Schumpeter comes to Wellington

(And what we can learn from the Luddites)

In 1987 Telecom New Zealand employed about 25,000 people. By 1997 it employed under 8,000. A single corporation shed 17,000 jobs in a decade, in a country of 3.3 million. The cost of Telecom’s long-distance calls fell by 60 per cent between 1987 and 1992. The decade that followed was, on the New Zealand Productivity Commission’s assessment, a period of historically high labour productivity growth.

Contrast Telecom’s headcount reduction with last week’s pre-Budget announcement proposing cuts to the public service. Finance Minister Nicola Willis’s proposal will reduce the core public service from 63,000 to 55,000 full-time equivalent positions by mid-2029 – 8,700 over four years, half the Telecom reduction, across an entire workforce in a country half as large again. The proposal merely returns the workforce to roughly its 2017 share of population. Yet the response treated the plan as catastrophic.

Duane Leo, national secretary of the Public Service Association, called the announcement an act of wilful destruction. Camilla Belich, Labour’s public services spokesperson, said public servants and their families would be “paying for the government’s budget.” Francois Febvre, a Wellington hospitality operator, called the announcement “bad news on top of bad news.” Andrew Little, the Mayor of Wellington and former Labour leader, said the cuts would spread anxiety as people wondered “when the axe is going to fall.”

Losing a job is a serious personal and family blow. The 8,700 positions are no exception. But the reactions themselves sit in a long tradition of similar reactions to similar adjustments. They count the jobs that will be lost. They cannot count the jobs that will be created in their place. The tradition has a name, an explanation and a record of having been wrong every time.

What the Luddites could not see

Joseph Schumpeter, writing in Capitalism, Socialism and Democracy in 1942, gave the modern name to what every wave of economic adjustment is fundamentally about. He called it creative destruction. He meant the phrase exactly: the destruction is real, the creation is real and the two are inseparable. New firms displace old ones. New technologies displace old ones. Labour and capital are continually released from less productive uses into more productive ones. The country that exists at any given moment conceals the country that is being formed within it.

Schumpeter did not invent the insight. Frédéric Bastiat, in his last essay Ce qu’on voit et ce qu’on ne voit pas published in 1850, argued that the difference between a good economist and a bad one is whether he sees only the visible effect or also the effects that follow but are not yet visible.

Nobel laureate Friedrich Hayek, in The Use of Knowledge in Society published in 1945, added the further point. The knowledge needed to redirect resources from less productive to more productive uses is not held by any central planner. It is dispersed across millions of producers, employers and workers, each of whom knows something specific about their own circumstances. When Telecom shed 17,000 workers between 1987 and 1997, no planner could have specified where they would go. The economy did not work through a master spreadsheet. The workers worked it out themselves, one by one, in conditions a central authority could not have anticipated.

These three ideas together account for the visible asymmetry of the current debate. The losses are concentrated in a known place, in a known set of buildings, with a single union to speak for them. The gains will be dispersed, over years, through thousands of separate hiring decisions in firms that have not yet identified the workers they need. There is no spokesperson for the developer in Tauranga who will hire her first project manager because a qualified one has become available. There is no union for the technology firm in Christchurch that will fill a data analyst position from the released supply. The architecture of public attention favours the visible side, and policy all too often follows.

The most famous example of this asymmetry came two centuries ago.

Between 1811 and 1816 the textile towns of Nottinghamshire, Yorkshire and Lancashire were torn by riots. Workers organising under the banner of the mythical Ned Ludd smashed the new stocking frames and power looms that were destroying handloom weaving as a trade. The distress of the “Luddites” (as they came to be known) was genuine: the new mills paid less, demanded longer hours and required skills different from those the weavers had spent their lives learning. Economist David Ricardo, in the third edition of his Principles of Political Economy and Taxation in 1821, added a new chapter titled “On Machinery” in which he conceded that the introduction of machinery could be injurious to the labouring class. It was one of the most famous reversals in economic thought.

What Ricardo did not concede, and what the Luddites could not have known in 1816, was the size of the country that came after. Real wages in industrial Britain roughly doubled between 1820 and 1900. Standards of living unimaginable to a Nottinghamshire stocking weaver became routine for his great-grandchildren. The mills did not stop. The displaced trades did not return. But the country into which the workers and their descendants moved was wealthier, healthier and freer than any that had preceded it.

Two centuries after Ned Ludd, the same argument is being made about the same kind of adjustment. Mr Leo, calling the cuts an act of wilful destruction, said that linking the use of artificial intelligence to a headcount target turns technology into a threat rather than a tool. The line is not new. The power looms were a threat rather than a tool. The motor cars were a threat rather than a tool. So were the personal computers, the spreadsheets, the email systems and every previous wave that displaced an existing way of working. Each generation finds the case persuasive until the country on the other side comes into view.

The layer that swelled

The Schumpeter argument explains why a national economy absorbs displaced labour over time. It does not by itself explain why a particular set of cuts is a sensible adjustment. A reasonable sceptic might accept the general principle and still ask why these workers should be released, and whether the work they did was valuable. The structure of the New Zealand public service over the past seven years answers that question.

Between 2017 and 2023, the core public service grew from 47,252 full-time equivalents to over 63,000 – a 34 per cent expansion in six years, against a labour force that grew by under 10 per cent in the same period. The growth did not occur where voters might have expected. As I argued in The austerity-proof public service earlier this year, managers rose by over 50 per cent, policy analysts by a similar figure and information professionals by 73 per cent. Frontline occupations shrank as a share of the workforce. The Ministry of Business, Innovation and Employment grew from 3,366 staff to more than 6,200.

The New Zealand Initiative report Unscrambling Government, which Jemma Stevenson and I published in September 2025, mapped the disorder in the executive. New Zealand operates with more than 80 ministerial portfolios across 28 ministers and 43 departments – three times as many portfolios and nearly twice as many departments as comparable nations such as Ireland, Norway and Singapore. MBIE reports to 20 different ministers. No one planned the ministerial proliferation; the structure accumulated through decisions that each made local sense but produced collective incoherence.

In Hidden Architecture, I set out the deeper reason the public service grew without delivering. When authority and consequences sit in different places – when those who make decisions do not bear the cost of them, and those who bear the cost cannot influence the decisions – failure becomes structural rather than accidental. The decisions about the size of the back office sat with the back office itself. The cost fell on taxpayers and on the frontline services that competed for the same fiscal envelope.

The cuts announced last month are properly targeted at the layer that swelled. The frontline agencies – health, education, police, defence, corrections and the intelligence services – are explicitly excluded from the savings exercise. The reductions fall on the back-office ministries where the 34 per cent growth was concentrated.

The AI element of last month’s announcement is the mechanism that makes the targeting achievable. The plan asks public service agencies to embed artificial intelligence as a basic expectation across customer-facing and back-office systems, and identifies routine cognitive work as the layer where productivity gains are concentrated: first-draft policy memos, basic procurement analysis, query-line responses, compliance form processing and the dozens of similar functions that consume time at the analyst grade. The very AI that Mr Leo calls a threat rather than a tool is part of what makes the 8,700 reduction achievable without touching the frontline.

A back-office analyst whose work is absorbed by a language model in 2027 is, in aggregate, a nurse who can be hired or a police officer whose post can be funded. Schumpeter’s process is at work inside the announcement, not only on the horizon.

The workers who will leave are not poorly equipped for the rest of the economy. The growth between 2017 and 2023 was overwhelmingly in managers, policy analysts and information professionals, whose skills transfer cleanly to consulting, professional services, banking, technology and corporate strategy. The match between the workers being released and the firms seeking to hire them is closer than any in recent reform history.

Reducing headcount is one thing. Repairing the structural disorder that allowed it to grow so rapidly is another thing entirely.

The new Ministry for Cities, Environment, Resources and Transport, due to stand up in July, consolidates housing, environment, transport and the local-government functions of internal affairs. The departmental merger is welcome. But without corresponding consolidation of the ministerial portfolios above it, MCERT risks repeating the MBIE mistake – one department reporting to a dozen masters, its energy consumed by managing the overlap. Unless the merger is followed by the consolidation of ministerial portfolios, the disorder identified in Unscrambling Government will remain.

Wellington and after

The 1990s Telecom story sat within a wider restructuring of the New Zealand state in which most of the contraction was reallocation rather than redundancy: functions moved from the core public service into corporatised entities or onto the open market, where the work continued and often improved. The Treasury study by Black, Guy and McLellan, published in 2003, found that average multifactor productivity growth in the measured sector rose from 0.09 per cent per annum between 1988 and 1993 to 1.32 per cent per annum between 1993 and 2002. As Bryce Wilkinson showed in his analysis of subsequent Statistics New Zealand data, communications services alone grew labour productivity at 11 per cent a year for the 20 years from 1988 to 2008.

The displaced workers moved into an economy that became, over a decade, wealthier than the one before. The political backlash at the time reflected real distress; the country on the other side was unmistakably better.

Last month’s adjustment is smaller by an order of magnitude than the reforms of the 1980s and 90s. The frontline is protected. The cuts fall on the layer that grew under the last government without delivering. The labour market into which the workers will move is tighter than the labour market of 1991, when unemployment briefly spiked above 10 per cent and the country took several years to absorb the released supply. The match between the skills being released and the firms seeking them is closer than the match then.

The 8,700 public service positions due to be lost by mid-2029 are real. They belong to people with names, incomes, mortgages and families. Their distress will be real. None of that is in doubt. And Mayor Little is right that Wellington will feel the adjustment first.

But Little speaks for Wellington as it is. Someone has to speak for the country as it could be. That country is being formed within this one now. It will appear in the data 15 years from now, in firms, sectors and roles that nobody has yet named. Joseph Schumpeter knew it would, Frédéric Bastiat knew it would and the New Zealand experience of the 1990s confirmed it on a scale several times larger than the adjustment announced last month.

The visible cost arrives first. The country that grows from it arrives later. Confusing the timing for proof is the oldest mistake in this argument, and the most expensive.