Correction, 20 July 2026. This article originally ran under the headline “Real wages in New Zealand are below their 1992 level” and described the Labour Cost Index deflated by the CPI as a measure of real wages. That was wrong. The Labour Cost Index is a fixed-quality price index. It tracks the pay rate for the same job done to the same standard, and by design excludes pay rises from promotion, merit, added responsibility and experience. It is not a measure of what workers actually earn. Measures of actual earnings show real wages have risen substantially since 1992. The headline, framing and conclusions have been rewritten. The underlying index numbers were correct and are unchanged. Details of what the LCI does and does not measure are set out below. We are grateful to the reader whose comment prompted the review.
Two things are true at once in New Zealand’s wage data. The pay rate for an unchanged job, tracked by Stats NZ’s Labour Cost Index, has only just kept pace with consumer prices since 1992. What workers actually earn has grown well ahead of prices over the same period, because people change jobs, win promotions, gain experience and move into better-paid work. Meanwhile labour productivity rose in every published industry aggregate. The genuinely striking story in the data is not that pay went backwards. It is that pay captured only part of the productivity gain.
This article originally read the first fact as if it were the second, and we have corrected it. The corrected version is, if anything, a more interesting story.
What the Labour Cost Index measures
Stats NZ’s Labour Cost Index has tracked salary and wage rates since the June 1992 quarter. It is built to answer one question — what does it cost an employer to have the same job done to the same standard? Stats NZ’s documentation is explicit that service increments, merit increases, promotions and pay changes that reflect an employee becoming more experienced, better qualified or more responsible are deliberately excluded from the index.
That design makes the LCI the right tool for measuring wage inflation, and the wrong tool for measuring living standards. Over a working life, most real income growth arrives through exactly the channels the LCI strips out. A flat real LCI over 33 years mostly tells you that pay rates for unchanged jobs rose at about the same pace as consumer prices. It does not tell you that workers in 2025 take home what they took home in 1992. They do not.
Wage rates and prices since 1992
Indexing to 1992 = 100:
| Year | Wage rates (LCI) | CPI all groups | Real wage rates, fixed job |
|---|---|---|---|
| 1992 | 100.0 | 100.0 | 100.0 |
| 2000 | 113.1 | 114.9 | 98.5 |
| 2005 | 126.7 | 129.8 | 97.6 |
| 2008 | 139.7 | 142.8 | 97.8 |
| 2015 | 158.9 | 161.0 | 98.7 |
| 2020 | 174.7 | 173.3 | 100.8 |
| 2024 | 200.2 | 210.1 | 95.3 |
| 2025 | 204.9 | 216.1 | 94.8 |
The series has barely moved in 33 years. The trough is 2023 at 94.4, the peak 2020 at 100.8, a total range of 6.3 points. Read correctly, this says the base rate for an unchanged job has tracked the CPI almost exactly since the index began, with the 2022 and 2023 inflation shock pushing it to the bottom of its historical band. That last part is real and current — pay rates for unchanged jobs lost about 6% of their purchasing power between 2020 and 2023 and have not recovered it.
What workers actually earned
Measures of actual pay tell a different story, because they include everything the LCI excludes.
Bill Rosenberg of the Council of Trade Unions, in a 2010 conference paper examining exactly this question, found the real ordinary-time average wage from the Quarterly Employment Survey rose 14% between 1989 and 2006, over a period in which the official LCI measures fell about 2% in real terms. Same economy, same workers, same years. The difference between the two numbers is promotions, job changes, experience and upskilling.
The Treasury’s 2023 analytical note Examining New Zealand’s increased rate of income growth puts the longer picture in international context. Using the OECD’s comparable measure of real annual wages, New Zealand’s real wage growth has been at the stronger end of the OECD since around 2000. Between the 1995 to 2002 average and 2019, New Zealand wages rose from 74% to 82% of Australia’s, from 61% to 66% of the United States’, and from 77% to 88% of the G7 average. Real wages here grew faster than in most rich countries, not slower, and the wage gap that drives trans-Tasman emigration narrowed.
Average ordinary-time hourly earnings in the Quarterly Employment Survey now stand above $44, against consumer prices that a little more than doubled since 1992. A clean full-length QES real-earnings series back to 1992 is a follow-up piece for this warehouse, but no serious measure of actual earnings puts real pay below its 1992 level.
Labour productivity since 1992
Over the same period, Stats NZ’s labour productivity index moved substantially:
| Sector | 1992 | 2000 | 2008 | 2015 | 2024 | Change 1992 → 2024 |
|---|---|---|---|---|---|---|
| Goods-producing industries | 100 | 115 | 120 | 124 | 118 | +17.7% |
| Manufacturing | 100 | 115 | 128 | 130 | 126 | +25.6% |
| Service industries | 100 | 120 | 137 | 148 | 160 | +60.0% |
Services productivity is the largest mover but also the measure most sensitive to methodology. A portion of measured services output is calculated from inputs, which builds productivity gains into the definition. The goods-producing comparison is the cleaner one.
The share of income going to labour
The defensible version of the wage-productivity story is about shares, not levels. Real wages rose, but by less than productivity, and the share of national income going to labour fell over the long run.
The Productivity Commission’s 2015 working paper Who benefits from productivity growth? found the labour income share of the measured sector fell from 65.5% of measured-sector GDP in 1981 to 53.7% in 2002, with most of the decline concentrated in three short episodes — the 1982 to 1984 wage and price freeze, 1992 to 1995 under the Employment Contracts Act, and 1999 to 2002. Notably, it also found the share has risen on average since 2002, in contrast to the continued decline in the United States and some other countries.
Rosenberg’s estimates point the same direction. Across the measures he tested, real wages captured between a quarter and a half of labour productivity growth from 1978 to 2006. The pie grew, workers’ slice of it thinned, and both statements fit in one sentence without either being the whole story.
The most recent window is visible directly in the warehouse. The Stats NZ quarterly GDP(I) tables publish compensation of employees as a share of total GDP income from 2012:
| Year | Compensation of employees | Gross operating surplus |
|---|---|---|
| 2012 | 43.8% | 43.3% |
| 2015 | 42.9% | 44.0% |
| 2017 | 42.2% | 44.7% |
| 2020 | 44.0% | 47.7% |
| 2024 | 44.4% | 43.2% |
| 2025 | 43.2% | 44.2% |
Within this window the share has moved in a narrow band, 42.2% to 44.4%, with no clean trend. The 2020 and 2021 bump reflects the Wage Subsidy Scheme flowing through compensation of employees. The long decline documented by the Productivity Commission happened earlier, mostly before 2002.
What the data settles
- Pay rates for an unchanged job have only tracked inflation since 1992. The real LCI stands at 94.8 against 100 in 1992, and lost about 6% of its purchasing power in the 2020 to 2023 inflation shock.
- Actual real earnings have risen substantially since the early 1990s. The real average wage rose 14% over 1989 to 2006 alone, and Treasury finds New Zealand’s real wage growth since 2000 has been at the stronger end of the OECD.
- Labour productivity rose in every published industry aggregate over the same 33 years. Goods-producing +18%, manufacturing +26%, services +60%.
- Wages captured only part of the productivity gain. The measured-sector labour income share fell from about 65% in 1981 to about 54% in 2002, though it has drifted up slightly since.
- The employees’ share of GDP income has been flat since 2012. The long fall in labour’s share is a 1980s and 1990s story, not a recent one.
What the data does not settle
- The mechanism of the long decoupling. Measurement conventions, compositional shift from manufacturing into services, changed bargaining power and the long fall in real interest rates are all consistent with the shape of the data, and aggregate series cannot separate them.
- Whether this is uniquely a New Zealand story. Wage-productivity divergence is documented across most developed economies. The Productivity Commission’s finding that New Zealand’s labour share has risen on average since 2002 suggests the recent New Zealand experience is milder than the American one.
- The full 1992 to 2025 path of real average earnings. Building a clean long-run QES real-earnings series is the natural follow-up piece, and would let levels be compared directly rather than through overlapping published studies.
Methods
Labour Cost Index series retrieved from Stats NZ Infoshare, All Sectors Combined × All Salary and Wage Rates × All Industries and Occupations Combined, base June 2017 quarter = 1000. Data covers 1992 Q4 through 2025 Q4. The LCI is a fixed-quality index of pay rates and is used here as a measure of wage-rate inflation, not of average earnings. Consumer Price Index from the Stats NZ Consumers Price Index quarterly release, all groups for New Zealand. Industry productivity series from Stats NZ Aotearoa Data Explorer, dataflow PRD_PRD_001, annual labour-productivity index (year ended March), aggregates for goods-producing industries, manufacturing, and services. Labour share computed from Stats NZ GDP(I) quarterly current-price series, compensation of employees over total GDP income, quarterly values summed to annual. Comparative earnings findings are drawn from Rosenberg (2010), Treasury Analytical Note 23/04 and the Productivity Commission’s 2015 labour income share working paper. Annual alignment uses calendar-year averages for quarterly series against year-ended-March productivity readings; this introduces a small timing mismatch (≤3 months) but does not affect point-to-point comparisons over multi-year horizons.
Your turn
Has your own pay over the years moved more like the flat line for an unchanged job, or the rising line for actual earnings? And where do you think the missing share of productivity growth went — measurement, composition, bargaining power, or something else? Add your view in the comments.
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