Reform Is an Answer. Nobody Checked the Question.

In short: reform will change the shape of local government. It won't change how quickly it moves, because what slows a council down is the gap between the call that's rational for the person deciding and the call that's best for the organisation. Redrawing boundaries doesn't reach that gap — and it matters more to how a council performs than shape does.

Reaction across the sector has been mixed. But notice what almost nobody is saying: that local government is working well and should be left alone. Everyone agrees there's a problem. The argument is only about the fix.

I have no view on whether amalgamation is the right fix. It might produce a better local government; it might produce a worse one. Whether your council should merge, and with whom, is a decision for elected members with a mandate to make it — not mine to have an opinion on. My question is narrower: whichever way it goes, will it have dealt with the problem everyone agrees is there?

Treating the wrong cause?

The performance problem is real, and it's measurable. Councils cost more to run than they used to, and residents aren't getting noticeably more for it.

Take what councils spend on staff per resident, adjusted for salary inflation. It has risen substantially over the past two decades — depending on the council, somewhere between a half and a doubling.

This isn't what happened everywhere. Over the same period the wider New Zealand services sector lifted its labour productivity by about a third — getting more output from the people it already had. Local government moved the other way: the number of residents each council employee serves fell to about three-quarters of what it was.

The obvious explanation is compliance, and it deserves to be taken seriously. Councils have been handed a great deal of new statutory work, much of it unfunded, and a good deal of it genuinely protective — drinking water is measurably safer than it was, and that is worth having whether or not a single resident notices. Anyone arguing the cost growth is purely self-inflicted isn't paying attention.

But the arithmetic doesn't work. The rest of the services sector absorbed two decades of new regulation as well — health and safety, employment law, privacy, financial reporting — and still found a third more output from the same people. Local government went backwards by a comparable margin. For compliance alone to explain a swing that size, the burden falling on councils would have to dwarf what the rest of that sector carried over the same period.

There's a second point, and it matters more. Being handed more work tells you nothing about how well an organisation absorbs it. Two councils handed the same obligation won't carry it at the same cost. Compliance explains the extra work. It doesn't explain the difference in cost.

So the concern driving reform is legitimate. Less settled is the explanation — and the reform has already assumed one.

Someone already asked whether size matters

The reform's implicit theory is that this is a problem of size and fragmentation: too many small councils, too much duplication, insufficient scale. It's a reasonable hypothesis, and it has been tested more than once — with results that don't support it.

New Zealand's own Infrastructure Commission went looking for exactly this effect. Its 2022 report Does Size Matter? examined whether larger councils are more cost-efficient and found that size has no bearing on it — including in the areas where you'd most expect scale to help, like road maintenance and building consents.

Denmark ran the experiment at national scale. Its 2007 reform merged 239 municipalities into 66, with roughly thirty left unchanged. The finding, published in the American Political Science Review, was a null net effect: savings in some areas were offset by deterioration in others, and for most services jurisdiction size didn't matter at all.

That's not evidence that merging saves nothing. Administrative costs did fall in Denmark. The point is that the gains appeared in one place and the losses in another, and they roughly cancelled.

You can't restructure an incentive

There's a deeper reason structural change underdelivers, and it isn't about cost at all. The behaviour that slows a council down is a rational response to what the organisation rewards — and reorganising doesn't change what it rewards.

Consider a manager who has the delegated authority to approve something and already knows the right answer. They convene three colleagues to confirm it anyway. This is extremely common, and it isn't timidity.

If they make the call alone and it goes badly, the consequence is specific and personal — their name is on it. Hold the meeting instead and the decision takes another fortnight and four people's time. That cost is entirely real, but it's spread thinly across a resident who waited and a colleague who chased it. It belongs to nobody, and it will never appear in anyone's performance conversation.

Being wrong is visible and attributable. Being slow is invisible and ownerless. Faced with that asymmetry, convening the meeting is the sensible move.

None of this applies to the complex and sensitive calls — a contentious RMA consent, a major capital commitment, anything with genuinely irreversible consequences. Those need considered deliberation, and rushing them would be its own kind of failure. What I'm describing plays out on the routine calls: approvals, variations, exceptions, the things that fill a working week. Individually none of them matter much, which is exactly why nobody examines them. Collectively they are where most of the organisation's capacity goes.

A second bias is stacked on that asymmetry, and this one is human rather than organisational. People weigh a potential loss more heavily than an equivalent gain — among the most reliably demonstrated findings in behavioural science. So even a manager reading the risks accurately will feel the downside of a wrong call more sharply than the upside of a fast one.

A well-designed system would correct for that. Most do the opposite: they add checks, which is to say they add weight to the side that was already over-weighted. The result is an organisation that quietly prices risk above performance — not by decision, but by accumulation.

This has been measured. Researchers surveyed 950 managers across all levels of a large public administration organisation, asking about the decisions that had mattered most to them over the previous year. Around a quarter of the most important decisions had been to protect the decision-maker rather than serve the organisation. Not through weakness — because the system had made the worse option the safer one to hold.

Now ask what a merger does to that. New reporting lines don't change the calculation. Centralising a function doesn't change it. If anything a larger organisation makes it worse: more people who could plausibly have made the call, more diffuse accountability, more distance between a decision and its consequence. You can redraw every box on the page, and the person deciding whether to check with someone first will do the same sum on the Monday afterwards.

The fuller version of this argument, with the rest of the sector evidence, sits in Many Arguments, One Underlying Cause.

We measured the shape, not the cost

Questions aren't neutral. What you ask determines what you measure, and what you measure determines what anyone can see — which is why the most expensive thing in a council may also be the thing least visible to the process reshaping it.

Look at the data assembled to inform this reform. Population per council. Rating base. Debt headroom. Back-office duplication. Boundaries. Scale economies. Every one of them is a measure of shape. All carefully gathered, all genuinely relevant — to the question that was asked.

Now try to find the number for what a routine operational decision actually costs to make. Elapsed time is measured in places — statutory clocks, service targets on consents and requests. What is almost never measured is the effort: the total staff hours a decision consumes before it's settled and the work can move. It certainly doesn't appear in any Service Level definition.

That's not an oversight. It's a direct consequence of the question. We collected structural data because we asked a structural question, and we didn't collect the cost of deciding because nobody asked. The familiar line — you can't manage what you don't measure — has a step in front of it that gets said far less often: you don't measure what you never thought to ask about.

The questions that aren't being asked

There are two questions available, and both can be asked in parallel while the structural reform question is worked through.

They're operational questions, but their reach extends well beyond anything a structural reform will deliver. And they cut to the heart of what prompted the Government's wider reform programme across the public sector: performance. The questions are how often is the rational call for the person deciding not the best call for the organisation? — and where would we catch that gap, aside from the bottom line?

This isn't an argument for pausing the reform. Whether your council ends up merged, and with whom, is running on its own timetable, and it sits with governance, as it should. Those two operational questions are governance's to ask, and the organisation's to answer.

Put them to your leaders — genuinely, not as a challenge. The honest answer is that almost nobody currently knows, and that isn't anyone's fault. The follow-up, though, is checkable. How many of our approval steps have ever changed the answer? Does anything we measure notice how many hours of staff meetings it took to agree a playground plan? What happens when we miss a project deadline?

No reform business case asks any of them. All will still be unanswered whatever shape the organisation ends up with.

The reform will settle the structural question, one way or another. The operational ones will still be waiting. Nobody is required to ask them, which is exactly why they're worth asking now.

If you're curious where this sits in your own organisation, the Friction Factor survey is a fast, free first read.