Why public financial management needs to shift from compliance to productivity

When I look at news on social media and read reports from public bodies, I see a constant drip of failures in sectors that seem to have nothing to do with each other. One day it's the HS2 project cutting sections because costs spiralled; another day it's the Ministry of Defence admitting its programmes aren't delivering promised capability; the next, you read about councils with unaudited accounts going back years, enthusiasm for using artificial intelligence, or the rush towards digital transformation. These issues seemed like isolated problems: lack of money here, poor management there, departmental silos, etc., the same old excuses… until a few days ago I listened to Gareth Davies, the head of the National Audit Office (NAO). In his speech, he wasn't listing errors; he was describing a "systemic disease" cutting across all sectors. It left me thinking that what connects a half-built hospital with AI software that doesn't work is a shared blind spot: government (and by extension, many governments) knows how much budget it has spent, but has no idea what it actually costs to produce a result. In both cases, they're flying blind because the system is designed to ensure budgets aren't exceeded, not to manage efficiency.

1. The counter-intuitive diagnosis

Davies mentions that the problem isn't a lack of information or willingness to do things properly. Government is full of data about how much is spent down to the last penny. The problem is the architecture, it doesn't organise data to make sense and serve decision-making; it only provides budgetary control, and therefore lacks understanding of what real value is being obtained for each pound invested.

He says it in these exact words:

"Government systems are too often set up to control spending within budgets, not to help managers manage"

It's an incredible paradox, especially when you consider that the United Kingdom is a world leader in financial reporting (IFRS standards and complex national accounts). They're excellent at "compliance"; they know perfectly well which budget line each pound falls into, but they suffer from absolute management blindness. Public managers know they haven't exceeded the budget, but they have no information about the real cost of their activity to inform decisions.

Consider these two examples from the speech:

Fees and charges: On paper, there's a policy that they should cover their costs, meaning government shouldn't make or lose money from the taxpayer. So, for example, Government says the price of a passport should cover exactly what it costs to produce it. The problem is that, because they don't know their real costs, Government charges you £50 when it actually costs £80 to produce. That £30 difference doesn't disappear; it becomes an "invisible transfer" and unplanned draw from general taxation, ultimately paid by the taxpayer. In the end, the taxpayer ends up paying twice: when buying the passport and when their taxes, which should go to a hospital, are used to plug that hole nobody saw coming.

Co-Pilot licences: Something similar is happening with artificial intelligence. Government is purchasing licences for tools such as Microsoft Copilot at significant scale, attracted by the promise of a “productivity revolution”. But even when pilots and evaluations are in place, the harder question remains: what counts as real value-for-money? If institutions do not have a robust baseline for how work is currently done, including time, cost, quality, and rework across entire workflows, it becomes difficult to tell whether AI is reducing costs, improving outcomes, or simply adding a new layer of spend. The risk is not only paying for technology licences, but paying for them while preserving the same slow processes underneath, with benefits that are visible in usage metrics yet difficult to prove in operational terms.

2. How this manifests across sectors

This architectural flaw manifests in every corner of government, reaching every sector:

In Mega-projects and Defence: Davies cites HS2 or Ministry of Defence programmes. New governance frameworks are created, but the fundamental is ignored: robust tracking of unit costs. Without financial management that makes visible the real cost of "capability", more money won't help at all. So it's not lack of budget; it's that, under the current design, government only sees the money that "goes out" today, but is blind to what that money is building (or destroying) for the future.

In Digital Transformation: The same thing happens. Police use AI tools to redact sensitive information in videos and police recordings that have the potential to save up to 11,000 working days per month if implemented nationally. Sounds good, but Davies warns that not all investments will bear fruit. Without a cost architecture, technological fascination becomes spending without discipline.

And then there's Local Government: There's a backlog of unaudited accounts going back years being closed with "disclaimed" opinions. By closing them this way, basically the auditor says: "I cannot vouch for these numbers". Government admits it has lost track of billions of pounds. It's not that councils are useless; it's that the oversight system has collapsed. There's lots of movement and lots of spending, but nobody has a torch to see whether that money actually turned into services or was wasted along the way.

Whilst listening to Davies, I recognised this "systemic disease" in my recent analysis on hospice financial sustainability presented to Parliament¹. The pattern is the same:

• Flying blind: Just as in major Government projects, in hospices there's "limited visibility", there's no national view of what the service actually costs, generating total disconnection between those with legal responsibility and those operating day-to-day.
• The invisible transfer of risk: Davies talks about how government moves costs from one place to another. In my analysis, I saw how Government depends on charity to absorb financial shocks, and this doesn't solve the problem, it only displaces risk towards individual providers, leaving the system constantly on a tightrope.
• The solution isn't just "more money": Just as Davies proposes for Defence or AI, my conclusion was that hospices need to be governed as a complete system. Budgetary patches aren't enough; what's needed is a delivery architecture where the NHS, Treasury and providers speak the same financial language.

The architectural flaw doesn't respect sectoral boundaries.

3. Proof of concept: when the architecture works

So that it doesn't seem we're facing an impossible problem, Davies gives us the star case: the Environment Agency.

In 2020, they were in the "penalty box" of audits because they couldn't value their flood defences as their data was a mess. New leadership didn't just seek the auditor's signature; they made asset data quality a strategic priority, so they cleaned up the information architecture.

The result? In 2024 they achieved a clean audit, but the real revelation was that they could now direct their finite resources much more efficiently. Having real data, they've protected more people, homes and businesses with the same money.

The equation would be:

Better data = Better asset management = Better results

This is the equation of an architecture that works.

Another example is the Border Change Delivery Unit at HMRC. There, finance teams work as business partners, coaching and upskilling budget holders, building financial accountability across the organisation. The result was better planning, better forecasting and better operational decisions.

As observed, the success pattern is consistent: when you integrate financial expertise into operational decisions, have real cost data and have leadership commitment, the consequence is a system that works.

4. The four pillars of correct architecture

In his speech, Davies gives us four principles which, if we reframe them, are the components of this new decision infrastructure:

1. Reporting for management (not just for compliance): Financial reporting cannot be an end in itself. Information is needed that serves to operate, not just to fill in compliance forms. The UK is already excellent at compliance reporting; the gap is in management information.

2. Management information vs. Budgetary control: If you don't understand the unit cost of what's being delivered, you won't know where the biggest bang for the buck is. This requires activity cost information, not just budget adherence tracking.

3. Integrated financial expertise: We don't need more accountants counting notes at the end of the month. What's needed are financial experts integrated into the operational decision-making process, as business partners. It's not about headcount; it's about expertise positioned correctly.

4. Leadership and Culture: The message that must come from the top is categorical: financial management isn't an office procedure for filling in forms; it's the dashboard that enables things to happen. If the leader sees finance as bureaucracy, they'll only care about not overspending; but if they see it as a tool, they'll use it to ensure each pound actually turns into better service for the citizen.

These aren't nice-to-haves, are infrastructure.

5. The risks of continuing to operate blind

Without this architecture, the challenges Davies sees on the horizon become exponentially more dangerous:

Defence increase: The Defence Investment Plan is eagerly awaited, but Davies is clear that without robust financial management, additional funds won't deliver the necessary capability. So it's not lack of budget; the risk is repeating the failures of HS2 and Ministry of Defence programmes on a larger scale.

AI adoption: Departments are identifying promising results, but Davies warns that AI comes with accuracy risks that must be managed to secure public trust. Without cost architecture, how can you distinguish high-value interventions from pure modernisation mirage? The risk of losing public trust is real.

Devolution: Government is moving resources from centre to regions and localities, but the local audit regime is broken (years of backlogs, accounts closed with disclaimed opinions). Davies notes this adds urgency to local audit improvement work. So the risk isn't just that money disperses; it's that, with the oversight system broken, nobody can see if those resources are being used for what was promised.

But these failures reveal something more structural than simple inefficiency. The current architecture is incapable of learning. I've explored this phenomenon from another perspective: how capital flows in sustainable systems². The pattern Davies identifies in public management is the same I see in dysfunctional capital architectures:

Immobilised capital: Davies uses HS2 to illustrate management blindness; it's a symptom of capital trapped in bureaucracy without generating real value. Money doesn't flow, it stagnates.

Capital that doesn't learn: Davies speaks of governments "flying blind". It's a system where money moves under constant monitoring but is incapable of adapting or generating knowledge about its own impact.

Towards regenerative architecture: Whilst Davies proposes efficient management, I propose going a step further. Information architecture shouldn't be just a control tool, but the system that allows capital to be regenerative, that learns from the environment and moves to the rhythm of society's real needs, not just budgetary cycles. British government today has an extractive architecture (spends resources) disguised as a control architecture (monitors spending). What it needs is regenerative architecture: where each pound invested generates information that improves the next decision… until the architecture learns to learn.

6. The necessary paradigm shift

In the end, what Davies is telling us is that the necessary change has nothing to do with what's always asked for. It's not about more resources, nor better political intentions, nor more technology in itself.

The real change isn't about budget, it's about mindset. We must stop seeing financial management as a simple compliance exercise (a "check" to see if rules are followed) and start seeing it as decision infrastructure. Think of it this way: moving from compliance to infrastructure is like stopping using financial data only to conduct autopsies on what's already happened, and starting to use them as the GPS that guides us in real time. Here it's not about filling in forms to keep the auditor happy; it's about building the foundations upon which you decide, with certainty, where to put each pound so it yields maximum returns.

The change is:

❌ From financial management as compliance ➡️ ✅ financial management as decision infrastructure

❌ From budgetary control ➡️ ✅ cost architecture

❌ From spending tracking ➡️ ✅ unit cost visibility

Davies closes his speech saying government will need structural reforms, digital skills and modernisation, but my conclusion is that none of these reforms will succeed without the information architecture to sustain them. Not because it's "good" to have it, but because it's the critical infrastructure upon which everything else's success is built.

---

Notes:

¹ Written evidence to Parliament Health and Social Care Committee on hospice financial sustainability (2026)

² Article: Funding the system: Rethinking how capital flows drive sustainable change (2025)