The compliance cost to the community of the upcoming Census has been magnified by the anxiety its botched implementation is causing Australians (see this Fairfax report on Census chaos). It appears many elderly Australians have been surprised by the letters they have received about the Census. Given many of them are not on the internet, they have tried to request paper copies, only to have had trouble getting through to the ABS which has not properly resourced its call centre. This has caused many elderly people to worry about the possibility of a $180/day fine.
Also, based on the intense conversation the Census generated yesterday in the office where I am based, it has had a productivity cost in workplaces across Australia. To make matters worse, due to privacy concerns through retention of identifying information and the linking of Census data to other data sets, it appears many Australians may boycott the Census, raising large concerns about the integrity of the data. The ABS has made a huge mistake here and should have avoided raising any doubts about the confidentiality and protection of Census data.
The ABS has clearly botched the implementation of the 2016 Census. It will be important for it and the Government, which is ultimately responsible for the Census, to learn some lessons. Obviously, dropping plans to retain personal information and linking Census data to other data sets would be a good start. Also, we should move away from having a Census every five years and have one every ten years, as is done in the US and UK. This would save significant amounts of taxpayers’ money, given the cost of the Census is estimated at around $400 million, and should allow for a better resourced and hopefully better managed Census, with the ABS able to take advantage of the longer lead time. Indeed, the ABS had previously proposed moving to a Census every ten years.
Last year, The Australian reported the ABS wanted to skip the 2016 Census so it could spend the money upgrading its old computer systems (see ABS wants Census every ten years). The ABS should have been listened to then. I agree with the ABS’s view that the Census is an inefficient and costly way of collecting the wide range of data that it is currently collecting. We could have a less frequent Census and use some of the money saved to boost sample sizes and the range of questions in household surveys, such as the Labour Force Survey which gives us the unemployment rate. We still need a Census from time-to-time, because its population data help the ABS target its surveys and develop population-level estimates, but we should be able to manage with one every ten years.
Queensland Economy Watch 
Having used the census data extensively it is very sad to read about these issues with the latest round of census collections. In my experience, the ABS goes to great lengths to safeguard data (randomisation below certain levels with micro data) – too much so sometimes for the well intentioned and data hungry social researchers among us! However, you are right that they should not jepordise the census with unwarranted attempts at data matching, we need the data for research so much more than for these other reasons!
In the era of data supremacy, the ABS could be a model agency (more like the ONS in the UK) – leading on information management best practice, providing us with clean, readily accessible data at low or no cost. Deploying well tested but innovative technologies. Providing massive amounts of value-add to our social progress and economic development. Instead we seem to have a grossly underfunded, creaking bureaucracy (how many attempts and ABS pilots to improve data accessibility have either stalled or failed completely in recent times?), producing highly questionable (see labour force data) and often politically influenced outputs (how many telling data sets and surveys have been quietly shelved over the last decade!).
Time for ABS v2.0 I suggest. Start a small parallel agency based on best practice (maybe buy in a few ONS staff on contract) and start a safe/staged transition of all valuable personnel and existing functions. Having contact with a few ABS staff over the years they have some great talent there – just oppressed by a failing management and governance structure. The stakes are high! With no independent data gathering agency, Facebook and Google will be telling us where to build our hospitals and schools as well as doing there own – much more intrusive – data matching activities!
Excellent points, Toby. Thanks!
How often should there be a census? I bet there’s room for some very interesting analysis to answer that question. Just pointing to the cost, or looking at what other countries do, doesn’t seem to be a very good justification for an investment decision. Shouldn’t we be comparing the value of the data (including the costs of having out-of-date data) with the costs of collecting the data?
Yes, it’s worth a solid cost-effectiveness analysis. My view (and the ABS’s it appears) is that much of the data could be collected more cost-effectively using regular population surveys. Some people would miss the detailed data for small areas that we currently get every five years, but I doubt it would have much of an adverse impact on policy making and planning if we only got it every ten years. Thanks for the comment, John.
I would much rather see the ABS reduce the frequency of the Census and channel the savings into two things. Firstly, greater utilisation of surveys to provide intra-census period estimates of key demographic data. Secondly, taking a leaf out of the worldwide work using ‘big data’ and establish tools where relationships between multiple ABS datasets can be analysed while still maintaining confidentiality. This would make the data collected and collated by the ABS more useful to decision makers and investors.
Yes, there are huge opportunities from big data. Thanks for the comment, Jim!