Going Local. Voting for independent local Parties in the Netherlands 1986-2010

ABSTRACT

This paper examines why the support of independent local parties has grown substantially in the Netherlands. These are parties that run in municipal council elections, but do not run in elections at higher levels, specifically the national level. Such parties saw their support double in the Netherlands between 1986 and 2010. Parties of this type have also grown in other Western European states. This paper examines two possible explanations: declining political trust on the level of voters and, on the supply side, the rise of parties that are not rooted at the local level. The evidence shows that the rise of independent local parties reflects the rise of national political parties that do not run in many municipal elections. This article examines the case of the Netherlands, pooling five surveys from the 1986–2010 period.

KEYWORDS:

• Independent local parties
• voting behaviour
• the Netherlands
• political trust
• new political parties

1. Introduction

In the Dutch municipal elections of 2018, independent local parties (parties which only run in municipal elections in a single municipality)¹ won 29% of the vote. They were by far the largest bloc in the municipal elections. In 1986 such parties won 12% of the vote. In a period of 32 years, this political family gradually more than doubled its votes. Surprisingly little is known about independent local parties. Little is known about what motivates citizens to vote for these parties (but see Otjes 2018). Studies of these parties have focused on organisational and institutional features (Reiser and Holtmann 2008; Boogers and Voerman 2010). The main question of this paper is: what explains the growth in support for these parties among Dutch voters?

There are four key reasons to study the growth of local parties in the Netherlands. Firstly, the growth of independent local parties cannot just be observed in the Netherlands alone. As will be discussed in greater detail below, such parties have grown between 1986 and 2010 in many other countries such as Germany, Norway, and Sweden. Therefore, a study of the growth of these parties in the Netherlands is relevant beyond the borders of the Netherlands. Secondly, the rise of independent local parties is often linked to growing dissatisfaction with national politics (Janssen and Korsten 2003). Political scientists have studied dissatisfaction with political parties extensively (Wattenberg and Dalton 2002). Independent local parties provide an uncharted form of dissatisfaction. Instead of turning away from politics entirely, or supporting new political parties (often led by charismatic politicians), voters turn towards politicians who promise to represent their local community. Evidence of an empirical link between growing political distrust and the growth of these parties has, however, hitherto not been established. Thirdly, understanding why people vote for non-state-wide parties, such as independent local parties, is important for the notion of ‘second order’ elections. Schakel and Jeffery (2013) propose that support for non-state-wide parties should not necessarily be understood as protest voting. This paper tests empirically to what extent voting for non-state-wide parties can be linked to citizens’ trust in the national political elite. Fourthly, this study further builds on the limited existing studies of voting for independent local parties (Otjes 2018) and tests whether the patterns found in recent years also hold for a period of almost a quarter of a century.

The growth of independent local parties in the Netherlands offers a puzzle for political scientists. One may expect that these parties mobilise voters who have become alienated from party politics at the national level. Yet despite stories of the decline of party politics, political trust has remained surprisingly constant (Bovens and Wille 2011). Country-specific explanations also do not suffice: these parties used to be strong in Catholic regions and in smaller municipalities (Voerman and Otjes 2018). Yet, Catholicism has decreased since the 1980s. Small municipalities have been integrated into larger municipalities.

This article studies the case of the Netherlands for a number of reasons. Firstly, because in the Netherlands independent local parties have grown significantly. Almost one in three Dutch voters votes for independent local parties in municipal elections. Still, the Netherlands is only the third country in Europe in terms of voting for independent local parties, after Belgium and Germany (Otjes 2018). Secondly, seven out of the most recent ten Dutch municipal elections were held in years in which there were also national elections. In five of those, questions about municipal voting were actually included in the Dutch Parliamentary Election Study. This allows for a study of the individual voting choice for independent local parties over a period of almost a quarter century. This is the exact period in which independent local parties went from being a relatively marginal phenomenon to becoming the largest political group in Dutch municipal councils if taken together.

This paper will have the following structure: the first section will examine general expectations about why voters may vote for independent local parties. The next sections will discuss the Dutch case from a historical and comparative perspective. The subsequent sections will look at the study’s methodology and the results. The final section will come to conclusion about the hypotheses and consider the implications of the study beyond the borders of the Dutch case.

2. Explanations for the support for independent local parties

The understanding of voting behaviour in subnational elections largely takes place within the debate about ‘second order’ elections. The central idea pioneered by Reif and Schmitt (1980, 8) is that some elections, including European, municipal and regional elections, are of secondary importance in the mind of voters and that their vote is structured by national considerations, rather than local ones (see Schakel and Jeffery 2013 for a critical discussion). Evidence from the Netherlands and neighbouring countries indicates that in these systems municipal elections are indeed second order elections (Lelieveldt and van der Does 2014; Marien, Dassonneville, and Hooghe 2015; Heath et al. 1999).

Schakel and Jeffery (2013) emphasise that non-state-wide parties can benefit in regional elections when they can capitalise on a distinct regional identity and regional autonomy and that their support should not necessarily be seen as a protest vote against the national political elite. In line with this, Otjes (2018) considered three reasons why voters supported independent local parties in the Netherlands in 2014: political distrust, the absence of preferred parties and localism. The first two are push-factors related to national politics. The latter is a local pull-factor. It encompasses the idea that voters vote for these independent local parties because they feel attached to their municipality, its specific needs or its autonomy (Copus and Wingfeld 2014, 664). Otjes (2018) finds that the first two factors are more important in voting for independent local parties than the last factor. As these two factors are related to national parties, one may conclude that is fruitful to explore the rising support of independent local parties within the context of second order elections, thus focusing on national level factors. This article looks at the demand for alternatives to national parties and the at times limited supply of national parties.

2.1. Demand side: political distrust

It may be the case that independent local parties are more successful when established alternatives fail (cf. Hauss and Rayside 1978). This means that increasing dissatisfaction among voters (‘the demand side’) with national political parties may fuel voter support for independent local parties. The political science literature has established a relationship between political trust and voting for new, populist or third parties: protest voting (Bergh 2004). This paper understands political distrust as the belief that in general government institutions, political parties and political decision-makers do not act according to the normative expectations of the citizen and specifically do not act in the public interest (Levi and Stoker 2000, 498). Voters who feel this way tend to vote differently from voters who do. How they vote differs from system to system: in two-party systems, the opposition party tends to attract voters with low levels of political trust or third parties may mobilise distrustful voters (Hetherington 1999; Bélanger and Nadeau 2005). In multi-party systems, populist parties, such as the Flemish Interest in Belgium or the Norwegian Progress Party, mobilise voters with low levels of political trust (Hooghe, Marien, and Pauwels 2011; Miller and Listhaug 1990). In multi-party systems, new parties may also attract these votes (Lago and Martínez 2011).

Indeed, evidence shows that independent local parties get support from voters with low political trust (Otjes 2018). This is reasonable considering the idea of protest voting: that political distrust can fuel voting for parties other than the national ones. It is also reasonable if one looks at these independent local parties themselves: they are often founded out of dissatisfaction with national parties (Zouridis and Tops 1994, 79; Boogers and Voerman 2010, 85). Politicians belonging to independent local parties are often more distrustful of national politics than other politicians (Angenendt 2018). Many independent local parties appeal to voters with an anti-political or populist message (Angenendt 2015, 135; Boogers, Lucardie, and Voerman 2007, 78; Holtman 2008, 13). In the Netherlands, the rise of populism at the national level is strongly linked to the success of local parties (Lucardie and Voerman 2012, discussed in greater detail below). One can expect that independent local parties mobilise voters by activating the division between establishment and the anti-establishment. A decline of political trust could explain the rise of independent local parties.

1. Political Trust Hypothesis: the more political distrust grows, the more support independent local parties will have.

2.2. Supply side: the absence of preferred parties

The first hypothesis focused on the demand side. The supply side may also matter: when national parties do not participate in municipal elections in every municipality, their voters need to consider alternatives (Lelieveldt and van der Does 2014; Marien, Dassonneville, and Hooghe 2015). That is, when voters cannot vote for the party of their first preference in their own municipality, they become ‘floating voters’.

National parties are more likely to compete in larger municipalities than in smaller municipalities. Their electoral niche might not be large enough to sustain them in those municipalities (Kjaer and Elklit 2010, 437). Therefore, the support for independent local parties may be larger in smaller municipalities (Kjaer and Elklit 2010, 439; Vampa 2016, 583; Göhlert et al. 2008; Aars and Ringkjøb 2007, 172; Janssen and Korsten 2003). Moreover, they may not have sufficient members to organise themselves there (Vampa 2016, 583). Supporters of new political parties may be more likely to be marooned by their party. These parties lack organisational capacity and may therefore not compete in all municipalities. At the same time, supporters of new political parties are more likely to have less alternatives. These parties are often not part of traditional party families, where if one party of the family does not compete, voters have the alternative of voting for a kindred party. Recent years have seen the rise of new parties all over Europe, but particularly in the Netherlands (Hobolt and Tilley 2016; Krouwel and Lucardie 2008).

When national parties do not compete in all municipalities, this leaves voters without the party of their first preference. Independent local parties may be a good alternative to the party of their first preference. This means that they do not have to vote for another national party, that is for a competitor of their party of first preference. Holtman (2008) and Otjes (2018) provide evidence that independent local parties indeed function as a surrogate for absent national parties. This pattern can also explain the growth of independent local parties over time: if there has been a rise in the support of national political parties which are weakly embedded on the national level and do not run in many municipal elections, the support of new parties may grow.

• (2) Supply-Side Hypothesis: the more voters’ support for national parties that did not run in municipal elections in their municipality grows, the more support independent local parties will have.

3. Comparative context

These two hypotheses are put to the test in the Dutch case. This country has been selected for this case study for methodological reasons: data on individual preferences for independent local parties is available from the 1980s onwards. Since then, the Netherlands has seen a considerable rise in independent local parties. The Netherlands is not a unique case: Figure 1 shows the growth of independent local parties in the Netherlands, eight German states,² Norway and Sweden.³ The figure shows that the share of votes for independent local parties in the Netherlands has grown considerably. In Bavaria, Rhineland-Palatinate and Baden-Württemberg, the support for independent local parties has grown in a similar way as it did in the Netherlands. In other German states, the support for independent local parties is smaller but their growth is larger: in North Rhine-Westphalia, Lower Saxony and Hessen, their support has more than doubled. Saarland is the only region in this sample in which this growth has declined during the period specified. In Norway and Sweden, the support for these parties shows more of a boom-and-bust cycle: support peaks to 7% (in Sweden in 2006) and 4% (in Norway in 1995). After that, the support weakens. Still, the share of voters that voted for these parties in the most recent election is greater than the share of voters who undertook the same action in the 1980s. Overall, the growth of independent local parties in the Netherlands is substantial, and similar trends can be seen in other Western European countries.

Figure 1. Share of votes for independent local parties.

Thick black line: the Netherlands.Thick long black stripes: Sweden.Thick short black stripes: Norway.Thick long and short black stripes: Rhineland-Palatinate.Thick grey line: North Rhine Westphalia.Thick long grey stripes: Bavaria.Thick short grey stripes: Lower Saxony.Thick long and short grey stripes: Baden-Württemberg.Thin black line: Hessen.Thin long black stripes: Schleswig-Holstein.Thin short black stripes: Saarland.Sources: (Voerman and Otjes 2018; SCB 2007; Valmyndigheten 2018; SSB 2018; LRP 2015; ITNW 2018; BLS 2018; LSN 2018; Hin and Eisenreich 2010; SLBW 2018; CDU 2006; HSL 2018; SAHSH 2013; FNS 2012; LZD 2009, 2014).

4. Historical context

In order to analyse the growth of independent local parties in the Netherlands, one needs to consider the history of Dutch independent local parties and voting behaviour more generally. There are three approaches to voting in the Netherlands: one based on religion and class, one based on issue positions and one based on political trust. These can be observed to apply to different time periods.

Firstly, until the mid-1960s, a sociological voting model based on religion and class could explain voting in the Netherlands (Lijphart 1968). It reflected the pillarised social structure in the Netherlands, in the context of which many citizens were embedded in tight-knit networks based on their religion and class. Catholic voters voted for the Catholic People’s Party (Katholieke Volkspartij, KVP) and (starting in 1977) for the Christian-Democratic Appeal (Christen-Democratisch Appèl, CDA). Protestant voters voted for one of a number of protestant parties, which would eventually merge into CDA as well. Out of the secular voters, the working class voted for the Labour Party (Partij van de Arbeid, PvdA), the middle and upper classes voted for the Liberal Party (Volkspartij voor Vrijheid en Democratie). The pillarisation of Dutch society actually contributed to the formation of independent local parties in some regions. The southern provinces of Limburg and North Brabant were almost uniformly Catholic until the mid-1960s (Janssen and Korsten 2003). Before 1970, the KVP was so dominant that other national parties did not participate in many municipal elections. In the southern provinces, the KVP opted for a policy of non-intervention: if no other national parties ran in the municipal elections, it also chose not to run (Janssen and Korsten 2003). This created the possibility for independent local parties to run successfully. Only when another national party ran in municipal elections would the KVP run its own list, ensuring that these municipalities would not fall into the hands of competitors. This allowed the diverging tendencies within the KVP to organise themselves into small local interest parties (Bornewasser 1995, 316). In four out of five municipalities in Limburg, only local parties participated in elections in the 1950s and 1960s (Kuiper 1994, 39). In 1974, 72% of the votes in Limburg and 51% of the votes in North Brabant were cast for independent local parties (Voerman and Otjes 2018). As can be seen in Figure 1, independent local parties are also strongest in southern, dominantly Catholic, states of Germany (Angenendt 2015). As the KVP began to lose voters in national elections in the 1960s and 1970s, other national parties began entering into municipal elections in the Catholic south. The KVP dropped its non-interference policy to the detriment of these local parties. Still, a legacy of the past persisted, and in many municipalities in the Catholic south, independent local parties remained strong. In general, one may expect Catholic voters and voters in the southern provinces to vote for independent local parties more often. This can, however, not fully explain their growth as the share of Catholics has declined due to secularisation.

The second model focuses on issue positions on an economic dimension and a moral dimension. Voters with conservative views on moral issues would vote for the CDA; voters with progressive views on these issues would vote for the Labour Party or the Liberal Party. Those electorates differed with regards to economic matters: those with right-wing views on economic issues would vote for the Liberal Party and those with left-wing views on economic issues for the Labour Party (Irwin and van Holsteyn 2008, 183–184).

Independent local parties are very diverse. They do not have a unified profile. Left-wing, right-wing, progressive and conservative views are all represented in these parties (Boogers, Lucardie, and Voerman 2007). Therefore, one would not expect that these dimensions are associated with voting for these parties in the aggregate.

This issue-based model was stable until 2002, when a new party entered the political arena: the List Pim Fortuyn (Lijst Pim Fortuyn, LPF). This party’s history is intrinsically linked to the development of independent local parties. In the 1990s, independent local parties expanded their vote share, in particular in larger municipalities outside of the Southern provinces. A number of these parties used the term ‘Leefbaar’ (Liveable). They shared a similar anti-establishment rhetoric. Representatives from a number of these parties came together to form a new party, Leefbaar Nederland (Liveable Netherlands, LN) in 1999 (Lucardie and Voerman 2012, 76–77). This new party rose in the polls. At the same time, disagreements within the party over immigration policy grew (Lucardie and Voerman 2012, 84). The party’s top candidate, Pim Fortuyn, split from LN and formed the LPF as a populist, anti-immigration party. Fortuyn, however, was killed just before the 2002 election. Still, the party won 17% in the 2002 election, which was unprecedented for a new party. The question to what extent voters of Fortuyn voted for the party because of specific issues or because of political trust is a matter of lively debate (Van der Brug 2003; Bélanger and Aarts 2006). Since then, most independent local parties in the Netherlands employ populist rhetoric (Boogers, Lucardie, and Voerman 2007). The rise of Fortuyn may have reinforced the relationship between low political trust and voting for independent local parties in recent elections.

5. Methodology

This paper seeks to explain the growth of local parties in the Netherlands between the 1980s and the 2010s. One of the key reasons this country has been selected is due to coincidence: a number of Dutch municipal and parliamentary elections were held in the same year. The Dutch Parliamentary Election Study (Van der Eijk, Irwin, and Niemöller 1986; Thomassen 1994, 1998; Skon et al. 2003, Brinkman et al. 2007; Mokrini et al. 2012) has asked respondents in many surveys about their voting behaviour in the municipal elections which were held a few months before the national election that these surveys focused on. In recent decades, municipal elections were held in the same year as parliamentary elections in 1982, 1986, 1994, 1998, 2002, 2006 and 2010. In five out of seven years, questions were included about the party choice in the municipal election (1986, 1994, 1998, 2002 and 2010). These surveys will be used as a basis for this paper.

The goal of this paper is to understand how the growth of independent local parties can be explained. Therefore, all five waves of the DPES are analysed together in a multilevel logistic regression analysis using the years as levels. This multilevel regression analysis consists of two parts: fixed effects and random intercepts (Bickel 2007, 126–128). The fixed effects are the coefficients for the variables derived from the hypotheses and control variables. In this case, the random intercepts account for the year-to-year variability in support for independent local parties. In an empty model with just random intercepts without any fixed effects, the random intercepts just reflect the difference in support for independent local parties between the surveys. If one were to add explanatory variables to the model, the random intercepts reflect the extent to which there is year-to-year variability, which the fixed effects cannot explain. The smaller the random intercepts are, the better the variables included in the model are able to explain the increasing support for independent local parties over time. By comparing models without the two hypothesised variables to models with either of these and a model with both of these, one can see to what extent the inclusion of a variable decreases the random intercepts. The variable that leads to the greatest reduction of the random intercepts does most to explain the increasing support of independent local parties over time. All analyses use the cases that have valid scores on all variables included in the most complete model. This way, differences in the random intercepts are not due to the inclusion or exclusion of specific cases.

The independent variable is voting for an independent local party. In 1994, 1998, 2002 and 2010, this was a response option. In 1986, this voting choice was combined with other options into ‘other’. A segment of these options is likely to be combined lists of smaller left-wing parties. This means that the 1986 results need to be interpreted with some caution.⁴ Descriptive data can be found in the Appendix. All variables are calculated so that their minimum is zero and their maximum is one, to allow easy comparison of effect sizes.

The political trust hypothesis proposes that voters who have less political trust are more likely to vote for independent local parties. Since at least 1986, the DPES has asked the same set of questions about the extent to which citizens trust (national) politicians and political parties. These are used to measure political trust.⁵ These five items scale sufficiently.

The second hypothesis concerns the supply side. Sadly, one cannot determine whether a respondent’s party of first preference ran in the municipal elections in the municipality the respondent lived in, because to ensure anonymity the municipality is not included in most surveys. To get a grasp on this phenomenon, this article follows Otjes (2018): firstly, it is determined what share of voters in the entire country can actually vote for each national political party in the municipal elections in their own municipality. Many national political parties only participate in a limited number of municipalities. In which municipalities one could actually vote for each party was determined based on existing overviews of municipal election results (Kiesraad 2018; Nlverkiezingen.com 2018). The share of municipalities a party ran in was weighted by the share of eligible voting population per municipality. In this way, every party is assigned a number between zero (not running in any municipality) and one (running in every municipality). These scores were linked to voters’ party preferences before the election. Every party preference was replaced with the share of the voters who could actually vote for the party. In each case, the municipal elections were held before the national elections, meaning that this is the party preference closest to the municipal election. The expectation is that the lower this number is, the more likely it is that a voter will vote for an independent local party.⁶ Otjes (2018) shows that this approach leads to the same substantive conclusions as an approach that relies on a more precise measure of whether a voter could vote for the party of their first preference in their municipality.

A number of control variables is included: the discussion above touched on religion, municipality size, class as well as on views on economic, moral and cultural matters. In addition, year of birth, gender and party identification are included. Firstly, one may expect that in smaller municipalities, more respondents vote for independent local parties. Because of differences in how municipalities were coded over time, this study uses a simple dichotomy between the four largest municipalities and the rest of the country.⁷

Secondly, a legacy of the past may be at play: as discussed above, until the mid-1970s, local parties were to a large extent a Catholic phenomenon. Catholic voters may consequently have been socialised in a setting where it is appropriate to vote for independent local parties. Therefore, a variable is included which reflects whether or not voters are Catholic.⁸ Class is another variable that used to strongly structure voting behaviour during the ‘pillarised’ period. A class dummy which divides between the working class on the one hand and middle and upper class voters on the other is included. Class is included in the models without an explicit expectation for the direction of the relationship. Thirdly, items on economic and moral matters, specifically income equality and euthanasia, are included.⁹ The diversity of independent local parties would lead one to expect no relationship to these variables.

Next, party identification is included. Voters that do not identify with a national political party are more likely to vote for an independent local party (Voerman and Otjes 2018): there can only be space for local parties if voters feel less attachment to the national ones. The renewed success of independent local parties in the Netherlands follows the thawing of the party system in the 1960s and 1970s. In every survey used, voters are asked to go through a step-by-step section in the questionnaire that determines the strength of their party identification.

Finally, gender and year of birth are included. Year of birth is used instead of age because it allows one to discern cohort effects as pillarisation weakened over time. One would expect younger voters to vote for independent local parties because they were socialised during a period when party identification was weaker. A binary gender variable is included without an explicit expectation for the direction of the relationship.

6. Results

Before the regression results, it is worthwhile to look at some aggregated patterns over time. These can indicate which factors can explain the growth of independent local parties. Table 1 shows the share of voters that voted for independent local parties, the share of respondents in the DPES, the average level of political trust (hypothesis 1) and party participation in municipal elections (hypothesis 2). Both in election results and in the DPES, the share of voters who voted for independent local parties grows significantly. Interestingly, the level of political trust stays much the same. Between 1986 and 2010, the average value of the scale increases by less than 0.01. This pattern is supported by other studies of political trust in the Netherlands, which observe oscillation but no decline (Bovens and Wille 2011). The level of party participation declined significantly from 0.87 to 0.45. More and more voters began to prefer parties than ran in less and less municipalities. So, out of these two factors, decreasing party participation appears to be the most likely explanation for the growth of independent local parties.

Table 1. Electoral results of independent local parties and descriptives 1986–2018.

Year	Election Result	Percentage DPES	Political Trust	Party Participation
1986	12%	14%	0.51	0.77
1990	13%	-	-	-
1994	16%	18%	0.51	0.61
1998	18%	16%	0.53	0.67
2002	26%	19%	0.49	0.60
2006	22%	-	0.51	0.52
2010	24%	22%	0.52	0.49
2014	28%	-	-	-
2018	29%	-	-	-

Source: Voerman and Otjes (2018)

Table 2 shows a number of regression models: an empty model (Model 1), a model with just the control variables (Model 2), a model with all variables except for party participation (Model 3), a model with all variables except for political trust (Model 4) and a ‘full’ model with both political trust and party participation (Model 5). Appendix 1 offers a number of additional models including those with a different measure of party participation and municipality size, which include a region variable and those that examine the differences between the 1986 data and the rest.

Table 2. Binomial regression results.

Model	1	2	3	4	5
Period	1986–2010	1986–2010	1986–2010	1986–2010	1986–2010
Political Trust	-	-	−0.74***	-	−0.66***
			(0.18)		(0.18)
Party Participation	-	-	-	−0.72***	−0.70***
				(0.10)	(0.10)
Party Identification	-	−1.43***	−1.36***	−1.12***	−1.07***
		(0.12)	(0.12)	(0.13)	(0.13)
Economic Dimension	-	0.14	0.14	0.05	0.05
		(0.15)	(0.15)	(0.14)	(0.15)
Moral Dimension	-	0.41**	0.41**	0.48***	0.48***
		(0.15)	(0.14)	(0.14)	(0.14)
Religion = Catholic	-	0.69***	0.68***	0.73***	0.72***
		(0.09)	(0.09)	(0.09)	(0.09)
Gender = Female	-	−0.20**	−0.20**	−0.21***	−0.21**
		(0.08)	(0.08)	(0.08)	(0.08)
Class = Working Class	-	0.14	0.08	0.12	0.06
		(0.09)	(0.09)	(0.09)	(0.09)
Year of Birth	-	0.75***	1.04***	0.62***	0.89***
		(0.26)	(0.27)	(0.26)	(0.27)
Urbanisation	-	−0.86***	−0.86***	−0.87***	−0.87***
		(0.14)	(0.14)	(0.14)	(0.14)
Intercept	−1.57***	−1.78***	−1.59***	−1.40***	−1.24***
	(0.08)	(0.23)	(0.24)	(0.23)	(0.24)
AIC	4573	4269	4254	4216	4205
N Years	5	5	5	5	5
N	4910	4910	4910	4910	4910
Variance of the Random Intercept	0.03	0.01	0.01	0.00	0.00
	(0.17)	(0.12)	(0.10)	(0.05)	(0.03)

0.1 < * < 0.05 < ** < 0.01 *** < 0.00

The first question is to what extent the different variables are significantly related to the chance of someone voting for an independent local party. The first hypothesis concerns the effect of political trust. The underlying idea is that voters with little faith in national politicians and parties are more likely to vote for independent local parties. Having the highest level of political trust cuts the chance of someone voting for an independent local party in half, compared to voters with the lowest level of political trust. In each one of the models in Table 2, the effect is observed to be strong, significant and in the expected direction.

Second, there is the extent to which one’s party of first preference participates in elections. If the party of one’s first preference participates in a limited number of municipalities, it is more likely that a voter is forced to vote for another party. A difference can be observed between voters who prefer a party that participates in all municipal elections to voters who prefer a party that participates in practically no municipal elections, and in the latter case the chance of someone voting for an independent local party doubles. Again this effect is observed to be strong, significant and in the expected direction. As mentioned above, this variable is somewhat problematic in how it deals with missing cases. Here the value zero was assigned to voters who did not yet know what they would be voting for. One can also remove those cases. In that case the N drops quite sharply. The models with this alternative specification, included in the Appendix, show that the patterns are similarly strong, significant and in the expected direction.

So, both variables are strongly and significantly related to voting for independent local parties. The question is, however, which factor or factors can explain the rise of independent local parties over time. The random intercepts show what level of unexplained variance there is between periods. Figure 2 visualises these intercepts. The empty model gives insight into the actual level of year-to-year variance in voting for independent local parties. The random intercepts in the empty model reflect the extent to which an election in a given year deviates from the share of support for independent local parties in the entire sample. The empty model indicates that in 1986, the share of the vote for independent local parties is 17% lower than average. In 2010, it is 27% higher. This is the variance that this paper sets out to explain. When one includes explanatory variables, the random intercepts reflect the extent to which there is unexplained variance between the years. The smaller they are, the better the variables included in the model explain differences between years. The model with only the control variables still needs considerable random intercepts: in 1998, the share of voters is 11% lower than the fixed effects would predict and in 2010, it is 13% higher. The control variables therefore do not explain the change over time. In the model where all variables are included, the random intercepts are smaller than 2%. This implies that the theoretically motivated variables that have been included in the model are good explanation of the differences between the periods. By dropping one of the two hypothesised variables in two separate models, one can get a sense of which variables are important for modelling the variance between periods. If one drops the party participation variable, considerable random intercepts are still necessary: in 1998, the share is 10% lower than the fixed effects would predict, while in 2010, they are still 12% higher. This shows a reduction compared to the empty model, albeit only a partial one. When the political trust variable is dropped, the random intercepts are relatively small: the correction of the random intercepts is less than 4%. This means that some unexplained variance between periods remains, but the bulk is explained. Overall, the differences between these two models imply that changes in party participation are far more important than changes in political trust when it comes to explaining the changes over time in voting patterns for independent local parties.

Figure 2. Random intercepts per year.

Finally, one can look at the effect of the control variables. Firstly, Catholicism. In every model, being Catholic had a strong and significant effect on voting for an independent local party. Catholics were twice as likely to vote for an independent local party compared to other voters. This reflects the Catholic roots of this phenomenon in the Netherlands.¹⁰ Class does not consistently or significantly affect voting for an independent local party. Two opinion items are included in the model: the economic and the moral dimension. As expected, voters of independent local parties do not have a consistent social-economic preference. Contrary to expectation, these voters do have a specific preference on moral issues: they tend to be more progressive on moral issues. The most morally progressive voters are about 60% more likely to vote for an independent local party compared to the most morally conservative voters.¹¹ There also is a strong effect on party identification. The idea is that those who no longer identify with a national party are more likely vote for an independent local party. Identifying strongly with a national party compared to not identifying with a national party decreased the chance of voting for an independent local party by around 70%. In each model the effect is observed to be strong, significant and in the expected direction. There is a consistent effect for gender between models: female respondents are about 20% less likely to vote for such a party compared to male respondents. Finally, there is a consistent effect for year of birth: the youngest voters in the data set are 90 to 180% more likely vote for an independent local party, compared to the oldest voters. The final variable looked at is urbanisation. Residents from the most urban municipalities are 60% less likely to vote for independent local parties compared to residents of the other municipalities. The Appendix includes alternative, richer operationalisations of urbanisation. This shows that in both the 1986 and 1994–2010 data sets, the richer measure shows that between the most and least urbanised municipalities, votes for independent local parties declined by more than three quarters.¹²

7. Conclusion

The central question in this paper is why the support for independent local parties has grown in the Netherlands between the mid-1980s and recent decades. Two explanations were examined: declining political trust, and the growth of parties that do not run in all municipalities. Given that many independent local parties are founded out of dissatisfaction with national political parties, the expectation that the growth of independent local parties reflects growing political distrust seems reasonable. While voters of independent local parties distrust national politics, political distrust has not grown significantly between 1986 and 2010: therefore, their growth cannot be explained by growing political distrust. The analyses indicate that the growing support of parties that do not run in all municipalities explains the growth of independent local parties. New parties developed on the left side of the political spectrum (e.g. the left-populist Socialist Party), on the right side of the political spectrum (e.g. the right-wing populist LPF) as well as in the centre (e.g. the pensioners’ party 50Plus) during this time period. Their voters tend to vote for independent local parties at the municipal level. There are only three parties that run candidates in more than 90% of the municipalities in every election year: the Labour Party, the Liberal Party and the Christian-Democrats. The support of these parties has decreased significantly in this period. The growth of independent local parties thus reflects a thawing of the party system at the supply side: it is a by-product of the growth of other new parties. These new and organisationally weak political parties are unable to run candidates in every municipality. Moreover, these parties are often not members of broader party families. Therefore, it is not even possible for voters to vote for a party of the same party family as the party of their first preference. This results in the availability of a large number of floating voters, which can be captured by independent local parties. Rather than dissatisfaction with national party politics, it is the diversification of national party politics that mainly fuels the rise of new political parties.

So what is the relevance of this study beyond the border of the Netherlands? The result shows the complexity of voting behaviour in multilevel systems. When it comes to independent local parties in the Netherlands, as an example of non-state-wide parties, Schakel and Jeffery (2013) are partially correct that their support cannot just be understood as protest votes. Yet this does not mean that national party preferences do not play a role in determining why people vote for independent local parties. In this case, voters who cannot vote for the party of their first preference are more likely to go for a local, non-state-wide option. The rise of new political parties can be seen all over Europe. Yet if these parties are municipally not embedded, they generate a number of voters who are politically homeless in the municipal elections. Other new political parties, in this case in the form of independent local parties, can profit from that. The strong results found for the party participation hypotheses makes this a reasonable expectation to explain the rise of independent local parties in other countries, in particular in Germany where these parties have grown. The continuous growth of independent local parties in Germany in the last 30 years coincides with the period during which Germany moved from the stable ‘two-and-a-half’ party system to the current formation with six parties in the Bundestag. Similar trends towards increasing fractionalization have been seen more recently all over Europe. This may be reflected locally in the support for independent local parties.

Further research may want to study voting patterns for independent local parties in a comparative fashion. A specific question here is whether the boom-and-bust cycle that can be seen in Sweden and Norway can be understood by changes on the side of national parties or developments in political distrust. Future research may also want to examine differences between municipalities more explicitly. Including variables at the municipality level may allow one to trace the effect of the supply side better: this includes tracing more precisely what the effect on voting behaviour is of the party of one’s first preference not running. It may allow one to see what the effect of the supply of independent local parties is on voting behaviour. Finally, one may want to examine how supply and demand interact by seeing whether the ideological profile of the independent local parties matters for which specific groups of voters support them.

Acknowledgement

The author would like to thank Michael Angenendt, Gerrit Voerman and anonymous reviewers for their useful comments and suggestions and Niki Haringsma for his editorial assistance in the preparation of this article.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Notes on contributors

Simon Otjes

Simon Otjes is a researcher in the Documentation Centre for Dutch Political Parties at Groningen University. His research looks at voting behaviour and party politics in Europe examining, amongst other topics, local political behaviour. He has previously published in the European Journal of Political Research, Party Politics and West European Politics.

Notes

1. This study uses the label ‘independent local party’ (Otjes 2018). These groups are parties in the sense that they participate in elections. They are local in that they are exclusively organised at the local level. They are independent in that they are not affiliated with a national party.

2. The eight West-German non-city states, because these actually have municipal elections and because data is available from the 1980s onwards, the same period studied for the Netherlands.

3. These have been included because their electoral system is similar to that of the Netherlands.

4. Because of this, in the Appendix some analyses will be run separately for the 1986 data and the other years.

5. ‘MPs do not care about the opinions of people like me.’; ‘Parties are only interested in my vote, not in my opinions.’; ‘Politicians promise more than they can deliver.’; ‘Ministers and junior ministers are primarily self-interested.’; and ‘Friends are more important than abilities to become MP.’

6. A large share of respondents does not have a party preference yet. One can either remove these cases or assign them the value zero (because they do not prefer a party which runs in municipalities). The latter variable is included in the main text. The first variable is included in the Appendix.

7. The 1986 DPES has a different indicator than the other DPESs. These could only be harmonised in a sample dichotomy which measures whether or not the respondent came from one of the four largest cities (Amsterdam, Rotterdam, The Hague and Utrecht). This measure is used in the article but the Appendix covers separate analyses for 1986 and the other years.

8. The Appendix also examines the effect of living in the three southern provinces (Limburg, North Brabant and Zeeland) on voting for independent local parties. This is not included in the main analysis for two reasons. Firstly, the region variable is not in the 1994 data, reducing the number of years. Since we are examining growth over time, losing an entire year is to be avoided. Secondly, the region variable is only available for the four remaining years as a fourfold north, south, east, west classification. This includes Zeeland in the south, despite the fact that it is predominantly Protestant. It also puts Catholic majority regions in Gelderland and Overijssel in the northern region, despite that fact that the principle of non-interference also applied here. In the Appendix, we run multiple models with this north-south division. This shows most importantly that the pattern over time as hypothesised stays in place. Second, it shows that both region and Catholicism have separate effects, although the effect of Catholicism decreases due to the inclusion of the region variable.

9. A variable concerning the civic integration of ethnic minorities is used to measure the migration dimension in the Appendix. It is only available for 1994–2010. It shows a significant effect.

10. This is also sustained by analyses in the Appendix that include a region variable. This shows that support for independent local parties is still stronger in southern provinces.

11. An item for the migration dimension is included in the Appendix. This was shown to be a significant predictor of voting for an independent local party (Otjes 2018). This aligns with expected results, as the LPF, whose rise is closely linked to the rise of independent local parties, mobilised voters with an anti-immigration agenda.

12. There is a big difference between the 1986 data set and the other data set in how they operationalised voting for an independent local party. The Appendix therefore also looks at the data for the elections without 1986 and only for 1986. The patterns discussed above are consistent if one drops 1986. The model for 1986 differs from the other models: it shows that voters of independent local parties are not less likely to trust politics but that they do have more left-wing views than other voters. This is likely caused by the different operationalisation of independent local parties in this data set: this group was thrown together with all ‘other’ voting options and is likely to encompass combined lists of left-wing parties. Overall, the 1986 data are different, but it is important to note that the results remain consistent if one excludes these data.

Appendix

Table A1. Descriptives.

Variable	Mean	Median	Min	Max	SD	N	High
Vote Choice = Independent Local Party	0.18	-	0	1	-	7345	Independent local party
Political Trust^a	0.50	0.53	0	1	0.24	7371	High trust
Party Participation	0.62	0.91	0	1	0.43	8549	High participation
Party Identification	0.44	0.33	0	1	0.36	9936	Higher identification
Gender = Female	0.51	-	0	1	-	10,071	Female
Class = Working Class	0.27	-	0	1	-	9819	Working Class
Year of Birth	0.61	0.60	0	1	0.18	10070	1993
Level of Urbanisation	0.16	-	0	1	-	10071	Big four municipalities
Economic Dimension	0.63	0.67	0	1	0.28	9699	Egalitarian
Moral Dimension	0.83	0.71	0	1	0.31	9641	Progressive
Religion = Protestant	0.19	-	0	1	-	10045	Protestant
Religion = Catholic	0.26	-	0	1	-	10045	Catholic

^aH = 0.46; Cronbach’s alpha = 0.74

Table A2. Alternative binomial regression results.

Model	1	2	3	4	5	6	7	8
Period	1986–2010	1986–2010	1994–2010	1994–2010	1994–2010	1994–2010	1994–2010	1994–2010
Political Trust	-	−0.82***	-	−0.81***	-	−0.73***	−0.64***	−0.70***
		(0.22)		(0.19)		(0.19)	(0.20)	(0.19)
Party ParticipationDK=NA	−0.61***	−0.53***	-	-	-	-	-	-
	(0.17)	(0.17)
Party Participation	-	-	-	-	−0.70***	−0.67***	−0.67**	−0.69***
					(0.10)	(0.10)	(0.10)	(0.10)
Party Identification	−0.99***	−0.92***	−1.48***	−1.36***	−1.16***	−1.10***	−1.10***	−1.14***
	(0.14)	(0.14)	(0.13)	(0.13)	(0.13)	(0.13)	(0.13)	(0.14)
Economic Dimension	0.04	0.06	−0.04	−0.00	−0.14	−0.11	−0.09	−0.10
	(0.18)	(0.18)	(0.17)	(0.17)	(0.17)	(0.17)	(0.17)	(0.17)
Moral Dimension	0.50***	0.50***	0.39**	0.39**	0.46***	0.46***	0.47**	0.58***
	(0.16)	(0.16)	(0.15)	(0.15)	(0.15)	(0.15)	(0.15)	(0.15)
Migration Dimension	-	-	-	-	-	-	0.32*	-
							(0.17)
Religion = Catholic	0.68***	0.67***	0.71***	0.70***	0.74***	0.74***	0.72***	0.70***
	(0.10)	(0.11)	(0.09)	(0.09)	(0.09)	(0.10)	(0.09)	(0.09)
Gender = Female	−0.28***	−0.28**	−0.16*	−0.15**	−0.17**	−0.16*	−0.15*	−0.17*
	(0.10)	(0.10)	(0.08)	(0.08)	(0.08)	(0.08)	(0.09)	(0.09)
Class = Working Class	−0.06	−0.12	0.23**	0.17*	0.21**	0.15	0.14	0.09
	(0.12)	(0.12)	(0.10)	(0.10)	(0.10)	(0.10)	(0.09)	(0.10)
Year of Birth	0.73**	1.04***	0.47*	0.79**	0.35	0.66**	0.69**	0.60**
	(0.31)	(0.32)	(0.28)	(0.29)	(0.28)	(0.29)	(0.29)	(0.29)
Urbanisation	−0.90***	−0.90***	−0.97***	−0.96***	−0.98***	−0.97***	−0.96***	-
	(0.18)	(0.18)	(0.15)	(0.15)	(0.15)	(0.15)	(0.15)
Urbanisation5-point scale	-	-	-	-	-	-	-	−1.47***
								(0.14)
Intercept	−1.55***	−1.40***	−1.48***	−1.30***	−1.12**	−0.98***	−1.28***	−0.46*
	(0.30)	(0.30)	(0.26)	(0.26)	(0.26)	(0.26)	(0.31)	(0.27)
AIC	2964	2951	3743	3727	3699	3686	3673	3612
N Years	5	5	4	4	4	4	4	4
N	3830	3830	4242	4242	4242	4242	4230	4242
Variance of the Random	0.00	0.00	0.02	0.01	0.01	0.00	0.00	0.00
Intercept	(0.01)	(0.00)	(0.13)	(0.12)	(0.08)	(0.06)	(0.04)	(0.09)
Model	9	10	11	12	13	14
Period	1986	1986	1986; 1998–2010	1986; 1998–2010	1986; 1998–2010	1986; 1998–2010
Political Trust	0.19	0.02	-	−0.78***	-	−0.75***
	(0.55)	(0.56)		(0.20)		(0.20)
Party Participation	−0.70**	−0.80***	-	-	−0.74***	−0.73***
	(0.30)	(0.31)			(0.11)	(0.11)
Party Identification	−0.77**	−0.89**	−1.39***	−1.32***	−1.08***	−1.02***
	(0.37)	(0.38)	(0.13)	(0.13)	(0.14)	(0.14)
Economic Dimension	0.78**	0.82**	0.16	0.15	0.07	0.07
	(0.36)	(0.36)	(0.17)	(0.17)	(0.17)	(0.17)
Moral Dimension	0.36	0.57*	0.35*	0.37**	0.43***	0.45***
	(0.39)	(0.39)	(0.16)	(0.16)	(0.16)	(0.15)
Religion = Catholic	0.54**	0.46*	0.37***	0.36***	0.42***	0.41***
	(0.25)	(0.25)	(0.10)	(0.10)	(0.10)	(0.10)
Region = South	-	-	0.75***	0.76***	0.76***	0.77***
			(0.10)	(0.10)	(0.10)	(0.10)
Gender = Female	−0.54**	−0.52**	−0.18**	−0.17**	−0.20**	−0.20
	(0.24)	(0.24)	(0.09)	(0.09)	(0.09)	(0.09)
Class = Working Class	−0.45	−0.50*	0.15	0.08	0.13	0.07
	(0.27)	(0.28)	(0.10)	(0.10)	(0.10)	(0.10)
Year of Birth	2.59***	2.58***	0.72**	1.02***	0.55*	0.85***
	(0.90)	(0.90)	(0.29)	(0.30)	(0.28)	(0.29)
Urbanisation	0.14	-	−0.60***	−0.60***	−0.62***	−0.61
	(0.39)		(0.15)	(0.15)	(0.16)	(0.16)
Urbanisation13-point scale	-	−1.65***	-	-	-	-
		(0.47)
Intercept	−2.83***	−1.81***	−1.90***	−1.71***	−1.49***	−1.31***
	(0.68)	(0.70)	(0.26)	(0.26)	(0.26)	(0.26)
AIC	512	499	3489	3476	3446	3433
N Years	-	-	4	4	4	4
N	763	763	4023	4023	4023	4023
Variance of the Random	-	-	0.02	0.01	0.00	0.00
Intercept			(0.14)	(0.11)	(0.06)	(0.00)