SPEECH - Linking the Dots

I acknowledge the Ngunnawal people, on whose lands you are meeting today, and pay respects to all First Nations people present.

I’m sorry to be appearing on screen rather than in the room. I’m in Tasmania, where the datasets are rich, the scenery is richer, and every regression should probably include a variable for sudden rain.

Let me begin by acknowledging and thanking Professor Philip Clarke, of the University of Melbourne and Oxford University.

Philip has probably had a bigger impact on Australian policy than any Oxford professor since Keith Hancock. That is because he brings intellectual horsepower, but also because he brings the rarer gift of applying frontier research to practical public problems.

Australia is lucky that Philip keeps returning home. He is one of those scholars who reminds us that research isn’t only about top journals. It can also help a minister, a department, a clinician or a service provider make a better decision.

That is the spirit of today’s workshop.

The question before you is simple to state and hard to answer: how can Australia make better use of linked and integrated administrative data to improve policy outcomes?

For much of the twentieth century, governments mostly saw aggregated data. Summary tax statistics. Tabulated employment figures. Average school test results. Mean hospital expenditures. Each dataset could tell us something useful. But too often, it was like trying to glean the wisdom of a library by looking at the book catalogue.

Linked data lets us see more of the story.

Handled safely and securely, de-identified linked data allows researchers and policymakers to follow pathways through education, work, health, housing and retirement. It can show which early interventions lift long-run outcomes and where systems are producing unintended gaps.

Over the past decade, Australia has made huge advances in de-identified panel datasets. Investment in these data assets is a classic public good. It is costly to build, but once created it can support hundreds of projects, thousands of researchers and millions of citizens who benefit from better policy decisions.

Much of this progress has been driven by the Australian Bureau of Statistics under the leadership of David Gruen, working with researchers across universities and the broader policy community. David has helped turn data linkage from an interesting possibility into a national capability.

The scale is big and getting bigger. As at 31 March 2026, there were 443 active projects in the DataLab accessing integrated data and standard microdata. There were 2,058 unique active researchers on DataLab projects, including 19 international researchers.

PLIDA, the Person Level Integrated Data Asset, now has more than 70,000 variables. It can link things that usually sit in separate drawers: visa pathways and traveller movements; apprenticeships, university study and payroll records; tax returns, rental property income and superannuation; Medicare claims and medicine purchases; disability supports, aged care and Centrelink; births, deaths and early childhood development.

Work is underway to link more housing, wealth and other financial data, helping us understand expenditure patterns and household resilience.

Integrated data also lowers the cost of evaluation. A randomised trial becomes much more useful when researchers can track long-term outcomes in administrative data, rather than relying only on expensive surveys or short-run follow-up. With the advent of the Australian Centre for Evaluation, that means we can test more, learn faster, improve programs and waste less.

Philip’s report looks at what it would it take for linked data to have an even more substantial impact on public policy.

That means reviewing existing initiatives, including work already underway in Treasury and the ABS. It means identifying what policy and decision makers need in order to understand the value of data assets. It means asking whether new structures are needed within departments or across government, so that data analysis is commissioned well and evidence is translated into policy rather than admired from a distance.

It also means getting the balance right between external expertise and internal capability. Consultants, think tanks and universities can all help. But government also needs its own analytic muscle. Otherwise, the public service risks becoming solely a purchaser of insight rather than a producer of it.

A further strength of this project is its focus on academic alignment. Australian researchers have extraordinary skills. The task is to connect those skills more directly to the policy questions that keep departments awake at night: how to improve health outcomes, support people with disability, strengthen aged care, lift mobility and make services more humane and more effective.

I’m especially pleased that Professor Emily Lancsar will contribute her expertise as Chief Health Economist in the Department of Health, Disability and Ageing. Health, disability and aged care are precisely the sectors where linked data can help us see the whole person, rather than the program silo.

Linked data will never replace judgement, values, democratic choice and ethical debate. But it can make those choices better informed. It can help us move from anecdotes and prejudice to analysis and pathways.

My thanks again to Philip, to David Gruen and the ABS, to Emily Lancsar, and to everyone taking part. You are helping build one of the foundations of better government: the capacity to know what works and what does not.

I wish you a productive workshop, and I look forward to seeing the roadmap that emerges.