Do you trust public procurement?

Barbara Allen
Deputy Head of School
Senior Lecturer in Public Management and Policy
School of Government
Victoria University Wellington

Every government spends public money (assuming it raises taxes), and the figure of 12-14% of GDP spent as public procurement expenditure is an oft used figure. This article points out some new reports related to the transparency of public procurement and raises new questions related to the field of artificial intelligence and procurement for future consideration.

Transparency International New Zealand (TINZ) has repeatedly encouraged the New Zealand government to improve its transparency in public spending through the procurement mechanism. They have supplied ample evidence of public procurement shortcomings. See for example another TINZ article Four years of Open Procurement Data.

One simple measure of transparency in this sphere is the extent to which the average knowledgeable person could find and understand how much public money is being spent on which programs, with which consultants and companies, and whether the spending has been evaluated with a view to its original outcomes. This is still palpably not possible in New Zealand.

This is not to say that there have been no improvements. In an area of bustling activity (yes really), there has been effort into gathering, analysing, and improving systems that utilise the procurement lever, including gathering information on the 2019 policy to pursue broader outcomes through purchasing

Not good enough for the OAG

Recently, the Office of the Auditor General published a short report ‘Getting it right: Supporting integrity in emergency procurement’. The OAG obtained information from six public organisations. They surveyed all public organisations required to use the Rules about the emergency procurement processes they undertook between June 2018 and July 2022. The OAG could only find published contract award notices for 14 of the 246 contracts for which they received records.

Given that the Rules provide a window of 30 days after all parties have signed to publish the award notice, the OAG has understandably said this is just not good enough. Other revealed issues include lack of declaration of conflicts of interest. There is very little room for forgiveness here. Yes emergencies are stressful but there is plenty of leniency built-in such that information does not even have to be declared right away.

From a researcher’s point of view, the lack of information about which organisations were studied and who responded, remains a highly non-transparent way in which to undertake research.

Infrastructure transparency

In other areas, the New Zealand Infrastructure Commission Te Waihanga recently published their review ‘Transparency within large publicly funded New Zealand infrastructure projects – How transparent are large infrastructure projects in Aotearoa?’ A huge part of the infrastructure eco-system involves procurement – from the commissioning of projects through to evaluation. 

The report notes that, ‘while large, public infrastructure projects are subject to normal official information processes, they’re not required to proactively publish key documents.’ 

But it’s worse than that. The research showed that about half of the Business Case and Assurance Case documents in these massive public projects were not accessible and reviews were not accessible for completed projects.

The new Infrastructure ‘Pipeline’ is an improvement in terms of publicly available information, at a high level, about planned projects by region and sector. Some useful findings have been published in the ‘City Rail Link Interim Review’ including a section on Procurement Strategy.

One notable finding is that inadequate consideration was given to the development of a cohesive Procurement Strategy as part of the Planning (Designation and Consenting) process and decision making appeared overly focussed on emerging issues such as commencing works around and under the Commercial Bay development. ‘The overall Alliance package created was the largest single contract ever procured in NZ and did not have the benefit of a robust development process from the onset.

These weaknesses both in process and communication of plans and strategies are worrying, but our public sector has the knowledge and underlying capability to improve the situation – should the desire be there.

Procurement and Artificial Intelligence

Even more problematic is the wave of questions emerging around artificial intelligence (AI) and procurement. We barely know the questions to ask.

There will be pressure on government departments and public procurers to purchase AI. In fact they are already purchasing it and have been for some time. Systems that create predictive models, such as those for the erstwhile social investment programs, already have the capacity to churn out vast arrays of information on public datasets.

Undoubtedly public officials are experimenting with tools such as ChatGPT for such things as draft policies, draft contracts, possibly even draft cabinet paper briefings. At the moment AI produces ‘starting points’, drafts that avoid the problem of the ‘blank page’ but are definitely not trustworthy in terms of content. 

Significant questions about governance and regulation of AI are already being asked but we can be sure we are already way behind the need for guidance. New Zealand’s Algorithm Charter is a step in the right direction, but remains limited and the New Zealand Government’s ‘Cloud First’ policy means that more and more information will be held in ‘hyper-scalers’ mostly off-shore.  Questions around procurement need to be asked. How does a procurement official deep in a government department know what AI is buried within the products the department wants? How is that transparent to the official, and transparent to a citizen?

Transparency is critical to democracy. What the government spends and how money is spent must be transparent as one of the key checks and balances to our system. Old problems remain and new problems are here in a tsunami of AI issues.

Blog Post written by:
No items found.