Save MMP!

The case for acting

Why the overhang must be fixed

For thirty years the overhang was a fault New Zealand could live with. Since 2026 it is not. This is less a fix we want than a fix we need.

What changed in 2026

From 1996 to 2023 an overhang was one or two seats, held by a very small party, and it never decided who governed. Leaving the rule alone was a fair choice.

In 2026 the overhang reached 11 seats on election-night figures. Those seats turned one side’s exact half of the seats into a majority, and could decide who governs (see what went wrong in 2026). Both sides of Parliament hold extra seats. This is a fault in the rules, not a charge against any party.

There are five reasons why it cannot be left as it is.

1. It decides who governs

The P in MMP stands for proportional. It is the point of the whole system: the party vote decides how strong each party is in Parliament. With an overhang, that stops being true.

And it stops being true exactly where it matters most, on the line between government and opposition. When an election is close, the extra seats create a majority that the party vote did not. That is what happened in 2026.

It has happened before. In the German state of Schleswig-Holstein in 2009, a government took office with a majority that rested on three uncorrected overhang seats, although the opposition parties had won more votes (see the German states compared).

The size matters too. Germany’s constitutional court has accepted an overhang with no correction only up to about 2.5 percent of parliament. On election-night figures, New Zealand’s overhang in 2026 is 8.4 percent, more than three times that limit.

The Gallagher index measures the P in MMP. It gives the gap between votes and seats: zero is a perfect match, and lower is better. We counted two German state elections, in parliaments close to our size, under four rules.

RuleBaden-Württemberg 2026Schleswig-Holstein 2009
Seats added until balanced0.360.86
Seats added up to a limitnot used1.98
New Zealand’s rule today5.577.21
Our fix0.220.64

Our calculation from the official results, counted among the parties that won seats. See how it was counted.

On the same votes, New Zealand’s present rule is many times further from the party vote than any of the others.

Our fix comes closer to the party vote, in both elections, than either kind of balancing the German states use. And it does so without adding a single balancing seat: the votes in Parliament stay at the planned number, where Baden-Württemberg needed 157 members for a parliament planned at 120.

2. It feeds itself, and it won’t stop

Once a party has an overhang, a further party vote for it adds nothing. Its electorate seats are fixed, whatever its party vote does. Its supporters do better to give their party vote to an ally.

Every party vote that moves that way makes the overhang larger and adds a seat to that side of Parliament. A party with an overhang strengthens its side by losing party votes to a partner. No voting system should work like that.

In both countries the overhang has grown over time. The party vote spread across more parties, more voters learned to split their two votes, and the large parties went on winning most electorates. Germany’s split voting rose from about 6 percent of voters in the 1970s to about 27 percent in 2017.

This is not voters misusing the system. They are using their two votes as MMP invites them to: one for a local MP, one for the party that speaks for them. More and more of them choose among a wide range of parties. In 2026 the two largest parties together took 58.6 percent of the party vote, the lowest share since MMP began, down from 81.3 percent in 2017 (2026 figures are preliminary). Meanwhile the party that leads in the electorates still wins about six in ten of them.

New Zealand needs a voting system built for a much more diverse Parliament, and ready for voters who use their voice in a much more nuanced way.

The law sets no limit. In theory every one of the 71 electorates could become an extra seat, for a Parliament of 191 members.

The experts’ open letter of 18 November 2026 warns of the same thing: an arms race between the two sides (see the letter).

3. It was known for decades, and left alone

The fault is not new, and it was never hidden. It was pointed out again and again, in Germany first and then here.

  • Germany, from 1949. Germany had overhang seats from its first election in 1949. In the forty years that followed there were 17 in all, never more than five at one election, and the rule was left as it was.
  • 1986, New Zealand. The Royal Commission saw that a large number of voters splitting their two votes, or two parties working together, could upset proportionality. It judged the risk remote and expected voters to punish any party that tried. Read the Commission’s own words.
  • 1988, Germany. The constitutional court stated the fault plainly: a voter who splits the two votes gains twice in the one case where overhang seats arise. That was five years before New Zealand passed the Electoral Act 1993.
  • 1993, New Zealand. During the referendum campaign an MP warned that a party could get round MMP by dividing into a “town” party and a “country” party, one to win electorates and one to collect party votes.
  • 1995, New Zealand. Parliament closed the formal version of that device, where one party stands as an umbrella over others. The Minister told the House that other ways round the rules would be found and that Parliament would need to keep watch. Nothing further was done.
  • 2012 and 2023, New Zealand. Two independent reviews recommended changing the overhang rule. Parliament did not act.

Germany’s numbers show why nobody saw a need to act, and what happened next. In its first forty years the overhang never passed 1 percent of parliament. After 1990 it grew, and by 2009 it was close to 4 percent.

ElectionOverhang seatsMembers of parliamentShare of members
Germany 194924020.5%
Germany 195334870.6%
Germany 195734970.6%
Germany 196154991.0%
Germany 1965 to 197604960%
Germany 198014970.2%
Germany 198324980.4%
Germany 198714970.2%
Germany 199066620.9%
Germany 1994166722.4%
Germany 1998136691.9%
Germany 200256030.8%
Germany 2005166142.6%
Germany 2009246223.9%
Germany 201346310.6%
Germany 2017467096.5%
Germany 2021347364.6%
New Zealand 2026111318.4%

From 2013 to 2021 Germany added seats for the other parties to balance the overhang, so the members shown for those years include the balancing seats. Since 2023 Germany has no overhang seats. German figures to 1987 leave out the West Berlin deputies, who sat without a full vote. Sources: the German parliament and the Federal Returning Officer. The New Zealand figures are from the election-night count.

Each time the answer was the same: it has not happened, so it does not need fixing. No party used the opening, and vote-splitting stayed small. For thirty years New Zealand’s proportionality rested on that restraint, and on nothing in the law.

What was known is the rule and the opening it leaves. What is new in 2026 is the scale: an overhang of 11 seats, part of it held by the party with the most MPs, and large enough to decide who has a majority.

Splitting your votes is every voter’s right, and nobody broke any rule in 2026. But a safeguard that works only while nobody tests it is not a safeguard. Now that the effect is known, every party on every side has a reason to make use of it.

4. Nobody else will fix it

In Germany a constitutional court forced parliament to act (see Germany’s three fixes). In Schleswig-Holstein the state’s constitutional court went further. In 2010 it held the election law unconstitutional, ordered a new law and sent the voters back to the polls early. New Zealand has no court with that power.

Independent reviews recommended a change in 2012 and again in 2023. Parliament left the rule as it was both times. The MPs who would vote on a fix today were elected under the present rule, and parties on both sides gained from it.

History shows that Parliament fixes voting rules when the public asks. MMP itself came about that way (see the history).

On 18 November 2026, 46 electoral experts asked the new Parliament to act this term (see what the experts’ open letter asks for).

5. Left alone, it puts MMP at risk

The more overhang seats there are, the more an election is decided the first-past-the-post way. A party with an overhang holds exactly the electorates it won, and its party vote decides nothing. In 2026 that was true of two parties, one of them the party with the most MPs.

New Zealanders have turned first past the post down three times.

  • 1992: 84.7 percent voted to change the voting system.
  • 1993: MMP won the binding referendum against first past the post, by 53.9 percent to 46.1 percent.
  • 2011: 57.8 percent voted to keep MMP, with first past the post on the ballot as an alternative.

Parliament as a whole is still far closer to the vote than it ever was under first past the post. But each extra seat moves it back in that direction. What an overhang party gets is closest to the supplementary member system: first-past-the-post electorates, with list seats for others added on top. Voters were offered that system in 1992 and in 2011, and turned it down both times.

If elections keep being decided by extra seats, people will lose trust in MMP itself. The call will then be to replace the system, not to repair it. We want to keep MMP, and that is why we want its one serious fault fixed.

The experts’ open letter sees the same danger: voters who lose trust and feel the system is rigged (see the letter).

What no number shows: Parliament loses good people

Many voters believe MMP was designed to give both candidates a chance, and they are right. A candidate can stand in an electorate and on the party list at the same time. A close local contest then decides who represents the area. It does not decide who is lost to Parliament: a strong candidate who loses the electorate can still come in on the list.

That is the spirit of MMP in a single electorate. Plurality decides who represents the electorate: the candidate with the most votes wins. Proportionality keeps the other voice in Parliament: the party vote fills the list.

An overhang takes the second half away, for the party that has one. A party that wins more electorates than its party vote earned gets no list seats at all. On 2026 election-night figures that is true of two parties, one on each side of the House: every one of their MPs is an electorate MP, and their lists brought in nobody.

  • For such a party the list is no longer a safety net. A candidate who loses the electorate is out of Parliament, however able, and however high the party placed them.
  • Every contest becomes win or vanish. Picture an electorate with two candidates of real standing, people most New Zealanders would want in Parliament whatever their party. Where the parties have overhangs, the electorate is not only choosing its local MP. It is choosing which of the two Parliament loses.
  • Nobody told the voters. They go to the polling booth believing the candidate who loses will come in on the list. For a party with an overhang that is no longer so.

It has already happened

On election-night figures the National Party won 42 electorates and no list seats. Four of its most senior people are not returning to Parliament:

  • Nicola Willis, deputy leader of the National Party since 2021 and Minister of Finance since 2023, also Minister for Economic Growth and for Social Investment. An MP since 2018, second on the party list. She stood on the list only.
  • Gerry Brownlee, Speaker of the House since 2023 and Father of the House, its longest continuously serving member: an MP since 1996. Twice deputy leader of his party, Leader of the House for nine years, and a former Minister of Foreign Affairs, Defence and Transport. He stood on the list only.
  • Chris Bishop, Attorney-General since April 2026 and, in the same Government, Minister for Housing, Infrastructure, Transport and RMA Reform. Leader of the House from 2023 until April 2026, and chair of National’s 2023 election campaign. An MP since 2014, third on the party list. He lost Hutt South to Labour’s Ginny Andersen by 889 votes, and the list could not bring him back.
  • Paul Goldsmith, Minister of Justice, for Treaty of Waitangi Negotiations, for Arts, Culture and Heritage and for Media and Communications, and since August 2026 Minister of Defence as well: ten portfolios in all, which RNZ reports is believed to be the most held by one Minister this century. An MP since 2011. He lost Epsom to ACT’s David Seymour, and the list could not bring him back.

Between them that is the Government’s finance, justice and defence Ministers, its Attorney-General and the Speaker, with more than sixty years in Parliament. Whatever one thinks of their politics, this is a real loss, to their party and to Parliament. National’s members and voters have as much reason as anyone to want this fixed.

Election-night figures; the official results are declared on 27 November 2026 and could change them. Sources: RNZ, The 16 MPs (likely) voted out of Parliament, 16 November 2026, and Nicola Willis has no regrets, 19 November 2026; 1News, Luxon farewells outgoing National heavyweights, 14 November 2026. Offices and dates are from the public record of their careers. We name them as examples of what an overhang costs a party’s own people. None of them has endorsed our proposal.

By chance, and by design

An overhang that comes about by chance is no fault. Voters in an electorate are free to choose the person they want, and now and then that gives a party one seat more than its party vote earned. We tolerate that: the electorate keeps its MP.

An overhang brought about on purpose is another matter. It is done to win votes in the House that the party vote did not give. Our rule does not have to judge which is which. It treats every overhang alike: the electorate keeps its MP, and the party gains no vote in the House. An overhang would gain a party one thing only, local representation. That leaves nothing to engineer one for. Parties would again ask for the party vote, their lists would work again, and Parliament could keep both candidates.

One limit, to be plain about it. The fix takes away the reason for an overhang, not the possibility of one. A party that still wins more electorates than its party vote earned would still have no list seats.

Germany knows both losses

The same thing happened in Germany in 2025. Wolfgang Schmidt was the Chancellor’s chief of staff and a Cabinet minister, and stood first on his party’s list in the state of Hamburg. He lost his electorate by 26.0 percent to 27.7. His party was owed three seats in that state and all three went to electorate winners, so its list brought in nobody, and he was out of Parliament.

Germany’s 2023 rule shows the loss running the other way. In one electorate in 2025 the winner was left without a seat, and the Vice-Chancellor she had beaten went to Parliament on his party’s list. What turning winners away looks like

What it costs the voter

Voters who helped bring about an overhang did what the rules rewarded. They won their side a few votes in the House. What they paid is harder to count: their own party’s list stopped working, and some of its best people are gone. Nobody adds that up, so it never appears in the result.

Nobody should have to choose between their party winning and Parliament keeping its best people. Change the rule, and the choice goes away.

What a fix has to do

A fix has to meet four tests.

  • Keep the local promise. Every electorate keeps the MP it chose.
  • Keep Parliament proportional. Each party votes with exactly what its party vote earned.
  • Hold whatever voters do with their two votes. It must not depend on anyone’s restraint, or on two large parties sharing the vote between them.
  • Last. It should be entrenched, protected in law, so that a bare majority cannot undo it.

We tolerate an overhang. We stop it from distorting proportionality.

See the fix