The slide finally stopped
For the first time in over two decades, New Zealand’s maths and science scores in the OECD’s PISA tests have not fallen. Science rose from 504 to 509 points and maths edged up one point to 480 in the results released on 8 September 2026. Reading was the exception, falling four points to a new low of 497.
For employers watching the skills pipeline, this matters more than another political victory lap. The OECD reckons 20 PISA points is worth roughly a year of learning, so a five-point science lift and a one-point maths gain are not transformative. But after successive declines, stopping the bleeding is itself a genuine data point. The direction of travel has changed, even if the distance travelled is small.
The number employers should actually watch
Averages hide the real problem. In maths, 29% of New Zealand 15-year-olds are now low performers, nearly double the 2003-09 figure, while just 11% are top performers, roughly half what it was. That halved high-performer cohort is the one that eventually feeds engineering, data science, finance and technical trades. A country cannot staff those roles from a shrinking tail.
Reading tells the same story from the other end. 23% of students are now poor readers, up from 14% in 2000, and the gap between the strongest and weakest readers has widened to 302 points, well above the OECD average of 264. From their early-2000s peaks, New Zealand has shed 43 points in maths, 32 in reading and 23 in science, the equivalent of roughly two years of maths learning wiped from the national average in a generation.
The caveat the victory lap will skip
Here is the detail most political coverage will gloss over. The Ministry of Education’s own report flags that New Zealand’s sample skewed toward higher-achieving students, potentially inflating scores by about 10-15 PISA points. In the Ministry’s words, more high-achieving students participated, so the results “will likely show New Zealand in a more positive light than if we had an unbiased sample.”
Strip that out and New Zealand’s true maths score could sit closer to 465-470, below the OECD average, and the one-point improvement is comfortably inside the statistical noise. The saving grace is that a similar bias affected the 2022 results, so the trend comparison is less distorted than the absolute figures. The slide has genuinely paused. The level it paused at is worse than the headline number suggests.
The frontier, not the average, is the benchmark
For a small open economy chasing high-value investment and skilled labour, the OECD average is the wrong yardstick. The frontier is what matters. China topped maths at 612 and science at 597, while Singapore led reading at 535. China’s maths result sits 132 points above New Zealand’s 480, more than six years of learning by the OECD’s own measure. Even Estonia, the strongest non-Asian nation, ran 28 points ahead of New Zealand in maths. New Zealand is stabilising at a level that leaves it well behind the economies it most needs to compete with for capital and talent.
Whether the policy is working
The 2026 results are the first real test of the government’s response. The February 2025 ‘Going for Growth’ framework set targets of 80% of Year 8 students at or above expected curriculum level in reading, writing and maths by December 2030, backed by structured literacy from April 2024, a mandatory hour a day for maths, reading and writing, and standardised assessment for earlier intervention. Those interventions had barely started when these students sat the test, so a science lift and a flat maths line are, at best, an early and cautious green light.
The historical baseline is sobering. The official PISA 2022 summary showed New Zealand’s maths average falling 15 points in a single 2018-2022 cycle, and the OECD’s 2022 country notes had New Zealand only modestly above the OECD average with every trend line pointing down.
What it means for hiring
The 15-year-olds who sat this test enter the workforce in five to ten years. A one-to-five-point lift in averages does nothing for the near-term hiring environment, and it does not rebuild the high-performer tail that technical employers depend on. The structural damage will take years to unwind even if the settings are right.
Treat the rebound as a leading indicator, not a solved problem. The slide has stopped. The question employers should keep asking is whether the government can rebuild the top of the distribution, not just flatten the average, because that is the cohort their future engineers and analysts come from.
Join the discussion
Add useful context, ask a good question, or challenge an idea — keep it specific and respectful.