfare benefits at higher costs (Takekawa, 2006; Yamada, 2006). Thus, we could conclude that relationships between self-interest and support for redistributive policies are more complicated than the original self-interest model assumes.

The trust model maintains that support for redistributive policies is influenced by levels of trust (Crepaz, 2008; Edlund, 1999; Rothstein, 1998; Svallfors, 2002). They investigate three different types of trust: generalized trust, primordial trust and trust in institutions, and these three types of trust relate to the legitimacy of a welfare state in different ways.

Generalized trust matters because it leads citizens to believe that welfare recipients deserve benefits. People support welfare policies that help the deserving needy (Larsen, 2008; Van Oorschot, 2000, 2006). According to Van Oorschot (2000), people use five criteria to judge who deserves welfare benefits: control, need, identity, attitudes, and reciprocity. Among these five criteria, the public think that three criteria – control, identity, and reciprocity – are the most important (Jæger, 2006). The control criterion considers whether people are responsible for their situations; the identity criterion concerns whether people can be regarded as ‘one of us’; and for the reciprocity criterion, the degree of reciprocation by people matters. Whether citizens assume that welfare recipients fulfill these criteria depends not only on the conditions of welfare recipients but also on their levels of trust (Crepaz, 2008; Van Oorschot, 2006; Voicu and Voicu, 2011).

Crepaz (2008) emphasizes the importance of separating generalized trust and primordial trust.³ Those who have strong generalized trust can have an idea that ‘ “we are all in the same boat” and see no contradiction in dutifully paying their taxes even though the benefits go to people they will never meet’ (Crepaz, 2008: 127). This statement implies that generalized trust facilitates fulfillment of the identity criterion for welfare recipients. Furthermore, people with high generalized trust expect that recipients of benefits will not cheat and will try hard to improve their situation (Van Oorschot, 2006). Thus, they support the policies. In contrast, according to Crepaz (2008), people who have high primordial trust may not support redistributive policies because they are concerned for the welfare of people they know well. They develop social norms that encourage them to prioritize people with whom they are familiar. Since ‘they know that many of their taxes go to people other than their immediate family clans and even to people of very different races, religions, or ethnic backgrounds,’ they ‘think twice whether they should pay their “fair” share’ (Crepaz, 2008: 127). This implies that primordial trust does not lead citizens to view welfare recipients as fulfilling the deservingness norm.

The third type of trust, trust in institutions, is related to a public view about what a state can do and what it cannot. As Rothstein (1998: 100) suggests, ‘it is quite possible for citizens to demand all the things offered by a general welfare state like the Swedish, yet reject the notion that it should be the responsibility of the state to provide them.’ In other words, when people regard governmental institutions and political systems as untrustworthy, they refuse state intervention. Thus, we can assume that redistributive policies cannot maintain legitimacy without public trust in institutions.

Although the three types of trust seem to be crucial factors that determine public support for redistributive policies, empirical studies show mixed results (e.g. Crepaz, 2008; Edlund, 1999; Svallfors, 2002; Voicu and Voicu, 2011). We maintain that a possible reason for this is that these studies miss the possibility that trust works differently according to one’s social status. In other words, self-interest and trust are interrelated when people judge whether they support redistributive policies.

Model of frame selection

We integrate the self-interest model and the trust model on the basis of the MFS. The MFS assumes that the decision-making process is composed of three selections of a frame (definition of the situation), a script (a set of appropriate actions), and an action (Kroneberg, 2006). Selection of action goes after selections of a frame and a script. If there is a frame which clearly fits the situation, people tend to choose their actions automatically and spontaneously. On the other hand, people will consider alternative utilities and costs and choose the one that has the maximum subjective expected utility if they cannot identify the frame that fits the situation. On the basis of the MFS, we can assume statistical interaction effects: ‘the effects of calculated incentives on the disposition for a certain behavior decrease, the more strongly a corresponding script is internalized, and that in the case of very strong internalization, other incentives are irrelevant’ (Kroneberg, 2006: 28).

From this viewpoint, we can assume that perceptions of welfare legitimacy regulate consideration of self-interests. If people strongly internalize legitimacy of redistribution by states, they support redistributive policies before calculating the individual incentives of the policies. In this case, trust in institutions and generalized trust weaken the effects of self-interests. If people believe that welfare recipients are the deserving needy and that government institutions work well, they do not doubt the legitimacy of redistributive policies even when they do not benefit from the policies. In contrast, we assume that primordial trust frames views about individual costs and benefits and thereby intensifies the effects of self-interest on support for redistributive policies.

In the next section, we focus on the Japanese case and investigate how a set of incentives is provided in Japanese welfare policies, and we establish hypotheses about the effects of self-interest and trust on support for redistributive policies in Japan.

Hypotheses about support for Japanese redistributive policies

As we mentioned before, the Japanese welfare system is largely segregated according to status in the labor market: ‘insiders’ and ‘outsiders.’ Furthermore, state welfare is small and residual. Given these welfare policies, we can assume that incentives of redistributive policies are constructed in the following ways. First, we can assume that ‘outsiders’ in the labor market (such as regular workers at small/medium-sized firms, non-regular workers, and the unemployed) have stronger support for redistributive policies than ‘insiders’ (such as regular workers and managers of large firms). We can also assume that people facing poverty receive more benefits from redistributive policies than wealthy people. Thus, we can establish the following hypothesis:

Hypothesis 1: People with lower socio-economic status, outsiders, and those facing poverty are more likely to support redistributive policies than insiders and those with higher socio-economic status.

Second, because a segregated social insurance system with a residual public benefit system clearly separates insiders and outsiders of a welfare system, primordial trust might affect support for redistributive policies differently according to one’s social position. According to institutionalism theory, a residual welfare state separates welfare recipients from tax payers (Crepaz, 2008; Rothstein, 1998). Means-tested welfare programs in liberal welfare states need to determine who belongs to the needy group and how much they need. Furthermore, such programs separate recipients from taxpayers by suggesting that ‘they’ (the recipients) live on ‘our’ (the taxpayers’) contributions. In this case, primordial trust strengthens the perceptions of costs and benefits of both insiders and outsiders. Thus, we can assume the following hypothesis:

Hypothesis 2: Primordial trust strengthens the effects of socio-economic status on support for redistributive policies.

Third, in contrast, generalized trust might alleviate the impact of self-interest because it weakens separation between ‘us’ and ‘them’ and creates the view that ‘we are in the same boat.’ In this case, those who cannot benefit from redistributive policies (i.e. those with stable socio-economic conditions) may still support redistributive policies. They have solidarity with poor people who they think do not cheat on the policies:

Hypothesis 3: Generalized trust alleviates the effects of socio-economic status on support for redistributive policies.

Fourth, as mentioned earlier, strong trust in institutions enables citizens to think state intervention legitimate. Thus, they support redistribution by a state before calculating their own incentives from the policies. Thus, we can assume that trust in institutions alleviates effects of socio-economic status on support for redistributive policies:

Hypothesis 4: Trust in institutions alleviates impact of socio-economic status on support for redistributive policies.

In the following section, we examine these hypotheses by analyzing public opinion survey data collected in 2012.

Data and methods

Data

We examine the above-mentioned hypotheses by analyzing data from the Web-Survey about Stratification and Social Psychology (SSP-W2012), which was collected in Japan in 2012. Samples were randomly selected from registered monitors of a research company. We first chose 250 areas from 1904 municipalities, and the registered monitors who lived in the areas were selected as samples. If areas with populations of fewer than 200,000 were chosen, we selected all the samples living in the same population-sized areas in the same prefecture. The final sample size was 2839 (Todoroki, 2012).⁴ When missing cases were removed, the new sample contained 1928 cases.

Dependent variables

Two variables on respondents’ attitudes towards redistributive policies are used as dependent variables in our analysis.⁵ The first variable is the response to the statement, ‘The government should commit more money to social welfare for disadvantaged people even though it increases taxes on wealthy people.’ Responses to this question are 5 = Agree, 4 = Somewhat agree, 3 = Neither agree nor disagree, 2 = Somewhat disagree, 1 = Disagree. We call this variable the support for selective redistributive policies.

The second dependent variable is the response to the statement, ‘It is more important to reduce inequality than to maintain the freedom of competition.’ Responses to this question are the same as the first variable. We call this variable the support for equality policies.

Correlation between these two variables is positive but not very high (.319, p < .01). This implies that these two variables capture different forms of redistributive policies. The former variable captures public opinion about redistribution from the rich to the poor, which can be residual and selective, while the latter variable captures public opinion about more general redistribution among citizens.

Independent variables

The major groups of independent variables are, as we argue above, trust and socio-economic status. Trust in our analysis consists of three types.⁶ The first type is trust in institutions. The survey asked respondents how much they trust the central government offices, local governments, the national government, courts, Diet members, local assembly members at the prefectural level, and local assembly members at the municipal level. Respondents were asked to choose one of the following alternatives for each question: 5 = Agree, 4 = Somewhat agree, 3 = Neither agree nor disagree, 2 = Somewhat disagree, 1 = Disagree. We use six trust variables, except for trust in courts, for factor analysis with the principle factor method and define the score of the first factor as the trust in institutions variable.⁷ Cronbach’s alpha of these six items is .898, and eigenvalue of the first principle component is 3.987.

The second type of trust is generalized trust. The survey listed six statements about generalized trust:

‘Most people are basically honest.’

‘Most people are trustworthy.’

‘Most people are basically good and kind.’

‘Most people are trustful of others.’

‘I am trustful.’

‘Most people will respond in kind when they are trusted by others.’

Respondents were asked to choose one of five alternatives, which were the same as those for trust in institutions. Cronbach’s alpha of these six items is .881. We apply factor analysis with the principle factor method to these variables and use the score of the first factor as generalized trust. The eigenvalue of the first factor is 3.779.

We also use primordial trust as the third type of trust. As argued above, primordial trust is defined as particularized trust in those whom people know well. It is measured by responses to the following four statements, and respondents were asked to choose one of five alternatives, which were the same as those for trust in institutions:

‘I trust a person I know well more than one whom I do not know.’

‘Whatever work I have to perform, I feel more secure when I work with someone I know well than with someone I do not know.’

‘Generally, a person with whom you have had a longer relationship is likely to help you when you need it.’

‘The people I trust are those with whom I have long-lasting relationships.’

Cronbach’s alpha of these four items is .821. We apply factor analysis with the principle factor method to these variables and use the score of the first factor as primordial trust. The eigenvalue of the first factor is 2.611.

The second group of major independent variables that represent self-interest is equivalent income, education, and employment status. We also use class identification because people’s political attitudes are often affected by their subjective positions in social stratification or class structure (Heys, 1995; Naoi and Tokuyasu, 1990). Furthermore, class identification synthesizes not only objective conditions, such as educational levels, occupations, and income, but also subjective socio-economic conditions, such as perceptions about opportunities to improve one’s own economic condition and life-satisfaction (Nakao, 2002). Because people take both of these factors into account when they consider costs and benefits of welfare policies, we use class identification as an independent variable. We categorize class identification into three groups: upper class (those who rank themselves as High or Upper Middle), middle class (those who rank themselves as Middle), and lower class (those who rank themselves as Lower Middle or Low). There are large similarities among objective socio-economic positions, as well as subjective socio-economic positions, between High and Upper Middle or between Lower Middle and Low (Nakao, 2002).

Education is classified into two categories: low/middle (those with upper-secondary education or less) and high (those with more than upper-secondary education). Employment status is classified into six categories: managers, regular workers in a large firm, regular workers in a small/medium-sized firm, non-regular workers, the self-employed, the unemployed, and those not in a labor market.

As shown later, among these variables equivalent income and class identification have coherent strong effects on support for redistributive policies. Thus, we focus only on these variables when we discuss how trust intervenes with effects of self-interest on support for redistributive policies.

Age, gender (male = 1, female = 0), marital status (married, divorced/bereaved, never married), having children or not, and religious belief are used as control variables. Previous research has pointed out that a religious attitude is an important determinant of support for redistribution (e.g. Ervasti, 2012; Yokoi and Kawabata, 2013). Therefore, we include it in the analysis as a control variable. It is defined by a question ‘how important do you think religious mind is?’, as in a previous study (Yokoi and Kawabata, 2013). Answers to the question are given on a five-point scale.

Results

Descriptive statistics

First, we look into the extent to which Japanese people support selective redistributive and equality policies (Figure 3). We can see that opinions about these policies are divided among the population. With regard to support for equality policies, more than 40% of respondents do not have clear opinions about the policies. Around 30% of respondents support the policies, while the same proportions of respondents are resistant to the policies. Japanese people are more supportive of selective redistributive policies: approximately 49% of respondents support these policies, while only 15% of respondents oppose them. From these figures we can assume that Japanese people tend to support redistribution from the rich to the poor but are rather careful when they consider the level of equality that is desired.

Figure 3.

Support for selective redistributive and equality policies (%).

Descriptive statistics of other variables in this article are summarized in Table 1.

Table 1.

Descriptive statistics of independent variables.

		Mean	SD
Generalized trust	Honest	2.3	0.8
	Trustworthy	2.3	0.8
	Good and kind	2.4	0.8
	Trustful of others	2.2	0.8
	Trustful	2.7	0.8
	Respond in kind	2.8	0.7
Primordial trust	Trust person I know	3.0	0.7
	Feel secure	3.1	0.7
	Longer relationship	3.0	0.7
	Long-lasting relationships	3.0	0.7
Trust in institutions	Central government offices	1.9	0.7
	Local governments	2.2	0.7
	The national government	1.7	0.7
	Diet members	1.6	0.6
	Local assembly members at prefectural level	1.7	0.7
	Local assembly members at municipal level	1.8	0.7
Age		43.1	10.0
Equivalent income (logged)		5.7	0.9
Religious attitudes		2.7	1.0
		%
Gender	Male	54.3
Education	Lower/middle	38.0
Occupational status	Managers	1.0
	Regular worker in a small/medium-sized firm	23.9
	Regular worker in a large firm	25.4
	Non-regular workers	18.9
	Self-employed	11.3
	Unemployed	6.2
	Not in a labor market	14.4
Marital status	Married	70.9
	Never married	22.7
	Divorced/bereaved	6.4
Having a child		62.0
Class identification	Upper class	21.3
	Middle class	45.7
	Lower class	33.0

N = 1928.

Results of multiple regression analyses

We conducted regression analyses with the support for selective redistributive policies and the support for equality policies as dependent variables,⁸ and three types of trust and socio-economic status variables, which represent self-interest, as major independent variables. We include interaction terms between trust variables and socio-economic status variables in the analyses to test the empirical validity of Hypotheses 2, 3, and 4.

Table 2 shows the results of a multivariate regression analysis of support for selective redistributive policies. We examine the main effects of socio-economic status and trust in Model 1 and the interaction effects between them in Models 2 and 3.

Table 2.

Determinants of support for selective redistributive policies.

	Model 1		Model 2		Model 3
	B	SE	B	SE	B	SE
Intercept	3.517**	.094	3.506**	.094	3.520**	.093
Age	.008**	.003	.008**	.003	.008**	.003
Male	.023	.055	.025	.055	.019	.055
Low education	.054	.048	.060	.048	.057	.048
Managers	−.036	.256	−.057	.252	−.038	.247
Regular workers in a small/medium-sized firm	.016	.066	.023	.066	.016	.066
Non-regular workers	.122	.079	.125	.079	.119	.079
Self-employed	−.040	.088	−.038	.088	−.045	.088
Unemployed	.222*	.109	.235*	.109	.224*	.109
Not in a labor market	.063	.084	.062	.084	.050	.084
Married	−.025	.078	−.026	.078	−.019	.078
Divorced/bereaved	.114	.121	.116	.120	.127	.121
Having a child	.073	.066	.073	.065	.065	.066
Religious attitudes	.079**	.025	.079**	.025	.078**	.024
Generalized trust	.091**	.028	.090**	.028	.101*	.050
Primordial trust	.125**	.028	.126**	.028	.186**	.045
Trust in institutions	−.024	.026	−.024	.026	−.082+	.046
Equivalent income	−.083**	.030	−.062*	.028	−.084**	.030
* generalized trust			.030	.029
* primordial trust			−.009	.025
* trust in institutions			.087**	.026
Middle class	−.192**	.054	−.189**	.054	−.191**	.053
* generalized trust					−.002	.062
* primordial trust					−.068	.060
* trust in institutions					.045	.059
Upper class	−.376**	.070	−.380**	.070	−.368**	.071
* generalized trust					−.038	.081
* primordial trust					−.173*	.079
* trust in institutions					.166*	.073
Adj. R²	.064**		.069**		.067**
AIC	5398.66		5391.76		5398.75

N = 1928, ** p < .01, * p < .05, + p < .1.

People who are in lower socio-economic positions – the unemployed, people with lower equivalent incomes, and those who perceive themselves as ‘lower class’ – are more likely to support selective redistributive policies in all models. These results support the idea that people consider self-interest when they decide whether they should support selective redistributive policies. Thus, Hypothesis 1 is supported. On the other hand, there is no significant difference in levels of support between regular workers in a large firm and regular workers in a small/medium-sized firm or non-regular workers, although coefficients of the latter are positive as we would expect. As in the previous research, a strong religious attitude increases support for selective redistributive policies.

Our primary interests lie in how the effects of socio-economic status differ according to levels of trust. Models 2 and 3 show positive interaction effects between socio-economic status and trust in institutions. Figure 4 illustrates the interaction effects between equivalent income and trust in institutions. As we can see, levels of support do not change according to equivalent income among those with high level of trust in institutions (above average by one standard deviation). In contrast, equivalent income affects levels of support among those with low levels of trust in institutions (below average by one standard deviation). A post hoc simple slope test shows the latter effects are significant while the former effects are not. Similar results can be seen with regard to the interaction effects between class identification and trust in institutions; trust in institutions alleviates effects of class identification. Hypothesis 4 is thus supported.

Figure 4.

Interaction effects of equivalent income and trust in institutions on support for selective redistributive policies (male, higher education, a regular worker in a large firm, middle class identification, never married, having no child, and other variables fixed at means).

What is more interesting is that there is a negative interaction effect between upper class perception and primordial trust. This interaction effect is illustrated in Figure 5. While levels of support differ according to one’s class identification with regard to high primordial trusters, it does not change levels of the support in the case of low primordial trusters. The figure implies that primordial trust strengthens self-interest especially among those who are in lower social positions. A simple slope test shows that effects of class identification are smaller among low primordial trusters than among high primordial trusters. Thus, Hypothesis 2 is supported.

Figure 5.

Interaction effects of class identification and primordial trust on selective redistributive policies (male, higher education, a regular worker in a large firm, never married, having no child, and other variables fixed at means).

Furthermore, there is no interaction effect between generalized trust and socio-economic status, which means that generalized trust cultivates support for selective redistributive policies regardless of one’s social position. From this viewpoint, Hypothesis 3 is not supported.

In the case of support for equality policies, trust and self-interest work in a similar way as support for redistributive policies, but there are a few differences (see Table 3). First, as in the case of support for selective redistributive policies, people with a lower socio-economic status – less-educated people, people with lower incomes, and those who rank themselves in lower social positions – are more likely to support equality policies than their counterparts in all models. These results support the self-interest model and thus Hypothesis 1. In addition, as in the analysis of support for selective redistributive policies, a religious attitude increases support for equality policies.

Table 3.

Determinants of support for equality policies.

	Model 1		Model 2		Model 3
	B	SE	B	SE	B	SE
Intercept	3.190**	.092	3.187**	.092	3.191**	.092
Age	.002	.003	.002	.003	.002	.003
Male	−.219**	.053	−.218**	.052	−.222**	.052
Low education	.143**	.047	.144**	.047	.146**	.047
Managers	−.133	.244	−.168	.240	−.149	.242
Regular workers in a small/medium-sized firm	−.035	.067	−.033	.067	−.036	.067
Non-regular workers	.090	.075	.089	.075	.082	.075
Self-employed	−.030	.085	−.023	.085	−.039	.085
Unemployed	.159	.107	.148	.107	.153	.109
Not in a labor market	−.195*	.082	−.198*	.082	−.206*	.082
Married	−.061	.075	−.063	.074	−.059	.075
Divorced/bereaved	−.129	.118	−.121	.116	−.123	.118
Having a child	.081	.062	.070	.062	.075	.062
Religious attitudes	.079**	.025	.078*	.025	.079**	.025
Generalized trust	.046	.029	.042	.029	.081+	.048
Primordial trust	.012	.027	.010	.027	.039	.042
Trust in institutions	.048+	.026	.049+	.026	−.038	.046
Equivalent income	−.123**	.032	−.123**	.034	−.125**	.032
* generalized trust			.088**	.033
* primordial trust			−.060*	.023
* trust in institutions			.038	.033
Middle class	−.151**	.053	−.140**	.053	−.147**	.053
* generalized trust					−.065	.062
* primordial trust					−.007	.059
* trust in institutions					.111+	.059
Upper class	−.307**	.069	−.300**	.069	−.299**	.069
* generalized trust					−.010	.078
* primordial trust					−.121+	.072
* trust in institutions					.141+	.072
Adj. R²	.069**		.075**		.071**
AIC	5312.57		5302.69		5315.22

N = 1928, ** p < .01, * p < .05, + p < .1.

Effects of occupational status do not seem to be consistent with Hypothesis 1. Those who are not in a labor market show less support for equality policies than regular workers in a large firm. However, the negative effects disappear if we exclude equivalent income from the model. It means that those who are not in the labor market are less supportive of equality policies than regular workers in a large firm if they are in similar economic conditions. Those who are not in the labor market are mainly homemakers. Eighty percent of them are married women. Therefore, we can assume that their protected social status (married women whose husbands work as a regular worker) might affect their weak support for equality policies.

When we look at the interaction effects of the three types of trust and socio-economic status, there is a positive interaction effect between equivalent income and generalized trust. Figure 6 illustrates this interaction effect. As we can see from Figure 6, equivalent income does not affect levels of support among high generalized trusters while it affects levels of support among low generalized trusters. The latter effect is statistically significant (B = –.207, p = .000) while the former effect is not statistically significant. It implies that generalized trust weakens the self-interest, which supports Hypothesis 3.

Figure 6.

Interaction effects of equivalent income and generalized trust on support for equality policies (male, higher education, a regular worker in a large firm, middle class identification, never married, having no child, and other variables fixed at means).

Furthermore, there is a negative interaction effect between equivalent income and primordial trust. The effects of equivalent income are larger among high primordial trusters (B = –.176, p = .000) than among low primordial trusters (B = –.068, p = .035). Primordial trust strengthens the self-interest of those who are in lower social positions. This result supports Hypothesis 2.

In Model 3, we examine interaction effects between class identification and trust. As a result, there are positive interaction effects between trust in institutions and upper class or middle class perception. A simple slope test shows that effects of class identification are larger among those with low trust in institutions than among those with high trust in institutions. Furthermore, there is a significant interaction effect between upper class identification and primordial trust. Figure 7 shows the interaction effect. It shows that effects of primordial trust clearly appear among upper class people. With regard to equality policies, strong primordial trust intensifies the self-interest of the upper strata of society. In contrast, there is no significant effect of an interaction term between class identification and generalized trust. Thus, we can conclude that generalized trust alleviates self-interest not of those who perceive themselves as rich but of those, in particular, who are actually rich.

Figure 7.

Interaction effects of class identification and primordial trust on support for equality policies (male, higher education, a regular worker in a large firm, never married, having no child, and other variables fixed at means).

Conclusions and discussion

This research examines how self-interest and trust affect support for redistributive policies. In particular, we focus on whether the effects of socio-economic position differ according to one’s levels of trust, which affect the impact of self-interest. The results of our analysis are summarized as follows.

First, self-interest has a strong impact on support for redistributive policies. Those with lower equivalent incomes and those perceiving themselves as lower class are more likely to support both selective redistributive and equality policies. However, education differentially affects two types of policies. Differences in support for equality policies can be found between those with lower/middle education and higher education, but differences in support for selective redistributive policies cannot be found between them. One possible explanation for these results is that low support for equality policies among those with higher education relates to preferences for free competition. Previous studies show that those with higher education have strong preferences for free competitions (Doba, 2011; Maruyama, 2011). Since they have won higher positions in society in free competition, they have strong preferences for it and thereby disapprove of equality policies that might risk it. At the same time, they do not necessarily disapprove of selective redistribution which eliminates inequality caused by competition.

Although the Japanese welfare system is highly segregated between ‘insiders’ and ‘outsiders,’ we find no clear difference between these two groups in support of the redistribution policies. This result is consistent with previous studies that show there is no clear relationship between welfare regime and difference in welfare attitudes (e.g. Taylor-Gooby, 2001). We need further investigation of how they are interrelated.

Second, as we predicted in Hypothesis 2, primordial trust strengthens the effect of self-interest. People with strong primordial trust tend to be concerned about whether redistributive policies benefit their own acquaintances rather than whether these policies are good for society as a whole. This might especially be the case in Japan, where the social security system is highly segmented by occupational status. Trust in people like ‘us’ is assumed to be a necessary bond that unites a society and maintains a comprehensive welfare state (e.g. Miller, 1995). However, our results show that if the range of ‘us’ is too narrow, it might disunite a society and cause withdrawal of the upper class from support for redistributive policies.

Third, generalized trust actually strengthens support for redistributive policies. The effects of equivalent income on the support for equality policies decrease as levels of generalized trust increase. Generalized trust alleviates the effects of self-interest. On the other hand, generalized trust increases support for selective redistributive policies regardless of one’s social position. It implies that generalized trust works as a basis of redistribution between the poor and the rich. It might be related to the evaluation of the deservingness of the needy. Generalized trust cultivates positive views of the needy, and thus makes it possible for citizens with high levels of generalized trust to accept the idea of redistribution from the rich to the poor. In other words, if they do not trust the needy, they are more reluctant to accept the idea of redistribution, even if these policies benefit themselves.

Fourth, as we expected in Hypothesis 4, trust in institutions alleviates the effects of self-interest. This suggests that trust in institutions is required to incline upper class people to accept redistribution. If they cannot trust the government to manage their money effectively, they hesitate to pay for the welfare state.

Our findings suggest that trust works differently according to one’s socio-economic position and that the three types of trust play different roles in support for redistribution policies. These results suggest that people’s calculations of costs and benefits are quite complicated: not only are direct interests related to their calculations, but also their perceptions about other citizens and government matter. Thus, in order to achieve a clearer understanding of the mechanism on how ‘rational choice’ is created, we need to consider both these factors.

Our primary consideration is why Japanese people do not support redistributive policies, even though the importance of these policies has increased in recent years. From these results, we can conclude that generalized trust and trust in institutions are the key. These factors encourage people to consider redistributive policies beyond self-interest. Especially, generalized trust works as a basis of support for redistributive policies. Low levels of generalized trust and low levels of institutional trust in the upper social strata are causes of Japanese hesitation in accepting the idea of redistribution. Miyamoto (2008) points out that Japanese attitudes towards welfare states are colored by distrust in government agencies. Our results support the validity of this statement.

However, our research has some limitations. First, as already mentioned, people with higher social status are overrepresented in our sample. Thus, we should be reserved about the generalizability of our results. We need to test the validity of our findings by analyzing a national representative random sample. Second, our research focuses on the Japanese case, so we did not examine how relationships between trust, socio-economic status, and support for redistributive policies are framed by welfare policies. According to institutionalism theory, welfare policies might affect people’s perceptions of costs and benefits. Thus, cross-national comparative studies are required to fully understand a mechanism in which people support or do not support redistributive policies.

Footnotes

Funding and acknowledgements

This work was supported by JSPS KAKENHI Grant Numbers 16001001, 23223002, 90196288. This work was conducted as a part of the SSP project (ssp.hus.osaka-u.ac.jp/). We use SSP-W data with the permission of the SSP project.

Notes

Author biographies

Kikuko Nagayoshi is Associate Professor of Sociology at Tohoku University. Her research interests include public opinions about welfare policies, immigration policies and immigrants. She has contributed a chapter to Japanese Perceptions of Foreigners (Trans Pacific Press, 2013).

Yoshimichi Sato is Professor of Sociology at Tohoku University. His current research focuses on inequality between regular and non-regular workers in Japan, and the results of his research have appeared in Japan’s New Inequality: Intersection of Employment Reforms and Welfare Arrangements (Trans Pacific Press, 2011).

References

Breen

Goldthorpe

(1997) Explaining educational differentials: Towards a formal rational action theory. Rationality and Society 9(3): 275–305.

Crepaz

MML

(2008) Trust Beyond Borders. Michigan: The University of Michigan Press.

Doba

(2011) Kakusa to seijiteki kachi: Meritokurashi shakai no rinen to shimin shakai no rinen. In: Saito

Misumi

(eds) Gendai no Kaiso Shakai 3 Ryudoka no Naka no Shakai Ishiki. Tokyo: Tokyo Daigaku Shuppankai, pp. 205–217.

Eardley

Ditch

Gough

Whiteford

(1996) Social Assistance in OECD Countries. London: HMSO.

Edlund

(1999) Trust in government and welfare regimes: Attitudes to redistribution and financial cheating in the USA and Norway. European Journal of Political Research 35(3): 341–370.

Edlund

(2006) Trust in the capability of the welfare state: Sweden 1997–2002. Acta Sociologica 49(4): 395–417.

Ervasti

(2012) Is there a religious factor involved in support for the welfare state in Europe? In: Ervasti

Andersen

Fridberg

Ringdal

(eds) The Future of the Welfare State: Social Policy Attitudes and Social Capital in Europe. Northampton, MA: Edward Elgar, pp. 214–229.

Esping-Andersen

(1997) Hybrid or unique? The Japanese welfare state between Europe and America. Journal of European Social Policy 7(3): 179–189.

Fujisawa

(2008) Nihon no teishotoku to seikatsuhogo seido: JGSS deta ni yoru shakaihujo jukyuu ketteiyouin bunseki wo tsujite. JGSS Research Series 4: 271–283.

10.

Fukuda

(1995) Seikatsuhogo seido no shakaishi. Tokyo: Tokyou Daigaku Shuppankai.

11.

Goodman

Peng

(1996) The East Asian welfare states. In: Esping-Andersen

(ed.) Welfare States in Transition. London: Sage, pp. 192–224.

12.

Heys

(1995) The impact of class on political attitudes: A comparative study of Great Britain, West Germany, Australia, and the United States. European Journal of Political Research 27(1): 69–91.

13.

Imai

(2011) Transformation of Japanese Employment Relations: Reform without Labor. Basingstoke: Palgrave Macmillan.

14.

Iversen

Soskice

(2001) An asset theory of social policy preference. American Political Science Review 95(4): 875–893.

15.

Jæger

(2006) Welfare regimes and attitudes toward redistribution: The regime hypothesis revisited. European Sociological Review 22(2): 157–170.

16.

Kamimura

(2000) Hukushi kokka ha ima nao sijisarete iruka. In: Satoh

Ishida

Ikeda

(eds) Shakai Chosa no Koukai Data. Tokyo: Tokyo Daigaku Shuppankai, pp. 211–226.

17.

Kohno

Takahashi

Hara

(2009) Nihonjin no ishiki henka no sanjugonen no kiseki (2): Daihakkai ‘Nihonjin no Ishiki 2008’ chosa kara. Houso Kenkyu to Chosa April: 1–39.

18.

Kroneberg

(2006) The Definition of the Situation and Variable Rationality: The Model of Frame Selection as a General Theory of Action. Sonderforschungsbereich 504 Publications 06–05. Mannheim: Universität Mannheim.

19.

Larsen

(2008) The institutional logic of welfare attitudes: How welfare regimes influence public support. Comparative Political Studies 41(2): 145–168.

20.

Maruyama

(2011) Neo riberarizumu: Sono tagensei to tairitsu jiku no kousaku. In: Tanabe

(ed.) Gaikokujin heno Manazashi to Seijiishiki. Tokyo: Keiso Shobou, pp. 119–140.

21.

Mehlkop

Neumann

(2012) Explaining preferences for redistribution: A unified framework to account for institutional approaches and economic self-interest for the case of monetary transfers for families and children. European Journal of Political Research 51(3): 350–381.

22.

Miller

(1995) On Nationality. Oxford: Oxford University Press.

23.

Miyamoto

(2008) Fukushi Seiji: Nihon no Seikatsu Hosho to Demokurashi. Tokyo: Yuhikaku.

24.

Nakao

(2002) Kaiso kizokuishiki to seikatsuishiki. Riron to Houhou 17(2): 135–149.

25.

Naoi

Tokuyasu

(1990) Seitousijiishiki. In: Hara

(ed.) Gendai Nihon no Kaisou Kouzou 2 Kaisou no Doutai. Tokyo: Tokyo Daigaku Shuppankai, pp. 149–172.

26.

Putnam

(2000) Bowling Alone: The Collapse and Revival of American Community. New York: Simon and Schuster.

27.

Rawls

(1999) A Theory of Justice, rev. edn. Oxford: Oxford University Press.

28.

Rehm

(2009) Risks and redistribution: An individual-level analysis. Comparative Political Studies 42(7): 855–881.

29.

Rothstein

(1998) Just Institutions Matter: The Moral and Political Logic of the Universal Welfare State. Cambridge: Cambridge University Press.

30.

Svallfors

(1997) Worlds of welfare and attitudes to redistribution: A comparison of eight western nations. European Sociological Review 13(3): 283–304.

31.

Svallfors

(2002) Political trust and support for the welfare state: A comparison of Sweden and Norway. In: Rothstein

Steinmo

(eds) Restructuring the Welfare State. New York: Palgrave, pp. 84–205.

32.

Takekawa

(2006) Fukushi kokka wo sasaeru kachiishiki. In: Takekawa

(ed.) Fukushi Shakai no Kachiishiki. Tokyo: Tokyo Daigaku Shuppankai, pp. 185–206.

33.

Taylor-Gooby

(2001) Sustaining state welfare in hard times: Who will foot the bill? Journal of European Social Policy 11(2): 133–147.

34.

Todoroki

(2012) SSP-W2012 Kodo Bukku Oyobi Kisoshukeihyo. Kanazawa: SSP Purojekuto Webuchosa Jisshibumon.

35.

Van Oorschot

(2000) Who should get what, and why? On deservingness criteria and the conditionality of solidarity among the public. Policy and Politics 28(1): 33–49.

36.

Van Oorschot

(2006) Making the difference in social Europe: Deservingness perceptions among citizens of European welfare states. Journal of European Social Policy 16(1): 23–42.

37.

Voicu

(2011) How sociability and trust impact on welfare attitudes: A cross-European analysis. Revista de Cercetare si Interventie Sociala 33: 72–90.

38.

Yokoi

Kawabata

(2013) Shukyosei no Sokutei: Kokusai Hikaku Kenkyu wo Mezashit. Shukyo to Shakai 19: 79–95.

39.

Yamada

(2006) Kazokukeitai to fukushi ishiki. In: Takekawa

(ed.) Fukushi Shakai no Kachiishiki. Tokyo: Tokyo Daigaku Shuppankai, pp. 139–150.

40.

Yosano

Hayashi

(2005) Social stratification, intermediary groups and creation of trustfulness. Sociological Theory and Methods 20(1): 27–44.