Does iPredict understand MMP? – UPDATED

iPredict’s weekly election update #7 had good news for those of us political tragics at the blue end of the spectrum:

John Key will lead a National/Act/UnitedFuture government with 62 seats and a two-seat majority in a 122-seat Parliament after the next General Election, this week’s snapshot of New Zealand’s prediction market, iPredict, suggests. Were the Maori Party to continue supporting this National-led government, the government would have 67 seats and a 12-seat majority. . .

. . .Forecast party vote shares are: National 44.5% (down from 45.6% last week), Labour 35.4% (up from 34.4% last week), Greens 7.8% (up from 7.5% last week), New Zealand First 4.1% (down from 4.6% last week), Act 3.6% (up from 2.8% last week), Maori Party 2.5% (down from 2.7% last week) and UnitedFuture 1.1% (up from 0.3% last week).

But I think there’s a mistake in the writer’s understanding of MMP: 

For the first time, UnitedFuture Leader Peter Dunne is forecast to be re-elected in Ohariu, with 37% probability compared with 35% probability for National to win the seat and 30% probability for Labour. . .

. . .   The result in Ohariu does not affect the likelihood of a National-led government, with National having 56 seats if Mr Dunne wins his seat and 57 seats if he does not.

It’s the party vote not electorate seats which determine the number of MPs a party has in parliament. If National got 44.5% of the vote it would get that percentage of seats in parliament regardless of whether or not it won Ohariu. Winning the electorate seat would mean it would have one less list seat and so would end up with the same number of MPs.

If Dunne lost the seat the party votes for United Future would be redistributed among all the other parties. That’s not many but if he commits to supporting  a National led government rather than a Labour one it would give the centre-right one more seat if he wins than if he doesn’t.

The iPredict website is here.

UPDATE: Comments from Matthew and Graeme below show iPredict was right, I hadn’t taken into account National getting an extra list seat if UF didn’t get an electorate or 5% of votes.

5 Responses to Does iPredict understand MMP? – UPDATED

  1. Matthew Hooton says:

    This is fully understood by the iPredict team. Under these particular numbers however, if Dunne wins then National’s total MPs would be down one, and if he loses then National’s total MPs would be up one. It’s just the way the maths turns out in this case – that whether he wins of loses makes no difference. Either National/Act/UF would have a total of 62 seats or National/Act would have the same 62 seats.


  2. homepaddock says:

    Matthew – You are talking about the total vote for the centre right – the quote “with National having 56 seats if Mr Dunne wins his seat and 57 seats if he does not” refers to the seats for National as a party not a National-led government with 62 seats.


  3. I haven’t done the numbers myself, but ipredict sounds right.

    If Dunne wins, then UF qualifies to be in the calculation of list seats (that is, their party vote will be within the first 120 quotients in the Sainte-Laguë calculation). They won’t get any because thye have an electorate, but they do count. If they don’t, this would mean Dunne would cause an overhang.

    It might sound counter-intuitive, but consider the last election. If you include NZF in the calculation, then five of the top 120 quotients were for NZF; but NZF wasn’t included, so five other MPs came in (Green, Labour, Nat x 3).

    If you take UF completely out of the calculation, one party is going to get one more of the top 120 quotients. On these numbers, it’s National.

    Consider the alternative, if Dunne loses, and all that happens is that Katrina Shanks changes from a list MP to an electorate MP, and no party gets an extra list MP then (ignoring MP overhang) there would only be 119 MPs in Parliament!

    That said, I’m going to take issue with iPredict’s understanding of probability; if their numbers are as they say:

    “Peter Dunne is forecast to be re-elected in Ohariu, with 37% probability compared with 35% probability for National to win the seat and 30% probability for Labour…”

    Then the prediction markets are saying there is a 63% chance Dunne *won’t be re-elected.* If what the markets suggest is that Dunne is likely to get 37% of the vote, vs 35% Nat, and 30% Lab, then they should have said.


  4. Matthew Hooton says:

    No, I am talking about both (and the point you make in the post is generally correct).

    However, under these particular numbers, National gets 56 seats in total if Dunne wins and 57 if he does not. This is not because of National winning Ohariu but just how the Sainte-Laguë allocation formula ends up allocating list seats under these numbers. Run the numbers through the calculator at if you would like to check.

    So, for these particular numbers National/Act or National/Act/UF both end up with 62 seats in a 122 parliament, regardless of what happens in Ohariu.

    I know this was counter-intuitive, and certainly came as a surprise, but there you go.

    Be assured everyone involved with iPredict does understand MMP and the point you are making.


  5. homepaddock says:

    Thank you both for taking the time to explain – I hadn’t taken into account National getting the extra list seat if UF’s votes were distributed among the other parties.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: