It's timely to take stock of the COAG performance reporting system, not least because the federal opposition has raised the possibility of overhauling Commonwealth-state relations. Its finance spokesman, Andrew Robb, suggested last month that a Coalition government would outsource to the states and territories the provision of services where the two levels of government presently share responsibility. Top of the list are services in education and health. In the same week, Coalition families spokesman Kevin Andrews proposed a new model of oversight of federally funded services delivered by community and business organisations. The existing reporting and compliance requirements would be significantly reduced, giving a greater supervisory role to local communities.

The Coalition is clearly following lines of argument developed by its counterpart in Britain, the Conservative Party, when it was in opposition. Indeed, one of British Prime Minister David Cameron's key advisers, and the architect of his ''big society'' vision, Phillip Blond, visited Australia recently to speak to Liberal leader Tony Abbott and his colleagues. The big society idea was intended to return control of local services to communities, thus allowing the dismantling of the elaborate structure of performance targets and reporting established under the Blair Labour government.

Britain's performance system applied to all departments and agencies but was most pervasive in the monitoring of education and health services. By the final years of the Labour government, the level of complaints about the compliance costs of performance reporting and the extent of micro-management from Whitehall had become overwhelming. As targets became more complex and required more extensive performance data, bureaucracies expanded but without any noticeable improvement in outcomes. With a change of government, the system was ripe for overhaul.

The parallels with Australia are not precise. Our federalism means that the central government is not directly responsible for many public services, particularly the key services of hospitals and schools. Nonetheless, the British performance targeting system has much in common with the current reporting regime that COAG is imposing on the states and territories. Both have provoked resistance from service providers; in Australia's case, particularly from state and territory governments. Both have given rise to complaints about compliance costs, particularly increased bureaucracy.

The similarities are not accidental, given the close contact between Blair's New Labour and the architects of COAG's reporting system, particularly the then prime minister, Kevin Rudd, and his departmental secretary, Terry Moran. Both systems spring from the same ''third way'' philosophy that seeks to combine the best of social democracy with the best of free-market capitalism. Social democrats in labour parties still accept that the government should be responsible for providing public services. But the third-way modernisers believe government service providers should be subjected to the private-sector disciplines of strategic management, such as reporting against precise targets.

Should the COAG reporting structure be defended, or should it be allowed to fall victim to cost-cutting in bureaucracies? The structure was part of the intergovernmental agreement on federal financial relations established by the Rudd government in its first term. This agreement was intended to rationalise Commonwealth payments to the states and territories, and to increase accountability for shared functions. As well as continuing to provide GST payments with no strings attached, the new policy allows for various types of tied payments. These include the more specific ''national partnerships'', which provide federal funds to deliver specified projects and may include reward payments for achieving designated targets, and the more general ''national agreements'', which cover six broad areas of policy, including education, health and indigenous affairs.

''Partnerships'' effectively continued the practice of the former specific-purpose payments, with the Commonwealth as paymaster paying for designated services. ''Agreements'', on the other hand, were intended to be more flexible, allowing states and territories the freedom to choose their own policies within the general areas, subject only to the obligation to report performance of agreed objectives, outcomes, outputs and performance indicators. To emphasise the states and territories' independence from the Commonwealth, responsibility for developing and overseeing the accountability structure for national agreements was placed in the hands of the COAG reform council, a body established and funded by both levels of government. The council was also tasked with monitoring performance under partnerships, to see whether states and territories achieved their predetermined targets.

Though state and territory governments agreed to the new reporting regime, they have found its requirements increasingly oppressive, as the federal government and the reform council have steadily extended the scope and level of detail of the performance data they require. This gradual expansion seems endemic in system-wide performance reporting regimes. It was a feature of the Blair government's system of targets and also of two major initiatives in the United States: the Congress-led Government Performance and Results Act 1993 and George W. Bush's administration's program assessment rating tool.

All such projects find the initially available data inadequate to convey an accurate account of performance. Their administrators therefore request more comprehensive and consistent data. As each reported snapshot of particular government programs appears to overlook or distort important aspects, the call goes out for more and more data in a potentially fruitless quest to construct a complete assessment. Adding targets or intended ''benchmarks'' (in COAG's softer language) often compounds the complexity without yielding any greater precision. Whether a program achieves its target is as much a judgment on the appropriateness of the initial target as it is a reasoned assessment of the program's performance. (The recent ''failure'' of the Australian Olympic team to reach its overambitious target of gold medals is a case in point. Aiming for a particular ranking is especially problematic, as it depends on the performance of others as well as one's own.)

Performance information can certainly be valuable. Indeed, effective program management depends on being able to assess success or failure in meeting intended results. The problem lies in the tendency to demand excessive levels of data with little obvious benefit for managing performance. It's unsurprising that the states and territories are complaining about the increasing compliance burden of the COAG reporting regime, or that federal cost-cutters should eye the Commonwealth's contribution as a potential source of savings.

The main rationale for the new performance regime has been to improve the accountability of jointly provided services. Like other federal systems in which different levels of government share responsibility for certain functions, Australian federalism has suffered from an accountability deficit in the sense that no single minister or government can be held accountable for the government's actions. When policies fail or mistakes are made, the natural temptation is always to pass the buck elsewhere. State and territory governments accuse the Commonwealth, typically on the ground of inadequate funding, while the Commonwealth points the finger at the alleged incompetence of state and territory governments. This is the legendary ''blame game'' that Rudd promised to end, especially in health, but also more generally.

Performance reporting, however, does not directly address the problem of shared responsibility and diverted blame. It can certainly help to clarify outcomes and identify weaknesses in joint programs. But this knowledge does not tell us who to hold to account for any shortfall. To be fully effective, accountability needs to move beyond the preliminary stage of ''giving an account''; that is, reporting to a designated forum or authority on the details of the action taken. It needs to include ''holding to account'', whereby the forum or authority can interrogate the accountable person or body and, where necessary, impose appropriate remedies.

Performance reporting can certainly produce more complete accountability when combined with the granting or withholding of reward payments, as in some of the so-called ''partnership'' agreements. In effect, such agreements seek to increase federal control and to reduce the autonomy of the states and territories, thus potentially shifting responsibility and potential blame to the Commonwealth. But where the reporting is not backed up by any sure remedial action, accountability is much weakened. ''Naming and shaming'' may trigger some voluntary improvements from state and territory governments. But for the most part, as McClintock has complained, state and territory governments take very little notice of the performance data that their public servants collect and which the COAG reform council promulgates. Again, the experience is international. The more performance data is published, the less attention it attracts. Performance federalism, if it is to survive as the ''cutting edge'' of modern federalism, needs to be aware of diminishing returns.

Richard Mulgan is an emeritus professor with the Crawford School of Public Policy at the Australian National University.

richard.mulgan@anu.edu.au