Cuts, Mergers and AI Won't Fix What's Actually Slowing the Public Sector

In short: the Government is pushing hard on public-sector productivity — roughly 9,000 fewer jobs by 2029, ministries merged into bigger entities, and AI drafted in to pick up the slack. Councils have been handed a parallel ultimatum: amalgamate or be amalgamated. Every one of these levers attacks cost or structure. None of them touches the thing that actually determines how much a public organisation gets done: how fast it can make a decision. Pull those levers without addressing decision friction and you get a leaner, consolidated organisation that is exactly as slow as before — sometimes slower.

This is one of several arguments behind the same diagnosis — the fuller case for improving council operational performance pulls them together in one place.

Three levers, all aimed at the same two things

Look past the announcements and the productivity drive comes down to three moves. Cut headcount — the core public service shrinking back toward 1% of the population, with the savings put at roughly $2.4 billion. Consolidate structure — separate ministries folded into larger ones. And automate — a standing bet that AI will lift efficiency across the board. Local government is being pointed down the same road, with councils told to bring forward amalgamation plans or have the decision made for them.

Notice what all three have in common. Two of them attack cost (fewer salaries, shared back-offices) and one attacks structure (fewer, larger entities). Both are real and legitimate targets. But neither is the same thing as output. An organisation's productivity isn't set by how many people it employs or how its org chart is drawn — it's set by how quickly those people can turn an input into a decision and move on. This idea is barely mentioned.

The variable nobody's pricing in

We've argued before that decision-making speed is the metric the public sector doesn't measure. It stays invisible because it never appears on a balance sheet, yet it drives most of what does. In a compliance-heavy environment, staff face a rational incentive: being seen to follow the correct process protects you, while moving fast and occasionally being wrong does not — even when the fast call would have served the public better. The result is defensive, slow decision-making. That is decision friction, and it is the real ceiling on public-sector output.

Here's the uncomfortable part: none of the three lowers that ceiling — and two of them can quietly raise it.

Fewer people, same queue

Cut headcount without changing how decisions get made and you haven't removed the work — you've removed the capacity to do it. The same approvals, sign-offs and defensive checks still have to happen; there are simply fewer people to route them through. A slow process with fewer hands is not a faster process. It's the same process with a longer queue. Productivity, measured honestly as value delivered per hour, can fall even as the wage bill drops.

Amalgamation redraws the chart, not the decision

Merging ministries or councils is a structural fix for a behavioural problem. A bigger unitary authority still runs on the same incentives that made the smaller ones cautious — and a larger organisation often adds coordination layers, not removes them. Reorganisation reshuffles who reports to whom; it does nothing, on its own, to change whether a manager feels safe making a judgement call without three rounds of cover. If the friction lived in the incentive structure, it survives the merger intact.

AI hits the human bottleneck faster

The AI bet is the most seductive and the most misread. AI is superb at the routine, high-volume work. But as we've written, it raises the floor without touching the ceiling: the moment a case needs genuine human judgement — exactly the cases a risk-averse culture is slowest on — AI hands it straight back to the same friction-heavy decision process. Faster processing just delivers you to the same bottleneck sooner.

Where the real lever is

There's a quieter finding worth sitting with. Recent NZIER work frames New Zealand's productivity problem less as a shortage of good ideas and more as a diffusion problem — the gap between the best-performing organisations and the median is unusually wide, and better practice doesn't spread. Decision-making speed is precisely that kind of under-diffused practice: rarely measured, so rarely improved, so it never travels.

That's the lever hiding in plain sight. Start by measuring how long decisions actually take and what they cost when they drag, then change the incentive so that a timely, sound call is rewarded rather than quietly punished. It is less visible than a headcount number and far more powerful — because it's the one change that makes every remaining person more productive rather than simply asking fewer of them to carry the same slow load.

Want to know where decision friction is quietly costing your organisation? The Friction Factor survey is a fast first read — or get in touch if you're a council or agency looking at this ahead of reform.