Know your coalitions — overview

In order to properly prepare you for what’s going to happen after the elections it’s time to talk about coalitions. Dutch parties and voters have been thinking about them from the start, and they are everyone’s number 2 priority (number 1 being “How do I get as many votes as possible?” or “Which party shall I vote for?”)

Today we’ll start with a broad overview. Separate posts will discuss each of the five most likely coalitions.

Neighbours and voter exchanges

This is the general left-to-right ordering of the major Dutch parties.

Left Right

Thus, D66 has PvdA and CDA as neighbours, while the VVD has CDA and PVV as neighbours. This concept plays an important role in coalition theory.

(The witness parties don’t fit into this scheme. For instance, on an economic left-to-right scale the CU would be between D66 and CDA, but orthodox protestant CU will never ever exchange voters with ultra-secular D66.)

Generally, a party exchanges voters not only with its neighbours, but also with the neighbours of those neighbours. Thus the PvdA will also draw SP and CDA voters, but traffic with the VVD is generally absent.

As a special rule, SP and PVV also exchange voters because both of them are extreme protest parties. (The SP is currently toning down its protestiness, but although that increases its chances during coalition negotiations, it makes it less attractive for protest voters. Wilders is doing exactly the opposite.)

All this leads to the following scheme:

Voters of will consider

It’s easy to see that part of the power of the centre parties PvdA, D66, and CDA stems from the fact they attract voters across much of the political spectrum. Wing parties SP, GL, VVD, and PVV have a less broad potential voter base.

(Remember that this is an abstraction only. Right now I myself am hesitating between SP and D66, and that does not fit into this nice, clean scheme of things. There will be quite a few more voters like me.)

Coalition theory

With that in mind, let’s see what coalition theory predicts.


  1. For all parties the point of participating in a coalition is to execute as much of their platform as possible.
  2. However, the formation of any coalition means negotiations and compromises. A party will never be able to execute its entire platform — there’s always something its coalition partners object to.
  3. The chance of losing part of one’s platform in coalition negotiations is less when a party forms a coalition with its ideological neighbours, whose platforms most resemble its own.
  4. Due to the need for compromises the profile of a party in a government coalition is always more bland and less outspoken than that of the opposition parties, who can afford to concentrate fully on their own platform. Disappointed platform voters may therefore switch from the governmental to the opposition party come the next elections.
  5. For both reasons parties vastly prefer a coalition with their ideological neighbours. That gives them the best chance to execute their platform and it binds their closest electoral competitors to the same governmental policies, lessening the danger of voter exodus.
  6. The power of a single party within a coalition is a direct function of its size relative to all other coalition parties. Therefore all parties have a vested interest in keeping the coalition as small as possible while still commanding an overall majority in parliament. (The so-called “minimal-size coalitions.”)
  7. The ideal position for an opposition party is in between two government parties. Thus it can wage opposition “from the centre,” which usually works very well. Disappointed government party voters who agree with most of the government’s actions but not all will flock to the the centrist opposition party come the next elections.

Thus coalition theory predicts a minimal-size coalition made up from ideological neighbours, preferably with no “holes” in it. In general this fits Dutch politics reasonably well.


Still, there are exceptions. Both Van Agt II (1981-1982; PvdA+CDA+D66) and Kok II (1998-2002; PvdA+VVD+D66) were oversized; D66 was not needed for a parliamentary majority.

In 1981 D66 was included because it had won seats while PvdA and CDA both had lost, and also in the hope that party leader Terlouw would keep the bitter rivals Van Agt (CDA) and Den Uyl (PvdA) from each other’s throats. (He didn’t, and that was that for Van Agt II. And for D66.)

The 1998 elections were about the continuation of the Purple coalition of which D66 was a vital ingredient, and the three parties had made their coalition preference known before the elections — an unusual occurrence. D66 simply had to be included.


Purple is also an example of a non-neighbouring coalition: between coalition partners D66 and VVD stood opposition party CDA. The coalition parties, especially D66, deliberately allowed this hole because their point was banishing the CDA from the halls of power.

Coalition theory predicts that the CDA would benefit hugely from Purple, being able to wage opposition “from the centre” and thus win back its voters. However, the CDA is an absolutely lousy opposition party, and in the 1998 elections it singularly failed to capitalise on its centre position.

The reasons it returned to power in 2002 had nothing to do with its brilliance in the opposition. In 2002 the CDA was available while being neither Purple nor Fortuyn. That was enough for centrist voters — any positive distinction was not necessary.


In similar situations, D66 did a lot better. The 1994 Purple coalition became possible only because D66 grew so huge in the elections, and that in turn was caused by the preceding CDA+PvdA coalition. Unlike the CDA, D66 is able to wage opposition “from the centre” and in 1994 it profited hugely.

Incidentally, D66 is in exactly the same position this cycle. If CDA+PvdA+CU Balkenende IV had crawled on for another year, D66 would have grown maybe even larger than the 1994 record of 24 seats. This very real danger was one of the reasons the PvdA blew up government. Despite that, D66 is still doing quite well in the polls. You now know why.

The five coalitions

With all that in mind, let’s take a look at the five most likely coalitions: left, centre-left, purple, centre-right, and right. The polls page contains their most recent scores.

Coalition Left Right Extra party
Left SP GL PvdA D66 (CU?)
Centre-left PvdA D66 CDA GL
Purple PvdA D66 VVD GL
Centre-right D66 CDA VVD (GL? CU?)

Here I assume that a two-party coalition will not be possible, but that voters will rally to the broad centre parties in the weeks directly before the elections. Thus I assume a three-party coalition will be possible. Still, to be on the safe side I tried to find an extra party to reinforce each coalition.

Four out of five coalitions wholly consist of neighbours. The fifth, Purple, does not, because the point of the Purple coalition is excluding the CDA, and because precedent suggests the CDA doesn’t have the faintest idea what to do in opposition and will not capitalise on its position “in the hole.”

Again D66

One thing is obvious: D66 figures in four out of five coalitions, and that makes its coalition preferences the most important ones in Dutch politics directly after the elections. This happened once before, in 1994, and the result was Purple.

Will D66 do the same this time? Nobody knows, likely not even the D66 leadership. There are just too many interesting options available. Centre-right is the least likely one, for reasons we’ll discuss in a moment. Left won’t have the seats, but that will still allow D66 to choose between Centre-left and Purple, between CDA and VVD.


GL’s position, too, is strong. Although it only figures formally in the Left coalition, which is unlikely due to a lack of majority, it is the party of choice if either Centre-left or Purple has to be reinforced for lack of majority. It even has a chance to participate in Centre-right, although that will bring government policies far from GL’s electoral base and may be dangerous.

GL knows all this full well, and right now no three-party coalition (except the impossible PvdA+CDA+PVV) has a majority.

Still, as I said, I expect voters to move to the centre in the weeks preceding the elections, which will seriously hamper GL’s chances, especially when the negotiations result in a Centre-Left coalition. Purple, on the other hand, is a few seats below Centre-Left in all polls, and will remain so unless the VVD grows larger than the CDA. Thus, reinforcing a new Purple coalition is GL’s best bet right now.

The forbidden coalition

You might notice that PvdA+CDA+VVD is missing. That’s because this “great coalition” is hideously forbidden in Dutch politics.

Despite all the complicated voter movements that this blog follows, despite the growth of and arguments with extremists parties, at heart the Dutch system is a three-party one with PvdA on the left, CDA in the centre, and VVD on the right.

The three-party system would implode if these three parties would participate in one coalition. If there’s no difference between the three, why choose between them? Better go to one of the extreme parties, who at least believe in what they say. Or to D66, comfortably ensconced “in the hole.”

The “great coalition” is only allowed in times of war or water-related disasters. At all other times it would destroy the Dutch party system from the inside. The last one dates from 1948, when Indonesia had to be given its independence. Political leaders from left to right wanted all parties to take responsibility for that shock to the national pride.

Historical overview

To close off, here’s a historical overview of all Dutch coalitions since 1973.

Year Government Coalition Parties
1973 Den Uyl
Left PPR PvdA D66 (ARP KVP)
1977 Van Agt
1981 Centre-left PvdA D66 CDA
1982 Lubbers
1986 Right CDA VVD
1989 Centre-left PvdA CDA
1994 Kok
Purple PvdA D66 VVD
1998 Purple PvdA D66 VVD
2002 Balkenende
2003 Centre-right D66 CDA VVD
2006 Centre-left PvdA CU CDA

As you see, all coalitions have been tried at least once. Back in the eighties CDA and VVD held a majority by themselves, and the right-wing coalition became the norm (as it also had been in the sixties). Traditional right-wing voters cannot resign themselves to the fact that the days of the right-wing majority are over.

The left-wing Den Uyl government was something of a special case which we’ll eventually treat in the article series. The left block has never won a majority; the 75 seats of 1998 were its best score ever.

Also note that D66 has entered a coalition without the PvdA only once, in 2003. When all’s said and done the PvdA is still D66’s main electoral competitor, and the 2006 elections punished the Democrats severely. That’s why I feel Centre-right is the worst choice for D66.

That said, the PvdA only entered a coalition without D66 twice, in 1989 and 2006. The 1994 elections were D66’s finest hour and decimated the PvdA. 2010 will likely be less drastic, mainly because the PvdA has caused the fall of government.

The same goes for CDA and VVD. CDA entered government without the VVD only three times, in 1981, 1989, and 2006. The 1982 and 1994 elections were the best ones ever for the VVD. 2010 shapes up to be less historical, mainly due to the presence of Wilders’s PVV on the VVD’s right flank.

The CDA is curiously exempt from all these rules. 1994 was a huge defeat, but you’d expect them to win back VVD voters in 1998. That didn’t happen — as I said, the CDA mainly returned to power in 2002 because it was there while being neither Purple nor Fortuyn.

Five shorter posts will treat the five coalitions in more detail.

<— Party profile — ToN | Wilders won’t go to US —>

This is the political blog of Peter-Paul Koch, mobile platform strategist, consultant, and trainer, in Amsterdam. It’s a hobby blog where he follows Dutch politics for the benefit of those twelve foreigners that are interested in such matters, as well as his Dutch readers.

If you like this blog, why not donate a little bit of money to help me pay my bills?


Comments (closed)

1 Posted by Frans on 24 March 2010 | Permalink

I figured I'd mention that I'm somewhat like you, at least in the sense that I don't fit in the schematic you drew up. D66 and VVD have my preference ideologically speaking (I'm somewhat of a social, secular libertarian), but I feel that the VVD drifted since Bolkestein left for Europe; I also just really, really don't like Mark Rutte and the direction VVD is taking these days. So I guess my current order of preference could be listed as D66, SP, VVD, PVV. For the PVV I'm ignoring the anti-Islam nonsense for its position in this list, but keeping in mind that almost all parties essentially seem to agree with Wilders and it's just the gradation with which they seem to disagree; as such it's not nearly as negative as it would otherwise be for Wilders since it's negative across the board. I'm wide open for a new Purple, at any rate. GL required to get the majority? Sounds awesome to me compared to what we've had to live with the past 8 years.

2 Posted by Bryan on 24 March 2010 | Permalink

You don’t get to vote for Prime Minister, only party, but as your analysis shows’ predicting the ultimate coalition of parties at this point is impossible. I think a common American perspective would be to assess whether you want Balkenende or "Yes we Cohen" and vote either for PvdA or CDA. I still have not been able to appreciate how voting say for D66 means a particular piece of legislation gets enacted or a specific government policy is changed or maintained “only if” D66 picks up more MP seats in June. At this point your not even sure D66 makes it into Cabinet. What am I missing?

3 Posted by ppk on 24 March 2010 | Permalink

PvdA and CDA hope to influence the campaign so that it will become a Cohen vs. Balkenende struggle. That tactic has historically worked quite well. Wilders is the third candidate, and that complicates matters, but doesn't change the basic dynamic.

But the average prime-minister voter is less motivated than followers of one of the smaller parties.

Why vote for a smaller party? Because you agree with it more than with either PvdA or CDA. And there's always a chance that they end up in government. Right now the chance of D66 making it into the coalition is good, VVD also good, GL somewhat less, and the rest not so good.

Besides, what if you're left but don't like what the PvdA did in government? Vote for another left-wing party. Thus these parties become so-called "whip parties" whose function it is to keep the main party on the straight and narrow path.

But the smallest parties are also the witness parties, who gain a small but fanatical following that is usually very motivated to turn out.

4 Posted by Eric Meyer on 25 March 2010 | Permalink

You say: "Right now I myself am hesitating between SP and D66, and that does not fit into this nice, clean scheme of things."

Looking at your scheme, your alternate choices would imply that you're naturally either a PvdA or GL voter, but for some reason aren't enamored of those parties in this cycle. Is that not the case?

I'm fascinated by the way the "neighbour" spectrum is in fact a loop, reminiscent of an old videogame.

5 Posted by ppk on 25 March 2010 | Permalink

Yes, after I published the entry I thought the same. I suppose deep down I'm a PvdA voter who doesn't much like the PvdA. Or something.

My guiding principle used to be that I never voted for a large party. That would mean I couldn't vote for the SP now, I suppose.

Anyway, it's good to have more than two options on the political market. I guess that's what many Dutch voters celebrate during the agonising making up of the mind.

6 Posted by Raphael on 25 March 2010 | Permalink

In addition to the "neighbours" system, I think you wrote in one of your earlier posts (I don't remember when) that there are three pairings of one right-wing and one left-wing party where the parties to some extent compete for the same voters- CDA/PvdA, VVD/D66, and PVV/SP. What about that?

7 Posted by Bryan on 25 March 2010 | Permalink

Having more choices is a luxury of political pluralism. While the two parties in America are not night and day different, the difference may be more acute than between most Dutch wing parties and the center. Obviously, there was not much of a veto group within the Democratic Party to moderate the recently passed National Health Care Program. All or nothing is seldom a great option for such an important issue. Only time will tell whether the Republicans can claw back some of the aspects of the program that appear inefficient. In America concern about those outside the “tent” happens most frequently when control of Congress and the Presidency are split between the two parties. Having PvdA and/or CDA in government might be the dutch equivalent. While there is a lot of noise surrounding both the dutch left and right wings, doesn’t the heavy lifting or best statesmanship ultimately have to come from the center?

8 Posted by Bryan on 25 March 2010 | Permalink

On the topic of choosing parties, what parties attract the most trade union members? Do most FNV members vote PvdA or SP, while CNV members divide up between PvdA and CDA with MHP members across the board?

9 Posted by Frans on 25 March 2010 | Permalink


I've always had the feeling that most FNV members vote PvdA. For that matter, now that I think about it, I have the feeling that the three you mentioned roughly align with the three parties. However, I have no data to back that up, so I wouldn't be surprised if I'm wrong on that one.

ppk? :D

10 Posted by ppk on 26 March 2010 | Permalink

Neighbour system (Rafael): this is mostly an abstraction; the conduits between left- and right-wing parties are the most important ones, and they form the core of the voter movements we'll see in the next two months.

11 Posted by ppk on 26 March 2010 | Permalink

Trade unions:

- CNV, protestant, aligned with CDA. Current CNV chairman Terpstra is part of the CDA's left wing.

- FNV (merger of socialists and catholics) is firmly aligned with the PvdA. Problem is that most common members voted SP last time. Currently the PvdA is on the defense within the FNV, and that fight would be a reason for the PvdA to shift leftward.

- MHP is new, and thus not denominationally aligned. I'd say members vote for safe centre parties, but that's only guesswork.

Of course the PvdA has to appeal both to FNV and to MHP members. That's the problem it's facing.

12 Posted by Sander Aarts on 26 March 2010 | Permalink

Political diagrams are always abstractions, but using only 1 axis might be just a little bit too much of a simplification. Leaving out the witness parties helps in this respect, but still.
Multidimensional diagrams are harder to make of course as the scaling becomes much more important (multiple one dimensional diagrams, one fore each axis, could solve that), but they may even clarify your own indecisiveness between D66 and SP.

13 Posted by ppk on 26 March 2010 | Permalink

A few months back I found an article with some very useful info about other axes, notably cultural issues and immigration. Unfortunately I can't find it now.

I do plan to publish something about other axes; maybe even as a nice JavaScript interactive graph or something. But first I have to find the info.

14 Posted by Frans on 26 March 2010 | Permalink

It seems to me that the easiest and most useful thing to add would be authoritarian vs. libertarian, while maintaining the left vs. right economically. PvdA and CDA would be more authoritarian while D66 and VVD would be more libertarian, for instance. One place that worked this out quite nicely is the Political Compass. Their site seems to suggest copying this is a bad thing, though I'm fairly sure I've seen similar graphs prior to seeing that site.

Another thing you could consider would be religiosity, although in my (biased) opinion libertarianism would align more with secularism while anything religious will easily jump to the authoritarian side. But that's a) biased and b) an oversimplification.

Stance on issues like immigration would also certainly be interesting to see.

15 Posted by Bryan on 26 March 2010 | Permalink

Check it out! is an interesting interactive site the allows visitors to answer 30 question surveys and it will analyze and match your responses to the different dutch parties. It even plots your profile on a 2 axis chart showing your profile in reference to the all the parties. For my profile, it showed the closest match as CU second closest CDA. Very impressive, if I could, I am pretty sure I would be voting CDA in June.