Would an independent costings unit fix election number fights?

Economist Michael Reddell says he is not sure he has ever changed his vote because of an election costing. His comment comes as National and Labour both promise an independent unit to cost party policies.

National's fiscal plan, its set-piece statement on tax, spending and borrowing, includes $1.2 million every three years for a unit to independently check the cost of policies put forward at election time, according to RNZ reporting on 12 October 2026. Labour's fiscal plan promises an "independent pricing authority" costing $6 million a year. The stated aim, in part, is to stop the usual campaign claims about fiscal holes and promises with no costings attached.

National tried to set up a costings unit in this term of Parliament but could not get New Zealand First or ACT to agree. Greens co-leader Chlöe Swarbrick said she still backed a costings body but was critical of National's version.

The row goes back to the 2023 election. Reddell and economist Sam Warburton looked at National's costings and found they "aren't credible". Christopher Luxon said he was "extremely confident" in the 2023 numbers but would not release the modelling behind them. After forming a government with ACT and New Zealand First, National dropped its plan to raise revenue through a 15 percent tax on house sales to foreigners.

Reddell said New Zealand is too small for an independent costings unit and asked whether it would add value. He pointed to Australia, where parties must put policies forward for costing, and said the fiscal debate there did not look much better. Writing on X as @MHReddell, he said that on the parties' giveaways alone, voters would think the Government was running surpluses.

The broader context here is familiar to Press Gallery staff. Each election, the big parties release a fiscal plan. Reporters and economists test the assumptions. Opponents look for a hole. The funding detail matters more than the headline. A payment of $1.2 million every three years points to a small team used only at campaign time. A $6 million-a-year body points to a permanent office with staff and its own economic models.

Looking at what this means for a future Parliament, the MMP problem is difficult to get around. Parties campaign as if they will govern alone. Then they negotiate a coalition agreement that rewrites tax and spending promises. No pre-election costing, however careful, can price a post-election deal that does not yet exist. Australia is instructive here. Making parties submit policies brings more numbers into the system. It does not stop selective release, or parties pushing unsubmitted ideas at the margins.

In my view, Reddell's doubt is less about accounting standards than about who uses the numbers. For campaign teams, lawyers and budget advisers, a shared baseline would change daily campaign work. It would limit exaggerated claims about holes and give finance spokespeople a neutral point to refer to. For voters, the effect is likely smaller. Few people read fiscal plans line by line. They read the coverage and judge credibility in the round, on tax, spending, borrowing and delivery.

That leaves two practical tests for either plan. First, access and timing. Without guaranteed access to Treasury data and clear deadlines in the last weeks of a campaign, a unit will struggle to stay relevant. Second, independence and use. If parties take part by choice, large parties may submit only what suits them. If they must take part, small parties will need help to comply. Either way, politicians will still argue about starting points, how people behave, and wider economic effects. Independence does not take judgment out of fiscal policy. It moves it somewhere else.