INTRODUCTORY ADDRESS TO STAFF AT THE AUSTRALIAN BUREAU OF STATISTICS
CANBERRA
THURSDAY, 9 JUNE 2022
As a stats nerd, this is a pretty exciting job to have.
I'm thrilled to have the opportunity to have Assistant Ministerial responsibility for the Bureau of Statistics, and I do so with a huge respect for the work that you all do. As you know, the great maxim ‘what gets measured gets managed’ really holds up. You determine many of the key things that Australians focus on. You shape the national conversation around inflation, unemployment and growth, but also deeper conversations too about the social statistics - about how we're tracking as a country in terms of our environmental measures, the social health in the nation, the levels of trust.
It's no coincidence that when dictators come to power, one of the first things they seek to do is to undermine their country's statistical agencies. Because the one thing that a dictator can't abide is an independent agency which really tells it like it is. And that's one of the great things about the Australian Bureau of Statistics - the independence you have from government, that extraordinary 116 year history since George Knibbs was appointed the first Statistician in 1906. George Knibbs, who was an acting professor of physics - and so it's in some way fitting that you have in David Gruen a PhD physicist, one of his two PhDs, now as your Australian statistician.
As David said, I'm in the past and actually in the relatively recent past a keen user of ABS statistics. I've made use of your CURFs. I’ve used the RADL. I've used the SIHC and the IDS, the HES. I've written a paper on Australia's record incarceration rate based on your data around prisoners. It did have some problems with the rebasing of the Indigenous share, which I had to fix, but apart from that I thought was really useful data to have. And my favourite paper that I did with ABS data was when we were interested in what happened when the baby bonus was introduced on 1 July 2004. Joshua Gans and I requested a special extract of daily births data from the ABS, and it was a real joy to open the Excel file and to discover that there were 30 years of daily births data there. And the number one births day was 1 July 2004, the day the baby bonus was put into place. We were able to show that around 1000 births were moved from June into July, in order that parents could qualify for the baby bonus. So again, the extraordinary things you can do when you have ABS data that goes back a whole generation, breaks it down on a daily basis and allows us to learn something meaningful, humorous, but also somewhat concerning, because not every birth that shifted is necessarily healthy for the mum and the bub.
I've really enjoyed participating in debates around the right measures of national accounts. We were having a discussion with your executive team before about whether GDP should be headline reported as real net national disposable income per capita or whether we ought to do the current thing we do of not adjusting for population, and really look forward to those constructive discussions with you. And I'm a huge admirer of the way in which you've pivoted during the pandemic to provide that regular real time data. For all public policymakers to have a better sense as to how businesses were travelling and what was happening to households was absolutely critical. I’ve written research using your provisional mortality statistics, which are extraordinary. The notion that you've now produced for us on a weekly basis mortality data as recent as 2021 is a real testament to the hard work of people in the room and the way in which you've engaged with the state and territory governments.
The ABS is now a much more outward looking institution than it's ever been before, working with so many partner agencies, and addressing some of the challenges that we face now, in an era of declining household survey response rates. The access to big data is one of the things that allows us to shape better policy. We took to the election a pledge of an Evaluator General, and one of the things that's going to enable that is being able to measure the outcomes of government policies without the additional cost of running bespoke surveys. You'll be absolutely critical in that mission to measure what matters. At the moment, I worry that too often we have government programs that are set up based on the gut intuition of policymakers, rather than a really evidence based grounding on what works. If we were doing that in the quest to cure cancer or to come up with a new vaccine, we'd be taking things straight out of the lab and putting them into the market. But we don't do that. We carefully evaluate new pharmaceuticals and new medical treatments, and we should take more of that scientific and political approach to policymaking. If we're going to do that, that's going to require the work that you do being brought in and integrated with public policy, and the linkages that you have with departments and other agencies will be absolutely critical.
I'm looking forward too to the opportunity of the ABS engaging with the government’s well-being agenda. You've got a strong tradition that having done that through Measures of Australia's Progress, and many of the social inclusion indicators. As a keen egalitarian and somebody who has written a book on social capital and struggled sometimes to find the right metrics for measuring the strength of community, I'm really keen to have a conversation with the Australian Bureau of Statistics about how we can do a better job of measuring the health of community and looking at the way in which Australia's income distribution is shaping up.
I do have a confession, though, which I think is not going to endear me to anyone in the room – I write my code in Stata. Now I know in confessing to using Stata, that's kind of put me firmly between two camps. The experienced hands will know how to code in SAS. They will have worked on mainframes, they'll think anyone that doesn't know how to put a set of punch cards in really isn't serious, and I just can't engage with them. And then all the cool kids in the room of course use R. And I've kept on telling myself I should learn R, but I just haven't found the time. It's, you know, a bit like learning Spanish - you kind of figure you ought to do it at some stage and you never get around to it. So I hope you won't hold it against me that you have a Stata user as your Assistant Minister.
I just wanted to wrap up where I began. I have a huge admiration for the Australian Bureau of Statistics, for the work that you do, for the integrity that you bring to that. I'm very much in the market for fresh ideas, and keen to hear your thoughts as to how the ABS can produce new metrics, things that you've traditionally done, that you think have reached their use by date, and new ways in which you can help inform the Australian community as to how we as a nation are travelling.
Thank you very much.
ENDS
Authorised by Paul Erickson, ALP, Canberra.
Showing 2 reactions
Sign in with
It is a good time to be here, because the Australian Bureau of Statistics 1 is about to undertake a new enterprise called measuring technology progress. Now, I have managed here to avoid, successfully, any political comments. I have wriggled and gone backwards and forwards, just as my colleagues have for the past two days, and failed to tell which way I sailed. However, I am going to make some political statements on one particular subject—I would say a cliché, but there is another a word if you want to call it that—and that is, the measurement of innovation and productivity.
The Australian Bureau of Statistics (ABS) has been exploring the use of machine learning and extreme machine learning to improve the quality and speed of its data collection and analysis 2.
In particular, the ABS has been investigating how these technologies can be used to improve the accuracy of its data collection, to improve the efficiency of its data analysis, and to reduce the time it takes to produce reliable statistics.
The results of the ABS’s research have been very positive, and the agency is now looking to further develop and implement these technologies across its operations.
References:
1 https://www.abs.gov.au/statistics
2 https://didatica.tech/o-que-e-machine-learning-aprendizado-de-maquina/