Tax and Transfer Policy Institute

Crunching official data helps evaluate policies. Potential embarrassment is the wrong reason for governments to keep it under wraps, Warwick McKibbin and Robert Breunig write.

As the engines of economic growth slow in Australia, there is a need for innovative policy changes to both raise economic growth but also to improve the wellbeing of Australia more generally. Design of good policy depends on a solid foundation of good data.

A recent workshop at The Australian National University convened by the Centre for Applied Macroeconomic Analysis, the Centre of Excellence in Population Aging Research, and the ANU Tax and Transfer Policy Institute explored this issue in great depth.

Governments hold vast banks of administrative data which, if mobilised, can be a potent force for policy development. In the hands of non-political quality research institutions, these data can fundamentally change the quality of policy design.

It’s not an outlandish proposition. Private companies routinely identify patterns from big data to pinpoint just about every conceivable aspect of consumer decisions. In Australia, there is a reluctance to put government data to better use outside of government, as experience from Australia and from around the world demonstrates that this reluctance causes valuable lost opportunities.

Some examples are useful. Australian administrative data reveals small businesses did make new hires following reductions in company tax rates. This evidence flies in the face of surveys suggesting they would not.

As the Australian automotive industry declined, data revealed that younger individuals laid off by Mitsubishi did better than those who stayed on with other auto companies in the late 2000s. Such insights help design and target assistance programs.

In the United States, a more pro-active approach called Opportunity Insightsproduces high-quality evidence about inequality, upward mobility and public policy. Top researchers given access to government administrative data sets generate insights about poverty and inequality that improve lives.

Making administrative data similarly accessible in Australia could dramatically and rapidly deepen our understanding of issues and of what and how policies are working. Analysis of administrative health data already benefits Australians. There needs to be a similar approach to childcare, education, employment, welfare, taxation, and business assistance.

Why is it that Australia, unlike Denmark and other successful developed economies, makes little administrative data available to researchers? In part, it is because of reasonable concerns about privacy. These concerns can be mitigated through careful procedures, like de-identifying data and vetting research and researchers.

There is also a more problematic reason for the inaccessibility of administrative data in Australia. Engagement with these data by qualified and impartial researchers will likely reveal that many of Australia’s public policies fail to meet their objectives. Such findings will embarrass governments and bureaucrats. But they would improve outcomes for all Australians by reallocating taxpayer funds to policies that work and reducing government waste.

Is it possible to have the analysis done by experts within the government without engaging the wider research community? Possibly, but there are reasons why this is a problem.

First, in-house analysis is expensive. Making administrative data available to researchers increases policy-relevant research at low cost. In Denmark, an explosion of quality research on tax, education, immigration, and policy followed the sharing of government administrative data, leaving little or no financial burden on the Danish government.

Second, academic researchers are subject to more rigorous and accurate standards than government researchers. High-quality analysts employed by governments are often capable of excellent analysis, but the breadth of expertise in academia, the incentives academics face, and the scholarly process of peer review encourage high-quality work.

Third, bureaucratic analysis is not transparent. In the absence of public oversight, politicians and administrators can and will claim that programs are working. Giving researchers access to administrative data provides a check on government.

Sir Michael Barber famously established the ‘Delivery Unit’ for Tony Blair’s UK government in the early 2000s. The unit, which required all facets of government to test the quality of policy implementation against the policy’s stated aims, was extremely successful and replicated elsewhere, including Australia.

But politicians, bureaucrats and administrators soon learned to game the system. Sir Michael today advocates data outside a government’s control as the best defence of good policy. And, if recent empirical evidence is correct and as much as 80 per cent of government interventions fail to achieve their intended outcomes, such data is the best defence against bad policy.

Transparent, accountable, crowdsourced policy evaluation from the research community will raise awareness of how hard it is to design and implement effective policies. We know Australians want better governance – in the last decade, the trust in political processes has collapsed. Widespread access to the insights of administrative data will improve policy processes, with large and sustained benefits to Australian society.

This piece was first published in the Australian Financial Review.