Getting Party Activists on Local Lists

• 1 Introduction

  The study of local politics in the Netherlands is an underdeveloped field (Vollaard, Boogaard, Van den Berg, & Cohen, 2018, p. 16). Dutch political scientists have tended to ignore this area; they left it to public administration scholars to study subnational tiers of government. Therefore, the emphasis has been on local administration rather than local politics (Andeweg & Irwin, 2005, p. 186). In particular, the role of political parties at the local level in structuring electoral competition and organising political cooperation has received scant attention (Grimberg & Vollaard, 2016). Exceptions are the national surveys into recruitment for municipal council elections that have been run since 2006 by Voerman and Boogers (2008, 2011, 2014; Boogers & Voerman, 2018).¹x As well as Grimberg and Vollaard’s (2016) survey of the municipality Voorschoten. They give us an important insight into how local party branches function and how they perform one of the crucial functions of political parties: the recruitment of candidates for public office. This study will examine the 2018 wave of these national surveys. Our specific goal is to explain the differences in how local party branches perform their recruitment function.
  Therefore, this study provides an insight into an understudied area in political science, the recruitment of candidates at the local level (De Winter, Erzeel, Vandeleene & Wauters, 2013, p. 95; Hazan & Rahat, 2010, fn. 1). Yet the results of this study are also relevant beyond that literature: the ability of a political system as a whole to recruit candidates for political office is often seen as an important indicator of the quality of democracy (Kjaer, 2007, pp. 195-196; Rallings & Trasher, 1999, p. 67). Fowler (1993, p. 19) goes so far as to write: “It is hard to imagine a regime that is both legitimate and dynamic that does not have a healthy supply of candidates.” A low number of candidates running for local elections is seen by many political scientists as a sign of a poorly functioning local tier of a democratic system (Grimberg & Vollaard, 2016, p. 39; Kjaer, 2007, p. 197; Voerman & Boogers, 2008): it may indicate that citizens are disinterested in what municipal councils decide and parties lack the connections in society to recruit candidates.
  Furthermore, the extent to which local party branches are able to recruit candidates for the municipal council is an important indicator of their strength and vitality. Existing evidence on recruitment points to a strong relationship between the number of members parties have and their ability to recruit candidates (Cross & Gauja, 2014a, p. 619; De Winter et al., 2013, pp. 102, 103; Ennser-Jedenastik & Hansen, 2013, p. 780; Norris & Lovenduski, 1995, p. 16; Voerman, 2000; Voerman & Boogers, 2014, p. 9). In comparison with other West European political systems, the Netherlands has a relatively low level of party membership (Van Biezen, Mair & Poguntke, 2012). The image of the organisational strength of local branches in the Netherlands is generally negative (Grimberg & Vollaard, 2016, pp. 20, 21; Voerman & Boogers, 2008, p. 190). Therefore, there are reasons to believe that party branches are no longer able to perform their basic recruitment function in the Netherlands. Given the low levels of party membership in the Netherlands, this may be a canary in the coal mine of what is to come in other European countries if party membership continues to decline. There is anecdotal evidence for this: local branches opt not to participate in municipal elections owing to the lack of a strong membership base from which to recruit potential candidates, even when they have a base of voters.²x Straub (2017), Bon (2017) and RTV Oost (2017). By studying the recruitment function of local party branches, we may get a grasp of the organisational strength of the party on the ground (Ennser-Jedenastik & Hansen, 2013).
  The article will have the following structure. First, we will discuss our central hypotheses. We focus on the relationship between the strength of local party branches, municipal size and the branches’ fulfilment of the recruitment function. Here we will also introduce some of the control variables that we will be employing. Next, we will discuss the specific Dutch context and the effect of the choice to study the Netherlands on the generalisability of the study. Then, we will introduce the survey and the specific measures that we employ. Subsequently, we will discuss the results of our analyses and what they mean for our hypotheses. Finally, we will draw a number of conclusions about the strength of party branches in the Netherlands and their role in municipal politics and sketch avenues for future research into party politics at the local level in the Netherlands.

• 2 Theory

  Recruitment is the only function of political parties that they have truly monopolised (Cross & Gauja, 2014a, p. 619; Dalton & Wattenberg, 2000, p. 7; Hazan & Rahat, 2010, p. 4; Norris, 2006; Norris & Lovenduski, 1995, p. 3). While, for instance, their role in electoral mobilisation is partially mediated through traditional and social media and their role in the articulation and aggregation of interests is also served by interest groups, by definition only political parties can nominate candidates for elections. Most research into recruitment, at the national and local levels, looks at what parties do when recruiting candidates (e.g. Cross & Gauja, 2014a; Devos, Reynaert & Verlet, 2008; De Winter et al., 2013; Hazan & Rahat, 2010; Norris & Lovenduski, 1995): who makes decisions? What kind of criteria do they use? Where do they recruit candidates? Who got selected? These studies ask how recruitment reflects internal democracy and affects democratic values such as responsiveness and representation (Hazan & Rahat, 2010). Studies in Denmark and the Netherlands (Kjaer, 2007; Voerman & Boogers, 2008) have pointed to a greater weakness: the question here is not who becomes a candidate but whether there are still enough potential candidates for municipal democracy to function well. That is, whether parties are able to perform their recruitment function in the democratic system. We build further on this last strand, looking systematically at what factors explain why local party branches actually field candidates.

  2.1 Recruitment

  Parties have to actively seek out, screen and select candidates to compete in elections (Dalton & Wattenberg, 2000, p. 7). At the national level, parties have structures in place to aid in scouting, screening and schooling political talents such as youth groups, scouting committees or party academies. At the local level such structures are less prominent. The local party branch has to perform all the functions of finding, screening, selecting and schooling candidates. The members who are active in the branch are likely to search for candidates among people they know (Grimberg & Vollaard, 2016; Rallings, Trasher, Borisyuk & Shears, 2010).
  Our study looks at four different dependent variables that reflect the ability of local party branches to perform this function: the first indicator is the decision whether party branches do or do not run as an independent list in municipal elections. The choice to field candidates or not is the most basic decision a party branches makes before elections. Party branches have essentially three options here: not to run at all, to run alone or to cooperate with other party branches on a joint list; these are called ‘cartel lists’ in Belgium, where they occur frequently (Devos et al., 2008, p. 58). In the Netherlands, local party branches often run on joint lists; in particular, cooperation between Christian parties and between parties on the left is quite common (Hippe & Voerman, 2010; Voerman, 2010). At least anecdotally, there is evidence that the choice not to run depends on the ability of parties to recruit candidates: there are examples from all over the Netherlands and from different party families of party branches deciding against participation in elections because they are unable to recruit what they consider sufficient candidates.³x E.g. Straub (2017), Bon (2017) and RTV Oost (2017).
  The second indicator is how many candidates local party branches put on their local lists. According to Kjaer (2007, p. 211), this is a good indicator for the health of local democracy. This is a clear objective measure of the extent to which party branches succeed in their function of recruiting candidates: a party branch that recruits only one candidate is less successful than one that is able to field a list of fifty candidates. It is important to note here that in the Netherlands (in contrast to Belgium or Denmark) the maximum number of candidates is independent of the size of municipal council.⁴x The limit is 50 candidates for political parties that have less than 15 seats in the previous legislative term; parties with more seats can field up to 80 candidates. So in a legal sense the range is 1-80, but empirically it is 1-50. In 2018, no party had more than 50 candidates. Parties use the full range: in 2018, 33 lists (out of 2,783 lists; 1%) ran with only one candidate, and 85 (3%) ran with 50 candidates. The average party had nineteen candidates with a standard deviation of eleven candidates. This diversity stands in contrast to Belgium, where branches generally aim to fill all these list slots, as incomplete lists are a sign of organisational and electoral weakness.⁵x We thank the anonymous reviewer for providing us with this contextual information. In Denmark, in contrast, parties also do not fill up the entire list (Kjaer, 2007).⁶x Kjaer (2007) suggests cultural reasons for this: “[P]utting up a list with the maximum number of candidates … may appear rather optimistic, not to say pompous, if the list cannot be expected to get more than one of a few of these candidates into office.” What explains how long local party lists are has been studied only in the Danish case (Kjaer, 2007).
  In addition, parties tend to put more candidates on the list than they would need to fill the seats they expect to win and possible vacancies that come up during the term. The reason for doing this is that they expect that people will vote for famous townspeople, friends or family; that is, additional candidates attract more votes. These candidates are referred to positively as lijstduwers (list pushers) or, more pejoratively, as nepkandidaten (fake-candidates) (Grimberg & Vollaard, 2016, p. 26; Van Holsteyn & Nagtzaam, 2018). Parties did not put them on the list because the party wanted them to become a local councillor, even when elected. While there are famous Dutch people who were lijstduwers and got sufficient preference votes to be elected,⁷x E.g. the levenslied-singer André Hazes in De Ronde Venen. we know of no empirical evidence in support of the idea that longer lists lead to more votes.⁸x The number of candidates elected merely on the basis of preference votes is low. Most preference votes are cast for the first candidates on the list. Those candidates would often be elected even without those preference votes (Boogers, Van Ostaaijen & Slagter, 2010, p. 44). The recipients of preference votes tend to be the first woman on the list, sitting members of local councils and candidates with a migration background (Boogers et al., 2010; Kranendonk, Michon, Schwarz & Vermeulen, 2014; Suijten, 2019).
  Internally, parties often differentiate between these list pushers and eligible candidates. Therefore, we can look at our third indicator, namely how many candidates on their lists party branches consider eligible. In this study, we did not tell respondents what we believe ‘eligible candidates’ were or ask them to define it. We asked them to report the number of eligible candidates. This means that what makes a candidate ‘eligible’ depends on how our respondents interpret this term. This is not merely subjective, as often local party boards may have decided how many candidates are considered eligible. On average five candidates (or 28% of the candidate lists) are considered eligible according to our respondents. As a final measure, we look at whether party branches report problems with recruiting candidates for eligible positions on the list.
  We expect that these indicators (electoral participation, the number of candidates and the number of eligible candidates, reported problems) all reflect the ability of local branches to recruit candidates for office. However, each one has drawbacks; for instance, the number of candidates and eligible candidates may be related to the number of seats a party has in the municipal council. The choice of a party branch to run in elections on a separate list may depend on the chance of getting a seat in a small municipal council where the electoral threshold is relatively high. Moreover, the three outcome measures (participation, list length, eligible seats) are not always a perfect indicator of effort: some party branches may set themselves the goal of recruiting thirty candidates and find that very challenging, while other party branches may find it very easy to recruit ten candidates and leave it at that. Compared with the existing studies of the performance of parties on their recruitment function (e.g. Kjaer, 2007, p. 211), our multi-indicator approach thus gives us a better grasp of what drives the ability of party branches to recruit candidates in general.

  2.2 Party Membership

  The most valuable resource that local party branches have is activists on the ground (Ennser-Jedenastik & Hansen, 2013, p. 780). The decline of party membership will make it more difficult for branches to recruit candidates. In traditional models of recruitment, the number of members of a party branch limits the pool of potential candidates (Norris & Lovenduski, 1995, p. 16). Parties view a vibrant membership as an important asset in recruitment because they supply a pool of prospective candidates who know the parties and whom the party knows (Cross & Gauja, 2014a, p. 619). As the number of party members of a local party branch declines, the number of candidates that a party branch can recruit from their own membership dwindles. If the number of candidates is limited, parties may have to accept candidates with lower qualifications (Ennser-Jedenastik & Hansen, 2013, p. 780; Voerman, 2000; Voerman & Boogers, 2014, p. 9), or their lists may lack the required diversity (Norris & Lovenduski, 1993). This is more likely to manifest itself in party branches rather than in the party at the national level (Ennser-Jedenastik & Hansen, 2013; Voerman & Boogers, 2011, p. 5). Research on recruitment in Belgium indicates that when party branches have more members, recruitment procedures are more formalised, while in smaller party branches the candidates themselves are more involved in recruitment (De Winter et al., 2013, pp. 102, 103). Our central expectation is that:

  1. Party membership hypothesis: the fewer members a local party branch has, the more candidate selection problems it faces.

  There are, however, theoretical and empirical reasons to doubt this hypothesis. Having a large pool of candidates is not sufficient. In a survey of candidate studies, Broockman (2014, pp. 106, 107) finds that the majority of candidates run because they are encouraged to do so. In recruiting candidates, party branches do not have to limit themselves to party members alone. They can recruit people who are not party members (Hazan & Rahat, 2010, pp. 24, 25). Research shows that party branches more often recruit candidates from among individuals who are not members of their party (Voerman & Boogers, 2014, p. 17). Still, the networks from which candidates can be recruited will depend on the networks of party members, and the more members a party has, the larger these networks are likely to be.
  Moreover, the empirical evidence, so far, shows a mixed image: Voerman and Boogers (2014, p. 19) and Boogers (2014) find that local party branches mention a lack of active members as the most important reason why they have problems recruiting members. Yet in a regression analysis the number of members, not the number of active members, is significantly related to whether local party branches report recruitment problems (Voerman & Boogers, 2014, p. 20). Instead of the absolute number of members, the recent changes in the membership appear to be an important predictor: local party branches that are losing members are more likely to experience problems recruiting candidates (Voerman & Boogers, 2014, p. 21).
  Therefore, we will look at a number of indicators to get a grasp of the relationship between the membership base of parties and their recruitment function: in addition to the actual number of members, we also look at the number of active members, the development of the membership base in the last few years and how local party branches themselves evaluate their performance in these functions. These indicators are listed in increasing order of expected predictive strength. We expect that the predictive strength of the number of active members is greater than the actual number of members: active members are likely to be on the radar of the local party branch, and this can better encourage them to run for office than passive members. Moreover, they are more likely to be motivated to run for office. Whether a branch experienced a declining membership is also likely to influence its ability to recruit candidates even more than the number of members as such: a party branch is likely to experience problems when things are not as they were before as it destabilises the status quo. Party branches develop practices to deal with the limited resources that they have. Branches with only a few members are likely to develop practices to cope with their lack of resources. Branches with more members, on the other hand, may be less economical with their resources. If either of them is hit by a decline in membership strength, a decline in their crucial resource, they will experience problems in dealing with this new reality. Finally, we expect that how branches function is the best predictor: a branch with a large number of members that functions poorly because of internal conflicts or weak organisation is likely to perform its recruitment function worse than a branch with a small number of members that lacks internal conflicts or organisational problems: this branch can focus all its resources on recruiting candidates instead of devoting resources to quashing internal conflicts or internal transaction costs.

  2.3 Municipal Size

  There is a strong relationship between the size of the municipality and the strength of local party branches (Ennser-Jedenastik & Hansen, 2013; Voerman & Boogers, 2014, pp. 16, 20): the smaller municipalities are, the weaker party branches are. This may influence candidate selection. Research from Belgium indicates that in smaller municipalities the recruitment process is organised in a less formal fashion with separate responsibilities for a recruitment committee and a larger role for the local party assembly (Devos et al., 2008, pp. 67, 68). All in all, because in smaller municipalities there are fewer people to be selected and to carry out the recruitment process, the process will be less successful. We therefore hypothesise that:

  2. Municipality size hypothesis: the smaller a municipality, the more candidate selection problems local party branches face.

  Yet there are also reasons to expect that branches in smaller municipalities do not face greater problems in recruitment than branches in larger municipalities. We know that the size of a polity has a strong effect on how democracy functions (Denters, Goldsmith, Ladner, Mouritzen & Rose, 2014; Van Houwelingen, 2017; Veenendaal, 2014; Weldon, 2006). Levels of political participation, such as turnout, are higher in smaller municipalities (Van Houwelingen, 2017). Parties in smaller polities have relatively larger and more active membership bases (Weldon, 2006). We can understand this both from a rational and from a sociological perspective: on the one hand, the likelihood that individuals contribute to a common project scales with the marginal value of their contribution. If there are fewer voters turning out, the chance that one’s vote is decisive increases. Therefore, people vote more often (cf. Downs, 1957). On the other hand, in smaller municipalities inhabitants have a greater ‘sense of community’ (Denters et al., 2014, p. 17): as communities grow larger, the share of inhabitants that individuals know personally decreases. Therefore, people’s social relations change. People identify less with their community and social distrust increases. This in turn decreases political engagements.

  2.4 Control variables

  There are other variables that correlate with municipality size that could influence the demand for and supply of candidates. For instance, larger municipalities have larger municipal councils. Kjaer (2007, p. 199) proposes that the number of candidates is likely to be determined by the chance of winning a seat. If there are fewer seats available, fewer individuals will want to become candidates because there is less of a chance to win a seat. In that case the effect should scale with the number of seats and not with the number of residents in that municipality. Another factor that scales with municipal size is the amount of time the work as a local councillor takes. This is the second most important aspect that local party branches mention when listing problems with recruitment in multiple studies (Grimberg & Vollaard, 2016, p. 39; Voerman & Boogers, 2014, p. 19). The amount of time the work as a councillor costs follows the size of the municipality. In the smallest municipalities, council work takes on average 51 hours per month. In the largest municipalities, it takes 126 hours per month (Flos, De Jager-De Lange, Maatkamp & Salentijn, 2017). One would therefore expect that in larger municipalities it is more difficult to recruit candidates, as the work takes more time. All in all, in order to get a grasp of the relationship between municipality size and recruitment, we also include the size of the municipal council and the workload of municipal councillors.
  Finally, we can include a number of indicators about the current status of the local party branch: the number of seats they currently have and whether they are currently in the local executive. The expectation would be that party branches that currently have seats experience recruitment problems less often, because they can recruit from the current councillors. Kjaer (2007, p. 204) finds a positive relationship between the current number of seats before the election and the length of party lists in Denmark. Local parties that participate in municipal executives are likely to experience fewer recruitment problems than those outside of it, because of the expectation that their return to office makes them more attractive as potential politicians (De Winter et al., 2013; Voerman & Boogers, 2014, p. 21).⁹x One factor that we are not including, but which is sometimes mentioned in the Belgian literature, is party ideology (Devos et al., 2008, p. 67; De Winter et al., 2013, p. 117). These studies show that parties with different ideologies focus on other aspects when recruiting (e.g. left-wing parties focusing on women’s representation) and run into different problems (e.g. right-wing parties experiencing difficulties with women). They do not indicate that Belgian parties of the left or right experience greater difficulty.

  Party members and elected representatives 1994-2018
  Party members of parties with representation in the national parliament.
  Source: Voerman and Boogers (2014) with additions on the basis of Kiesraad (2019) and DNPP (2019).

• 3 Local Party Branches in the Dutch Context

  The process of recruitment is shaped by specific legal and partisan institutions. The selection of party candidates is mostly an extralegal, party-internal process (Hazan & Rahat, 2010, p. 4), where the election law plays only a minor role. Therefore, we need to know something about the specific institutional context in the Netherlands.
  Municipal councils in the Netherlands are elected every 4 years. In 2018, when the survey was held, the Netherlands had 380 municipalities. As can be seen in Figure 1, the number of municipalities, and therefore the number of municipal councillors, has declined: in 1994 almost 12,000 municipal councillors were elected; in 2018 this was less than 8,000 (see Figure 1). All municipal councils are elected on the same day. The exceptions are municipalities that have recently been or are expected to be amalgamated, resulting in early or postponed elections. We disregard those municipalities in this analysis.
  The key element to understand how parties field candidates for elections is ballot structure (Däubler & Hix, 2018; Norris & Lovenduski, 1995, pp. 28, 29). In the Netherlands, semi-open list proportional representation is used for all its elections at all levels (municipal, provincial, national and European). There is a maximum to the number of candidates on the list: this is fifty unless the party group has more than fifteen seats in the relevant council (Grimberg & Vollaard, 2016, p. 30).¹⁰x It is eighty for larger parties, but this requirement is irrelevant here: no party in the survey had more than 15 seats.
  If the list wants to bear the name of a political party registered at the national level, the voter that hands in the list needs a statement of the authorised representative of that party at the national electoral council. This last point brings us to the issue of the autonomy of subnational party branches, an element that has been of considerable interest to researchers of subnational party politics (Bolleyer, 2012; Deschouwer, 2006, p. 294; De Winter et al., 2013; Fabre, 2008, p. 321; Moon & Bratberg, 2010; Müller, 2013, pp. 183-185). In essence, this requirement gives national parties a veto on whether their party branches can participate in municipal elections. To what extent national parties use this veto differs.¹¹x The Socialist Party, which did not participate in this survey, in particular uses this veto power when it believes that local party branches lack the necessary integration in local communities (De Vries, 2017; Marcelissen, 2018; Omroep West, 2017).
  It is well known in the literature regarding candidate selection that these procedures are very diverse. They differ within parties and between parties (Cross & Gauja, 2014b; Devos et al., 2008, p. 59; De Winter et al., 2013, p. 94). This is not different in the Netherlands. Parties are free to set up their own procedures (Leijenaar & Niemöller, 1997), and within parties, different branches can develop different procedures (Grimberg & Vollaard, 2016, pp. 26, 27). The board of the local party branch may appoint a separate selection committee that scouts, recruits and screens candidates and drafts a proposed list.¹²x These are often called a ‘kandidatencommissie’ (candidates committee), but referred to as ‘ad hoc comité’ (ad hoc committee), ‘comité van wijzen’ (committee of wise people) of ‘ontwerpkandidatenlijstwerkgroep’ (working group for the concept candidate list) in Belgium (De Winter et al., 2013, p. 94). Current members of the municipal council may be excluded from this selection committee. In other cases, the sitting party group leader may play an advisory or even a leading role (Grimberg & Vollaard, 2016, p. 35). In smaller municipalities, the board may take this responsibility because there are insufficient volunteers to fill a selection committee. Some parties limit who can be a candidate, for instance by having term limits or requiring a period of party membership before becoming a candidate (Grimberg & Vollaard, 2016, pp. 30, 31). This proposed list then needs to be ratified by the party members, often in a local party meeting but sometimes via e-voting (Grimberg & Vollaard, 2016, pp. 26, 29).¹³x An exception here is Party for Freedom (which also did not participate in this survey), where provincial representatives of the party selected the candidates for municipal elections (Van der Noordaa, 2017). The work in party branches is done entirely by volunteers. In general, Dutch local branches lack the financial capacity to hire long-term paid staff; although municipal council groups may have some paid assistance for a few days a week.
  The size of the party membership base is often seen as an important predictor of how branches perform in this recruitment function (Voerman & Boogers, 2014, p. 9). Party membership figures in the Netherlands have been relatively stable in the last 25 years (see Figure 1). In the Netherlands, about 2.5% of the voting population is a member of a political party. This is a major decline compared with the middle of the last century. While the total number of party members has been stable in the last few decades, the disaggregate picture shows a clear difference between the main established parties that continue to lose members and new parties that are seeing increasing membership.
  What could the results of this study say beyond the borders of the Netherlands? Table 1 shows the electoral system used in municipal elections (Van der Kolk, 2007) and the relative size of the party membership (Scarrow & Gezgor, 2010; Van Biezen et al., 2012) in West European countries. Our study looks at the number of candidates party branches field under a list system. That means that our results cannot really travel to countries that do not use a list-system, such as France, Ireland, Luxembourg or the United Kingdom do in some municipalities. Dutch political parties have lower party membership levels than the European average. Within the category of countries that use a list system in the elections, the Netherlands even has the lowest membership. So the Netherlands is a most likely case to find problems with finding candidates in elections. On the one hand, this means that the generalisability of the study is limited: if we expect to find problems in selecting candidates for a local party list, we expect to find it in the Netherlands. On the other hand, the Netherlands may serve as an early warning in the coal mine for the comparative literature: if the decline in party membership in Europe persists, as it has done in the previous years (Van Biezen et al., 2012), the problems that are found in the Netherlands may soon be found elsewhere too.

  Table 1 Relevant characteristics of western European countries

  Country	Electoral systems	Membership/electorate (%)
  Austria	List system	17
  Finland	List system	8
  Belgium	List system	6
  Norway	List system	5
  Denmark	List system	4
  Portugal	List system	4
  Spain	List system	4
  Sweden	List system	4
  Netherlands	List system	2
  Greece	Mixed system	7
  Italy	Mixed system	6
  Ireland	Majority system	2
  Luxembourg	Multiple systems	8a
  Switzerland	Multiple systems	5
  Germany	Multiple systems	2
  France	Multiple systems	2
  United Kingdom	Multiple systems	1
  Mean		5

  ^a Scarrow and Gezgor (2010) based on survey; the rest on the basis of party reports.
  Source: Van Biezen et al. (2012); Van der Kolk (2007).

• 4 Methods

  In this study, we use the 2018 wave of the Dutch survey among local party secretaries by Boogers and Voerman (2018). The details of this survey are discussed in the introduction to this special issue. Given the scope of this special issue, we will limit ourselves to local party branches and exclude independent local parties.
  As mentioned in the introduction, we will look at four dependent variables. The first is whether the party branches run as an independent list. This is derived from the survey: 12% of the branches report not running independently. Of these, 90% (44 branches) report running on combined lists with national or local parties. Only 1 in 10 (6 branches) report not running at all. This number is likely lower than the actual number of branches that decided not to run: if branches decide not to run, their organisations are likely to be weaker, and they might be less likely to answer the survey. The second is the actual number of candidates on the party list. We have obtained the actual party lists from the Open State Foundation (2018). The third is the number of eligible candidates on those branches. This was asked in the survey. The last dependent variable is whether branches experience problems getting sufficient candidates for eligible positions on the list. Respondents could answer on a four-point scale. The analysis looking at the decision to run is a logistic regression model. The two analyses looking at the number of candidates use ordinary least squares regression. The analysis looking at the level of difficulties branches face uses ordered logistic regression, as this variable has a limited number of answer options that are ordered.
  As mentioned earlier, we look at a number of indicators concerning party membership: first, we look at the membership that party branches report. The local secretaries were asked about their actual membership and the number of members that are active (participating in local party meetings, being members of the board, or members or advisers of the local party group or, managing the website or being active in the election campaign). Next, party branches were asked to indicate the direction of movement of the membership numbers: did they increase, decrease or stay stable in the last few years? Finally, secretaries were asked to rate the functioning of their branches in general on a five-point scale.
  We use three variables related to municipal size. The first is the actual number of inhabitants in each municipality, which we derive from the Central Bureau of Statistics (CBS, 2019). As control variables, we look at the size of the council and workload of councillors. We use the number of inhabitants to assign each municipality the number of municipal councillors (which depends on the size of the population) and the number of hours municipal councillors on average report performing their work as councillors, which Flos et al. (2017) collected on the basis of a survey just before the municipal election. Finally, in the survey we also asked party secretaries to report the current number of seats and whether their party branch is part of the municipal executive in their municipality. Table 2 provides some descriptive statistics of the employed ratio-interval variables. The ordinal variable reported that problems consist of four categories (‘No problems’ – 28%, ‘Scarcely any problems’ – 31%, ‘Some problems’ – 32% and ‘A lot of problems’ – 8%). Note that we use the actual numbers for nearly all variables but that the number of inhabitants is of such a different order of magnitude that we divided this number by 1,000 to get comparable numbers to the other variables.

  Table 2 Descriptives

  Variable	Mean	Median	Standard deviation	Minimum	Maximum
  Electoral participation	0.88	-	-	0	1
  Number of candidates	21.59	20.00	9.18	5	50
  Number of eligible candidates	5.33	4.00	3.98	1	20
  Number of active members	28.46	20.00	29.12	3	400
  Number of members	161.23	95.00	287.33	3	4,300
  Branch functioning	0.65	0.75	0.19	0.00	1.00
  Membership development	0.08	0.00	0.79	-1.00	1.00
  Population size (in 1,000 of inhabitants)	63.29	35.84	103.72	5.18	885.90
  Number of current seats	3.23	3.00	2.05	0.00	13.00
  Council workload (in hours)	71.1	65	10.77	61	98
  Participation in municipal executive	0.44	-	-	0	1
  Size of the municipal council	26.28	25	8.14	13	45

  Table 3 Problems branches report

  Reason	Percentages (%)
  Too few (active) members	63
  Council work takes too much time	60
  Complexity of the new responsibilities that fall within municipal remit	15
  Insufficient pay for municipal councillors	12
  Our party has term limits	3
  Growing number of threats to local politicians	2
  N	211

• 5 Results

  Before we look at different regression models, it may be useful to look at the problems parties themselves report. We list them in Table 3. The problems they mention the most are closely related to indicators we will use in the rest of our studies: more than three out of five local secretaries mention a lack of active members as the source of their selection problems. A similar percentage mentions the amount of time the council work takes. The other reasons, the complexity of municipal responsibilities and insufficient pay, are mentioned only by less than half of the respondents.
  Next, we will discuss the results of the multiple regressions. We ran four regressions (in Table 4) for each of our dependent variables: a binary logistic regression for the choice to participate, an Ordinary Least Squares (OLS) regression for the number of candidates and the number of eligible candidates and ordered logistic regression for the reported problems.
  Our first expectation is that a vibrant membership base is necessary for parties to perform their recruitment function. In all four regressions, branch functioning is a strong and significant predictor of the recruitment function. Branches that report optimal functioning are fifty times more likely to participate in elections than branches that report the worst functioning. An electoral list in a branch that functions optimally is four candidates longer than a branch that reports the worst functioning; in the case of optimal functioning, the list also has two more eligible candidates on it compared with the situation of the worst functioning. If one were to move a branch from the worst functioning to the best functioning, they are likely to go from reporting a lot of problems in candidate recruitment to no problems.
  The number of active members is a significant predictor of the choice to participate and the number of actual candidates: every active member makes a branch 6% more likely to participate in elections. For every 25 active members a branch has, an additional candidate is put on the list. In this multivariate analysis, it is not significantly related to the number of eligible candidate or experiencing problems.
  The relationships for members in general are weaker: the only significant relationship that is in the expected direction is for the number of candidates. For every one hundred members a party branch has, it gets an additional candidate on its list. The relationship with the choice to participate is negative in this multivariate analysis. This means that the number of active members is the driver of participation: a branch that has ten active members out of twenty-five members is more likely to enter into elections than a branch that has five active members out of fifty. There is no significant relationship between the number of members and the number of eligible candidates or reported problems.

  Table 4 Regressions

  Model	1	2	3	4
  Dependent variable	Electoral participation	No. of candidates	No. of eligible candidates	Reported problems
  Model type	Logistic	OLS	OLS	Ordered logistic
  No of active members	0.06** (0.02)	0.04* (0.02)	0.01 (0.01)	0.00 (0.01)
  No of members	-0.01** (0.00)	0.01*** (0.00)	-0.00 (0.00)	0.00 (0.00)
  Branch functioning	3.95** (1.07)	4.22*** (2.09)	1.92** (1.00)	2.22*** (0.50)
  Membership development	-0.47* (0.25)	0.18 (0.47)	0.21 (0.23)	0.27** (0.12)
  Population size	-0.01 (0.02)	-0.05* (0.02)	0.03* (0.02)	0.01 (0.01)
  Size of the municipal council	0.01 (0.08)	0.09 (0.14)	-0.04 (0.07)	0.02 (0.03)
  Council workload	0.01 (0.08)	0.25** (0.12)	-0.10 (0.07)	-0.01 (0.03)
  Number of current seats	0.78*** (0.19)	1.15*** (0.24)	1.08*** (0.12)	-0.05 (0.06)
  Participation in municipal executive	1.81*** (0.42)	-0.39 (0.82)	-0.07 (0.40)	-0.22 (0.20)
  Intercept	-4.06 (4.06)	-6.12 (6.20)	5.57 (3.65)	-
  R2	-	0.37	0.31	-
  AIC	242	-	-	1,131
  N	451	396	376	450

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

  Finally, we can look at the direction in which the membership developed: this significantly affects the choice to participate and the reported problems. Compared with a branch that reported a stable membership, a branch that saw its membership decrease had a 37% lower chance of participating in elections. Secretaries of branches that see their membership decline are more likely to observe problems.
  We have put in multiple closely related indicators. This allows us to pinpoint exactly which elements of a well-functioning branch translate into a good performance in the function of candidate selection. The results indicate that the most important predictor of how party branches perform in their recruitment function is how the branch as a whole functions. More than the number of members, their mutual relationships matter. Moreover, the number of active members and the direction of this development are more important than the absolute number of members.
  The second hypothesis was that municipal size would affect the recruitment of candidates. We find a direct effect of population size on the number of eligible candidates. For every 30,000 inhabitants a municipality has, the number of eligible candidates grows by one. We find this effect even when we control for the size of the municipal council. For the actual number of candidates, we do not find a positive relationship with population size; rather population size has a negative effect. The positive effect is picked up by the council workload. The size of the municipal council as such does not have a significant effect on the recruitment function. All in all, we find paradoxical results for the number of candidates. Contrary to our hypothesis, we find that the more inhabitants a municipality has, the lower the number of candidates, and the greater the council workload the more candidates are on candidate lists. These results likely indicate that the relationship between municipality size and the number of candidates is non-linear. The variable for population size is simply the number of inhabitants divided by a thousand. The workload variable is more truncated and roughly follows the square root of the number of inhabitants. Combined, this means that they are likely to capture a non-linear effect of the number of inhabitants: the number of candidates does not increase linearly for every inhabitant, but, rather, as the number of inhabitants grows, the marginal effect of an additional inhabitant decreases.¹⁴x We get a similar R-squared and pattern for the variables when, instead of the council workload and size of the municipal council variables, we include the square root of the population term.^,15x The size of municipal council, the number of inhabitants and workload are correlated. The Pearson’s R for the size of the council and workload is 0.94, for the number of inhabitants and workload 0.92, and for the number of inhabitants and council 0.63. All are significant at the 0.01 level. The Variance Inflation Factors for these variables are also over the benchmark of 4 in all regressions. Yet this is a feature, not a bug: we wanted to get a grasp on which of these is the driving factor between the assumed relationship between municipality size and the length of the list. In the end, the evidence points to the functioning of the branches, which has a much weaker relationship with these factors as well (R is 0.06 for number of inhabitants, 0.07 for municipal council size and 0.09 for workload, the latter two being significant at the 0.1 level).
  Next, we look at the seats in the council a party branch already has. For every seat a party branch has, the chance of their participation in municipal election doubles. Moreover, for every seat they have, a party branch puts an additional candidate on their list. We find an effect of similar strength for the eligible candidates. The current number of seats does not affect branches reporting problems. Currently, being in the local executive makes a party branch five times more likely to participate in local elections but does not affect the other indicators.

• 6 Conclusion

  In this study, we examined four indicators of how local party branches performed their recruitment function: whether party branches choose to participate, the number of candidates they field, the number of eligible candidates and, finally, whether the branches reported problems. Our research was guided by two hypotheses: first, that the stronger the membership base of a party branch is, the better it can perform its recruitment function. The underlying idea is that if the membership base that parties can draw their candidates from is narrow, they will have more problems with recruitment. The evidence showed that the absolute number of members only impacts the absolute number of candidates on the party list. A far more important predictor is how branches functioned. This matters for all our indicators of recruitment. Having a large number of members does not mean party branches are able to perform their function when these members get along. When faced with a hypothetical choice between a large number of rowdy members and a small number of members who do as they are asked, party branches may prefer the latter (cf. Epstein, 1980, p. 258). More than the absolute number of members, we find that the active members are correlated with the choice to participate in the elections and with the number of candidates on party lists. Local lists reflect the organisational capacity of local party branches: the fact that branches have a long list shows that they have many active members. These are members that can also be active in other functions in the party such as the mobilisation of voters. The development of the membership of a branch also affected the choice to participate and the problems these branches experience. Party branches have developed procedures to ensure that they have sufficient candidates: branches with a few members are likely to have more intensive procedures than branches with more members. If their resources decline, these procedures will yield less candidates.
  Our second hypothesis was that in smaller populations local party branches had more difficulty recruiting candidates. The limited evidence does show that when it comes to the number of (eligible) candidates, population size matters. A number of notes are necessary to understand this outcome. First, this effect occurs when controlling for the size of municipal council, so the effect does not just reflect the number of available seats. Moreover, the increased workload in municipal councils does not appear to prevent candidates from filling their lists, even though branches do report this as a major problem. Finally, the relationship with the number of candidates and the number of inhabitants appears to be non-linear.
  So what can we conclude about how local party branches function in the Netherlands?¹⁶x One question is, how do these results relate to earlier studies of recruitment in the Netherlands? In 2006, the number of branches that reported problems was 15%, and in 2018 this was only 8%. There are three possible explanations: the first of these is that party branches may have started to recruit non-party members for office (Grimberg & Vollaard, 2016; Voerman & Boogers, 2014). That is, they may have adapted to the new reality. The second explanation may be that during this period the Netherlands saw a decrease in the number of municipalities as smaller municipalities amalgamated. The effect of municipal amalgamation on the strength of party branches is unknown. On the one hand, the larger number of party members for a lower number of seats may make it easier to recruit candidates. On the other hand, fewer citizens may participate in larger municipalities. The third explanation of the pattern we find may be that between 2006 and 2018 the response rate of the survey declined: from 62% to 30%. It may be the case that the municipalities that suffered problems in recruitment no longer participated in the survey. It may be that the problems of the limited number of volunteers and decreasing time for political activities was such that branch secretaries no longer had the time to answer the survey. Because the response rate of the surveys declined over time, these results are not comparable. All in all, a comparison of the patterns of party recruitment over time may be a valuable avenue for further research. If we look at whether party branches are able to field candidates, we find that most party branches are able to compile a list with a sufficient number of candidates: the average local list covered in our study has twenty-one candidates (with the average municipal council these party branches ran in having only twenty-six members). On average, five of these candidates are considered ‘eligible’ compared with the three seats the party branches currently have. In purely numerical terms party branches are not struggling to recruit candidates. Branches that reported a lot of problems were still able to come up with on average seventeen candidates. The impressions that party secretaries report are likely to be relative to expectations they had. A party branch that wanted to field ten eligible candidates but found only five people fit for office is likely to report great problems; yet a party branch that aimed for four eligible people and found five candidates may report no problems at all.
  Further research may want to examine how local party branches function more generally and how they perform in other functions of political parties, such as the mobilisation of voters, the aggregation and articulation of political interests and the organisation of council politics. The question of how the limited resources they have aid them in these functions is a pressing issue. In particular, researchers may want to study how party branches use their membership base to mobilise voters and whether party branches that have more (active) members are also able to perform better in municipal elections compared with branches with less (active) members. As longer local lists are a ‘symptom’ of healthy branches with more cooperative and active members, students of local party politics may be interested in their impact on election results.
  The results for municipal size were less strong and consistent. Future research may want to examine the relationship between municipal size and party politics in greater detail. While we know that at the national level, there is a negative relationship between the number of inhabitants and how many members parties have and how active they are, we do not know this at the local level (Weldon, 2006). If as municipalities grow, smaller party members become more active, local party branches in smaller municipalities may be inoculated against the negative effect of membership decline.

References

• Andeweg, R. B. & Irwin, G. A. (2005). Governance and politics of the Netherlands. London: Macmillan International Higher Education.

• Bolleyer, N. (2012). New party organization in Western Europe: Of party hierarchies, stratarchies and federations. Party Politics, 18(3), 315-336.

• Bon, M. (2017, October 26). Ook komende vier jaar geen VVD in Laarbeek. Eindhovens Dagblad.

• Boogers, M. J. G. J. A. (2014). Rekrutering en selectie op lokaal niveau. In S. De Lange, M. Leyenaar & P. De Jong (Eds.), Politieke partijen: overbodig of nodig? (pp. 63-70). Amsterdam: Raad voor het Openbaar Bestuur.

• Boogers, M. J. G. J. A. & Voerman, G. (2018). Rekrutering en selectie van kandidaten voor de gemeenteraadsverkiezingen in 2018: ontwikkelingen vanaf 2006 geduid. Enschede/Groningen: Technische Universiteit Twente/Rijksuniversiteit Groningen.

• Boogers, M. J. G. J. A., Van Ostaaijen, J. & Slagter, L. (2010). Lokale kiezers: lokale keuzes? Onderzoek naar de achtergronden en de betekenis van het stemgedrag bij de gemeenteraadsverkiezingen van 2010. Tilburg: Universiteit van Tilburg.

• Broockman, D. E. (2014). Mobilizing candidates: Political actors strategically shape the candidate pool with personal appeals. Journal of Experimental Political Science, 1(2), 104-119.

• Central Bureau of Statistics. (2019). Regionale Kerncijfers Nederland [Regional Key Figures The Netherlands]. Rijswijk: Author.

• Cross, W. & Gauja, A. (2014a). Designing candidate selection methods: Exploring diversity in Australian political parties. Australian Journal of Political Science, 49(1), 22-39.

• Cross, W. & Gauja, A. (2014b). Evolving membership strategies in Australian political parties. Australian Journal of Political Science, 49(4), 611-625.

• Dalton, R. J. & Wattenberg, M. P. (2000). Unthinkable democracy: Political change in advanced industrial democracies. In R. J. Dalton & M. P. Wattenberg (Eds.), Parties without partisans: Political change in advanced industrial democracies (pp. 3-16). Oxford: Oxford University Press.

• Däubler, T. & Hix, S. (2018). Ballot structure, list flexibility and policy representation. Journal of European Public Policy, 25(12), 1798-1816.

• Denters, B., Goldsmith, M., Ladner, A., Mouritzen, P. E. & Rose, L. E. (2014). Size and local democracy. Cheltenham: Edward Elgar.

• Deschouwer, K. (2006). Political parties as multi-level organizations. In R. S. Katz & W. J. Crotty (Eds.), Handbook of party politics (pp. 291-300). Thousand Oaks: SAGE.

• Devos, C., Reynaert, H. & Verlet, D. (2008). Lijstvorming op lokaal vlak: het spannende voorspel van de electorale strijd. In J. Buelens, B. Rihoux & K. Deschouwer (Eds.), Tussen kiezer en hoofdkwartier. De lokale partijafdelingen en de gemeenteraadsverkiezingen van 2006 (pp. 47-72). Brussels: VUB Press.

• De Vries. (2017, September 20). SP in Hoorn mag niet meedoen aan raadsverkiezingen. De Volkskrant.

• De Winter, L., Erzeel, S., Vandeleene, A. & Wauters, W. (2013). Nationale Bemoeienis of Lokale Autonomie? Het Lijstvormingsproces Bij Gemeenteraadsverkiezingen. In K. Deschouwer, T. Verthé & B. Rihoux (Eds.), Op Zoek Naar de Kiezers: Lokale Partijafdelingen En de Gemeenteraadsverkiezingen van Oktober 2012 (pp. 93-121). Brussels: ASP.

• DNPP. (2018). Ledentallen van Politieke Partijen. Groningen: Rijksuniversiteit Groningen. Retrieved from https://dnpp.nl/themas/leden.

• Downs, A. (1957). An economic theory of democracy. New York: Harper.

• Ennser-Jedenastik, L. & Hansen, M. E. (2013). The contingent nature of local party system nationalisation: The case of Austria 1985-2009. Local Government Studies, 39(6), 777-791.

• Epstein, L. (1980). Political parties in Western democracies. New Brunswick: Transaction Publishers.

• Fabre, E. (2008). Party organization in a multi-level system: Party organizational change in Spain and the UK. Regional and Federal Studies, 18(4), 309-329.

• Flos, B.-J., De Jager-De Lange, A., Maatkamp, J. & Salentijn, H. (2017). Nationaal Raadsledenonderzoek. Vijfde trendonderzoek naar de tijdsbesteding en werkzaamheden van raadsleden. Nijmegen: Daadkracht.

• Fowler, L. L. (1993). Candidates, Congress and the American democracy. Ann Arbor: University of Michigan Press.

• Grimberg, M. & Vollaard, H. (2016). De rekrutering en selectie van kandidaat-gemeenteraadsleden. Bestuurswetenschappen, 70(2), 20-42.

• Hazan, R. Y. & Rahat, G. (2010). Democracy within parties: Candidate selection methods and their political consequences. Oxford: Oxford University Press.

• Hippe, J. & Voerman, G. (2010). Reformatorisch Staatkundig Verbond? Over de samenwerking tussen SGP, GPV en RPF (1975-2000). In J. Hippe & G. Voerman (Eds.), Van de Marge naar de Macht. De ChristenUnie 2000-2010. (pp. 51-70). Amsterdam: Boom.

• Kiesraad. (2019). Verkiezingsuitslagen [Election result]. Den Haag: Author. Retrieved from https://verkiezingsuitslagen.nl/.

• Kjaer, U. (2007). The decreasing number of candidates at Danish local elections: Local democracy in crisis. Local Government Studies, 33(2), 195-213.

• Kranendonk, M., Michon, L., Schwarz, H. & Vermeulen, F. (2014). Opkomst en stemgedrag van Amsterdammers met een migratie-achtergrond tijdens de gemeenteraadsverkiezingen van 19 maart 2014. Amsterdam: IMES Report Series.

• Leijenaar, M. & Niemöller, K. (1997). The Netherlands. In P. Norris (Ed.), Passages to power: Legislative recruitment in advanced democracies (pp. 114-136). Cambridge: Cambridge University Press.

• Marcelissen, S. (2018, June 1). SP doet ook in Geertruidenberg niet mee aan gemeenteraadsverkiezingen. BN/De Stem.

• Moon, D. S. & Bratberg, Ø. (2010). Conceptualising the multi-level party: Two complementary approaches. Politics, 30(1), 52-60.

• Müller, J. (2013). On a short leash? Sub-national party positions between regional context and national party unity. Journal of Elections, Public Opinion and Parties, 23(2), 177-199.

• Norris, P. (2006). Recruitment. In R. S. Katz & W. J. Crotty (Eds.), Handbook of party politics (pp. 89-108). Thousand Oaks: SAGE.

• Norris, P. & Lovenduski, J. (1993). ‘If Only More Candidates Came Forward’: Supply-Side Explanations of Candidate Selectionin Britain. British Journal of Political Science 23(3): 373-408

• Norris, P. & Lovenduski, J. (1995). Political recruitment: Gender, race and class in the British Parliament. Cambridge: Cambridge University Press.

• Omroep West. (2017, September 22). SP Rijswijk mag niet meedoen aan gemeenteraadsverkiezingen. Retrieved from https://www.omroepwest.nl/nieuws/3523629/SP-Rijswijk-mag-niet-meedoen-aan-gemeenteraadsverkiezingen.

• Open State Foundation. (2018). 2018 Gemeenteraadsverkiezingen [2018 Municipal elections]. Amsterdam: Open State Foundation. Retrieved from https://data.openstate.eu/dataset/kandidatenlijsten/resource/8945fb2d-73cc-4a09-b443-76a22e93980d.

• Rallings, C. & Trasher, M. (1999). An audit of local democracy in Britain: The evidence from local elections. Parliamentary Affairs, 52, 58-76.

• Rallings, C., Trasher, M., Borisyuk, G. & Shears, M. (2010). Parties, recruitment and modernisation. Evidence from local election candidates. Local Government Studies, 36(3), 361-379.

• RTV Oost. (2017, November 13). D66 doet niet mee aan gemeenteraadsverkiezingen in Tubbergen. RTV Oost. Retrieved from https://www.rtvoost.nl/nieuws/279845/D66-doet-niet-mee-aan-gemeenteraadsverkiezingen-in-Tubbergen.

• Scarrow, S. E. & Gezgor, B. (2010). Declining memberships, changing members? European political party members in a new era. Party Politics, 16(6), 823-843.

• Straub, M. (2017, November 22). De PvdA doet niet meer mee in Hardinxveld-Giessendam. Algemeen Dagblad.

• Suijten, L. (2019). Media zichtbaarheid en voorkeurstemmen. De invloed van media zichtbaarheid op voorkeurstemmen bij de 2018 gemeenteraadsverkiezingen. BA-Scriptie. Leiden: Universiteit Leiden.

• Van Biezen, I., Mair, P. & Poguntke, T. (2012). Going, going,…… gone? Party membership in the 21st century. European Journal of Political Research, 51(1), 24-56.

• Van der Kolk, H. (2007). Local electoral systems in Western Europe. Local Government Studies, 33(2), 159-180.

• Van der Noordaa, R. (2017, April 29). Selectie gemeenteraadsleden PVV in volle gang. De Volkskrant.

• Van Holsteyn, J. & Nagtzaam, M. (2018, March 20). Stem op een echte kandidaat, niet op een lijstduwer. NRC Handelsblad.

• Van Houwelingen, P. (2017). Political participation and municipal population size: A meta-study. Local Government Studies, 43(3), 408-428.

• Veenendaal, W. (2014). Politics and democracy in microstates. Abingdon-on-Thames: Routledge.

• Voerman, G. (2000, January 13). Ledenverlies plaatst politieke partijen voor groeiend rekruteringsprobleem. Nederlandsche Staatscourant.

• Voerman, G. (2010). Communisten, pacifistisch-socialisten, radicalen en progressieve christenen. De voorlopers van Groen Links. In P. Lucardie & G. Voerman (Eds.), Van de straat naar de staat? GroenLinks, 1990-2010. (pp. 15-36). Amsterdam: Boom.

• Voerman, G. & Boogers, M. J. G. J. A. (2008). Het rekruteringsvermogen van de politieke partijen bij de gemeenteraadsverkiezingen van 2006. Een onderzoeksverslag. In G. Voerman (Ed.), Jaarboek Documentatiecentrum Nederlandse Politieke Partijen 2006. (pp. 175-194). Groningen: Rijksuniversiteit Groningen.

• Voerman, G. & Boogers, M. J. G. J. A. (2011). Rekrutering revisited: de kandidaatstelling voor de gemeenteraadsverkiezingen van 2010 vergeleken met 2006. Groningen/Enschede: Rijksuniversiteit Groningen/Technische Universiteit Twente.

• Voerman, G. & Boogers, M. J. G. J. A. (2014). Rekrutering van kandidaten voor de gemeenteraadsverkiezingen in 2006, 2010 en 2014: Kandidaatstellingsproblemen vergeleken en verklaard. Groningen/Enschede: Rijksuniversiteit Groningen/Technische Universiteit Twente.

• Vollaard, H., Boogaard, G., van den Berg, J. & Cohen, J. (2018). De Gemeenteraad. Ontstaan en ontwikkeling van de lokale democratie. Amsterdam: Boom.

• Weldon, S. (2006). Downsize my polity? The impact of size on party membership and member activism. Party Politics, 12(4), 467-481.

Noten

• 1 As well as Grimberg and Vollaard’s (2016) survey of the municipality Voorschoten.

• 2 Straub (2017), Bon (2017) and RTV Oost (2017).

• 3 E.g. Straub (2017), Bon (2017) and RTV Oost (2017).

• 4 The limit is 50 candidates for political parties that have less than 15 seats in the previous legislative term; parties with more seats can field up to 80 candidates. So in a legal sense the range is 1-80, but empirically it is 1-50. In 2018, no party had more than 50 candidates.

• 5 We thank the anonymous reviewer for providing us with this contextual information.

• 6 Kjaer (2007) suggests cultural reasons for this: “[P]utting up a list with the maximum number of candidates … may appear rather optimistic, not to say pompous, if the list cannot be expected to get more than one of a few of these candidates into office.”

• 7 E.g. the levenslied-singer André Hazes in De Ronde Venen.

• 8 The number of candidates elected merely on the basis of preference votes is low. Most preference votes are cast for the first candidates on the list. Those candidates would often be elected even without those preference votes (Boogers, Van Ostaaijen & Slagter, 2010, p. 44). The recipients of preference votes tend to be the first woman on the list, sitting members of local councils and candidates with a migration background (Boogers et al., 2010; Kranendonk, Michon, Schwarz & Vermeulen, 2014; Suijten, 2019).

• 9 One factor that we are not including, but which is sometimes mentioned in the Belgian literature, is party ideology (Devos et al., 2008, p. 67; De Winter et al., 2013, p. 117). These studies show that parties with different ideologies focus on other aspects when recruiting (e.g. left-wing parties focusing on women’s representation) and run into different problems (e.g. right-wing parties experiencing difficulties with women). They do not indicate that Belgian parties of the left or right experience greater difficulty.

• 10 It is eighty for larger parties, but this requirement is irrelevant here: no party in the survey had more than 15 seats.

• 11 The Socialist Party, which did not participate in this survey, in particular uses this veto power when it believes that local party branches lack the necessary integration in local communities (De Vries, 2017; Marcelissen, 2018; Omroep West, 2017).

• 12 These are often called a ‘kandidatencommissie’ (candidates committee), but referred to as ‘ad hoc comité’ (ad hoc committee), ‘comité van wijzen’ (committee of wise people) of ‘ontwerpkandidatenlijstwerkgroep’ (working group for the concept candidate list) in Belgium (De Winter et al., 2013, p. 94).

• 13 An exception here is Party for Freedom (which also did not participate in this survey), where provincial representatives of the party selected the candidates for municipal elections (Van der Noordaa, 2017).

• 14 We get a similar R-squared and pattern for the variables when, instead of the council workload and size of the municipal council variables, we include the square root of the population term.

• 15 The size of municipal council, the number of inhabitants and workload are correlated. The Pearson’s R for the size of the council and workload is 0.94, for the number of inhabitants and workload 0.92, and for the number of inhabitants and council 0.63. All are significant at the 0.01 level. The Variance Inflation Factors for these variables are also over the benchmark of 4 in all regressions. Yet this is a feature, not a bug: we wanted to get a grasp on which of these is the driving factor between the assumed relationship between municipality size and the length of the list. In the end, the evidence points to the functioning of the branches, which has a much weaker relationship with these factors as well (R is 0.06 for number of inhabitants, 0.07 for municipal council size and 0.09 for workload, the latter two being significant at the 0.1 level).

• 16 One question is, how do these results relate to earlier studies of recruitment in the Netherlands? In 2006, the number of branches that reported problems was 15%, and in 2018 this was only 8%. There are three possible explanations: the first of these is that party branches may have started to recruit non-party members for office (Grimberg & Vollaard, 2016; Voerman & Boogers, 2014). That is, they may have adapted to the new reality. The second explanation may be that during this period the Netherlands saw a decrease in the number of municipalities as smaller municipalities amalgamated. The effect of municipal amalgamation on the strength of party branches is unknown. On the one hand, the larger number of party members for a lower number of seats may make it easier to recruit candidates. On the other hand, fewer citizens may participate in larger municipalities. The third explanation of the pattern we find may be that between 2006 and 2018 the response rate of the survey declined: from 62% to 30%. It may be the case that the municipalities that suffered problems in recruitment no longer participated in the survey. It may be that the problems of the limited number of volunteers and decreasing time for political activities was such that branch secretaries no longer had the time to answer the survey. Because the response rate of the surveys declined over time, these results are not comparable. All in all, a comparison of the patterns of party recruitment over time may be a valuable avenue for further research.