Public Sector Reform: The Performance Paradox and Pulling on Better Threads

02/12/2025

The Public Service Amendment Bill promises better performance through tighter control. This post argues that the diagnosis is incomplete: what looks like a solution may, in fact, deepen the pathology it claims to treat. Drawing on comparative public management scholarship and recent developments in both Wellington and Whitehall, the analysis shows why the centralising instinct has reached its limits, how the Simplifying Local Government proposals create a rare structural opening for principled localism, and what specific amendments to the Public Service Act might look like if Parliament wished to place devolution formally on the table as a tool for performance improvement rather than a risk to be managed.

The Public Service Amendment Bill is presented as a tool for improving performance. It seeks to sharpen the public service’s focus, clarify its purpose, and strengthen accountability to the government of the day.

In the intricate tukutuku panel of our public administration, these reforms are an attempt to tighten the weave, pulling the threads of control firmly towards the centre.

Tukutuku, of course, is not merely decorative. The patterns encode knowledge about relationships, about balance, about obligations, about how parts hold together across time and distance.

The logic of the Bill is intuitive: if we want better performance, we need tighter control. But what if this diagnosis is wrong? What if the relentless pull towards Pōneke is not the solution, but a core part of the problem?

This is the central paradox of the Bill: in the name of performance, it doubles down on a model of centralised control that public management scholarship has spent decades showing to be fraught with unintended consequences that undermine performance over time.

As I said in my first read of the policy papers, which are very well written, “excessive centralisation can pose risks to performance.” The Bill’s focus on top-down accountability overlooks a more potent driver of performance: enabling it from the ground up through a principled commitment to localism.

Understanding the Appeal

One must, however, understand why ministers have reached for these levers. The Public Service Act 2020 sent mixed signals. It celebrated itself, but its practical effect was to focus attention on new arrangements in Wellington: interdepartmental executive boards, functional chief executives, system leadership. The public service became skilled at describing its internal architecture whilst losing sight of what that architecture was supposed to deliver.

Furthermore, years of Performance Improvement Framework reviews tell a consistent story: agencies that perform well have clarity of role, clarity of strategy, and clarity about the outcomes they are pursuing. Where that coherence is absent, performance drifts.

Ministers, observing a system that seemed to have lost this clarity, have understandably concluded that reasserting control is the path back to performance. The instinct is sound. The logic is laid out clearly in the policy papers. But the diagnosis is incomplete. The missing insight is that clarity at the centre does not automatically produce clarity at the point of delivery, and that genuine performance improvement requires not just more transparent accountability upward, but potentially greater authority devolved downward.

The Illusion of Tighter Control

The Bill’s approach reflects a classic New Public Management belief in contractualism and managerial accountability. Yet, as Tom Christensen and Per Lægreid argued in their seminal 2001 analysis of NPM reforms in Aotearoa and Norway, this path is deceptive.

They found that while such reforms aim to increase political control, they often have the opposite effect. Structural devolution, separating functions into arms-length central government agencies, inherently “increases the distance between the political leadership and subordinate units,” thereby decreasing and diluting the influence of central political signals.

More fundamentally, the contractualist approach replaces a system based on professional ethics and mutual trust with one that, in their words, “potentially furthers distrust.” It assumes that the right set of external incentives and key performance indicators can guarantee performance. It cannot guarantee performance because it instead creates a trade-off between accountability and responsibility.

A manager can be held accountable for a narrow set of pre-defined targets, but this can lead to a “shrinkage of a sense of responsibility” for the complex, unquantifiable, and often more important aspects of public service. As Robert Gregory observed, “Efficiency is no guarantor of good political and social judgement, which is essential in securing genuine political responsibility and legitimacy” (1998, p. 536). The Bill risks creating a public service that is excellent at meeting its targets, but less capable of exercising the wise and careful judgment our country needs. A child protection system might hit every placement target whilst systematically failing to exercise judgment about the enduring importance of safety as well as whakapapa connections: precisely because the target measures the speed of placement, not its wisdom.

The Alternative: A System That Learns

So, for the sake of argument, let’s assume contractualism corrodes responsibility and flattens judgement. A credible alternative is a system that distributes authority closer to where knowledge and consequences sit, and that can therefore learn faster.

In their 2015 report In the Zone, Eric Crampton and Khyaati Acharya provide a useful articulation of this insight, though the insight itself is, of course, not new. See my Waitangi Tribunal Thursday writing for a parallel.

Crampton and Acharya argue that Aotearoa’s unitary state, for all its tidiness, “does not handle change well.” Because policy change is typically rolled out nationwide, the stakes are incredibly high. If it fails, the whole country wears the cost, and the government risks its future. This makes our system inherently risk-averse and stifles innovation.

Crampton and Acharya propose a localist alternative, using Special Economic Zones not as tax havens, but as policy laboratories. Their core observation is that learning requires variation. By allowing different regions (or localities, as I call them) to experiment with different rules and service delivery models, we can see what actually works. Their chapter ‘Failing Fast and Failing Slow’ captures this key insight: we must “fail fast” on a small scale, rather than “fail slow” on a national one. As they put it, “Changing policy in one region, and not in other regions, makes it easier to tell whether the policy has succeeded.”

This approach turns regions from passive recipients of central policy into active participants in a local and national learning system. The devolution I am arguing for here is not the arm’s-length fragmentation of NPM to state agents further and further away from political control, but the principled delegation of functions and powers closer to communities, potentially along the lines proposed in the Simplifying Local Government set of ideas, but within a clear constitutional frame set by Cabinet itself, to be overseen by the Public Service Commission in their institutional design and assurance role.

The question remains: what makes localised decision-making trustworthy? Crampton and Acharya offer competitive pressure and the discipline of exit: the possibility that the Crown might disinvest from what does not work. In practice, that “competition” in a unitary state like ours is less about being brave enough to disinvest than about the Crown having real options to redirect investment and delivery models when evidence and communities show that something is not working.

I would add that our geography and scale add a second discipline: relational knowledge. This kind of knowledge comes from proximity to consequences, and from having to look one’s community in the face. Decision-makers bound by relationships that precede and outlast any particular government are accountable in ways that distant officials cannot be. But accountability for what? This is where the logic must be completed. Local accountability should be for outcomes and learning: not for compliance with processes designed elsewhere. The combination of competitive discipline, relational accountability, and a focus on what actually works creates the conditions for a system that genuinely learns.

The Hard Work of Letting Go

This is not a call for a chaotic free-for-all. Devolution is difficult. A 2025 report from the United Kingdom’s Institute for Government on the challenges of implementing local government reorganisation and devolution simultaneously, in what they term “dual delivery”, provides a sober reminder of the complexities. The research concludes that such reforms are “complex and time-consuming,” and that the “risk of collapse or delay cannot be ruled out.” Poor implementation, they warn, can lead to the anticipated benefits failing to materialise.

Crucially, their work highlights that success depends on what they call “strong and trusting relationships at both political and official levels.” This echoes the foreword to Crampton and Acharya’s report, where former LGNZ CEO Malcolm Alexander identified the “historic lack of trust between local and central government” as a key barrier to progress in Aotearoa.

But we must be careful about causality here. As others have observed, when incentives between central and local government work at cross-purposes, poor relationships are often the consequence rather than the cause. Councils whose structural position makes compliance irrational will resist; central government, frustrated that its signals are not heeded, will reach for tighter control. The mistrust is real, but it is downstream. Getting the structures right makes it more likely that collaborative, learning cultures develop.

A turn to localism, therefore, cannot simply be a plea for better relationships. It must begin with a structural change that makes trust rational: a framework within which cooperation serves both parties’ interests.

This insight finds striking confirmation in recent work from the United Kingdom. In a July 2025 guest paper for the Institute for Government, Peter Makeham, a former director general in the Department for Education and the Home Office, offers a diagnosis of Whitehall that reads as an uncanny parallel to our own situation. Makeham argues that the present departmental structure at the heart of government is based solely on the needs of government itself, with a command-and-control attitude that focuses on what government wants rather than what people and communities need. The allocation of policy responsibilities between departments, he observes, does not align with the complex, interconnected systems of society and the economy. Those systems are not owned by government; central government departments often fail to understand them or engage with them properly.

His proposed solution is worth noting: decisions and resources should move from central government departments to the frontline and to places. This is not a reduction in the importance of government, but a fundamental change in how it operates: enabling decision-making by local communities and frontline professionals, supported by good performance measures and data. Broad government-set outcomes would enable more devolved decisions that better match personal and local needs.

The parallel is instructive. What we are witnessing is not a peculiarly Aotearoa problem, but potentially a Westminster-wide recognition that the centralising instinct of the past several decades has reached its limits. The question is not whether to devolve, but how to do so responsibly.

A Structural Opening: Simplifying Local Government

It is here that last week’s announcements became significant. On 25 November 2025, the Government released its Simplifying Local Government proposals for public consultation, described by Ministers Bishop and Watts as the most significant changes to local government since 1989.

The proposals would replace elected regional councillors with Combined Territories Boards made up of mayors from each region’s city and district councils. Each CTB would then be asked to prepare a regional reorganisation plan within two years, assessing how councils can best work together to deliver efficient and effective local infrastructure, public services, and regulatory functions.

The Government’s stated rationale is to cut duplication, reduce costs, and streamline decision-making. Critics have raised legitimate concerns about democratic accountability and the potential loss of regional voice. These are serious matters deserving careful attention. But for this analysis, what matters is the structural opportunity the proposals create.

If implemented, Aotearoa would have a new tier of regional governance: Combined Territories Boards, led by mayors directly elected by their communities, that could serve as a receiving structure for devolved authority from the public service. The regional reorganisation plans would provide a mechanism for determining which functions might be better exercised locally, which require regional coordination, and which must remain with central government.

Weaving a Stronger Panel

This is where the Public Service Amendment Bill could make a profound contribution. Instead of simply tightening the central weave, it could provide the legislative framework for a responsible and principled localism: one that works in concert with the Government’s local government reforms.

The Simplifying Local Government proposals would reorganise the receiving end: local and regional governance capable of accepting devolved authority. What is missing is a corresponding provision on the public service side: a formal mechanism enabling chief executives and ministers to direct them to devolve functions and powers to these new arrangements.

I therefore propose an amendment to Schedule 6 of the Public Service Act to do just that. After the existing delegation provisions, the following clauses would create the necessary framework:

2A Purpose of devolution provisions

(1) The purpose of clauses 2B to 2D is to improve the performance, responsiveness, and effectiveness of the public service by formally enabling and encouraging the devolution of functions and powers.

(2) These clauses are based on the principle that specific functions and powers may, where appropriate, be exercised at the level closest to a problem, where the best information exists, and the impact of the action is most directly felt.

(3) These clauses acknowledge that excessive centralisation can pose risks to performance and that the devolution of functions and powers is a tool for managing that risk.

2B Power to devolve functions and powers

(1) A public service chief executive may, in the performance of their role, delegate or otherwise provide for the exercise of their functions and powers—including financial powers—to other levels within their agency or to appropriate external partners in order to improve performance.

(2) In deciding whether to devolve a function or power, a chief executive may have regard to whether the performance benefits of devolution outweigh the assessed risks.

(3) A responsible minister may, through a letter of expectations or similar instrument, direct a chief executive to give particular regard to devolution options when exercising their functions under this clause as a means of improving performance and responsiveness.

(4) This power is in addition to, and does not limit, the delegation of powers in clause 2.

The effect is to place devolution formally on the table as a performance-improvement tool, available to chief executives exercising professional judgement, and to ministers who wish to direct their agencies toward more localised delivery models.

It creates neither obligation nor constraint, but permission and encouragement. The Public Service Commissioner could issue guidance on responsible devolution under existing powers, and provide the capability-building framework that the United Kingdom experience shows to be essential.

Together with the Simplifying Local Government proposals, this amendment would create a coherent framework for principled localism. One reorganises who can receive devolved authority; the other enables the public service to extend it. It provides the means to manage the risks identified by Christensen and Lægreid by building a system based on earned trust and shared responsibility, not merely contracts.

It provides a legislative vehicle for experimentation and learning: a goal many have advocated for a long time. And it acknowledges the complex realities of implementation highlighted by the United Kingdom experience, by creating a framework for building capability and managing the process of letting go. It also speaks directly to Iwi and Hapū’s desire to see public policy shaped by local conditions, with the communities that have to carry the burden of the final decisions¹.

The choice before Parliament is not between control and chaos. It is between the illusion of control offered by a rigid, centralised system, and the genuine resilience of an adaptive, intelligent, and localised one.

A tukutuku panel gains its strength and beauty not from a single, central point of tension, but from the balanced interplay of threads anchored across the entire frame: each strand bearing weight, each connection holding the others in place. By amending the Public Service Amendment Bill to enable localism formally, we can begin to weave a stronger, more responsive, and more effective public service for all of Aotearoa, honouring both our constitutional commitments and the practical realities of delivery.

References

Christensen, T., & Lægreid, P. (2001). New public management: The effects of contractualism and devolution on political control. Public Management Review, 3(1), 73–94. https://doi.org/10.1080/14616670010009469

Crampton, E., & Acharya, K. (2015). In the zone: Creating a toolbox for regional prosperity. The New Zealand Initiative.

Department of Internal Affairs. (2025). Simplifying Local Government: A draft proposal. New Zealand Government.

Fright, M., & Routley, S. (2025). Dual delivery: How can areas successfully reorganise local government and implement devolution at the same time? Institute for Government.

Gregory, R. (1998). Political responsibility for bureaucratic incompetence: Tragedy at Cave Creek. Public Administration, 76(3), 519–538. https://doi.org/10.1111/1467-9299.00115

Makeham, P. (2025). Rewiring government. Institute for Government.

¹ I should acknowledge a tension I am still working through, both in my own thinking and with clients navigating these questions in practice. Devolution, as framed here, the delegation of state functions to local receiving structures, is not the same as genuine power-sharing as imagined under Te Tiriti o Waitangi. Delegation keeps authority with the Crown; it merely permits others to exercise it. Rangatiratanga, by contrast, is not a gift of the Crown to be extended or withdrawn at ministerial discretion. Combined Territories Boards composed of mayors are not, and cannot be, vehicles for the exercise of rangatiratanga; they are creatures of statute, accountable upward to the Crown, not laterally to iwi and hapū as treaty partners. And yet a system in which all significant decisions flow through Wellington, via arms-length entities and companies, serves Iwi and Hapū interests even less: for currently they structurally discount local knowledge, whakapapa connections, and place-based relationships. The proposed amendments do not resolve this constitutional question. They may, however, create conditions more hospitable to its resolution: a public service in a narrow role, habituated to sharing authority and experienced in the disciplines of learning and not making work. Whether that is sufficient, or whether it risks becoming yet another mechanism through which the Crown retains control and demands compliance whilst appearing to share it, is something I continue to think about. I welcome correspondence from readers grappling with similar questions.