ISSN 0080-6757 Doi: 10.1111/1467-9477.12089 C 2017 Nordic Political Science Association V
Inter-level Trust in a Multilevel Political System David Karlsson* The aim of this article is to study political representatives in a multilevel government system and their level of political trust in relation to institutions in their own and other tiers of government. The results show that two factors are essential for understanding variations in their trust: Representatives tend to trust institutions in other tiers less than those in their own tier (the tier-association effect), and they trust institutions more where their own party is in a governing position (the home-team effect). Of the two, the home-team effect is by far the most dominant factor. Multiple mandates serve to increase inter-level trust, but divided loyalties possibly make representatives with multiple mandates less trusting towards political institutions in their own tiers than more partisan single mandate representatives. The study builds on data from surveys conducted among all elected representatives in Sweden in 2012 (local and regional councillors) and 2014 (national MPs).
Introduction The Scandinavian countries are some of the most ambitious and decentralised welfare states in the world as their extensive public services are largely assigned to local and regional authorities (Heinelt & Hlepas 2006; Loughlin et al. 2010). Public responsibilities are distributed among the tiers of government in an attempt to balance the values of cost-effective service production and local self-governance. Arguably, the decentralisation strategy has been a key part of the success of the Nordic welfare state model (Sellers & Lidstr€ om 2007). However, the reliance on a multitude of self-governing entities also poses serious challenges for a political system as national, regional and local political leaders often have diverging agendas. Although political representatives are interdependent on one another for achieving political goals in their respective autonomous tiers of government, they cannot easily control decisions that are made in tiers other than their own. In any situation where actors need to rely on other autonomous actors, mutual trust is key to successful collaboration. For actors in a multilevel * David Karlsson, School of Public Administration, University of Gothenburg, Post Box 712, SE40530, G€ oteborg, Sweden. E-mail:
[email protected] Scandinavian Political Studies, Vol. 00 – No. 00, 2017
1
C 2017 Nordic Political Science Association V
government system, this entails mutual confidence in the abilities of actors and institutions in other tiers of government. In this study, this special kind of confidence will be referred to as ‘inter-level political trust’. Fortunately, the Scandinavian countries are high-trust societies (Delhey & Newton 2005; Catterberg & Moreno 2006), with – as we shall see – decent levels of inter-level political trust among political representatives (see also Aars 2014). To maintain inter-level political trust at acceptable levels, however, it is vital to understand which factors strengthen and weaken this form of trust. Factors affecting inter-level political trust that relate to the institutional setting and actors’ roles in political organisations are of particular interest, as such factors may be easily altered by organisational reforms. The structure of political organisations define the interests and goals of its actors (Egeberg 2003), and changes in the institutional settings in which political representatives are positioned are therefore likely to fundamentally affect their perceived roles, interests, loyalties – and trust. The main aim of this article is to study political representatives in a multilevel government system and the factors affecting their level of political trust in institutions in their own and other tiers of government. The focus will be on inter-level trust, with intra-level trust as an essential reference point. The analysis is built on the combined data from two surveys that cover all elected political representatives in Sweden in the local, regional and national tiers of government. The European tier of government is not included in this analysis due to the lack of relevant data, and not a lack of relevance. The Research Questions It is quite likely that the trust of political representatives is affected by the same factors as citizens’ trust. Social sciences have put a lot of effort into studying citizens’ political trust in institutions in all tiers of government (e.g., Denters 2002; Noren Bretzer 2005; Berg 2007; Aars & Christensen €derlund 2015), and when citizens evaluate 2013; Kestil€ a-Kekkonen & So political institutions, it is often found that legitimacy based on procedures and performance is central to the generation of trust (e.g., Gustavsen et al. 2014; Roos & Lidstr€ om 2014). However, the roles and interests of representatives differ from those of citizens in a number of significant ways that could affect their trust levels. Compared with the average citizen, political representatives are much more involved and informed about political processes, and they are able to relate, from their own experiences, to the challenges of governing faced by others. Moreover, insider familiarity with some institutions in different tiers of government may influence their political trust. This study will especially focus 2
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
C 2017 Nordic Political Science Association V
on two factors relating to representatives’ positions within the political system that distinguish them most from citizens: their association with a particular tier of government; and their associations with a particular party. First, whereas a citizen is no more a citizen in one tier of government than another (even if they may feel more attached to certain tiers; Berg 2007), representatives are much more strongly associated with the tier (or tiers) in which they have been elected to office. Representatives are also directly and personally responsible for the outcome of political processes in this tier. This study’s first line of research questions is therefore concerned with if and how political representatives’ tier-association affects their trust. Do representatives trust institutions in other tiers of government more or less than institutions in their own tier? And does it matter whether representatives are elected into only one tier of government (single mandate representatives) or in more than one tier (multiple mandate representatives)? Second, as party operatives, representatives are far more invested in the success of their own party, both from an ideological and a personal career point of view. Their assessments of political institutions are therefore more likely to be coloured by their parties’ roles in those institutions. This gives rise to a second series of questions relating to representatives’ party affiliation. Do representatives trust institutions more in tiers where their own party is in government? And if this is the case, does this ‘home-team effect’ (Holmberg 1999) differ between different types of institutions? And if there is a home-team effect, is this effect more consequential than the tier-association effect?
The Factors behind Inter-level and Intra-level Political Trust: Five Hypotheses What is Inter-level and Intra-level Political Trust, and Why is It Important? Political trust generally refers to people’s ‘confidence in institutions such as parliament, the state bureaucracy and the courts, based upon their sense of how these institutions work’ (Newton 2007). Studies of political trust tend to focus on citizens, but inter-level political trust specifically relates to the confidence political actors – especially elected political representatives – in different tiers of government have in the abilities of institutions in other tiers. Correspondingly, intra-level political trust is the confidence of political actors in the abilities of institutions in their own tier. Mutual trust between political actors is crucial to successful policy implementation, especially in situations when implementers need autonomy to deal with uncertainties by flexibility and learning (Lane 1987). Confidence in Scandinavian Political Studies, Vol. 00 – No. 00, 2017
3
C 2017 Nordic Political Science Association V
each other’s abilities and a shared understanding are essential in situations where implementation relies on collaboration between actors (Warm 2011; Head & Alford 2015). Mutual trust is also an essential mechanism for overcoming potentially dysfunctional relationships – for example, by increasing the willingness to disclose rather than hide information (Jacobsen 1999). And when leading administrators trust political leaders, they are more inclined to engage in political advice and service (Mouritzen & Svara 2002). Mutual trust is especially important when policy implementation depends on cooperation between institutions in different tiers of government (Lundin 2007). Ideally, political actors in all the tiers have faith in each other’s good intentions and abilities to fulfill the requirements of their roles, and they work together in a spirit of harmony for the benefit of the entire country. Inter-level distrust, on the other hand, may be a disruptive force that threatens to undermine the legitimacy and functionality of a political system (Aguirre et al. 2014). The risk for political failure is high when actors in different tiers are unable to rely on each other’s ability to fulfill their respective responsibilities. Local actors may not adhere to national policies and regulations if they lack confidence in the institutions behind these policies. And local service providers cannot implement national policy goals if national actors do not trust municipalities and regions with sufficient autonomy and resources. A breach in trust between national and local institutions could lead to actions by the state where local self-governance is reduced by factors such as increasing regulation, performance scrutiny (Johansson & Montin 2014) or centralisation of responsibilities. It could be very hard to heal such rifts of distrust between tiers (Lowndes 1999). Tier-association Every multilevel political system has a number of inherent and probably unavoidable tensions and conflicts of interest between different tiers of government, which may lead to breaches of trust (Hansen & Klausen 2002; Aars et al. 2008; Karlsson & Gilljam 2014). The primary source of these tensions derives from conflicts of interest among territorially defined groups of citizens, between or within tiers of government. What is good for your country or your neighbouring municipality is not necessarily good for your own municipality or region. A political representative is expected to be a fierce advocate for the interests of their self-governing unit or tier (Karlsson 2015). And when national, regional and local representatives all strive to implement their own preferred policies in the tier in which they have been elected, inter-level tensions could turn into a veritable struggle for political power. Actors in other tiers of government could be seen as meddling interlopers, limiting representatives’ scope of action in their own €bler & Michel 2006). domain (cf. Ku 4
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
C 2017 Nordic Political Science Association V
These multilevel conflicts are potentially manifest within political parties too (Deschouwer 2006; Allern & Saglie 2012). In extreme cases, a representative elected in one tier may come to regard a fellow party member elected in another tier as a political opponent. In this respect, the tier-association of political representatives is a potential source of political loyalty and trust (for institutions in their own tier) and distrust (for institutions in other tiers). In a previous study, a key finding was that political representatives are more satisfied with democracy in the tier where they are elected, and less satisfied with democracy in other tiers (Esaiasson et al. 2013). The study suggested a couple of possible explanations for this tier-association effect. First, representatives’ awareness of how democracy works is much stronger in relation to their own tier. Even though political representatives may have good knowledge of political life in general, it is the particular facts and realities of their own day-to-day work that they know best. And it is perhaps easier to appreciate what you know well and to be a bit more suspicious of the unknown. However, the argument for familiarity may be dubious since experience and closeness often, but by no means always, increase trust in public actors and institutions (Weibull 2012). Some studies even suggest that familiarity may breed contempt (Mondak et al. 2007). Second, the representatives are personally responsible to some extent for how democracy works in their own tier. And, since they have been elected to their offices in the existing system, the system has obviously worked in their favour. It is probably easier to appreciate processes that you are responsible for and which have benefitted you in the past, and harder to appreciate processes that are fully out of your control and bring little personal gain. It is a well-established fact that representatives’ views on democratic practices are affected by personal and party interests (e.g. Bowler et al. 2006; Gilljam & Karlsson 2015). Democratic satisfaction is a broader concept than political trust. Using analogous reasoning, however, it is likely that political representatives’ trust will be stronger in relation to institutions in their own tier of government and lower in relation to institutions in other tiers: H1: The tier-association hypothesis: Political representatives trust political institutions in their own tier of government more than representatives elected in other tiers do.
Multiple Mandates Where conflicts between tiers may create distrust, shared greater interests, familiarity and widened horizons may increase trust. One such potentially trust-generating force is the presence of elected representatives with multiple mandates. For example, the long tradition of multiple mandates in France – cumul des mandats – is often seen as one of the factors contributing to the country’s image as a centralised unitary state (Knapp 1991). When Scandinavian Political Studies, Vol. 00 – No. 00, 2017
5
C 2017 Nordic Political Science Association V
representatives have experience of political work from more than one tier of government, they might be advocates for the one tier in the other, spreading inter-level understanding and respect. In a study of members of parliament (MPs) with a background as local representatives, Jacob Aars suggests a number of possible mechanisms for why this could be the case: Information and knowledge are being passed on through the movement of personnel between tiers; MPs may have been socialised into identifying with the problems and concerns of local government; and there is likely to be stronger personal ties between MPs and the people back home (Aars 2014).1 Earlier studies give no clue as to how multiple mandates may affect inter-level political trust of individual representatives. The potential effects of single and multiple mandates are therefore an open question with at least two conceivable answers. The first possibility is that just as representatives with multiple mandates divide their time between their offices (Francois & Weill 2016), they also divide their loyalties. Single mandate representatives, in contrast, would then be more likely to be partisan and thus more trusting towards their sole source of loyalty. Following this logic, a hypothesis could be formulated in the following way: H2: The single mandate partisan hypothesis: Political representatives with single mandates trust political institutions in their own tier of government more than representatives with multiple mandates do.
However, one cannot take for granted that distribution of loyalty and trust is a zero-sum game. Multiple mandate representatives may be equally loyal to all their tiers – in which case H2 is incorrect. Furthermore, it is well established that travel broadens the mind and increases generalised trust in other people (Cao et al. 2013). Perhaps the experience of having mandates on several tiers of government will break down representatives’ partisanship and strengthen their generalised political trust? With this assumption, multiple mandates would provide representatives with insights that increase their political trust in all institutions, even in tiers where they have not been elected. This optimistic prediction will be tested in our third hypothesis: H3: The multiple mandate, widened horizons hypothesis: Political representatives with multiple mandates trust political institutions in tiers where they are not elected more than representatives with single mandates do.
The Home-team Effect Another potential source of trust is the presence of representatives from the same parties in all tiers of government. Inter-level, intra-party camaraderie could potentially ease tensions between tiers just like multiple 6
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
C 2017 Nordic Political Science Association V
mandates, even though the integrative role of parties may vary between countries due to the structure of the multilevel system, the composition of the party system, ethnical conflicts between regions and so on (Caramani 2004; B€ ack 2010; Detterbeck 2012). Intra-party networks and channels of communication are a foundation for inter-level political activism in multilevel systems (Wonka & Rittberger 2014). As ‘we tend to trust those we agree with’ (Citrin 1974), it can be assumed that representatives have more trust in institutions where their own party is in power since they share the ideology and political goals that these institutions strive to achieve. S€ oren Holmberg (1999) calls this assumption ‘the home-team hypothesis’ (Anderson & Tverdova 2001). The home-team effect has been identified repeatedly in studies of citizens. For example, results from a longitudinal study of citizens’ trust in the Swedish national parliament and the national government showed sharp and immediate changes in trust levels before and after an election. After the election, supporters of the party in the outgoing government lost trust in both the government and parliament, while trust in both institutions increased among supporters of parties in the new government (Holmberg & Weibull 2014). The home-team hypothesis has also been supported for citizens’ trust in local political institutions (Noren Bretzer 2005). In a multilevel system based on the principles of parliamentary democracy, parties in each tier-related political unit (municipality, region, state) can be divided into government parties and opposition parties. With three tiers of government, representatives from a particular party could be in positions where the party is in government in all three tiers; in opposition in all three tiers; or in a combination of government and opposition roles in different tiers. In relation to all these tiers, a hypothesis could be formulated in the following way: H4: The home-team hypothesis: Political representatives trust political institutions where their own party is in government more than representatives whose parties are in opposition do.
The home-team effect may also vary in relation to different kinds of political institutions. Studies of citizens show that political ideology is a very important factor for explaining trust in the national government as citizens with an ideology similar to the governing parties are much more inclined to trust the government as an institution. Ideology is less important in relation to trust in the parliament, and even less important in relation to unpolitical administrative agencies (Holmberg & Weibull 2014). With analogous reasoning for representatives, this hypothesis follows: H5: The differentiated home-team hypothesis: The predicted home-team effect in H4 will be stronger in relation to trust in executive institutions, less strong in relation to elected Scandinavian Political Studies, Vol. 00 – No. 00, 2017
7
C 2017 Nordic Political Science Association V
assemblies, and have little or no effect in relation to unpolitical administrative institutions.
Representative Democracy in a Multilevel System: Sweden The five hypotheses formulated above will be tested on data from the case of Sweden. The Swedish multilevel government system is based on three tiers of government where political assemblies are elected on the same day every fourth year: 290 municipalities with close to 13,000 elected representatives in local councils; 20 regions with 1,662 members in regional councils; and the national level with 349 members in the national parliament – The Riksdag.2 In 2012, 5 percent of local councillors (665 individuals) were also members of the regional council. This means that 40 percent of regional councillors are also local councillors. About 1 percent of local and regional councillors claimed to be national MPs in 2012. Conversely, 36 percent of the national MPs in 2014 claimed to be members of local councils and 6 percent of regional councils. It is possible to be an elected representative in all three tiers of government simultaneously, and six individual representatives in the KOLFU 2012 survey and 13 in the RDU 2014 survey claimed to be so. Unlike mandate-holders in a local or regional council, the role of an MP is a full-time job. Many MPs therefore end their local and regional career when they enter a national office. Results from the RDU 2014 show that, in addition to the MPs who hold multiple mandates today, another 44 percent of MPs had been elected to a local council previously and 29 percent had previously been elected to a regional council. Less than 20 percent of Swedish MPs in 2014 had never been elected in a subnational election. All elected representatives in Sweden could easily be classified as members of parties either in government or in opposition in the tier where they are elected (Gilljam & Karlsson 2012). Furthermore, using the information about which municipality and which region a representative lives in, it has been possible to classify representatives as members of a party either in government or in opposition in tiers where they are not elected.3 The Swedish political system is built on parliamentary principles in all tiers of government. The political executive at the national level is the national government and its members are not members of the Riksdag while serving as ministers. At local and regional level, executive power is mainly invested in the executive boards (a ‘committee-leader model’ of local governance; Mouritzen & Svara 2002). The councils elect the members of the boards, but the board members are not necessarily councillors 8
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
C 2017 Nordic Political Science Association V
themselves. The party affiliation of the board members mirrors the percentage of parties in the council – that is, both representatives of the ruling parties and of the opposition are board members (the system has been called a ‘quasi-parliamentary model’; B€ack & Johansson 2000). The chair of the board, the Swedish mayor equivalent, is always the leader of the ruling party or coalition (Karlsson 2006).
Methodology and Data The data used in this article come from two surveys that together cover all elected local, regional and national representatives in Sweden. The first of these – KOLFU 20124 – was a survey of all local and regional councillors in the 290 municipalities and 20 regions in Sweden. The survey, which was carried out from October 2012 to March 2013, accumulated 10,491 responses, which is a response rate of 79 percent. Analyses show no differences in response rates between male and female representatives. The response rate of seven of the eight national parties varies between 77 and 82 percent, while the rate of Sweden Democrats is somewhat lower at 62 percent. The response rate exceeded 60 percent in all 20 regions and in 266 of 290 municipalities (Karlsson & Gilljam 2014). The second survey – RDU 20145 – targeted all 349 MPs in the Riksdag. It was carried out from November 2014 to May 2015 and accumulated 306 responses, which is a response rate of 88 percent. Analyses show that more female MPs responded (92 percent) than male MPs (86 percent). The response rate of the eight national parties varies between 76 percent (Sweden Democrats) and 96 percent (The Green Party) (Karlsson & Nordin 2015). The two surveys were conducted two years apart and it should be noted that since RDU started after the general elections of 2014, the results of the two surveys refer to different election periods.6 The dependent variables of this study all relate to the representatives’ trust in political institutions. These variables are built on responses to a question that was included in both the RDU and the KOLFU survey. The question was: ‘How much trust do you have in the way the following institutions conduct their work?’ The representatives were asked to rate their trust in the political executive, the elected assembly and the administrative authorities in each tier of government. The representatives responded on a five-graded scale, from a very low to a very high level of trust.7 In this study, the dependent variables are recoded into a 0–100 scale. The general picture is that Swedish political representatives have high levels of trust in political institutions in all tiers of government. But there is also significant variation. For example, the overall tendency is that executive institutions have somewhat lower trust levels than other institutions, Scandinavian Political Studies, Vol. 00 – No. 00, 2017
9
C 2017 Nordic Political Science Association V
and inter-level trust in regional institutions is lower than inter-level trust in local and national institutions. Model Strategy As mentioned above, we know from earlier studies that political trust is often related to the procedures and performance of the institutions in question. Trust is likely to be stronger in institutions that function well. Indirectly, structural factors like size, demographic and economic situations may have causal relationships with trust levels. The questions refer to specific institutions in 290 municipalities and 20 regions. As factors relating to procedures and performance in these 310 political units vary, some of the variations in trust levels among representatives may be explained by these systemic factors. Trust in national institutions may, of course, also depend on procedures and performance, but since there is only one state, there is no variation in procedures and performance that could explain variation in trust levels among representatives. Fixed-effects linear multilevel regression models are used when analysing trust in local and regional institutions to distinguish individual factors from systemic factors. Ordinary least squares (OLS) regression is used when analysing trust in national institutions. It is possible to assess the proportion of variation that lies between municipalities and regions (the intra-class correlation) by estimating a random-intercept-only model (null model) in the models for local and regional institutions. Results of null model analyses show that between 3 and 5 percent of the variation in trust levels is found among the 290 municipalities and 20 regions in which representatives are clustered, while 95–97 percent of the variations is found among individual representatives. Although statistically significant, the systemic factors potentially relating to procedures and performance of institutions are minute in relation to factors on the individual level and they will not be included in the models in this article. These tests will be carried out in three steps. In the first step, model 1, the tier-association effect and the home-team effect are singled out as the only independent variables. In model 2, a number of control variables are introduced. Finally, in model 3, the tier-association variable is replaced by a set of dummy variables to distinguish between effects of single and multiple mandates. The control variables in models 2 and 3 include gender, education level and age. As representatives of the Sweden Democrats are in opposition everywhere, affiliation to this party is also controlled to ensure that the effect of belonging to a government party is separate from the effect of being a Sweden Democrat. 10
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
C 2017 Nordic Political Science Association V
One important lesson from earlier studies (Esaiasson et al. 2013) is that an individual’s perception of a specific political institution may in part be a result of the individual’s general perception of all political institutions. Some people are more satisfied or trusting than others. In order to separate the effects of the main independent variables on each specific institution, an indicator for the representatives’ general perception of public institutions is also included as a control variable: general democratic satisfaction (GDS).8 The intention is that, under control for GDS, the tierassociation effects and the home-team effects in models 2 and 3 will be institution-specific. Finally, a survey question on interpersonal trust (IPT) is introduced as a control variable.9 Many earlier studies have concluded that this factor is of some importance when explaining trust in political institutions (e.g., Newton & Zmerli 2011). In all analyses, the dependent variables (the trust questions) are coded as 0–100. Most of the independent variables are dichotomous 0–1, while GDS and IPT are linear and recoded, with 0 being the lowest and 1 the highest possible value.
Results In this section, we return to the five hypotheses and summarise their accuracy. The results of the analyses are presented in Tables 1 (trust in local institutions), 2 (trust in regional institutions) and 3 (trust in national institutions). The first hypothesis to be tested is the tier-association hypothesis (H1), which states that political representatives trust political institutions in their own tier of government more than representatives elected in other tiers do. This hypothesis is supported in relation to a specific institution if there is a statistically significant positive effect on trust by the variable ‘mandate in this tier’ in model 2. The results show that a positive tier-association effect could be confirmed in relation to seven out of nine institutions. In two cases (trust in the national executive and in the local elected assembly), the effect is positive but not statistically significant. The strongest tier-association effects are found in relation to the national parliament and national administrative authorities. H1 is thus largely supported. The second hypothesis is the single mandate partisan hypothesis (H2), which predicts that political representatives with single mandates trust political institutions in their own tier of government more than representatives with multiple mandates do. H2 is supported if the effect of ‘single mandate in this tier’ is significantly stronger than the effect of ‘dual mandate in this and other tiers’ in model 3. The results show that this predicted result is found in relation to seven out of nine institutions. For one Scandinavian Political Studies, Vol. 00 – No. 00, 2017
11
12 M3
M1
M2
M3
M1
M2
Administrative authorities M3
47*** (1.8) 290 9,202
11.3** (0.4) 28.1** (2.6) 20.4 (0.6) 22.5** (1.0) 17.4*** (1.3) 149*** (1.3)
70*** (3.0) 290 9,045
70*** (3.0) 290 9,045
51*** (1.7) 290 9,217
11.7*** (0.4) 26.5* (2.6) 10.8 (0.5) 23.7*** (1.0) 17.4*** (1.3) 147*** (1.3)
70*** (2.9) 290 9,059
71*** (3.0) 290 9,059
110*** (1.2) 111*** (1.2)
11.9*** (0.4) 24.9 (2.5) 10.9 (0.5) 24.3*** (1.0) 16.6*** (1.3) 147*** (1.3)
21.3 (10)
20.1 (10)
19.2*** (1.3) 19.4*** (1.3)
11.4** (0.4) 26.8* (2.6) 20.3 (0.6) 23.0** (1.0) 16.8*** (1.3) 149*** (1.3)
14.7* (1.9) 20.4 (1.9)
15.5** (2.0) 11.4 (2.0)
54*** (1.6) 290 9,177
10.8 (0.4) 23.1 (2.5) 20.5 (0.5) 21.7 (0.9) 15.1*** (1.2) 135*** (1.2)
61*** (2.8) 290 9,022
62*** (2.9) 290 9,022
115*** (1.2) 115*** (1.2)
10.8* (0.4) 22.2 (2.5) 20.4 (0.5) 22.1* (0.9) 14.7*** (1.2) 134*** (1.2)
221* (9.7)
15.6** (1.9) 12.6 (1.9)
125*** (0.5) 121*** (0.5) 121*** (0.4) 121*** (0.5) 117*** (0.4) 117*** (0.4) 111*** (0.4) 18.0*** (0.4) 17.9*** (0.4) 15** (1.7) 14* (1.9) 16.4*** (1.7) 13.3 (1.9) 16.1*** (1.6) 15.2** (1.8)
M2
Elected assembly
Notes: An underlined dual mandate effect (of ‘in this and other tiers’) signifies that this effect differs significantly (p < 0.05) from the effect of ‘single mandate in this tier’. The N numbers in this table are lower than in Tables 2 and 3 because single mandate regional councillors are excluded as we do not have information about their home municipality and thus whether their party is in government there. ***p < 0.001; **p < 0.01; *p < 0.05. Source: RDU 2014 and KOLFU 2012 combined.
Constant N of municipalities N of representatives
Gender: Female Education: Higher Age: 651 Age: 30 Party: SD General democratic satisfaction (GDS) Interpersonal trust (IPT)
Single mandate in this tier Dual mandate in this and other tiers Dual mandate in other tiers (Control: Single mandate in other tiers)
In government in this tier Mandate in this tier
M1
Political executive
Table 2. Swedish Political Representatives’ Trust in Regional Institutions (Fixed Effects, Standard Errors)
C 2017 Nordic Political Science Association V
13
14 M3
35*** (0.3) 0.55 9,944
40*** (3.2) 0.56 9,776
20.2 (0.9)
20.1 (1.3) 53*** (0.3) 0.21 9,895
50*** (3.0) 0.27 9,728
110*** (1.2)
11.3** (0.4) 16.2* (2.6) 22.1*** (0.5) 13.3** (0.9) 12.4 (1.2) 127*** (1.2)
51*** (3.0) 0.27 9,728
110*** (1.2)
11.2** (0.4) 15.3* (2.7) 22.1*** (0.5) 13.7*** (1.0) 12.3 (1.2) 127*** (1.2)
21.1 (1.0)
11.0 (1.0)
10.5 (0.4) 13.1 (2.9) 11.2* (0.6) 20.7 (1.0) 16.1*** (1.3) 120*** (1.3)
M1
M2
53*** (0.3) 0.14 9,657
53*** (2.8) 0.20 9,498
113*** (1.1)
20.9* (0.4) 12.5 (2.5) 21.0* (0.5) 15.3*** (0.9) 13.0** (1.1) 124** (1.2)
M3
53*** (2.8) 0.20 9,498
113*** (1.1)
20.8* (0.4) 13.3 (2.5) 21.1* (0.5) 15.1*** (0.9) 13.2** (1.2) 124*** (1.2)
0.3 (0.9)
116*** (1.8) 112*** (1.5)
113*** (0.4)
Administrative authorities
120*** (0.4) 114*** (0.4) 113*** (0.4) 118*** (1.0) 113*** (1.3)
M3
110*** (2.0) 114*** (1.6)
10.4 (0.4) 12.3 (2.9) 11.2* (0.6) 20.4 (1.0) 16.1*** (1.3) 120*** (1.3)
41*** (3) 0.56 9,776
M2
121*** (0.4) 120*** (0.4) 114*** (1.1) 113*** (1.4)
M1
15.3* (2.1) 10.7 (1.7)
150*** (0.5) 148*** (0.5) 148*** (0.5) 11.7 (1.2) 12.3 (1.5)
M2
Elected assembly
Notes: An underlined dual mandate effect (of ‘in this and other tiers’) signifies that this effect differs significantly (p < 0.05) from the effect of ‘single mandate in this tier’. ***p < 0.001; **p < 0.01; *p < 0.05. Source: RDU 2014 and KOLFU 2012 combined.
Constant Adjusted R2 N
Gender: Female Education: Higher Age: 651 Age: >30 Party: SD General democratic satisfaction (GDS) Interpersonal trust (IPT)
Single mandate in this tier Dual mandate in this and other tiers Dual mandate in other tiers (Control: Single mandate in other tiers)
In government in this tier Mandate in this tier
M1
Political executive
Table 3. Swedish Political Representatives’ Trust in National Institutions (OLS Regression, b-values and Standard Errors)
C 2017 Nordic Political Science Association V
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
C 2017 Nordic Political Science Association V
institution (the local administrative authorities), the single mandate effect is stronger than the multiple mandate effect, but not significantly so. For one institution (the national parliament), the effect is the opposite to that expected: dual mandate MPs have greater trust in the national parliament than single mandate MPs. Nevertheless, H2 must be deemed as largely supported. The multiple mandate, widened horizons hypothesis (H3) predicts that political representatives with multiple mandates trust political institutions in tiers where they are not elected more than representatives with single mandates do. In the set of dummy variables used as indictors for tier-association in model 3, representatives with ‘single mandate on other tiers’ are the control group. H3 is therefore supported when the effect of ‘dual mandate in other tiers’ in model 3 is positive and statistically significant. The results show that the hypothesis is supported in only one out of nine cases (in relation to trust in regional administrative authorities). In relation to local administrative authorities, the effect is negative and statistically significant – a result that goes against the hypothesis. In the other seven cases, the effects of ‘dual mandates in other tiers’ are not statistically significant. H3 is thus decidedly not supported. The home-team hypothesis (H4) predicts that political representatives trust political institutions where their own party is in government more than representatives whose parties are in opposition do. H4 is supported when the effect of ‘in government in this tier’ is positive and statistically significant in model 2. The results show that the home-team effect is supported in relation to all nine institutions. In all cases, the effect is statistically significant on the 0.001-level. H4 is thus fully supported. Finally, the differentiated home-team hypothesis (H5) states that the home-team effect will be stronger in relation to executive institutions, less strong in relation to elected assemblies, and weak or non-existent in relation to unpolitical administrative institutions. H5 is supported when the effect of ‘in government in this tier’ is significantly stronger in relation to the political executive than in relation to the elected assembly, where the latter effect in turn is significantly stronger in relation to the administrative authorities. The results show that this predicted effect pattern is found and statistically significant in all three tiers of government. H5 is thus fully supported. The magnitude of the effect becomes particularly evident in relation to unpolitical administrative authorities. The administrators have no political colour but representatives’ confidence in their ability to do their job is still significantly affected by the home-team effect. Furthermore, the fact that the home-team effect is strongest in relation to the national political executive, where partisan distinctions and divisions are most likely to be pronounced, underscores that trust among representatives appears to be Scandinavian Political Studies, Vol. 00 – No. 00, 2017
15
C 2017 Nordic Political Science Association V
most strongly driven by partisan factors where the stakes are highest and the need for compromise is less salient. As the tests show that both the tier-association and the home-team hypotheses are supported, which of the two effects is the most important source of trust? The effect sizes are compared in Figure 1. The results leave us in no doubt about the answer: The home-team effect is by far the strongest factor for trust in elected assemblies and the political executive in all tiers of government. The home-team effect is also somewhat stronger in relation to trust in unpolitical local administrative authorities, but not significantly so. For trust in administrative authorities in the national and regional tiers of government, the strength of the two effects is more or less equal. It is worth noting that even though some of the control variables have strong effects on institutional trust, the introduction of these variables in models 2 and 3 has only a limited impact on the effects of the principal independent variables ‘in government in this tier’ and ‘mandate in this tier’, as compared with model 1. The effect of control variables relating to representatives’ personal characteristics is limited. Female representatives generally have greater trust in Figure 1. Tier-association and Home-team Hypotheses: Comparison of Effect Sizes.
Notes: The bars illustrate the effects (fixed effects and b-values in model 2 of Tables 1–3) of ‘mandate in this tier’ (the tier-association effect) and ‘in government in this tier’ (the home-team effect) on representatives’ political trust for different institutions in different tiers of government, and a 95 percent confidence interval of each effect.
16
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
C 2017 Nordic Political Science Association V
political institutions than their male colleagues. Representatives under 30 year of age have greater trust in the national parliament and national administrative authorities, but less trust in local institutions. Representatives over 65 have greater trust in the national government but less trust in national parliament and national regional administrative authorities. Representatives with a high level of education have greater trust in the national parliament but less trust in local and regional executives and elected assemblies. The indicators for IPT and GDS have stronger effects. IPT has expected positive effects on trust in eight out of the nine institutions (the exception being the national government, where IPT has no effect). GDS has expected positive effects on trust in all institutions, enabling us to distinguish effects of the other factors as being specific for the analysed institutions. The strongest effects of GDS are found in relation to regional and local institutions while GDS has weaker effects in relation to national institutions. The indicator most related to GDS is affiliation to the Sweden Democrats (SD). As SD representatives are always in opposition, affiliation to this party was introduced as a control variable in models 2 and 3 in order to separate the home-team effect from an SD effect. As an antiestablishment party, SD is well-known for being critical towards political institutions. It is therefore initially a surprising result that the SD effects in models 2 and 3 are positive in relation to trust in almost all cases. Are Sweden Democrats really more trusting towards political institutions than other representatives? The answer is ‘no’. Control analyses reveal that the main effect of being an SD representative on trust is channeled through a substantially lower degree of general democratic satisfaction and, to a lesser extent, a lower degree of interpersonal trust among representatives affiliated with this party. It is only under control for these factors that the SD effect on trust is positive.
Discussion It is often said that trust is hard to build and easy to lose. Luckily, both inter-level and intra-level political trust are relatively high in Sweden. This is especially fortunate, as the results of this study indicate that strong forces generating both trust and distrust are in play among political representatives in different tiers of government: Tier-association, multiple mandates and the home-team effect are all factors that impact the levels of inter-level political trust among representatives. And by far the most important of these factors is the home-team effect. However, one should not assume that a high level of inter-level trust is everlasting. Small negative changes could lead to unpredictable Scandinavian Political Studies, Vol. 00 – No. 00, 2017
17
C 2017 Nordic Political Science Association V
consequences, as increased distrust will probably make it harder for political actors to marshal resources and solve problems. These reduced abilities may, in turn, increase the distrust in a negative spiral (cf. Hetherington 1998). The managerial trend to increase performance scrutiny activities of national authorities vis-a-vis local government in Sweden could be one such effect. A result of this trend, which in itself is a sign of distrust in local authorities’ abilities, is a change in the focus of local actors from local citizen responsiveness to addressing deviations from national indicators of service quality. This distracts them from their democratic duties and may fuel further distrust in local government (Johansson & Montin 2014). The factors identified as affecting inter-level and intra-level trust are not static either. Managerial, structural and constitutional reforms could easily alter the conditions of tier-association and home-team loyalty. For example, France is currently involved in a process to limit its traditional system of cumul des mandats. Members of the European Parliament have been prohibited from becoming members of national parliaments since 2002. The results of this study indicate that reforms involving abolishment or encouragement of multiple mandates could have considerable effects on inter-level trust since multiple mandates are a factor that decidedly generates inter-level trust. However, it is paradoxically not a given that the overall political trust in institutions in the multilevel system would increase by an increase in multiple mandates. Most representatives today are only elected in one tier of government and single mandate representatives have decidedly more trust in the institutions of their own tier than dual mandate colleagues. Dual mandates are apparently less partisan and thus less trusting in relation to the tiers where they are elected. And there is no evidence whatsoever that multiple mandate representatives have a greater degree of general political trust as a result of their broader experience and widened horizons. Adjusting the timing of elections may also have substantial effects on inter-level trust. In Sweden, national, regional and local elections are held at the same time, which probably contributes to the relatively high level of inter-level trust. As voters are influenced by the same election trends in all three elections, the political colour of regimes in all tiers is more likely to be similar in Sweden than in countries where the elections in different tiers are separated. Inter-level trust among political representatives would probably decrease if national and local/regional elections were held separately. Furthermore, increasing and decreasing the presence of parliamentary principles in a political system may have a major impact on inter-level political trust. Parliamentary principles have been gradually introduced into Swedish local and regional government since the 1970s, largely based on informal changes in norms rather than formal 18
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
C 2017 Nordic Political Science Association V
constitutional reforms (Karlsson 2013). Until the 1990s, it would have been hard to determine exactly which parties were in government in Swedish municipalities since majorities could alter from issue to issue. And the single most important factor in all tiers of government for determining political trust – the home-team effect – would have no meaning in systems without clear distinctions between governing regimes and oppositions. The results of this study are all derived from the case of Sweden, and it remains an open question whether it is possible to generalise the findings to other cases. It is most likely that the conclusions are valid for other Scandinavian countries whose models of multilevel government and parliamentary democracy are very similar to Sweden. Future studies will have to address whether the findings are relevant in political systems with other norms and rules concerning parliamentary principles and multiple mandates, and with more detached or more integrated tiers of government. This study has not analysed trust in relation to the European Union, a tier of government which relies heavily on inter-level trust for it to function (Weller 2015). We already know that citizens in many countries are losing trust in the EU (Armingeon & Ceka 2013), but little or nothing is known of inter-level trust among political representatives at the EU level and in other tiers of government. At present, the home-team effect is not likely to be a factor in relation to trust in European institutions since the EU is not a parliamentary system. But the election of the President of the European Commission in 2015 indicated a break from norm in that respect (Hobolt 2014), and if the EU takes further steps in that direction this could put the home-team effect into play. NOTES 1.
2.
3.
4. 5.
However, Aars (who studied the cases of Norway and Sweden) discovered that the relationship between previous local experiences and MPs’ priorities might not be so clearcut. Norwegian MPs with a background in local politics seemed to be more sceptical towards local government than colleagues who do not share this experience. For Swedish MPs, a local experience seemed to have the opposite effect. Differences in statelocal relationships between the two countries may partly explain this divergence. A table showing the percentage of representatives in each tier of government with dual mandate (i.e., who simultaneously have mandates in other tiers) is presented in the Online Appendix. The EU level and the 20 Swedish members of the European Parliament are not included in this study. Unfortunately, data on inter-level political trust among MEPs and on the trust for European institutions among MPs and local and regional councillors are not available. This is possible for all representatives except single mandate regional councillors, since there is no information on home municipality for this group. The percentage of members of government/opposition parties by tier of government in 2012 (local and regional tiers) and 2014 (national tier) is included in the Online Appendix. ‘Kommun- och landstingsfullm€ aktigeunders€ okningen 2012’. ‘Riksdagsunders€ okningen 2014’.
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
19
C 2017 Nordic Political Science Association V
6.
7.
8.
9.
KOLFU 2012 started in the middle of the 2010–14 election period, while the RDU 2014 survey was conducted early in the following period. This means that new regimes established after the 2014 election had had very little time to make a mark at the time of the RDU survey. The most important of these regime changes was the shift at the national level from a centre-right to a left-green coalition. The MPs’ trust levels might have been affected by the party political turbulence after the election, but no other contextual change – economic, political or otherwise – could be identified as possibly impacting the results from 2012 and 2014. Tables presenting descriptive results and the full wording of this question in both English and the original Swedish are included in the Online Appendix. Included also is a table presenting mean value of trust in each institution for representatives in the three tiers of government. This indicator is built on another set of questions included in the KOLFU and RDU surveys concerning satisfaction with democracy in four tiers of government: EU, national, regional and local tiers. The GDS indicator is the mean value of representatives’ satisfaction with democracy in all four tiers: 0 represents great dissatisfaction and 1 represents great satisfaction with democracy in all four tiers. In Swedish: ‘Pa˚ det hela taget, hur n€ ojd €ar du med det s€ att pa˚ vilket demokratin fungerar i: EU; Sverige; Landstinget/regionen d€ar du bor; Den kommun d€ ar du bor.’ The question was: ‘In your opinion, to what extent is it possible to trust people in general?’ The responses were given on an 11-graded scale ranging from ‘It is not possible to trust people in general’ to ‘It is possible to trust people in general’.
REFERENCES Aars, J. 2014. ‘Ties That Bind? Effects of Inter-level Mobility on MP Attitude to Local Government in Two Scandinavian Countries’, Scandinavian Political Studies, 37, 345–65. Aars, J. & Christensen, D. A. 2013. ‘The Impact of Local Associational Life on Citizens’ Attitudes towards Local Government’, Policy and Politics, 41, 371–87. Aars, J., Fimreite, A. L. & Flo, Y. 2008. ‘Statlig lokalpolitik och lokalt sj€ alvstyre: En j€ amf€ orelse €ver kommunerna. Stockholm: SKL. mellan Sverige och Norge’, in Makten o Aguirre, J. F., Lara, J. I. & Casetti, M. M. 2014. ‘Alta Direcci on P ublica Municipal’, Revista Del Clad Reforma y Democracia, 59, 119–50. Allern, E. H. & Saglie, J. 2012. ‘Inside the Black Box: Parties as Multi-level Organisations in a Unitary State’, West European Politics, 35, 947–70. Anderson, C. J. & Tverdova, Y. V. 2001. ‘Winners, Losers and Attitudes about Government in Contemporary Democracies’, International Political Science Review, 22, 321–38. Armingeon, K. & Ceka, B. 2013. ‘The Loss of Trust in the European Union during the Great Recession since 2007’, European Union Politics, 15, 82–107. B€ack, H. 2010. ‘Partiv€asendet i stat och kommun: fra˚n integration till dualism?’, in Johansson, F., ed., Kommunvalet 2006. Gothenburg: CEFOS, University of Gothenburg. B€ack, H. & Johansson, F. 2000. Mellan samlingsregering och parlamentarism. Stockholm: University of Stockholm. Berg, L. 2007. Multi-level Europeans: The Influence of Territorial Attachments on Political Trust and Welfare Attitudes. Gothenburg: University of Gothenburg. Bowler, S., Donovan, T. & Karp, J. A. 2006. ‘Why Politicians Like Electoral Institutions: Selfinterests, Values or Ideology?’, Journal of Politics, 68, 434–46. Caramani, D. 2004. The Nationalization of Politics: The Formation of National Electorates and Party Systems in Western Europe. Cambridge: Cambridge University Press. Cao, J., Galinsky, A. D. & Maddux, W. W. 2013. ‘Does Travel Broaden the Mind? Breadth of Foreign Experiences Increases Generalized Trust’, Social Psychological and Personality Science. Available online at: http://journals.sagepub.com/doi/abs/10.1177/1948550613514456 Catterberg, G. & Moreno, A. 2006. ‘The Individual Bases of Political Trust: Trends in New and Established Democracies’, International Journal of Public Opinion Research, 18, 31–48. Citrin, J. 1974. ‘Comment: The Political Relevance of Trust in Government’, American Political Science Review, 68, 973–88.
20
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
C 2017 Nordic Political Science Association V
Delhey, J. & Newton, K. 2005. ‘Predicting Cross-national Levels of Social Trust: Global Pattern or Nordic Exceptionalism?’, European Sociological Review, 21, 311–27. Denters, B. 2002. ‘Size and Political Trust: Evidence from Denmark, the Netherlands, Norway and the United Kingdom’, Environment and Planning C: Government and Policy, 20, 793– 812. Deschouwer, K. 2006. ‘Political Parties as Multi-level Organizations’, in Katz, R. S. & Crotty, W., eds, Handbook of Party Politics. London: Sage. Detterbeck, K. 2012. Multi-level Party Politics in Western Europe. Basingstoke: Palgrave Macmillan. Egeberg, M. 2003. ‘How Bureaucraritc Structure Matters: An Organizational Perspective’, in Peters, B. G. & Pierre, J., eds, The Sage Handbook of Public Administration. London: Sage. Esaiasson, P., Gilljam, M. & Karlsson, D. 2013. ’Sources of Elite Democratic Satisfaction: How Elected Representatives Evaluate Their Political System’, in Dahlberg, S., Oscarsson, H. & W€angnerud, L., eds, Stepping Stones. Gothenburg: Department of Political Science, University of Gothenburg. Francois, A. & Weill, L. 2016. ‘Does Holding a Local Mandate Alter the Activities of Deputies? Evidence from the French Assemblee Nationale’, French Politics, 14, 30–54. Gilljam, M. & Karlsson, D. 2012. ‘Tungan pa˚ va˚gen-partier i svenska kommuner’, in Bj€ orkman, J. & Fjaestad, B., eds, Tungan pa˚ va˚gen. Va˚gm€ astare och balanspartier. G€ oteborg: Makdam f€ orlag. Gilljam, M. & Karlsson, D. 2015. ‘Ruling Majority and Opposition: How Parliamentary Position Affects the Attitudes of Political Representatives’, Parliamentary Affairs, 68, 555–72. Gustavsen, A., Røiseland, A. & Pierre, J. 2014. ‘Procedure or Performance? Assessing Citizen’s Attitudes toward Legitimacy in Swedish and Norwegian Local Government’, Urban Research and Practice, 7, 200–12. Hansen, T. & Klausen, J. E. 2002. ‘Between the Welfare State and Local Government Autonomy’, Local Government Studies, 28, 47–66. Head, B. W. & Alford, J. 2015. ‘Wicked Problems: Implications for Public Policy and Management’, Administration and Society, 47, 711–39. Heinelt, H. & Hlepas, N.-K. 2006. ‘Typologies of Local Government Systems’, in B€ ack, H., Heinelt, H. & Magnier, A., eds, The European Mayor: Political Leaders in the Changing Context of Local Democracy. Wiesbaden: VS Verlag. Hetherington, M. J. 1998. ‘The Political Relevance of Political Trust’, American Political Science Review, 92, 791–808. Hobolt, S. B. 2014. ‘A Vote for the President? The Role of Spitzenkandidaten in the 2014 European Parliament Elections’, Journal of European Public Policy, 21, 1528–1540. Holmberg, S. 1999. ‘Down and Down We Go: Political Trust in Sweden’, in Norris, P., ed., Critcal Citizens: Global Support for Democratic Governance. Oxford: Oxford University Press. Holmberg, S. & Weibull, L. 2014. ‘Institutionsf€ ortroende mellan ber€ ordhet och mediebevakning’, in Oscarsson, H. & Bergstr€ om, A., eds, Mittfa˚ra och marginal. Gothenburg: SOM-institutet, University of Gothenburg. Jacobsen, D. I. 1999. ‘Trust in Political-Administrative Relations: The Case of Local Authorities in Norway and Tanzania’, World Development, 27, 839–53. Johansson, V. & Montin, S. 2014. ‘What If Performance Accountability Mechanisms Engender Distrust?’, Urban Research and Practice, 7, 213–27. Karlsson, D. 2006. Den svenske Borgm€ astaren. Kommunstyrelsens ordf€ orande i den lokala demokratin. F€ orvaltningsh€ ogskolans rapporter 36. Gothenburg: F€ orvaltningsh€ ogskolan, University of Gothenburg. Karlsson, D. 2013. ‘The Hidden Constitutions: How Informal Political Institutions Affect the Representation Style of Local Councils’, Local Government Studies, 39, 681–702. Karlsson, D. 2015. ‘Local Autonomy or National Equality? A Conflict of Values and Interests for Political Leaders’, Regional and Federal Studies, 25, 19–44. Karlsson, D. & Gilljam, M., eds. 2014. Svenska politiker. Om folkvalda i riksdag, landsting och kommun. Stockholm: Sant erus f€ orlag. Karlsson, D. & Nordin, L. 2015. Riksdagsunders€ okningen 2014. Gothenburg: University of Gothenburg.
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
21
C 2017 Nordic Political Science Association V
Kestil€a-Kekkonen, E. & S€ oderlund, P. 2015. ’Political Trust, Individual-level Characteristics and Institutional Performance: Evidence from Finland, 2004–13’, Scandinavian Political Studies, 39, 138260. Knapp, A. 1991. ‘The Cumul des Mandats, Local Power and Political Parties in France’, West European Politics, 14, 18–40. K€ ubler, D. & Michel, P. 2006. ‘Mayors in Vertical Power Relations’, in B€ ack, H., Heinelt, H. & Magnier, A., eds, The European Mayor. Wiesbaden: VS Verlag. Lane, J.-E. 1987. ‘Implementation, Accountability and Trust’, European Journal of Political Research, 15, 527–46. Loughlin, J., Hendriks, F. & Lidstr€ om, A. eds. 2010. The Oxford Handbook of Local and Regional Democracy in Europe. Oxford: Oxford University Press. Lowndes, V. 1999. ‘Rebuilding Trust in Central/Local Relations: Policy or Passion?’, Local Government Studies, 25, 116–36. Lundin, M. 2007. ‘Explaining Cooperation: How Resource Interdependence, Goal Congruence and Trust Affect Joint Actions in Policy Implementation’, Journal of Public Administration Research and Theory, 17, 651–72. Mondak, J. J., Carmines, E. G., Huckfeldt, R., Mitchell, D. G. & Schraufnagel, S. 2007. ‘Does Familiarity Breed Contempt? The Impact of Information on Mass Attitudes toward Congress’, American Journal of Political Science, 51, 34–48. Mouritzen, P. E. & Svara, J. H. 2002. Leadership at the Apex: Politicians and Administrators in Western Local Government. Pittsburgh: University of Pittsburgh Press. Newton, K. 2007. ‘Social and Political Trust’, in Dalton, R. J. D. & Klingemann, H.-D., eds, The Oxford Handbook of Political Behavior. Oxford; Oxford University Press. Newton, K. & Zmerli, S. 2011. ‘Three Forms of Trust and Their Association’, European Political Science Review, 3, 169–200. Noren Bretzer, Y. 2005. Att f€ orklara politiskt f€ ortroende: betydelsen av socialt kapital och r€attvisa procedurer. Gothenburg: University of Gothenburg. Roos, K. & Lidstr€ om, A. 2014. ‘Local Policies and Local Government Legitimacy: The Swedish Case’, Urban Research and Practice, 7, 137–52. Sellers, J. M. & Lidstr€ om, A. 2007. ‘Decentralization, Local Government and the Welfare State’, Governance, 20, 609–32. Warm, D. 2011. ‘Local Government Collaboration for a New Decade: Risk, Trust and Effectiveness’, State and Local Government Review, 43, 60–65. Weibull, L. 2012. ‘F€ ortroende f€ or olika grupper’, in Bergstr€ om, A., ed., V€ astsvensk vardag. Gothenburg: SOM-institutet, University of Gothenburg. Weller, M. 2015. ‘Mutual Trust: In Search of the Future of European Union Private International Law’, Journal of Private International Law, 11, 64–102. Wonka, A. & Rittberger, B. 2014. ‘The Ties that Bind? Intra-party Information Exchanges of German MPs in EU Multi-level Politics’, West European Politics, 37, 624–43.
Supporting Information Internet Appendix: Inter-level trust in a multi-level political system Table A1: Multiple-mandates among elected political representatives in Sweden by tier of government (percent within each tier) Table A2: Swedish political representatives who are members of parties in government or in opposition in different tiers of government (percent) Table A3: The dependent variables: Trust in political institutions among national MPs (percent): How much trust do you have in the way the following institutions conduct their work? Table A4: The dependent variables: Trust in political institutions among regional councillors (percent): How much trust do you have in the way the following institutions conduct their work?
22
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
C 2017 Nordic Political Science Association V
Table A5: The dependent variables: Trust in political institutions among local councillors (percent): How much trust do you have in the way the following institutions conduct their work? Table A6: Trust in political institutions of elected representatives by tier of government and parliamentary position (mean values 0–100 and mean differences)
Scandinavian Political Studies, Vol. 00 – No. 00, 2017
23
