The Reserve Bank adjusts interest rates to keep inflation under control. This does not work with perfect precision, but it works. There is one measure, one lever and a well-understood relationship between the two.

Not every policy goal is like that. Education Minister Erica Stanford took office promising to reverse a decades-long decline in New Zealand school education, especially in literacy and mathematics.

Stanford was faced with many potential levers, including curriculum, teacher training and school regulation. She decided to use as many of those levers as she could. She has reformed the curriculum, provided professional development for teachers, mandated an hour per day of reading, writing and mathematics for primary students and more.

Last week, the New Zealand Council for Educational Research reported on data from its Progressive Achievement Tests (PATs). PATs are widely used in New Zealand's primary schools to measure reading and mathematics learning.

The report includes data gathered over 15 years, from 2012 to 2026, from more than 1,300 schools and nearly 650,000 Years 4-8 students. Between 2012 and 2018 the mathematics data didn't change.

Scores slipped slightly in 2019 and 2020 and then dipped further during the pandemic. By 2023, when Stanford took office, they had recovered to pre-covid levels. In 2025 there was a lift, and a bigger one in 2026. Now, PAT mathematics attainment sits about a third of an average year's learning higher than in 2024.

Is the lift attributable to Stanford's reforms or to something else?

The compulsory hour per day was implemented for the 2024 school year. The new mathematics curriculum was published in late 2024 and mandated from Term 1, 2025. Throughout, primary teachers were given professional development in teaching maths, which a 2021 report from the Royal Society Te Apārangi had shown to be a weakness.

Other factors may be in play too. On the back of the Royal Society report, the Ministry of Education implemented a new strategy in 2022.

The mathematics PAT was refreshed in 2024, so the later results are not completely comparable with earlier ones. But the big gains in 2025 and 2026 are both on the new version.

The evidence in favour of Stanford's reforms is as good as it gets this soon after implementation. The timing fits, even if other potential causes can't be cleanly separated in the data.

Whoever is education minister after November's election should stay the course.