Where Do Our Tax Dollars Go: A Case Study (Part 12) - The Cost of Not Counting

‍ Note: The views expressed in this article are my own. They are based on publicly available information, published government data, and my professional experience in public sector transformation.

‍ ‍

Every problem worth solving starts the same way. Before anyone builds anything, before anyone spends a dollar, before anyone writes a policy or drafts a framework, you define the problem. Not broadly. Precisely. What are we actually trying to fix, and how will we know when we have fixed it.

This is not a complicated idea. It is how a building gets built. It is how a bridge gets designed. It is how any serious project begins. You define the scope. You identify the gaps between where you are and where you need to be. Then you work out what it will take to close them.

The New Zealand public sector has two problems that are currently being treated as separate issues. They are not separate. They are the same problem wearing different clothes.

The first is that public money is being wasted. The government has committed to $3.9 billion in savings from centralising back-office operations. That number belongs to every New Zealander who pays tax. Whether it is achievable depends entirely on whether the systems and people responsible for delivering it are functioning as they should.

The second is that people are being harmed at work. Not occasionally. Consistently, across agencies, in patterns that repeat because nobody is measuring them. The Health and Safety at Work Act 2015 places a clear legal obligation on every employer to manage psychosocial risk. The obligation is there. No one is assessing whether it is actually being met.

These two problems connect directly. A workforce that does not feel safe does not speak up. A workforce that does not speak up does not flag the governance failures, the procurement irregularities, or the operational risks that cost the public money. The $3.9 billion in savings the government has promised cannot be delivered by a workforce that has learned, through hard experience, that raising a concern is more likely to end a career than fix a problem.

So the first question is not how we solve this. The first question is what exactly we are trying to solve.

In March 2026, Allan Halse of CultureShift NZ (https://www.allanhalse.com/) filed an Official Information Act request with the Public Service Commission. It asked for fifteen categories of data on workplace bullying and harassment across the public sector, covering the last five financial years. Total complaints received by agencies. Complaints escalated beyond internal HR handling. Independent investigations initiated. Substantiated findings. Disciplinary outcomes. Complaints involving senior leaders. Repeat complaints within the same agency or under the same Chief Executive. Employment related legal costs. Mediation costs. Settlement payments. External investigation costs. Contractor backfill costs. Recruitment and onboarding costs for replacing staff who left after complaints. Data on employees who exited within months of filing a complaint. Mental health support accessed during or after complaint processes.

These fifteen categories are the floor. They are the minimum any employer would need to know whether its own processes are working. A complete system would generate dozens more data points from these fifteen alone.

The Commission's response, dated 23 April 2026, was consistent across every one of the fifteen. No centralised record exists. The Commission does not hold one and has no grounds to believe any other agency does. It has produced no internal advice explaining why this information is not collected. It has no policy rationale for the absence. It has issued no guidance to agencies on how they should monitor psychosocial harm.

Every one of those complaints was filed somewhere. Every investigation produced a report that sat in someone's system. Every settlement was paid from a budget line. Every legal invoice was processed through a finance team. Every person who left after raising a concern went through a payroll system and an exit process. The information exists. It sits in HR systems, legal files, finance records, and exit paperwork across dozens of agencies that were never designed to talk to each other. It is unassembled. And because it is unassembled, nobody at the centre of the system can see the pattern that all of it together would show.

A new Code of Conduct took effect on 30 March 2026. When the Commission was asked whether it had developed any evaluation framework, any monitoring framework, any outcomes framework, or any key performance indicators to measure whether the Code is working, the answer was the same. The information does not exist.

A standard requiring agencies to declare serious misconduct findings when hiring from other public service employers has been in place since 2022. It was cited in dismissal letters across the sector as a consequence employees should understand. When you ask how many declarations were made last year, how many were requested, how many were refused, or whether the system is functioning as designed, there is no answer. The data is not collected.

This is what an undefined problem looks like. Not chaos. Something quieter and more durable than chaos. A system that produces activity at every level without producing the information that would tell you whether the activity is working.

New Zealand's public sector workforce numbers approximately 60,000 people across core public service agencies. The Te Taunaki Public Service Census in 2025 included questions on unacceptable behaviour including bullying, racial harassment, and sexual harassment. The results are publicly available. They tell us that a significant proportion of the workforce reported experiencing these things. They do not tell us what happened next. They do not tell us whether anyone was held accountable. They do not tell us whether the same people or the same environments produced the same results the following year.

A survey that asks people whether they were harmed and then does nothing systematic with the answer is not an accountability mechanism. It is a listening exercise. Listening is necessary. It is not sufficient.

Both obligations require the same thing. A system designed to see what is happening, record it consistently, measure whether it is changing, and report on what the measurement shows. That system does not currently exist. The question is what it would take to build it.

A system that works would know how many complaints were filed across the public sector last year and what happened to every one of them. It would know whether the person who raised the concern is still in their role twelve months later. It would know whether the same environment produced the same harm again the following year under the same management. It would know what the legal costs were, what the settlements cost, and whether the money spent on process actually changed the conditions that produced the complaint in the first place.

Right now, none of that is visible. One agency records an event as a bullying allegation. Another records identical conduct as an interpersonal conflict requiring mediation. A third has no applicable category and absorbs it into general HR correspondence. Each record makes sense inside its own agency. None of them can sit alongside the others and reveal the pattern that all of them together would show.

Building that visibility means designing the system at the level of the data itself. What gets recorded, at what point in the process, by whom, in what form, against what definitions. A complaint filed in Christchurch and a complaint filed in Wellington need to produce records that are structurally comparable. The same fields. The same categories. The same definitions. Applied consistently across every agency covered by the framework. This is the work most reform efforts skip because it feels slow relative to the visible work of building something. It is also the work that determines whether everything built afterwards produces insight or produces data that cannot answer the questions it was meant to answer.

The OIA response is the starting point for the gap identification work. The fifteen categories were designed to map what a functioning accountability system would need to see. The Commission's response confirmed in writing that none of it is visible at the system level. Those fifteen are the floor. Once assembly begins, more data points will surface, because the information already exists inside agency systems in different forms. The records are there. The work is assembly and standardisation, and it can start now.

I have written the implementation blueprint for delivering the back-office savings this government has promised. It draws on published government data, parliamentary reports, and the United Kingdom's extensively documented shared services failures. It is publicly available at https://doi.org/10.5281/zenodo.20635176. The framework in that paper and the framework needed to address workforce harm are built on the same foundation.

If done right, this is a two to three year problem. The funding already exists. It is currently being spent on legal costs, settlement payments, the loss of experienced staff and the cost of replacing them, and the compounding harm of re-investigating the same conduct in the same environments because nobody built a system capable of seeing the pattern the first time. Those costs are real. They are simply never counted against the cost of prevention.

I know what it costs to work inside a system that cannot see its own harm. I also know what it takes to build a system that can.

New Zealand has the data, the framework, and the obligation. What it needs now is the will to build it. I am ready to help do that.