The Government’s plan to replace New Zealand’s five-yearly census with an annual approach built mostly on records agencies already hold has stalled in Parliament, and Stats NZ is now working out how it would run a traditional count in 2028 instead.
Labour pulled its support for the Data and Statistics (Census) Amendment Bill on 3 August, which removed the 75 percent majority the Government needs to get the changes through. Because the timing of the census is tied to the redrawing of electorate boundaries under the Electoral Act 1993, part of the package sits inside what the law calls reserved provisions. Those can only be amended with the votes of three quarters of all MPs, or by a referendum, so no government can change them on its own numbers no matter how large its majority.
Statistics Minister Scott Simpson said the bill would now have to wait. “Considering the feedback and the limited time left before the House rises, the Bill will now need to be prioritised in the next Parliamentary term,” he said in comments reported by interest.co.nz. Simpson also said Stats NZ has a statutory obligation to prepare a census for 2028 and would “need to get on with that job” now the law change cannot pass before the election.
Labour’s statistics spokesperson Lemauga Lydia Sosene said the party could not sign up to a change of that size while basic questions about it were still open. “Reliable data is essential to plan for hospitals, housing and schools, and we can’t afford to get it wrong,” she said. Labour’s stated position is that a reform of this significance should not proceed while so many important questions remain unanswered, particularly about whether an administrative data first model would be accurate enough to replace a direct count.
The two sides have spent the days since trading blame. The Government says Labour walked away at the eleventh hour after months of process, and the Opposition says it was never given a credible plan to say yes to. Both things can be true at once, because the bill went through a select committee stage in which submitters raised the same questions Labour is now pointing to, and the Government pressed on regardless.
The practical consequence lands on Stats NZ rather than on either party. Newsroom reported that the agency is now figuring out how it can deliver a census in 2028, despite having previously said it had neither the time nor the money to run one the old way. Government Statistician Mark Sowden has described the 2028 count as likely to use a mixed method drawing on three sources, being administrative data, surveys and targeted collection in places where the records are thin. That was designed as the first step of the reform. It now has to work as the fallback as well.
The reason the Government wanted out of the old model is not hard to follow. The 2013 census cost about $72 million. The 2018 one cost about $120 million and produced a response rate of 85.8 percent, low enough that Stats NZ had to fill gaps from administrative records. The 2023 census cost about $325 million, more than four times the 2013 figure, and lifted the response rate only to 88.3 percent, still well short of the 93.2 percent achieved in 2013. RNZ reported that coverage reached 99.1 percent once administrative enumeration was added, which is the polite way of saying the census is already partly built from other agencies’ files. Former statistics minister Shane Reti pointed to rising costs, falling response rates and disruption from events such as cyclones when the reform was first put forward.
Submitters were not persuaded that the answer was to stop counting. The New Zealand Council for Civil Liberties argued in April that administrative records are least reliable for exactly the groups that most need to be counted properly, including people who move often, people not engaged with government services and people in insecure housing. It also argued that shifting the timetable for boundary reviews has representation effects, since electorate boundaries are redrawn by the Representation Commission using census population figures. Change when the numbers arrive and you change when the map is redrawn.
That link is the whole reason the 75 percent threshold applies. It is one of the few genuine constitutional handbrakes in a system with no upper house and no written constitution, and it exists so that the rules governing elections cannot be rewritten by whoever happens to hold 61 seats. It has done its job here whether or not you think the census reform was sound policy, and it means the next version of this bill will have to be written to survive scrutiny from both sides of the House rather than just one.
None of that is settled by the election. Whoever forms the next government still faces a census bill that costs more each time it runs, a response rate that keeps drifting down, and a statistics agency that has been asked to prepare for a count it was told it would not have to hold. The parties that will have to find 75 percent between them are the ones on the ballot in a few months, and their positions on how the country counts itself are part of what voters are choosing. If you want to work out where each party sits on this and everything else before you vote, our voting tool lets you compare policies side by side.
For now the census of 2028 is back on, in a form nobody has fully specified, run by an agency that had been planning for something else. Stats NZ has just over two years to work out what that looks like.
Do you think New Zealand should keep the traditional census or move to a data driven count? Tell us what you think in the comments below.
This article was written by AI, briefed to report the facts, hopefully without some of the bias people bring to the job 🙂
Leave a Reply