The complete balls-up of the census on Tuesday night should be absolutely no surprise to anyone who has been paying attention to the way the Australian Bureau of Statistics has been run and how it has been treated by the government. It is the story of what happens when governments believe cutting funding has no consequence and that paying fewer people to do the same work is an “efficiency” regardless of quality.
I am probably among the top 0.1% users of the ABS website. I spend whole days looking at data, and as one who will mine the census data when it is finally released I had no issue with completing my census form. I wasn’t very much persuaded by those arguing I should boycott the census due to privacy fears, mostly because I found the arguments pretty weak and covered in layers of smugness towards anyone who dared to suggest the concerns were overstated.
But the failure of the census site last night unfortunately will only serve to bolster those arguments. It rather easily raises the question: if the ABS can’t even be trusted to properly collect our data, how can we trust them to keep it?
The reality is almost no one outside of the ABS was surprised the site crashed. It was utterly foreseeable.
And suggestions of a DDoS attack are completely beside the point. When shifting to an online survey of such a size the first item on the risk management list would have been the ability to cope with large numbers, and the second would be the ability to cope with a DDoS attack.
Being told that a router overloaded just screams incompetence.
The signs were evident long ago that the ABS was struggling.
Early last year – in what was mostly understood to be a desperate attempt to wake up the government – the ABS floated the idea of moving the census from every five years to every 10 years.
The problem had nothing to do with efficacy of a five-yearly census, but was all about funding. After years of “efficiency dividends” the ABS was holding itself together with duct tape and prayers.
The census in 2011 cost $440m, so for an agency under severe budget constraint, stopping a half a billion dollar exercise is not just one way to save money, but given the importance of the census, is also a good way to make the issue of ABS funding very public.
And it worked – in the 2015 budget the government included $234.7m in funding over five years to the ABS.
But it was too late to help this year’s census and too late to undo what has been a long running and bipartisan approach.
The ALP under Kevin Rudd in 2008 cut funding to the ABS along with other government agencies as part of its “efficiency dividends”. As a result the ABS cut the job vacancy survey.
It meant they cut the survey in August 2008 just as the global financial crisis was about to hit, and thus at the precise moment the figures would have been most useful, given job vacancies are a leading indicator of employment:
Other funding cuts meant the ABS had to find savings elsewhere.
One saving was in the labour force figures. In 2013 it reduced the sample size of the survey, which meant that whereas from February 1978 to June 2013 the unemployment rate had a 95% confidence interval of plus or minus 0.2 percentage points, since July 2013 it has been plus or minus 0.4 percentage points:
Soon after the change a number of pretty bizarre employment figures began to get released – something I noted in March 2014.
The funding cuts continued under the Abbott government. In June 2014, the ABS announced it would cut a number of surveys in order to reduce its spending by $50m over three years.
The funding cuts were bad enough, but the ABS was also becoming rather cavalier as well – never a good thing given data is meant to be dull and sensible (and trustworthy).
In the middle of 2014, the ABS, presumably not satisfied with tinkering with the labour force figures enough, began to change the way it collected the data – shifting to an online survey with a lower response rate, and also slightly changing some of the questions being asked.
The ABS didn’t first run the old and new systems together to test the changes. The result was a complete shemozzle.
In August 2013, the ABS reported that employment in one month grew by 121,000 – the biggest jump ever recorded:
No one believed it, and neither did the ABS. The next month it admitted its seasonally adjusted figures from July to September 2014 were complete bollocks and to be ignored.
It was a major indictment for an organisation to admit that the most high profile of its surveys was not to be trusted. Perhaps the only survey the ABS does which is more high profile is the census, and this time round we have a similar result.
Due to abject failure to consult openly about its proposed changes of the holding of people’s names, and an utter inability to explain the reasons for those changes, a mistrust of the census began to arise.
Now mistrust of the census is almost an Australian tradition, but this time round there was little heed of what were legitimate (if, in my view, overstated) concerns.
While the ABS itself was shown to be incapable of adequately responding, so too was the government. From December 2014 to September 2015 Kelly O’Dwyer had oversight of the ABS. In her time on the front bench, O’Dwyer has not exactly met expectations. Her abilities, shall we say, seemed much more sparkling when they were not being used while she sat on the backbench, but she did at least secure that extra funding for the ABS. Prior to the election, Alex Hawke had carriage.
But after the election, the prime minister also showed a complete disregard of the census. It wasn’t until Wednesday last week that it was clear that Michael McCormack was the minister responsible – less than a week before the census.
And even worse, his attempts to assuage public fears were pathetic.
He suggested that worries about the privacy fears about the census were not a big deal given the people worried were also using Facebook. Such a statement indicates that McCormack is woefully out of his depth and totally unqualified to handle this responsibility. His attempts this morning to suggests the DDoS attack was not an attack because the attack didn’t succeed, even though it did, does nothing to allay fears that our government is run by fools.
But really when you get down to it, the failure of the census on Tuesday night is not just about poor risk management, poor government oversight, poor administration; rather it is as the ABS’s head, David Kalisch suggested part of a “confluence of events”.
But that confluence of events is not about DDoS attacks and the failure of a router (god help us, really?) but of a generation of thinking that the public sector can be cut every year and somehow continue to fulfil all its functions. It is the culmination of thinking that paying fewer people to do the same work and to put off spending to upgrade IT systems to “save money” is good budgeting.
It is part of the belief that a generation of privatising and outsourcing work to the private sector does not have institutional impacts on talent and quality of service in the public sector.
The ABS does work that for the most part could not be done with the same level of trust by the private sector. Yes the funding cuts and “efficiency dividends” have helped save some money over the past few years, but the cost to the reputation of the ABS from this bungle is enormous.
And the ABS, above all, relies on its reputation.
There will no doubt be the desire to call for someone to be sacked due to this failure – and rightly so. But unless there is a change in how governments treat and regard the public sector, such failures will only continue.
