Aotearoa New Zealand’s government is attempting one of the country’s largest public service reforms in decades – and betting artificial intelligence (AI) can help offset thousands of planned job cuts.
By any measure, the reforms announced by Finance Minister Nicola Willis and Prime Minister Christopher Luxon this week are wide-ranging. They would cut some 8,700 roles, merge departments and rapidly embed AI across government, making its use a basic expectation.
All of this begs several important questions.
To what extent is this planned overhaul a symptom of a new governing logic? How effective might these initiatives be? And what can be expected for the public service in the near to mid future?
A familiar strategy
Far from being a novel idea, the reforms tap into a tried and tested classic for public services: restructuring. It has been a recurring feature of public service change in Aotearoa and its appeal to governments has rarely wavered.
As research by School of Government PhD graduate Annika Naschitzki shows, the country’s public sector underwent 484 separate restructuring initiatives between 2018 and 2021 alone.
The latest reforms are not particularly new, even by this coalition government’s own standard.
Public service job cuts were a key part of its initial 100-day action plan, even if the most recent workforce data showed approximately 500 more full-time equivalent roles than in 2023.
Many of these newer roles also differ from traditional perceptions of the public service. They are concentrated in support functions such as information and communications technology, legal services, human resources, procurement, finance and management – or what our colleague Karl Lofgren has described as a “new public bureaucracy”. By contrast, only 5.7% of public service workers are policy analysts.
Other commentators have argued there has been little significant long-term shift in New Zealand’s broader public sector structure since the reforms of the 1980s, concluding: “plus ça change, plus c’est la même chose” – the more things change, the more they stay the same.
The logic underpinning the reforms is also far from unique to New Zealand. Only a few months ago, for example, the Victorian government in Australia announced public service job cuts aimed at saving A$4 billion.
In New Zealand’s case, the reforms are also being framed as a return to a smaller pre-pandemic public service, with its target of 55,000 full-time equivalent roles equating to around 1% of the population.
The historical basis for this 1% figure, however, is at best unclear and does not appear to reflect the country’s administrative experience since the 1890s.
In 1985, for instance, when the population was 3.3 million, the public service stood as 88,000 full-time equivalent roles.
In 1995, the public service had a far smaller number of employees, with only 35,000 full-time equivalent roles, even though the population had only shifted marginally to 3.67 million.
The AI productivity gamble
The logic behind the government’s AI push also requires closer scrutiny, because there is still limited evidence about how it will improve public service delivery.
Increasingly, downsizing and digital transformation has morphed into a new discourse, in which AI is presented as the mechanism that makes a smaller public service administratively possible.
But there are many uncertainties. Some highly specific trials have demonstrated promising results in narrow contexts. But the idea that AI can simultaneously support workforce reductions, organisational mergers and improved public service performance rests on a series of assumptions that are highly questionable.
As well, the reforms are likely to involve a huge disruption to institutional memory. This means that, rather than a simple number of jobs, the public sector would lose institutional knowledge about how government functions – as it has been for the past three decades.
Further cuts to full-time staff will reduce in-house capability, while efforts to cut consultancy spending will reduce external expertise. Attempting to simultaneously restructure agencies, shrink the workforce and scale up AI could therefore create capability gaps, just as demand for specialised knowledge and oversight increases.
It might simply be decided that AI vendors become the new public service, with a full shift to dependence on these platforms.
This creates obvious risks. And right now, the public service still lacks the legislative, regulatory and governance foundations needed to support such a transition.
Efficiency is not the same as cost cutting
At this point, what the public service arguably requires is more internal capability, not less. If improved coordination, integration, productivity and efficiency are the objective, these latest plans are unlikely to deliver them.
Of course, public services should always be designed and delivered as efficiently and effectively as possible. But efficiency does not necessarily mean simple cost cutting.
As the head of the United Kingdom’s Office for Statistics Regulation, Ed Humpherson, recently argued, efficiency is not simply about the cost of inputs but also the quality of outputs:
Perhaps cost cutting makes for a simpler message than the more complex ‘efficiency can be more outputs for the same cost’ story.
Projected job losses naturally generate headlines. But the more important concern is the cost they will have on the quality and delivery of public services for New Zealanders.
