Australian Statistician David Kalisch won’t forget his birthday in 2016 any time soon. About the time he might have expected to blow out a few candles, a cyberattack was launched against the online Census. Almost three years on, #CensusFail is still conjured for its stark lessons to enterprise about cybersecurity, the need for unwavering vigilance and a belt-and-braces approach to risk management.
As head of the Australian Bureau of Statistics (ABS), Kalisch had no option but to close down the site until the problem could be fixed, and the integrity of Australians’ personal data assured. It was a stark reminder of exactly how valuable data is — both to the government and the private sector.
Lies, damn lies… and statistics
From a peak of $380m in 2009–10, ABS funding is predicted to fall to less than $250m by 2020. The ABS currently employs about 2700 staff. Allocating resources means assessing five guiding principles to maximise public value, says Kalisch.
- What are the core national official statistics people expect from the ABS? How can they be delivered with quality and in a timely fashion?
- What new forms of data and information do government, business and the community expect? What new insights are needed and how can that be achieved?
- Is there effective and safe access to the data? Not just in aggregate forms, but so researchers can access more detailed data resources without compromising the privacy and secrecy of the information provided by households, persons and businesses.
- Making use of new data sources, including some of the data emerging from government — such as the ATO, Medicare and Customs — and, where possible, substituting that for direct data collection from individuals and businesses.
- Identifying the sorts of data and statistics that Australia and Australian business will need in the future.
A landmark study by management consulting firm McKinsey & Co has estimated that using open data — that is data available for public access — could unlock about US$3 trillion of global value each year by breaking down information gaps across industries. This would potentially unleash insights that could help boost productivity, drive innovation and replace gut feel with data-driven decision-making.
In Australia, Nicholas Gruen, CEO of Lateral Economics, calculated that Australian government open data — freely available from the ABS and other agencies — is worth $25b a year. Why? Because insights from that data alone, or combined with private data sources, could reveal where a retail chain should locate its next store based on housing trends; it could give insurers information about agricultural risk based on weather data; or deliver important insights to health providers about the future needs of an ageing demographic.
While many have tried, Kalisch believes that it is still hard to put an accurate monetary value on data because of the way it has become pervasive in our lifestyles. It affects the way we live and work, how society operates and the way business performs. He says that, in general, Australian enterprise recognises the power of data to support quality decision-making, but warns, “I suspect companies are only using a fraction of the data available and relevant to them. There is much more they could exploit from our available data resources and other data services that are publicly available.”
The latest ABS annual report reveals that the private sector makes up just one per cent of the cohort using its DataLab analysis platform, currently available across 248 different data sets. Only 24 per cent of the people using TableBuilder, a self-serve analytical tool for interpreting census data, come from the commercial sector, with another four per cent coming from not-for-profits. To fully exploit data, enterprise needs to engage data analysts and personnel with “curious minds”. If this happens, Kalisch says he will be “really enthusiastic about the opportunities we have as a nation, about better use of data in the future”.
Meanwhile ABS has tough decisions to take. In his preface to the ABS Forward Work Program 2018–19, Kalisch notes ongoing funding cuts have already forced reductions in the ABS statistical program in the years 2008–09, 2013–2014, 2017, and a slight rework to the current year. This year it will receive $255m from the government, and an additional $46m from clients.
”Do you want data available today or do you want quality data? Not all data provides insight. Data might be distorted, it might be biased, it might not be truly representative.”
This means the ABS needs to consider which surveys can be automated; and which delayed? Which data needs to be real-time? The overarching requirement is accuracy.
“Do you want data available today or do you want quality data?” asks Kalisch. “You can have both in some instances, and that is where there is a nice accommodation, but not all data provides insight. Data might be distorted, it might be biased, it might not be truly representative.”
Lessons from Census Fail 2016
- Understand expectations are changing and that there is a reduced tolerance for any error of judgement or delay in service.
- Be prepared. Plan and practise for things to go wrong, including identifying clear roles and responsibilities and agreeing to them up front with key players.
- Be transparent. Tell people what you are planning to do in detail. Act with integrity.
- When things go wrong, act quickly. Apologise, acknowledge what happened, and explain what action you are taking to remedy the situation.
- Manage risk and reward – assess the right balance in each circumstance. Innovation does inevitably carry risk.
- Recognise the importance of leadership and good people management at challenging times – care for staff.
- Manage cyber risk more extensively.
Kalisch warns of the risks of some forms of internet “analysis” that is seemingly conducted in an echo chamber of like minds. He says the ABS must be more rigorous. It does source some real-time data — such as real time scanner data from retailers to update the CPI — although that figure is only updated quarterly. It also takes data feeds from the commercial housing sector and is looking at how to use satellite data to provide agricultural statistics that don’t rely on surveys by farmers and industry bodies.
“Where we are paying a bit more attention at the ABS, is the way that we present the data,” he says. “There is probably a little bit more mixed performance from the media in terms of how they can accurately delve into the data, given the nature of the disruption in traditional media and the limited time journalists have to focus on some of these things. We still find a number of key economic journalists dig behind and beyond the numbers and understand more of what is being released. However, increasingly, others are taking the headline data, repeating it and not capturing its full value.”
Kalisch believes business leaders are relatively data-literate and data-aware, adding, “To date, Australia is one of those countries that has been quite well served by the quality of data available and accessible. That’s not to say more can’t be done. There is always a desire for more and more data, but when you look internationally, Australia is in the top group of countries with its national statistical office and the data we produce.”
He sounds one note of caution. While Australia has ranked well in terms of its open-data credentials — making information widely available to support business, society and government — Kalisch notes there are protections that need to be maintained around some data. Private data, for example, and commercially valuable data.
“There need to be some protections, at least at the ABS,” says Kalisch. “We need to be assured that the personal dimensions of business and households won’t be compromised. If that level of trust was to be broken by us being too cavalier around providing open access to data — including compromising some of that personal and business information — then that is a bond of trust we would break. That is something we’re not willing to do.”