What if New Zealand’s electorates used preferential voting?
I re-ran all 71 electorates from the 2026 preliminary count as if voters had ranked their candidates. Three seats change hands, a fourth becomes a coin toss, and the overhang barely moves.
First past the post still decides 71 of the 120 seats in Parliament. On the preliminary count, 35 of those 71 winners took less than half the vote. Aperahama Edwards won Te Tai Tokerau with 30.5%, and David Seymour held Epsom with 36.8%.
Yesterday’s feature tested the overhang against the rules other countries use. This one asks a different question. If electorate MPs had to win a majority after preferences, as they do in Australia’s House of Representatives, who would have won, and would the overhang have survived?
Contents
The method
No 2026 ballot recorded a second preference, so preferences have to be modelled. I did it in five steps.
Step one: the first-preference count. For the first 23 electorates alphabetically, from Auckland Central to Manurewa, I transcribed every candidate’s vote from the Commission’s electorate pages. For the other 48, the winner’s and runner-up’s votes are exact. The remaining candidate votes are known only as a total, so I estimated them.
Step two: the estimate. In each of those 48 electorates, the leftover candidate vote is shared among the other parties in proportion to their party vote in that electorate. Each share is weighted by how well that party’s candidates ran against its party vote in the 23 known seats. Green candidates won 0.66 candidate votes per party vote there, NZ First 0.55, Opportunity 0.29 and ACT 0.25. Te Paati Maori’s electorate-only campaign makes its ratio far higher: the median across the six Maori seats where it finished first or second is about seven. No estimated candidate is allowed to outpoll the actual runner-up. Tested against the 23 seats where the real figures are known, the estimate is out by an average of 435 votes per minor candidate.
Step three: the preferences. Each party’s voters are given a ranking of the other parties, shown in the next section. When a candidate is eliminated, their ballots move on according to the ranking of the party those voters first chose.
Step four: the count. The lowest candidate is eliminated and their votes transferred, round by round, until two remain. I ran this two ways. Compulsory preferential voting, as in Australia’s House, transfers every ballot. Optional preferential voting, which New Zealand already uses for STV local elections, lets voters stop ranking. That version assumes 25–55% of transferred ballots exhaust, depending on the party.
Step five: the stress test. I re-ran both versions after doubling and then halving how strongly voters stay within their own bloc.
Five scenarios, 71 electorates. Most seats give the same answer every time. A handful do not.
The preference assumptions
The assumptions draw on the closest real-world evidence: Australia.
At Australia’s 2025 federal election, Greens preferences favoured Labor over the Coalition by 88.2% to 11.8%. One Nation’s favoured the Coalition by 74.5% to 24.5%. The Australian Electoral Commission’s flow tables publish these splits for every party. The New Zealand assumptions follow the same pattern. The table shows the share of transferred ballots going to the first-named finalist when only those two candidates remain.
When the final two are… | From NAT | From LAB | From GRN | From ACT | From NZF | From Opp | From TPM | From others |
Labour v National (to Labour) | – | – | 87% | 8% | 29% | 50% | 90% | 44% |
Labour v Green (to Labour) | 83% | – | – | 83% | 91% | 60% | 53% | 67% |
National v ACT (to National) | – | 87% | 89% | – | 74% | 79% | 91% | 71% |
Te Paati Maori v Green (to TPM) | 14% | 30% | – | 33% | 71% | 20% | – | 17% |
Te Paati Maori v Labour (to TPM) | 3% | – | 27% | 9% | 20% | 14% | – | 9% |
Author’s assumptions, informed by Australian 2025 federal preference flows. Each figure is the share of that party’s transferred ballots going to the first-named finalist.
Three judgement calls matter most. NZ First voters are assumed to favour National over Labour by 71% to 29%, close to One Nation’s Australian split. National voters choosing between Labour and the Greens are assumed to prefer Labour by 83% to 17%. Labour voters choosing between Te Paati Maori and the Greens are assumed to favour the Greens by 70% to 30%.
Change those assumptions and a few seats change with them. The stress test shows which.
The seats that change hands
Three electorates change hands in every scenario. A fourth flips only under weaker bloc loyalty.
Electorate | First-past-the-post winner | Preferential winner | Final count | Scenarios won (of 5) |
Epsom | David Seymour, ACT (36.8%) | Paul Goldsmith, National | 58.6–41.4 | 5 |
Wellington North | Tamatha Paul, Green (39.4%) | Ayesha Verrall, Labour | 54.8–45.2 | 5 |
Te Tai Tokerau | Aperahama Edwards, Te Paati Maori (30.5%) | Labour candidate (v Huhana Lyndon, Green) | 53.6–46.4 | 5* |
Wairarapa | Mike Butterick, National (43.5%) | Butterick holds; Kieran McAnulty, Labour, under weaker loyalty | 50.4–49.6 | 2 for Labour |
* Te Paati Maori loses in all five scenarios. The Labour candidate’s first-preference vote is estimated; below about 6,500 with optional preferences, the Greens win. Base scenario: compulsory preferences.
Epsom. Labour’s Camilla Belich polled 5,798 votes, the Greens’ Lawrence Xu-Nan 3,105. Once Xu-Nan is out, Belich is eliminated third. Labour and Green voters overwhelmingly rank National’s Paul Goldsmith above ACT’s David Seymour, and Goldsmith wins 58.6% to 41.4%. ACT loses its electorate lifeline. Because it cleared 5% of the party vote, it keeps its 13 seats regardless.
Wellington North. Tamatha Paul led Ayesha Verrall by 3,006 votes. The National candidate’s vote, estimated at about 8,700, decides it. On the assumed 83–17 split, Verrall wins 54.8% to 45.2%. Paul would hold only if National voters split close to evenly between Labour and the Greens.
Te Tai Tokerau. Edwards’ 30.5% of first preferences attracts almost no transfers. Once the minor candidates’ preferences are distributed, Te Paati Maori loses in every scenario. Labour wins all five on the estimated Labour candidate vote of about 7,700. That figure is an estimate. If the real Labour vote was below about 6,500 and preferences were optional, the Greens’ Huhana Lyndon would win instead. Read the seat as Labour or Green, but not Te Paati Maori.
Wairarapa. Mike Butterick’s 139-vote lead holds in the base case, 50.4% to 49.6%, because NZ First and ACT preferences outweigh the Greens’. Weaken bloc loyalty and Kieran McAnulty wins it back.
Under preferential voting, a seat won on 30.5% of the vote no longer looks safe.
The knife-edge seats
National’s marginal seats tighten but hold.
In the National–Labour marginals, Green preferences flow strongly to Labour, but NZ First and ACT preferences flow just as strongly to National. On these assumptions, the two roughly cancel out. Eight seats finish within four percentage points after preferences:
Electorate | Preferential winner | Final count | Margin after preferences | First-preference margin |
Wairarapa | National | 50.4% v 49.6% | 323 | 139 |
Waiariki | Te Paati Maori | 50.5% v 49.5% | 265 | 2,274 |
Nelson | National | 50.8% v 49.2% | 577 | 1,210 |
Hamilton West | National | 51.0% v 49.0% | 656 | 2,181 |
Mt Roskill | Labour | 51.5% v 48.5% | 900 | 262 |
Kapiti | National | 51.6% v 48.4% | 1,360 | 2,077 |
Whangarei | National | 51.6% v 48.4% | 1,253 | 1,092 |
Hamilton East | National | 51.9% v 48.1% | 1,155 | 456 |
Base scenario (compulsory preferences). Final counts in seats outside the 23 fully transcribed electorates rely on estimated minor-candidate votes.
Waiariki is the one Maori seat that stays genuinely close. Rawiri Waititi holds on by about 265 votes after Labour preferences flow mostly to the Greens’ Tania Waikato. Ikaroa-Rawhiti, Te Tai Tonga, Te Tai Hauauru and Tamaki Makaurau all stay with their first-preference leaders.
The marginal seats stay marginal. Preferences do not turn them into landslides.
What Parliament would look like
Preferential voting changes who wins some electorates. It does not remove the overhang.
The party vote is untouched, so each party’s 120-seat entitlement stays at National 35, Labour 36, the Greens 16, ACT 13, NZ First 12, Opportunity seven and Te Paati Maori one. Only the electorate results move.
Party | Electorates (actual) | Seats (actual) | Electorates (pref., base) | Seats (pref., base) | Electorates (pref., weaker) | Seats (pref., weaker) |
National | 42 | 42 | 43 | 43 | 42 | 42 |
Labour | 21 | 36 | 23 | 36 | 24 | 36 |
Green | 2 | 16 | 1 | 16 | 1 | 16 |
ACT | 1 | 13 | 0 | 13 | 0 | 13 |
NZ First | 0 | 12 | 0 | 12 | 0 | 12 |
Opportunity | 0 | 7 | 0 | 7 | 0 | 7 |
Te Paati Maori | 5 | 5 | 4 | 4 | 4 | 4 |
House size | 71 | 131 | 71 | 131 | 71 | 130 |
National + ACT + NZ First |
| 67 |
| 68 |
| 67 |
Majority needed |
| 66 |
| 66 |
| 66 |
Party entitlements from Sainte-Laguë on the preliminary party vote; electorate seats in excess of entitlement are overhang seats.
In the base case, National wins 43 electorates instead of 42, so its overhang grows to eight. Te Paati Maori’s falls to three. The House stays at 131, and the coalition goes from 67 seats to 68. Under weaker bloc loyalty, National gives Wairarapa back, the House shrinks to 130 and the coalition holds 67.
The overhang survives because it does not come from plurality voting. It comes from voters splitting their electorate vote and party vote, and preferential voting does nothing to that.
A majority rule for electorates leaves the proportionality problem exactly where it was.
Preferential voting without MMP
Without the party vote, the distortion would be far starker.
If the 71 electorates were the whole Parliament, as in Australia’s House, here is how first past the post and preferential voting would compare:
Party | First past the post (71 seats) | Preferential, base (71 seats) | Preferential, weaker loyalty (71 seats) | Party vote |
National | 42 | 43 | 42 | 29.2% |
Labour | 21 | 23 | 24 | 29.4% |
Green | 2 | 1 | 1 | 13.3% |
ACT | 1 | 0 | 0 | 10.9% |
Te Paati Maori | 5 | 4 | 4 | 0.6% |
NZ First / Opportunity | 0 | 0 | 0 | 15.7% |
Party vote shares from the Electoral Commission preliminary count, shown for comparison only.
National won 29.2% of the party vote. Under preferential voting, it would take 43 of the 71 electorates, or 61% of the seats. Majority-preferential voting guarantees each MP a local majority. It does not make Parliament proportional, and nothing in these results suggests it would.
Preferential voting answers the question of who each electorate prefers. It does not answer how the country voted.
A preferential party vote?
Ranking the party vote would barely matter this time.
Parties below the threshold took 20,759 party votes, 0.86% of the total. Some have proposed letting voters name a back-up party in case their first choice misses out. Transferring those votes on the same assumptions changes no seat, whether or not ballots exhaust.
In 2026 the wasted party vote was too small to matter. The overhang was not.
What this tells us
Preferential voting fixes the plurality problem, not the overhang.
It ends the Epsom arrangement, because a seat a party keeps through its partner’s tactical voting does not survive Labour and Green preferences. It removes Te Paati Maori’s seat won on 30.5%. It hands Wellington North to the candidate most voters there would rather have. Those are real gains for local legitimacy.
What it cannot do is stop a party winning more electorates than its party vote earns. That still needs one of the rules tested in yesterday’s feature.
The usual caveats apply. These are preliminary figures, with about 399,365 special votes still to come. The estimates cover 48 electorates where the minor candidates’ votes were not separately captured. And preferences are modelled, not observed: under a real preferential system, parties would issue how-to-vote cards and voters would behave differently.
Preferential voting changes the electorate count. The overhang comes from the party vote.
Ranked ballots would have cost ACT Epsom and Te Paati Maori Te Tai Tokerau, and still left New Zealand with a 131-seat Parliament. The overhang is a separate problem and needs its own rule.
Appendix: all 71 electorates
The full modelled result for every electorate under the base scenario, with a check on whether all five scenarios agree.
Electorate | FPP winner | Preferential winner | Final two | Winner’s final share | All 5 agree? |
Auckland Central | Green | Green | Green v National | 54.0% | Yes |
Banks Peninsula | Labour | Labour | Labour v National | 55.2% | Yes |
Botany | National | National | National v Labour | 73.6% | Yes |
Christchurch Central | Labour | Labour | Labour v National | 59.4% | Yes |
Christchurch East | Labour | Labour | Labour v National | 63.1% | Yes |
Coromandel | National | National | National v Labour | 69.2% | Yes |
Dunedin | Labour | Labour | Labour v National | 69.3% | Yes |
East Cape | National | National | National v Labour | 57.3% | Yes |
East Coast Bays | National | National | National v Labour | 72.1% | Yes |
Epsom | ACT | National | National v ACT | 58.6% | Yes |
Glendene | Labour | Labour | Labour v National | 62.1% | Yes |
Hamilton East | National | National | National v Labour | 51.9% | Yes |
Hamilton West | National | National | National v Labour | 51.0% | Yes |
Henderson | Labour | Labour | Labour v National | 61.2% | Yes |
Hutt South | Labour | Labour | Labour v National | 52.8% | Yes |
Ilam | National | National | National v Labour | 56.9% | Yes |
Invercargill | National | National | National v Labour | 62.4% | Yes |
Kaikoura | National | National | National v Labour | 66.6% | Yes |
Kaipara ki Mahurangi | National | National | National v Labour | 69.0% | Yes |
Kapiti | National | National | National v Labour | 51.6% | Yes |
Kenepuru | Labour | Labour | Labour v National | 67.3% | Yes |
Mangere | Labour | Labour | Labour v National | 77.1% | Yes |
Manurewa | Labour | Labour | Labour v National | 73.2% | Yes |
Maungakiekie | National | National | National v Labour | 53.1% | Yes |
Mt Albert | Labour | Labour | Labour v National | 62.0% | Yes |
Mt Maunganui | National | National | National v Labour | 69.7% | Yes |
Mt Roskill | Labour | Labour | Labour v National | 51.5% | Yes |
Napier | National | National | National v Labour | 59.8% | Yes |
Nelson | National | National | National v Labour | 50.8% | Yes |
New Plymouth | National | National | National v Labour | 60.3% | Yes |
North Shore | National | National | National v Labour | 66.2% | Yes |
Northcote | National | National | National v Labour | 53.6% | Yes |
Northland | National | National | National v Labour | 61.2% | Yes |
Otahuhu | Labour | Labour | Labour v National | 77.3% | Yes |
Pakuranga | National | National | National v Labour | 71.7% | Yes |
Palmerston North | Labour | Labour | Labour v National | 54.8% | Yes |
Papakura | National | National | National v Labour | 65.2% | Yes |
Port Waikato | National | National | National v Labour | 66.2% | Yes |
Rangitata | National | National | National v Labour | 70.7% | Yes |
Rangitikei | National | National | National v Labour | 58.5% | Yes |
Remutaka | Labour | Labour | Labour v National | 70.6% | Yes |
Rotorua | National | National | National v Labour | 58.3% | Yes |
Selwyn | National | National | National v Labour | 68.8% | Yes |
Southland | National | National | National v Labour | 67.5% | Yes |
Taieri | Labour | Labour | Labour v National | 57.4% | Yes |
Takanini | National | National | National v Labour | 52.6% | Yes |
Tamaki | National | National | National v Labour | 57.4% | Yes |
Taranaki-King Country | National | National | National v Labour | 67.3% | Yes |
Taupo | National | National | National v Labour | 67.8% | Yes |
Tauranga | National | National | National v Labour | 65.6% | Yes |
Tukituki | National | National | National v Labour | 63.9% | Yes |
Upper Harbour | National | National | National v Labour | 64.2% | Yes |
Waikato | National | National | National v Labour | 68.2% | Yes |
Waimakariri | National | National | National v Labour | 63.6% | Yes |
Wairarapa | National | National | National v Labour | 50.4% | No |
Waitakere | Labour | Labour | Labour v National | 57.2% | Yes |
Waitaki | National | National | National v Labour | 61.0% | Yes |
Wellington Bays | Labour | Labour | Labour v Green | 57.5% | Yes |
Wellington North | Green | Labour | Labour v Green | 54.8% | Yes |
West Coast-Tasman | National | National | National v Labour | 59.4% | Yes |
Whanganui | National | National | National v Labour | 59.5% | Yes |
Whangaparaoa | National | National | National v Labour | 71.4% | Yes |
Whangarei | National | National | National v Labour | 51.6% | Yes |
Wigram | Labour | Labour | Labour v National | 54.0% | Yes |
Hauraki-Waikato | Te Paati Maori | Te Paati Maori | Te Paati Maori v Labour | 69.2% | Yes |
Ikaroa-Rawhiti | Labour | Labour | Labour v Te Paati Maori | 62.1% | Yes |
Tamaki Makaurau | Te Paati Maori | Te Paati Maori | Te Paati Maori v Labour | 59.1% | Yes |
Te Tai Hauauru | Te Paati Maori | Te Paati Maori | Te Paati Maori v Labour | 56.1% | Yes |
Te Tai Tokerau | Te Paati Maori | Labour | Labour v Green | 53.6% | Yes |
Te Tai Tonga | Labour | Labour | Labour v Green | 62.5% | Yes |
Waiariki | Te Paati Maori | Te Paati Maori | Te Paati Maori v Green | 50.5% | Yes |
FPP = first past the post. “All 5 agree?” = the same winner under compulsory and optional preferences, and with bloc loyalty doubled and halved.
