The Public Sector Informant

CensusFail: should Australian Statistician David Kalisch's head roll to protect the census?

If the census data turns out to be generally reliable for policymakers, much will be forgiven.

The reasons for the failure of the census website on census night remain obscure. What went wrong and who was responsible are questions that won't be satisfactorily answered for months, even years, as experts pore over the records and interview the participants. In the meantime, however, we can assess the immediate reaction to the census shutdown and whether the processes of public accountability are working well.

The Australian Bureau of Statistics is an independent statutory authority at arm's length from government, a status that affects its accountability. Though it reports to its portfolio minister, the treasurer, and one of the junior treasury ministers, it doesn't take instructions from them. In this case, the relevant ministers could have expected to be informed promptly about the shutdown so they could, in turn, inform the public. But they couldn't be expected to take full responsibility for the bureau's actions as they would for a normal department under ministerial control.

The ministerial response was initially somewhat uncertain, partly because the relevant junior minister, Michael McCormack, had held the job for only a few weeks. Confusion also surrounded the immediate cause of the shutdown and the possible role of hackers, in Australia or overseas. The next day, Prime Minster Malcolm Turnbull quickly took charge of the government's response, claiming a massive failure on the part of the ABS and promising an inquiry to be conducted by government cyber-security adviser Alastair MacGibbon, to be supported by the Australian Signals Directorate, the Treasury and the ABS. The inquiry would discover who was to blame and "which heads should roll".

Turnbull's attribution of responsibility to the ABS was constitutionally quite correct. In the words of a former Australian statistician, Bill McLennan (1995-2000), "the buck stops with the Australian statistician on everything". Responsibility, McLennan insisted, did not lie with the minister. This point was missed by many media observers, who slipped easily and lazily into the familiar grooves of "massive government failure – minister avoids taking responsibility".

The opposition was happy to muddy the waters; its spokesman, Andrew Leigh, trotted out the "Westminster system of ministerial responsibility" and demanded the government take full responsibility for its failure. Opposition Leader Bill Shorten described the census as "a complete Turnbull train wreck". One can only speculate what havoc Tony Abbott, as opposition leader, would have caused if the census had been held on the Gillard government's watch. However, the combination of an inexperienced minister, who couldn't reasonably be held to blame, and the fact that Parliament wasn't sitting at the time helped to dull the political attack.


Certainly, from a broader perspective, the government may share some responsibility. Relentless squeezing of the budget, by both Labor and Coalition governments, has reduced the ABS's capacity, though it has yet to be established how far that was a factor in the census failure. In addition, the Turnbull government's enthusiastic support for innovation and agility could be said to lend support for such high-risk ventures as an online census, though, again, that may or may not have been a contributing factor. But the government, on the whole, has successfully cast itself in the role of solution-finder rather than problem-maker.

In terms of accountability, the fact that no minister is in the frontline can have advantages in that it allows public criticism to focus on a wider range of relevant issues. When the opposition is baying for a ministerial scalp – as, for example, Abbott did against Peter Garrett over home insulation – the press gallery becomes obsessed with the gladiatorial contest and leaves no oxygen for anything else. In that case, it took months before critics started to look elsewhere for the major culprits among the economics ministers, in the central agencies and in the Environment Department and its consultants. Garrett eventually emerged much later as a minor player who had loyally taken the heat for his colleagues and staff. With the census, the absence of a ministerial lightning rod means we can discuss the failure more dispassionately.

In this public discussion, the key figure taking responsibility must be the Australian Statistician, David Kalisch. In retrospect, Kalisch and his colleagues, in the months before the census, didn't do enough to explain their reasons for changing the census process, particularly the reasons for holding on to people's names for longer. The level of public disquiet took them by surprise and they paid the price for not taking the public more into their confidence. The conduct of the census is no longer an issue that can be safely left to discussion among policy insiders but must involve the wider public.

Since the crisis struck, however, Kalisch has been scrupulous in exercising his accountability responsibilities surrounding the conduct of the census. He has made himself available to the media, including appearing on the ABC's 7.30, and hasn't sought to hide behind the minister.

Kalisch has also refrained from publicly passing the buck to the ABS's main contractor, IBM, which was operating the online census. He will no doubt engage in some tough talking with IBM behind closed doors. But he has properly accepted that the ABS retains ultimate responsibility. He has acknowledged that, in the eyes of the public, the ABS owns the census and the ABS must repair the damage and restore public trust. This is the correct response. Outsourcing of government services does not absolve governments from responsibility for the outcome. If governments choose to use outside contractors rather than in-house staff, they need to manage the risks associated with diminished control. If the contractors fail for some reason, the blame will be sheeted home to the government that drew up the contract and chose the contractor. Media commentary again sometimes missed the point, speculating whether blame rested with the ABS or with IBM and noting that IBM had yet to comment publicly. But the ABS and IBM were not equal partners sharing responsibility for the census, in spite of the fashionable weasel word "partnership" used to describe their relationship. Instead, they are related more as principal and agent, with the ABS the buyer and IBM the contracted provider. In this case, the providing agent, IBM, is accountable to the ABS but the ABS is accountable to the government and the public.

If the Australian Statistician and the ABS are to be held responsible, should Kalisch's head be the one to roll? Certainly, some critics called for his immediate resignation. They linked his failure to resign with the general unwillingness of ministers to resign in similar circumstances. Kalisch, instead, has judged that he is the right person to oversee the ABS's immediate response and to begin the process of restoring trust in the ABS.

Kalisch didn't do enough to explain the reasons for changing the census process, particularly holding on to people's names for longer.

Whether Kalisch's position is tenable in the longer term remains to be seen. A useful parallel can be found in the actions of another statutory officer, Ed Killesteyn, the Australian electoral commissioner, who presided over the West Australian debacle in the 2013 federal election. He continued in the post for some months until the High Court ruled the election was invalid. He then quietly resigned, along with the West Australian commissioner, leaving the way open for new leadership that could rebuild public trust in the office's integrity.

A similar course of action may eventually appeal to Kalisch as the best means of rebuilding public confidence in the ABS and the census. Much will depend on the general verdict about the 2016 census that emerges over the coming months. Here, there are signs that public hostility may be relenting. Privacy concerns over retaining names for longer seems likely to abate in the light of better public understanding of the policy benefits from cross-matching data and further assurances from the Privacy Commissioner. No lives were lost and there are no bereaved relatives angrily demanding retribution.

The e-government lobby and innovation spruikers have been quick to defend the general policies of data-matching and of moving the census online. The shutdown on census night, they argue, should be seen as an unfortunate, one-off failure in implementation rather than an indictment of the whole enterprise. Ironically, these views echo the Prime Minister's own advice to public servants that they should embrace change and risk failure. Turnbull himself, however, was too busy surfing the wave of public anger to play his preferred role as champion of innovation and agility. When he could have called for calm and offered support for those pioneering a complex new electronic process, he chose to bluster about rolling heads. Observant public servants will draw their own conclusions about whether failure is worth risking.

As the days and weeks pass, and more people heed the ABS's soothing advertisements, urging them to complete the census late without fear of penalty, the census data will become more complete. If the 2016 census turns out, in the long run, to be generally reliable for policymakers, much will be forgiven and Kalisch's reputation may recover. If it does, he and his ABS colleagues will be able to thank the pressures of public accountability, which, through a baptism of fire, forced them to explain themselves better to the public and to adjust their actions accordingly.

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