Four of New Zealand’s most energy-intensive sectors are producing substantially less output per worker today than they did at the turn of the century. Three of them peaked during the final years of the Maui gas field, and have not recovered in the quarter-century since. The boom in renewable generation — which reached a record 84% of electricity supply in 2024 — has not reversed that slide.
A Newswire analysis of 47 years of Stats NZ industry productivity data, joined to MBIE’s electricity generation records, shows the following peak-to-2024 labour productivity changes across the most electricity-intensive sectors of the economy.
| Industry | Peak year | Index at peak | Index in 2024 | Change |
|---|---|---|---|---|
| Mining (incl. oil & gas extraction) | 2001 | 3,310 | 1,658 | −49.9% |
| Petroleum, chemical, polymer & rubber mfg | 2018 | 1,840 | 932 | −49.3% |
| Electricity, gas, water & waste services | 2000 | 3,005 | 1,618 | −46.2% |
| Metal product manufacturing | 2004 | 1,568 | 1,202 | −23.3% |
| Non-metallic mineral product mfg | 2007 | 2,005 | 1,725 | −14.0% |
| Wood & paper products mfg | 2022 | 2,402 | 2,116 | −11.9% |
| Manufacturing (aggregate) | 2022 | 1,823 | 1,619 | −11.2% |
Three of the seven industries in the table — mining, petrochemicals, and the utilities sector itself — have lost close to half of their peak labour productivity. The rest have lost between 11% and 23%. No heavy-industry sector in the Stats NZ panel is at or above its all-time peak.
The Maui inflection
The Māui gas field — discovered 1969, offshore Taranaki, first production 1979 — reshaped New Zealand’s energy economy. Between 1985 and 2001, gas-fired generation rose from 20% of total electricity supply to a peak of 24.6% in 2001. Over the same span, thermal generation (gas + coal + oil combined) climbed from 23% to 28%. Hydro’s share fell from 70% to 56% — not because there was less hydro, but because total generation was rising faster on the back of cheap, abundant natural gas.
Three sector productivity peaks cluster tightly around this moment.
- Utilities (electricity, gas, water) peaked in 2000 at index 3,005
- Mining (which in Stats NZ’s ANZSIC classification includes oil and gas extraction) peaked in 2001 at 3,310
- Metal product manufacturing (which includes the Tiwai Point aluminium smelter) peaked in 2004 at 1,568
Gas production from Māui declined sharply after 2005 as the field matured. New Zealand’s industrial base had spent twenty years becoming structurally gas-dependent. When the cheap gas ran out, productivity — measured as output per worker — fell, and has not returned.
Fuel mix and heavy-industry productivity at milestone years
gas% hydro% renew% metal_lp petro_lp util_lp
1978 17 70 75 1000 1000 1000
1985 20 70 74 1032 1008 1247
1995 12 77 82 1182 1368 2090
2001 25 56 63 1383 1587 2973 ← Māui peak
2008 19 52 64 1347 1497 2349
2015 12 56 80 1257 1716 2092
2024 8 54 84 1202 932 1618
The renewables paradox
From 2001 to 2024, the renewable share of New Zealand’s electricity supply rose from 63% to 84% — a 21-percentage-point swing, achieved primarily through new wind (nearly zero in 2001) and continued hydro expansion. Coal-fired generation collapsed. Gas-fired generation more than halved in absolute terms. By 2024, electricity was cleaner than at any time since 1995.
And yet productivity in the sectors that use that electricity most intensively — metals, chemicals, mineral products — has not recovered.
One hypothesis in energy-sector commentary is that renewables have “caused” heavy-industry decline by making electricity more expensive or less reliable. This article does not support that claim. The correlation between renewable-share shifts and subsequent heavy-industry productivity changes, after removing common trends, is weak. Across 75 tested combinations of fuel-share changes and industry-productivity changes, at lags of 0 to 5 years, only two pairs survive multiple-comparison correction — and both are measurement artifacts. (See methodology below.)
A more parsimonious reading is that Māui was a one-off windfall. The gas was cheap because it was abundant and already paid for. When it ran down, the structural advantage of locating energy-intensive industry in New Zealand went with it. Renewables are filling the megawatt-hour gap, but cannot replicate the input-price conditions of the 1990s. The industries that boomed during those conditions are now adjusting to a higher, global-market price of energy.
Three specific cases
Electricity utilities — down 46% since 2000
The utilities sector — which in this classification covers electricity generation, transmission, distribution, plus gas, water, and waste — saw its labour productivity peak in 2000 and is now at 1,618, roughly the same level as 1990. Three-and-a-half decades of industry reform, privatisation, vertical disaggregation, and investment in renewables have not lifted output per worker back to Māui-era levels.
Caveats apply. Stats NZ’s methodology for measuring utilities productivity changed in the late 1990s, which may inflate the 2000 peak. The decline nonetheless remains marked on any consistent-methodology sub-period.
Petrochemicals — a structural collapse compressed into two years
The petroleum, chemical, polymer and rubber manufacturing sector peaked in 2018 and has since halved. Most of the decline happened between 2022 and 2024, during which the index fell from 1,661 to 932 — a 44% drop.
This needs to be read carefully. Marsden Point Oil Refinery — New Zealand’s only oil refinery — ceased refining operations on 1 April 2022, converting to an import-only terminal. That single closure removed a large, high-productivity subsector from the aggregate in near-real-time and explains much of the 2022-2024 drop. The pre-closure petrochemical sector was also declining from its 2018 peak but at a gentler rate.
Conclusion. The headline 49% peak-to-2024 decline in petrochemicals is a blend of a genuine gradual decline (2018-2022) and a mechanical one-off (Marsden Point, 2022). Both are real, but they have different stories.
Metal products and Tiwai
Metal product manufacturing includes the Tiwai Point aluminium smelter — a single facility that consumes roughly 12-13% of New Zealand’s electricity. The sector’s labour productivity peaked in 2004 at 1,568 and has declined around 23% since, to 1,202 in 2024.
Tiwai’s future has been the subject of multiple near-closure episodes since 2013. New Zealand Aluminium Smelters Ltd has repeatedly negotiated power pricing concessions. Each of those episodes has coincided with a dip in metal product manufacturing productivity. The sector trend is dominated by one facility’s operating decisions as much as by the wider electricity cost environment.
What the statistical test looked like — and what it missed
To test the direct question — does short-run change in electricity fuel mix predict short-run change in industrial productivity? — we ran a cross-correlation analysis.
- Annual panel 1978-2024 with five fuel-share variables (hydro, gas, coal, renewable, thermal) and 15 industry labour-productivity indexes.
- First-differenced both sides to remove the common upward trend that creates spurious correlation.
- Tested all 75 (fuel × industry) pairs at lags 0-5 years, recording the best lag for each.
- Applied Benjamini-Hochberg FDR correction at α = 0.10 to avoid false positives from multiple testing.
Of 75 pairs, only two survived correction.
- Gas share → Mining productivity (lag 1 year, r = +0.478)
- Coal share → Mining productivity (lag 5 years, r = +0.472)
Both are almost certainly mechanical. “Mining” in ANZSIC includes oil and gas extraction, so the same gas that shows up in electricity generation shows up in mining output.
So the differenced, year-to-year test finds no evidence that fuel mix predicts subsequent heavy-industry productivity. But the raw level data shows a clear pre/post-Māui regime shift that the differenced test is not designed to pick up. Both results are correct. They answer different questions.
Methods and data
Sources.
- Electricity generation by fuel type, annual GWh, 1974-2024 — MBIE Energy Statistics,
electricity-dec-2025-q4.xlsx, sheet “6 – Fuel type (GWh)”, retrieved from mbie.govt.nz. - Industry productivity labour-productivity indexes, annual, 1978-2024 — Stats NZ Aotearoa Data Explorer, dataflow
PRD_PRD_001(Industry Productivity Statistics ANZSIC06),SERIESTYPE=I, VAR=2, by ANZSIC industry code. - Consumer Price Index, quarterly, 1914-2025 — Stats NZ bulk CSV via consumers-price-index-december-2025-quarter release.
Panel construction. Each MBIE fuel-share series was aligned to calendar year with its ratio of sector-specific to total generation. Each Stats NZ productivity series was left at its published annual frequency. Panel joined on calendar year. 47 matching years.
Differencing. For the lead-lag test, all variables were converted to year-on-year percentage change prior to correlation, removing common trend.
Multiple-comparison correction. Benjamini-Hochberg FDR procedure applied to the 75 p-values (calculated from Pearson r via Fisher-z transform), at α = 0.10.
Classification caveat. Stats NZ uses ANZSIC 2006 classification. “Mining” includes oil and gas extraction. “Electricity, gas, water and waste services” is a single ANZSIC class that blends distinct activities. Productivity measures at this level of aggregation will reflect sub-industry shifts that cannot be identified without finer-grained data.
Known limitations.
- 47 annual observations is thin for a lead-lag test with multiple lag orders. Power is low. The correlation findings should be treated as exploratory.
- Productivity is affected by many factors beyond electricity — capital investment, exchange rates, global commodity prices, trade policy, firm-level entry and exit. This article describes an association, not a causal relationship.
- “Measured sector” coverage in PRD_PRD_001 begins in 1996, not 1978. The aggregate labour-productivity comparison uses the longer “Goods-producing” series where possible.
- Stats NZ revises productivity statistics annually. Figures reported here are as-of the 2024-release vintage. Historical numbers may shift in future releases.
Your turn
Does the “Māui windfall” reading fit what you’ve seen on the ground in Taranaki, in manufacturing, or in the electricity market? Are there sectors or facility-level stories this aggregate view misses? Add your thoughts 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