24
The Netherlands: The
Sanctity of Proportionality
Rudy B. Andeweg
THE INSTITUTIONAL CONTEXT
The Dutch political system does little to concentrate political power in the hands
of a democratically legitimated majority (for an overview, see Andeweg and
Irwin 2005). Lijphart (1999), for example, uses ten institutional criteria to classify
political systems as more majoritarian democracies or more consensus-oriented
democracies on two dimensions, and on both dimensions the Netherlands is close
to the consensus end of the scale. With regard to the executives–parties dimension,
the Dutch political system is characterized by a high number of relevant political
parties, most of which originated in one of the social segments, known as ‘pillars’,
that were divided by deep social cleavages of class and religion. Hence, these parties
can be classified into three different political ‘families’. Within each of these families
the main party (the Labour Party (PvdA) for the socialist left, the Christian Democratic
Appeal (CDA) for the Christian democrats, and the conservative Liberals
(VVD) for the liberals) face competition from smaller ideologically more radical or
populist parties. None of these parties is close to obtaining a parliamentary majority,
and the country is always governed by coalitions and relatively often by ‘oversized’
coalitions that include more parties than numerically necessary to obtain a parliamentary
majority.
Although there are elements of separation of powers in the Dutch system (government
ministers are not allowed to be members of parliament, for example), the
government is dependent on the confidence of a parliamentary majority in order to
survive. Several mechanisms (a detailed coalition programme, an elaborate system
of consultation and coordination among the governing parties, and party discipline)
result in a high degree of governmental stability. Still, opposition parties are not
entirely without influence, especially through relatively powerful and specialized
parliamentary committees. There is a tradition of consulting interest groups and
preferably incorporating them into policy-making through formal and informal
corporatist arrangements.
On Lijphart’s federal–unitary dimension, the Netherlands has a bicameral parliament.
An electoral council composed of all members of the country’s twelve
provincial legislatures elects the upper house, confusingly referred to as the Eerste
Kamer (First Chamber). It consists of part-time politicians, meets less frequently
than the Tweede Kamer (Second Chamber), and lacks the power of amendment, but
it has an absolute veto over all legislation. Procedural and numerical hurdles must be
overcome to change the constitution, which serves to protect minority rights. The
central bank is independent (even more so now that the most important of its
functions have been taken over by the European Central Bank). In a consensus
democracy the courts are expected to review the constitutionality of statutes. Although,
formally, such judicial review does not exist in the Netherlands, in practice
this does little to constrain the growing political role of the judiciary. Lijphart even
classifies the Netherlands as a semi-federal country (Lijphart 1999: 191). He does so
not because of a high degree of territorial decentralization in the Dutch political
system; despite the country’s plural name in many languages (Netherlands, Pays
Bas, Niederla¨nde) and its origins as a confederacy, today it is hard to find a more
centralized political system. Instead, Lijphart bases his classification on the considerable
degree of delegation to private associations (‘sociological federalism’). This
would seem to confound federalism with corporatism, but even if we disregard this
controversial classification, the Netherlands remains very much a consensus democracy.
Of the thirty-six countries that are included in Lijphart’s analysis, the
Netherlands is the fifth most consensus-oriented polity on the executives–parties
dimension, and it takes ninth place on the federal–unitary dimension.
Not all of these characteristics are enshrined in the constitution, but those that are
date back at least to 1848. Since that year, the Dutch political system has undergone
very little change. There is one important exception to this picture of stability,
however, and that is the electoral system that is the focus of this chapter.
ORIGINS OF THE ELECTORAL SYSTEM
The electoral system was changed radically from a majoritarian system to proportional
representation in 1917. Before 1917, members of parliament (MPs) in the
Second Chamber of parliament were elected by absolute majority from electoral
districts. If no candidate obtained an absolute majority, a second round would be
held in which the choice was restricted to the two candidates who received most
votes in the first round. Originally, most districts elected two MPs, but it was unusual
for them to be elected simultaneously, as every two years half of the MPs were
elected. The number of MPs was linked to the size of the population (one MP for
every 45,000 inhabitants). In 1888, the number of MPs was fixed at 100, to be elected
simultaneously. In 1896 all districts became single-member districts (SMDs).
This system was abolished in what is known in Dutch history books as the Great
Pacification of 1917 (Loots 2004). In the years preceding the First World War,
intense political conflicts along the lines of the social cleavages mentioned above
divided the country, primarily over the issues of universal suffrage and public
funding of religious schools. At the beginning of the First World War, an attempt
was made to resolve these issues by setting up two committees composed of the
492 The Politics of Electoral Systems
parliamentary leaders of the main parties. It would take until 1917 for a package deal
to be worked out. In this compromise the religious parties received full state
financing of religious schools, and the Labour party got the universal (male) suffrage
that it had campaigned for. Although not all Liberals rejected a widening of the
suffrage, most had opposed both elements of the package deal. Moreover, the Liberal
parties represented more affluent voters. The abolition of the re´gime censitaire
without a change in the electoral system would most likely have wiped out the
Liberal parties: the enfranchisement of less wealthy voters would have reduced the
Liberals to a minority in most if not all SMDs. To win their consent to the package
deal, a third element was added: a replacement of the electoral system of absolute
majority in SMDs by nationwide proportional representation. Under that system, the
Liberal minorities in various regions could be aggregated in order to obtain a share of
the seats in the Second Chamber. In order not to endanger this fragile compromise,
the parties agreed not to contest the last elections that had to be held under the old
system in order to make the necessary constitutional amendments!
Apart from widening the suffrage to female citizens only two years later, in 1919,
the most important change to the electoral system was the abolition of compulsory
attendance at the polls (customarily, if incorrectly, referred to as compulsory voting)
in 1970. Compulsory voting was introduced in 1917 to guarantee that proportional
representation (PR) of all segments in society would not be affected by low turnout.
This aspect of the electoral system never obtained the legitimacy that the other
aspects would soon acquire, and several attempts were made to abolish it. Nevertheless
it was not until 1970 that the eighth abolition attempt was successful.
Paradoxically, protection of existing political parties was an important motivation
for both the introduction and the abolition of compulsory voting. In 1967, the
established political parties had been shocked to see a right-wing populist party
(the Farmers’ Party) and a centre-left reformist party (Democrats ’66) together take
fourteen seats, a landslide in those days of still unquestioned party loyalty. Although
this was not admitted openly, the established parties calculated that the voters of
these parties were ‘protest voters’, most of whom would have abstained had it not
been for compulsory voting. By abolishing compulsory voting, it was argued, the
risk of such electoral upsets would be reduced. When turnout decline in more recent
years prompted proposals for the reintroduction of compulsory voting, some opponents
again argued that forcing alienated citizens to come to the polls would only
benefit the extreme right.
HOW THE ELECTORAL SYSTEM WORKS
Voting is made exceedingly simple in the Netherlands. Registration is automatic
(through the population register) and, some weeks before election day, voters
receive a list of candidates and a card informing them of the opening hours and
location of their polling station (the procedure for Dutch citizens residing abroad is
more complicated). The ballot structure (or rather the display on the voting machines
that have replaced paper ballots in all but a few municipalities) and the
The Netherlands 493
electoral system in general is virtually identical for the elections to the Second
Chamber of the national parliament, the provincial legislatures, the local councils,
and the Dutch delegation to the European Parliament (the description that follows
focuses on the Second Chamber and ignores the small differences regarding the
other elections). A ballot paper—or, to be precise, part of a ballot paper—from the
days of manual voting is shown in Figure 24.1.
Yet, this simple electoral system is also misleading. Voters can only express a
choice for an individual candidate, but the electoral system treats their vote primarily
as a choice for a particular party. Voters vote in one of nineteen electoral districts,
and the candidates or even the parties on the ballot are rarely identical in all districts,
but, as we shall see in a moment, the electoral system does not result in any form of
geographical representation.
Political parties present lists of candidates in one, several, or all of the nineteen
districts. The lists of parties already represented in parliament are put on the ballot in
the order of their support in the previous elections. They are followed by the lists of
new parties, ordered according to the number of electoral districts in which they
participate. In case of a tie, lot determines the order. Today, parties can register a
name that is then printed on top of the party’s list. Before 1956 this was not possible,
which explains why Dutch election posters prominently displayed the list number
under which a party’s candidates were to be found on the ballot that year, a tradition
that continued until quite recently.
There is no provision on the ballot for casting a vote for the party as such. Each
voter has one vote, which is cast for one of the candidates on one of the lists by
marking the white dot next to the candidate’s name with a red pencil, or by pushing
the button for that candidate. Voters who have a preference for a party but not for
any particular candidate usually cast their vote for the first candidate on the list,
the so-called ‘list-puller’ (lijsttrekker). Customarily, this position is reserved for the
party leader. All votes for other candidates lower on the list are known as preference
votes (voorkeurstemmen), as they indicate a preference for a particular candidate
over all other candidates on the list. Such preference votes, however, have
little impact on the original ranking by the party. The votes that are cast for all
candidates on a list are aggregated, and this total number of votes determines a
party’s total number of seats. These seats are allocated to the candidates in the order
in which they appear on the list. The only exception is that any candidate who has
received at least 25 per cent of the Hare quota will be elected regardless of his or her
position on the list, provided the party has won sufficient seats. It is relatively rare
for a candidate to be elected in defiance of the party’s ranking: since the threshold
for preference votes to take effect was lowered to this 25 per cent of the electoral
quota, two out of the 150 MPs would not have been elected in 1998 had it not been
for preference votes, in 2002 only one candidate was thus elected, and in 2003 two.
Obviously, political parties do not encourage preference voting and in the past some
parties even made their candidates sign an undated letter of resignation, or a pledge
not to accept election by preference votes without the approval of the party executive.
Thus, votes cast for persons are effectively counted as votes for parties.
494 The Politics of Electoral Systems
Figure 24.1 Netherlands ballot paper, 1989
Note: This is just a part (approximately a quarter) of the ballot paper facing voters in The Hague in 1989.
The full ballot paper contained 23 party lists and the names of 499 candidates. Since then paper ballots
have been replaced by an electronic interface, which looks very similar.
The Netherlands 495
As mentioned above, voters cast their votes in one of nineteen electoral districts.
Although most parties contest the elections in all districts, there are usually a few
parties participating in only one or in a few districts. Even the lists of the parties that
take part in all districts usually vary from one district to another. Until the 1960s,
some parties had different list-pullers in different regional clusters of electoral
districts to facilitate campaigning in those days of canvassing, local rallies, and
personal contact. With the arrival of television, campaigns became increasingly
nationalized, and all parties now present lists with the same list-puller, and often
with the same top three to five candidates, in all districts. It makes sense, however, to
have variation lower down on the district lists: the Second Chamber of parliament
has 150 seats, but the ballot papers or voting machines cannot accommodate
between fifteen and twenty-five party lists with 150 or more candidates each. For
that reason, the Electoral Law only allows a party to put no more candidates on one
of its district lists than thirty or (for incumbent parties) twice its current number of
seats in the Second Chamber, with a maximum of eighty candidates. By varying the
composition of at least part of its lists across electoral districts, a party can field more
candidates nationwide than if it were to submit an identical list in each district.
Voters thus face a different ballot from one district to another. The potential this has
for geographical representation, however, is thwarted by the use that political parties
make of other options in the Electoral Law.
The number of seats that is to be allocated to a district list is based on the number
of times that list has obtained the national electoral quota (the Hare quota). Usually
not all seats are allotted on the basis of multiples of the electoral quota. The
‘remainder seats’ (restzetels) are allocated to the district lists on the basis of the
system of highest averages (D’Hondt). If a party presents lists in all nineteen
electoral districts, the electoral system does not treat it as a single party in the
nationwide allocation of seats, but as nineteen separate parties. However, the
Electoral Law does allow a party to combine its district lists so that all its district
votes are aggregated into a national vote. Parties that participate in more than one
district always make use of this option because the D’Hondt formula is slightly more
advantageous to larger parties (see Appendix A). By combining its district lists, the
party thus increases its chances of winning one or more of the remainder seats.
As an aside it is worth mentioning that two or more parties can even decide to
combine their respective lists (apparentement). The existence of such an alliance is
mentioned on the ballot paper and serves the same purpose as pooling district lists
within a party: obtaining remainder seats. Curiously, when the seats that have been
allocated to such a party alliance are distributed to the various parties within the
alliance, the system of largest remainders (LR–Hare) instead of highest averages is
used for allotting any remainder seats. This system does not advantage larger parties,
which is probably the reason why it is exceptional for one of the bigger parties to
join such an alliance. Instead, it is usual to see alliances of small parties on the left
(Green Left and Socialist Party) and the religious right (Christian Union and
Political Reformed Party (SGP)). In neither the 2002 nor the 2003 elections did
these alliances affect the distribution of seats.
496 The Politics of Electoral Systems
If we return to the pooling of district lists within a party, it is clear that this
practice limits geographical representation. Moreover, parties do not seem to make
widespread use of the possibility to field regionally popular candidates in a particular
district (and preference votes appear to be motivated more by a candidate’s
gender than place of residence, although the latter is mentioned on the ballot). As a
result, the nineteen electoral districts primarily serve administrative purposes
(the organization of the elections, the coordination of the counting procedures,
etc.). Effectively, the electoral system treats the whole country as a single 150member
district.
Although no legal threshold exists for the other elections, there is one for
the elections to the Second Chamber: a party that falls below the electoral quota
is also excluded from the distribution of remainder seats (even if the party is
part of an apparentement that does cross the threshold). By linking the legal
threshold to the electoral quota, it is determined by district magnitude, and by
treating the whole country as a single district for the distribution of seats to parties,
assembly size becomes the crucial factor. For a long time the electoral quota/
threshold was 1 per cent, but with the enlargement of the Second Chamber from
100 to 150 in 1956 it automatically dropped to 100/150, or 0.67 per cent of the vote,
which explains why the Dutch electoral system is often described as an extremely
open one.
Using the 2003 elections to the Second Chamber as an example, Table 24.1 shows
the procedure for allocating seats to parties and candidates in detail. Under this
system, by-elections are not necessary. When a seat in parliament becomes vacant
because an incumbent MP dies, resigns, or is appointed to the cabinet, the next
candidate on the district list from which the ex-MP was elected is declared elected.
POLITICAL CONSEQUENCES OF
THE ELECTORAL SYSTEM
Impact on the party system
‘Duverger’s law’ states that systems of PR have a tendance multiplicatrice, that is,
that they lead to a multiplication of the number of parties (Duverger 1951). Although
the electoral system of the Netherlands described above is an extremely proportional
one with a very low electoral threshold, it should be emphasized that it was
introduced to protect a multiparty system that had already established itself under
a system of absolute majority in SMDs: in 1913, the last elections that were actually
contested under the old electoral system, seven political parties won seats; in 2003,
the latest elections under the current electoral system, nine parties won seats. In the
first elections under PR the number of legislative parties did go up to 17, but since
then the average number of parties in parliament is 10.7. The effective number of
parties (Laakso and Taagepera 1979) fluctuates between three and six, with an
average of 4.7; in 2003 the effective number of parties was five. The Dutch case
seems to confirm Lijphart’s finding that proportional representation is correlated
The Netherlands 497
Table 24.1 Distributing parliamentary seats: the 2003 Dutch election
Number of valid votes:
Electoral quota/threshold:
Number of competing parties:
9,654,475 (79.9% of the electorate)
64,363.17 (number of valid votes divided by 150)
19
Parties Votes % vote Seats % seats
CDA 2,763,480 28.6 44 29.3
PvdA 2,631,363 27.3 42 28.0
VVD 1,728,707 17.9 28 18.7
SP/GL 1,105,525 11.5 17 11.3
[SP 609,723 6.3 9 6.0]
[GL 495,802 5.1 8 5.3]
LPF 549,975 5.7 8 5.3
D66 393,333 4.1 6 4.0
CU/SGP 354,999 3.7 5 3.3
[CU 204,694 2.1 3 2.0]
[SGP 150,305 1.6 2 1.3]
PvdD 47,754 0.5 — —
Others 79,339 0.7 — —
(Party alliances italicized)
Distributing seats to parties/alliances
Step 1 The electoral quota is determined by dividing the total number of valid votes by the 150 seats in
the Second Chamber (Hare quota). In 2003 the electoral quota was 64,363.17. This is the electoral
threshold. The PvdD (Party for the Animals), as well as nine even smaller parties, receive no seats.
Step 2 Seats to parties and alliances that did cross the electoral threshold are distributed according to the
D’Hondt system.
Distributing seats to parties within an alliance
Step 3 For those parties that have linked their lists in an alliance (in this case SP and Green Left, and
Christian Union and SGP), the seats that have been allocated to their alliance are distributed among them.
An intraalliance electoral quota is calculated based on the number of votes cast for the alliance and the
number of seats just allocated to the alliance (for SP/GL, for example, it was 65,030.88). Each party in an
alliance receives a seat for each multiple of the intra-alliance electoral quota (in the SP/GL alliance, nine
seats go to SP and seven to Green Left, with one remainder seat). Any remainder seat is allocated to the
party with the largest remaining number of votes after subtracting its multiple of the electoral quota from
its number of votes (in the case of the SP/GL alliance, the remainder seat goes to Green Left).
Distributing seats to district lists within a party
Step 4 For all parties that have linked their district lists (all parties in the 2003 case), the seats that have
been allocated to them are distributed among their district lists. The procedure is similar to the one
described above for allocating seats within party alliances, each district list being treated as a party within
the national party.
Distributing seats to candidates on a district list
Step 5 Now that we know the number of seats that a party has won in a particular district, candidates can
be assigned to those seats. As long as their party has won enough seats, candidates who have received
more than 25 per cent of the national electoral quotient (16,091 votes in 2003) nationwide are declared
elected regardless of their position on the list in the district in which they received most votes (twentyseven
candidates, or 18 per cent of all MPs, but twenty-five of them would also have been elected because
of their position on the list).
Step 6 The party’s remaining seats in a district are assigned to the remaining candidates on the district
list, in the order of that list. Candidates who win a seat in more than one district are declared elected in the
district in which they received most votes.
Note: Based on Andeweg and Irwin 2005: Box 4.1.
498 The Politics of Electoral Systems
with higher numbers of parliamentary parties, but that it is not a very strong
correlation (Lijphart 1994).
The electoral system affects the number and the size of the parties in the Second
Chamber. Even the low legal threshold acts as a barrier. Immediately after the first
elections under PR, the threshold was raised from 0.5 per cent to 0.75 per cent of the
electoral quota and in the next elections the number of parties winning seats dropped
from seventeen to ten (in 1933 it was further raised to equal the Hare quota). New
political parties that seek representation face more barriers than just this low
electoral threshold:
. party labels of new parties that resemble those of existing parties are not permit-
ted;
. district lists of new parties must be supported by at least thirty voters in the district
who have to declare their support in person at the town hall of their municipality.
No such requirement exists for incumbent parties;
. new parties have to pay a deposit of e11,250 if they have combined their district
lists (if not, the same deposit is required for each district list). The deposit is
forfeited if the party obtains less than 75 per cent of the electoral quota (in 2003
nine parties lost their deposit). No deposit is required from incumbent parties (in
2003 the ‘Livable Netherlands’ party obtained only 60 per cent of the electoral
quota, but as it was an incumbent party at the time, it had not paid a deposit);
. new parties do not receive state subventions whereas incumbent parties have
received (modest) financial support during the preceding parliament. One form
of state support (broadcasting time) does not differentiate between new and
incumbent parties during the campaign: parties competing in all nineteen districts
are given six blocks of three minutes on public television and a total of twenty
minutes on public radio. They are also equally provided with funds to cover some
of the production costs;
. new parties are allowed no more than thirty candidates on a list whereas incumbent
parties may put forward thirty candidates or twice the number of seats
currently held with a maximum of eighty candidates on a list;
. the lists of incumbent parties (and especially of the larger ones) are given the most
prominent positions on the ballot;
. in the absence of compulsory voting, incumbent parties, which have already
developed a loyal following, benefit from higher turnout of their supporters
compared to new parties.
Despite such differential treatment, however, the fact remains that the Dutch system
is extremely open: one or more new parties entered the Second Chamber in eleven
out of the seventeen elections between 1948 and 2003.
The electoral system also affects the relative strength of the parties in parliament.
The Dutch electoral system is no exception to Rae’s assertion that ‘Like the Sheriff
of Nottingham, electoral systems are apt to steal from the poor and give to the rich’
(Rae 1971: 86; also known as the ‘Matthew effect’ after the Gospel according to
Matthew 13: 12), but the distortion of proportional representation in favour of bigger
The Netherlands 499
parties is limited, as a comparison of the percentages of the vote and the percentages
of the seats (Table 24.1) for 2003 illustrates: the largest deviation from perfect
proportionality in that year was 0.8 per cent for the conservative Liberal VVD. In the
2003 elections that serve as the main example in this chapter, Gallagher’s least
squares (LSq) index of disproportionality was 1.19 (including all participating
parties). For the eleven Dutch parliamentary elections between 1956 and 1989,
Lijphart calculated the LSq index to average 1.32 (Lijphart 1994: 162, excluding
some of the smaller parties that did not win any seats). These figures are quite low
and, compared to other countries, the disproportionality of the Dutch electoral
system is less than almost anywhere else, or—in the technical language of Taagepera
and Shugart (1989: 196)—‘for the huge magnitude of the Netherlands’ countrywide
district, the calculated curve is practically indistinguishable from the ideal PR
line’. What little disproportionality exists is accounted for entirely by the allocation
of remainder seats: in 2003 all over-represented parties were recipients of remainder
seats (CDA, PvdA, VVD) or of a remainder seat within an alliance (Green Left), and
all underrepresented parties did not obtain remainder seats. Historically, there have
been two opposing trends: the shift from LR–Hare to the D’Hondt formula in 1933
has increased disproportionality, and the later enlargement of the Second Chamber
from 100 to 150 seats in 1956 has decreased it.
In combination, the two features (few barriers to new parties, high degree of
proportionality) result in a high sensitivity of the composition of the Second
Chamber to shifts in the electorate. On average, 12 per cent of the seats changed
hands in each of the seventeen elections between 1948 and 2003, but as Figure 24.2
shows, there has been considerable variation. From 1948 to 1963 (the heyday of
voters’ loyalty to their social segment, or ‘pillarization’), 5 per cent of the seats went
from one party to another in an average election. The onset of depillarization pushed
the average proportion of seats changing hands up to more than 11 per cent. After a
curious drop in 1989, an average of more than one-fifth of the seats changed parties
in subsequent elections. Note that this increase is not caused by changes in the
electoral system but by higher electoral volatility.
0
5
10
15
20
25
30
35
48 56 63 71 77 82 89 98 '03
Election years
Figure 24.2 Percentage of seats changing hands at elections
500 The Politics of Electoral Systems
Impact on the parties
Under the Dutch electoral system, individuals have a choice of two strategies for
getting elected to parliament: they can either convince the party selectorate to give
them an eligible (verkiesbare, i.e. high) position on the list, or they can convince
party voters to give a preference vote to them rather than to any of their party’s other
candidates. The first strategy is more likely to be successful: preference votes have
limited impact, and for voters to be able to cast a preference vote for you, you still
depend on the party selectorate putting you on the list in the first place. Campaigning
for preference votes does occur occasionally, but it is at a relatively modest level
(small ads in local newspapers, displaying posters, and distributing flyers). Parties
sometimes recruit candidates whom they expect will attract large numbers of
preference votes from people who would not normally vote for the party, but this
tends to be a mixed blessing when such candidates demand a special position in
return. In 1998 the CDA put the hitherto non-partisan leader of the Dutch branch of
Me´decins sans Frontie`res on its list to attract additional votes, but after the election,
he often ignored party discipline and eventually left the CDA. In 2002, the VVD
recruited a prominent anti-Islam campaigner who had received death threats while
she worked for the PvdA, but after her election several conflicts occurred within the
parliamentary party when she publicly took positions without consulting the parliamentary
party. The party selectorates therefore do not necessarily regard the ability
to attract numerous preferences as an unqualified asset in an aspiring candidate.
Thus, whatever intraparty conflict occurs tends to take place at the nomination stage,
rather than during the election campaign.
Although the parties are autonomous in the ways in which they draw up their lists
of candidates, the electoral system does have an indirect effect: the internal organization
of the major parties is a reflection of the way in which the electoral system
operates. Most parties have, for example, organizations at the level of the electoral
district. The role of these district organizations in the nomination process varies by
party and across time. In general, the parties seek to present a balanced list in terms
of candidates’ background (gender, region) and policy expertise. This goal may
conflict with the democratic requirement of giving party members influence in the
nomination process. Responding to calls for democratization in the 1960s, the major
parties used a decentralized procedure in which the organizations at the district level
played an important role during the 1970s and 1980s (Koole and Leijenaar 1988). In
practice, decentralization reinforced the position of regional ‘party barons’ rather
than produce high levels of rank-and-file involvement in the nomination process
(Hillebrand 1992). Dissatisfaction with this result, together with calls for more
female and young candidates, and for placing more emphasis on professional
expertise and less on rewarding party service, led to a return to more centralized
procedures in the major parties (Leijenaar and Niemo¨ller 1997). This procedure
usually entails the creation of a selection committee by the party’s national executive
board. This committee invites nominations from party members (sometimes
The Netherlands 501
even from non-partisans who are willing to join the party, by putting an advertisement
in the national newspapers) and proposes a rank ordering of candidates to the
executive board. Local branches may be consulted, but eventually the executive
board proposes a rank ordering to a national party congress for formal endorsement.
For a long time the progressive liberal D66 formed an exception, by giving its
members the opportunity to determine the ranking of the candidates by postal ballot.
However, in 2002 the PvdA started to ballot its members on the designation of its
list-puller, and other parties have announced similar plans.
Impact on parliament
The electoral system (and in particular the party lists and the limited impact of
preference votes) allows the political parties to control the composition of their
parliamentary party through the rank ordering of the candidates. In the past, the
religious parties, in particular, sought to widen their appeal by reserving so-called
‘quality seats’ for representatives of affiliated interest associations. Together with a
system of specialized parliamentary committees, this led to a tradition of MPs
specializing in narrowly defined policy fields. However, representativeness with
regard to gender, age, and so on (microcosmic or descriptive representation) is
increasingly drawing attention.
Compared to countries with plurality SMDs, the Dutch Second Chamber may be
quite representative, but the comparison with a number of Nordic parliaments in
Table 24.2 shows that some other countries with PR systems do better with respect
to descriptive representation. The fact that the periphery is considerably less well
represented in the Netherlands than in the Nordic countries may be explained by the
fact that the Dutch electoral system does not give more impact to its electoral
districts, whereas the districts are important in the Nordic countries. However,
Table 24.2 The representativeness of the Dutch parliament in comparative perspective
Population category The Netherlands Denmark Iceland Norway Sweden
Women 69.8 70.2 50.0 76.0 78.8
Periphery 74.7 100.1 148.0 104.8 99.7
Young (18–35) 25.3 21.9 8.6 28.9 22.2
Elderly (55þ) 75.7 103.4 85.7 112.5 116.7
Blue-collar workers 7.2 10.5 0.0 30.3 40.0
Private sector 58.5 n.a. 77.3 86.9 77.4
Note: Entries are the proportion of MPs belonging to a particular population category as a percentage of
the proportion of the population/electorate belonging to the same category. 100 per cent would indicate
perfect representativeness, lower percentages stand for under-representation, and higher percentages for
over-representation.
Periphery is defined as other-than-Randstad provinces (The Netherlands), other than the greater
Copenhagen area (Denmark), the Districts (Iceland), Mid-Norway and the North (Norway), Go¨taland
and Norrland (Sweden).
Source for the Netherlands: Dutch Parliamentary Study 2001 (Leiden and Twente Universities).
Sources for the Nordic countries: Narud and Valen 2000: 88 and Valen, Narud and Hardarson 2000: 116.
502 The Politics of Electoral Systems
nothing in the electoral system leads us to expect the more pronounced underrepresentation
of the elderly, of blue-collar workers, or of the private sector in the
Netherlands. Apparently, the electoral system allows the parties to strive towards a
socially or even geographically representative parliament, but it does not force them
to do so. Under very similar systems, Nordic parties put a greater effort into
descriptive representation than Dutch parties.
The combination of the electoral system and the highly centralized nature of the
Dutch polity generally also leads us to expect that a high saliency of party and a low
saliency of the districts are the most important consequences for legislative behaviour.
This is borne out by the data presented in Table 24.3: compared to the Nordic
countries where the political system does emphasize geographical representation,
representing one’s own region is seen as least important by Dutch MPs, whereas
representing one’s political party is seen as most important. Party cohesion is strong
in the Netherlands, to such an extent that parliamentary votes are recorded by party
rather than by individual MP, unless a roll-call is specifically requested. Parties
clearly dominate life in parliament, although it should be noted that party discipline
seems more relaxed for MPs in their role of specialist in a particular policy area, or
for MPs participating in parliamentary inquiries (Andeweg 2000).
Another consequence of the absence of effective electoral districts is that there is
no incumbency effect: there are no strong constituency feelings to be taken into
account by the party selectorates when they decide whether or not to reselect a
sitting MP. Turnover is therefore relatively high: in the seventeen elections between
1948 and 2003, on average 23 per cent of all elected MPs had not served in
parliament before (Secker 2000: 301–2 (the permission to use the original data is
gratefully acknowledged) and own calculations for 2002 and 2003). Turnover is
partly accounted for by the accuracy of the electoral system in translating electoral
volatility into seats changing hands, in particular when new parties enter parliament
(e.g. in 2002, when a record 45 per cent of MPs were newly elected). However, the
fact that the average percentage of new MPs is almost twice the average percentage
of seats changing hands indicates that the nomination process of the parties has an
effect on turnover independent from that of the electoral system’s faithful translation
of electoral volatility. The party nominations are probably the more important effect:
under the pre-1917 electoral system, the average percentage of new MPs was
already 19 per cent, which is only 4 per cent lower than in the post-war elections
under the current system.
Table 24.3 MPs’ views on the importance of representing regions and parties (% very
important)
The Netherlands Denmark Iceland Norway Sweden
Own region/constituency 15 18 24 22 39
Own political party 79 56 52 68 77
Source for the Netherlands: Dutch Parliamentary Study 2001 (Leiden & Twente Universities).
Source for the Nordic Countries: Esaiasson 2000: 59.
The Netherlands 503
Government formation
I already mentioned that the introduction of PR in 1917 consolidated rather than
created the multiparty system. Similarly, the fact that the country has always been
governed by coalitions, even under the pre-1917 electoral system (at least from the
time when the organization of parties makes it possible to distinguish single-party
and coalition governments), should prevent any rash conclusion that the electoral
system is the cause of coalition government. The difference is rather that the country
was governed by either secular or religious coalitions before 1917, and by a much
greater variety of coalitions (with less complete alternation of governing coalitions)
thereafter. The likely cause is that the two-round majority system encouraged preelection
(or at least pre-second round) coalitions, whereas no such incentive exists
under the current system. The possibility of apparentement may induce smaller
parties into forming an electoral alliance, but it is widely regarded as a technicality
rather than a political coalition with post-election consequences.
In some respects, there are even disincentives to form pre-election coalitions. If
they are formed, they are usually (but not exclusively) incumbent coalitions that
announce their intention to continue their cooperation in government. When such a
pre-election coalition then falls short of the seventy-six seats needed for a parliamentary
majority, the subsequent government formation tends to be complicated
and cumbersome, inevitably resulting in at least one of the pre-election coalition
partners breaking its campaign promise (as happened, for example, in 1972 and
1981). Usually, the parties keep their options open and do not advertise their
coalition preference during the election campaign. This further weakens the impact
of the election outcome on the composition of the government. There is also no
tradition that winning or losing seats in the election should affect a party’s fate in
coalition formation. All that elections may do is deny a majority to some coalitions,
but in general they leave several alternative majority combinations to choose from.
THE FAILURE OF ELECTORAL REFORM?
Apart from the abolition of compulsory voting and the lowering of the electoral
quota (and threshold) that was the side effect of the enlargement of the Second
Chamber, changes in the electoral system have been relatively minor, such as the
successive lowering of the voting age from twenty-five before 1946 to the current
eighteen, the introduction of (postal) absentee voting for Dutch expats, and giving
preference votes more effect (more about that last reform later). This lack of change
can be explained in part by satisfaction with the current electoral system: electoral
reform is not an issue that mobilizes voters in any significant way. It continues to be
debated by politicians and political commentators, and this might well have resulted
in changes, had it not been for the widespread consensus that reforms should remedy
PR’s disadvantages without affecting the high degree of proportionality of the Dutch
system, a circle that is difficult to square.
504 The Politics of Electoral Systems
Originally, the problem that most concerned reformers was the tenuous link
between elections and the composition of the government (also see Andeweg
1989: 48–52). It was realized that the electoral system was not the only or even
the main culprit in denying voters any direct influence over the government’s
composition, and most attention focused on reforms other than changes in the
electoral system. Some, primarily in the PvdA, sought to reform the party system
by advocating the formation of two pre-election coalitions. On the left such a preelection
coalition was formed in 1971, presenting a shadow cabinet, and again in
1972, presenting a joint manifesto, but this ‘progressive’ coalition came nowhere
near a parliamentary majority. In 1972 there was also no majority for a centre-right
coalition and after protracted negotiations (163 days) the ‘progressives’ were forced
to accept Christian Democrats in ‘their’ government. Since that episode, reform of
the party system appears to have been shelved. Others advocated reform of the
constitutional system, by introducing a directly elected prime minister, a reform that
was later implemented (and quickly abandoned again) in Israel (see Chapter 16).
The reformist party, D66, was founded expressly to press for this reform. The idea
was also attractive to the Labour Party because it would induce parties to form preelection
coalitions (as prime-ministerial candidates nominated by just a single
party would stand little chance of winning an overall majority). The reforms of
the party system and/or the constitutional system were to be backed up by reform
of the electoral system. D66 suggested the introduction of two- to three-member
districts but later sided with the Labour Party in pressing for PR within 12–15
electoral districts (of 10–13 seats each). Even in the latter proposal, it was hoped
that the higher electoral quota and threshold would give parties an incentive to join
forces and reduce the fragmentation in parliament (e.g. PR in fifteen districts of ten
seats rather than in one district of 150 seats would result in a legal threshold based
on an electoral quota of 10 per cent within each district rather than 0.67 per cent
nationally).
In 1969, an all-party Government Advisory Committee on Constitutional and
Electoral Law (known as the Cals–Donner Committee) published a majority report
recommending direct elections for the office of formateur and PR within electoral
districts. When the governing coalition of Christian Democrats and conservative
Liberals refused to endorse the recommendations, the three parties in the ‘progressive’
pre-election coalition (Labour, D66, and the Radicals (now part of Green
Left)) introduced a private members’ bill containing the Committee’s proposals in
1971. The parliamentary debate that followed was probably the moment that reform
of the electoral system came closest to being realized. It is likely that a majority of
the Second Chamber would have voted in favour of introducing electoral districts
had that proposal been put to a vote. However, the sponsors of the bill linked
electoral reform to the directly elected prime minister. When all other parties
rejected the latter proposal, the entire bill was withdrawn. When the three ‘progressive’
parties entered government with the Christian Democrats in 1972, the same
combination of constitutional and electoral reform was introduced once more in
1974, and defeated in the same way.
The Netherlands 505
Raising the electoral threshold could also have reduced the fragmentation of
parliament, and thresholds of 1–2 per cent have been proposed repeatedly, primarily
by the Catholic Party (now part of the Christian Democrats), but without success.
When the enlargement of parliament effectively lowered the threshold in 1956, a
majority in the Second Chamber actually favoured retaining it at the old 1 per cent
level, but, curiously, they failed to follow this up with legislation on two occasions
(Daalder 1975: 229).
Since the 1970s, democratic reformers seem to have given up on the idea of
establishing a direct link between electorate and government composition, and now
focus on establishing a direct link between electorate and government policy through
the introduction of an abrogative referendum or popular veto. This does not mean
that reform of the electoral system is no longer on the agenda, but it is now seen as a
solution to a different problem: a widening confidence gap between voters and
representatives. Discontent with the democratic institutions and cynicism about
politicians is actually rather low in the Netherlands compared to other EU member
states, and it shows no sign of a dramatic increase, not even in the turbulent election
year of 2002. In the perception of politicians, however, the confidence gap is a
reality: in the 2001 Parliamentary Study, 63 per cent of all MPs interviewed agreed
that such a gap does exist. To bridge this gap, it is felt, MPs should be brought
into closer contact with their voters and voters should be given more influence over
the choice of MPs. There can be no doubt that voters value influence not just over the
partisan composition of parliament, but also over its personal composition: the
proportion of votes cast for candidates other than the list-pullers (preference
votes) gradually increased from 3 to 4 per cent in the 1950s, around 10 per cent in
the 1960s and 1970s, to about 20 per cent in the 1990s. In 2002 preference votes
reached a record 27 per cent of the valid vote, to drop back to 19 per cent in 2003.
One way to reinforce the link between voters and individual MPs would be to give
more weight to such preference votes, and this has actually been done in a limited
way. In 1998 the threshold for preference votes to take effect was lowered from 50
per cent of the electoral quota to 25 per cent. Under the more stringent rules, only
three candidates had ever been elected in defiance of the party’s rank ordering in the
fourteen elections between 1948 and 1994, whereas five candidates were elected by
preference votes alone in the three elections between 1998 and 2003. Significant as
this increase may be, the impact of preference votes remains quite marginal and in
each election candidates with only a few hundred preference votes are declared
elected on the basis of their position on the list while other candidates on the same
list with several thousand votes are not elected. So far, attempts to give preference
votes more impact have failed, primarily out of fear that it would lead to intraparty
competition at the electoral level.
Instead, the debate concentrates on introducing some form of geographical
representation. In order to satisfy the requirement of undiluted proportionality, the
most prominent of the current proposals is to adopt some form of mixed-member
proportional (MMP) system following the example of the German or New Zealand
electoral systems (see Andeweg 1997: 239–44). Under this system, the voter would
506 The Politics of Electoral Systems
receive two votes: one vote for a national party list of candidates, as under the
current system, but without the preference vote option, and one for a candidate in
one of seventy-five (or fewer) SMDs. The 150 seats in the Second Chamber would
continue to be allocated to the parties in the same way as under the current system.
Within parties, the seats would then be allocated first to the party’s candidates who
have won a plurality in their SMD. A party’s remaining seats would be allocated to
candidates on the party list, in the order of that list. This proposal had been
considered, but rejected, by the Cals–Donner Committee in 1969, because it
would not strengthen the voters’ influence over the composition of the government.
In 1990 the idea was relaunched by an all-party ad hoc parliamentary committee of
party leaders, the Deetman Committee, as part of a comprehensive package of
suggestions for constitutional and administrative reform. Proponents particularly
emphasized that introduction of some form of regional representation in Dutch
politics would bring voters and MPs into closer contact, and expressed the expectation
that MPs who seek to win a district will be generalists rather than specialists in
a narrowly defined policy area, as most Dutch MPs are now. Since the Deetman
Committee’s suggestion, several varieties of such a system have been proposed,
with one or with two votes, with single-member districts or with multimember
districts, etc. (see e.g. Van Schagen and Kummeling 1998).
The Balkenende II government of Christian Democrats and conservative- and
progressive-Liberals that took office in 2003 initially committed itself in its coalition
agreement to introducing such a change in the electoral system before the next
election. In a white paper the government outlined the details of its proposal, which
differed in two crucial respects from MMP as it is used in Germany and New
Zealand.1
Under MMP it is possible that a party wins more district mandates than
it is entitled to seats on the basis of its national list vote, and in Germany (where this
occurs regularly) and in New Zealand (where it has not yet occurred), such district
candidates are declared elected nevertheless, and U¨berhangmandate (supplementary
seats) are added to the regular number of seats in parliament (see Chapter 10). The
Dutch constitution, however, fixes the number of seats in the Second Chamber of
parliament at 150. The government sought to avoid changing the constitution, as the
procedure is cumbersome (two readings interrupted by elections) and a two-thirds
majority is required in the second reading; adopting a process that necessitates
amending the constitution would significantly reduce the chances of success.
In order to minimize the risk of additional seats, MPs would not be elected in
seventy-five SMDs, but in twenty districts of two to six seats. Within these districts,
voters would cast a single non-transferable vote for a candidate. Candidates would
not be elected if they did not receive a number of votes equal to or higher than half
the national electoral quota, even if that would mean that a district seat would not be
filled by a district candidate. In most cases this would have made it unattractive for
parties to field more than one candidate, because a fragmentation of the party vote
1
For an English language version of the white paper, see www.nieuwkiesstelsel.nl/contents/pages/
1250/engels.pdf.
The Netherlands 507
over several candidates may result in none of the party’s candidates crossing the
electoral threshold. This would have reduced the chances of a party’s winning more
district mandates than it has won seats nationwide, but critics argued that it also
would have meant that the election of a large party’s candidate would have been a
foregone conclusion in most districts, and that it would have been likely that several
district seats would not be filled by a district candidate. Even under this system,
excess seats cannot be completely ruled out. If a party were to have more successful
district candidates than it had won seats, remainder seats that were allocated to other
parties as part of the national proportional distribution of seats were to be reallocated
to accommodate those winning district candidates. In the unlikely case that there
would not be enough remainder seats to be reallocated in this way, winning district
candidates whose party had not won sufficient seats would be denied a seat in order
to avoid supernumerary seats. Critics also pointed out that the proposed districts
would be so large (between 450,000 and 1.2 million inhabitants) that it would have
been almost impossible for MPs elected in such a district to maintain contacts with
voters in their constituency.
A second important way in which this differed from the German and New Zealand
systems is that the Dutch voter would have had more influence over the personal
composition of parliament: voters would have continued to vote for individual
candidates on the national lists, and their preference votes might have led to listcandidates
being elected out of list order. In the government’s original proposal,
candidates would not have been allowed both to stand for election in a district and to
be nominated on the national list. In Germany and New Zealand, district candidates
are often ‘re-insured’ by their party against the risk of losing in the district by a
position on the national party list, with the result that losing district candidates still
enter parliament, taking away the effect of a district vote (Shugart and Wattenberg
2001: 593–5). This aspect of the original proposal met strong opposition because
most parties felt that it would deprive them of the power to ensure a balanced
composition of their parliamentary party. The government then decided to drop this
aspect of its proposal and to allow double candidacies. This concession was not
enough to stem the growing tide of criticism, even from within the governing parties.
In March 2005, when the cabinet minister who had authored the bill was defeated
over another proposal and resigned, the coalition parties quickly shelved the plans.
Instead, the government announced that preference votes would be given more
weight under the current electoral system at the next elections, while new proposals
for more radical reform at a later stage will be developed.
Notwithstanding the failure of the government’s initial proposal, however, if
radical reform of the Dutch electoral system comes about, it would probably entail
the introduction of some form of mixed member proportional system, precisely
because that presents a relatively minor deviation from the current level of proportionality.
Over a quarter of a century ago, Lijphart (1978: 131–2) wrote that ‘any
reform proposal that deviates from nation-wide proportional representation or from
a low threshold has to contend with the pervasive and persistent notion that these
deviations are violations of the principle of proportionality. This fundamentalist
508 The Politics of Electoral Systems
and—in view of the interpretation of proportionality in most other PR systems—
clearly ethnocentric attitude is likely to preserve extreme PR in the Netherlands in
the foreseeable future’. Despite the fact that the primary raison d’eˆtre of proportional
representation—to mirror in parliament the strong and stable groups in
society—no longer applies in a dealigned electorate and individualized society,
Lijphart’s conclusion still seems valid.
REFERENCES
Andeweg, R. B. (1989). ‘Institutional Conservatism in the Netherlands: Proposals for and
Resistance to Change’, West European Politics, 12: 42–60.
—— (1997). ‘Institutional Reform in Dutch Politics: Elected Prime Minister, Personalized
PR, and Popular Veto in Comparative Perspective’, Acta Politica, 32: 227–57.
—— (2000). ‘Fractiocracy? Limits to the Ascendancy of the Parliamentary Party Group in
the Netherlands’, in K. Heidar and R. A. Koole (eds.), Parliamentary Party Groups in
European Democracies. London: Routledge, 89–105.
—— and Irwin, G. A. (2005). Governance and Politics of the Netherlands, 2nd edn. Houndmills/New
York: Palgrave Macmillan.
Daalder, H. (1975). ‘Extreme Proportional Representation: The Dutch Experience’ in
S. E. Finer (ed.), Adversary Politics and Electoral Reform. London: Anthony Wigram,
223–48.
Duverger, M. (1951). Les Partis Politiques. Paris: Librairie Armand Colin.
Esaiasson, P. (2000). ‘How Members of Parliament Define Their Task’, in P. Esaiasson and
K. Heidar (eds.), Beyond Westminster and Congress: The Nordic Experience. Columbus:
Ohio State University Press, 51–82.
Hillebrand, R. (1992). De Antichambre van het Parlement; Kandidaatstelling in Nederlandse
Politieke Partijen. Leiden: DSWO Press.
Koole, R. A. and Leijenaar, M. (1988). ‘The Netherlands: The Predominance of Regionalism’,
in, M. Gallagher and M. Marsh (eds.), Candidate Selection in Comparative Perspective:
The Secret Garden of Politics. London: Sage, 190–209.
Laakso, M. and Taagepera, R. (1979). ‘ ‘‘Effective’’ Number of Parties: A Measure with
Application to West Europe’, Comparative Political Studies, 12: 3–27.
Leijenaar, M. and Niemo¨ller, K. (1997). ‘The Netherlands’, in P. Norris (ed.), Passages to
Power: Legislative Recruitment in Advanced Democracies. Cambridge: Cambridge University
Press, 114–36.
Lijphart, A. (1978). ‘The Dutch Electoral System in Comparative Perspective: Extreme
Proportional Representation, Multipartism and the Failure of Electoral Reform’, Netherlands
Journal of Sociology, 14: 115–33.
—— (1994). Electoral Systems and Party Systems. Oxford: Oxford University Press.
—— (1999). Patterns of Democracy: Government Forms and Performance in 36 Countries.
New Haven, CT: Yale University Press.
Loots, J. (2004). Voor het Volk, Van het Volk; van districtenstelsel naar evenredige vertegenwoordiging.
Amsterdam: Wereldbibliotheek.
Narud, H. M. and Valen, H. (2000). ‘Does Social Background Matter?’ in P. Esaiasson and
K. Heidar (eds.), Beyond Westminster and Congress: The Nordic Experience. Columbus:
Ohio State University Press, 83–106.
The Netherlands 509
Rae, D. W. (1971). The Political Consequences of Electoral Laws, 2nd edn. New Haven, CT:
Yale University Press.
Secker, I. (2000). ‘Representatives of the Dutch People: The Smooth Transformation of the
Parliamentary Elite in a Consociational Democracy 1849–1998’, in H. Best and M. Cotta
(eds.), Parliamentary Representatives in Europe 1848–2000: Legislative Recruitment and
Careers in Eleven European Countries. Oxford: Oxford University Press, 270–309.
Shugart, M. S. and Wattenberg, M. W. (2001). Mixed-Member Electoral Systems: The Best of
Both Worlds? Oxford: Oxford University Press.
Taagepera, R. and Shugart, M. S. (1989). Seats and Votes: the Effects and Determinants of
Electoral Systems. New Haven, CT: Yale University Press.
Valen, H., Narud, H. M., and Hardarson, O. Th. (2000). ‘Geography and Political Representation’,
in P. Esaiasson and K. Heidar (eds.), Beyond Westminster and Congress: The
Nordic Experience. Columbus: Ohio State University Press, 83–106.
Van Schagen, J. A. and Kummeling, H. (eds.) (1998). Proeve van een Nieuw Kiesstelsel.
Deventer: W.E.J. Tjeenk Willink.
510 The Politics of Electoral Systems