IPAA ACT Conference - Maximising Value from Data: Navigating the Opportunities and the Challenges
17 October 2019
Closing remarks from David W. Kalisch, Australian Statistician
Can I start by recognising this audience is likely to be atypical, in that we do not need to convince each of the value of data. We all appreciate the value of data, and this is why we have devoted a morning to attending and participating in this event.
We do need to manage data within a resource constraint. In the ABS context, we seek to maximise the public value we can deliver across five areas of:
- delivering quality official statistics;
- ensuring continuing relevance of our official statistics for contemporary and future use;
- enabling effective but still safe use of data;
- minimising respondent burden to surveys/Census, opportunities for data substitution; and
- building future capability (ongoing stewardship of continuing institution).
These public value dimensions are used by the ABS to guide our internal resource prioritisation and our Forward Work Program, within available funding.
This morning’s presentations, and the associated Q&A, have all highlighted that we are operating in an environment of considerable change around what is possible with data.
This is also bringing a number of challenges. The early poll of conference participants who were asked their view on the main contemporary challenge to maximising the value from data emphasised data skills capability gaps as the main constraint, suggested by around half of today’s participants.
Other challenges that also received responses in the 10-20% range included the three dimensions of funding constraints, culture and enabling legislation, and each of these should not be ignored.
The perspective I would bring is that ultimately all of these dimensions are important if we do want to maximise value from data. The insight I also drew from this straw poll of today’s participants is that building data capability needs to get more attention in the future, noting that this is something the ABS has been talking about over the past year.
There was some discussion of good and bad data. I prefer not to use such subjective terms, but would suggest we do need to be smart, informed users of data that is fit for purpose.
- We need to pay particular attention to where and how data is being used. More sensitive and important uses of data, where data is informing really important decisions, requires more care and understanding around the quality of the data. Nature of the decision using the data needs to be brought into the considerations around data use.
- New data opportunities and technology opportunities are emerging and will continue to emerge over time. We need to be open to, and rigorously consider, use of new data opportunities and analytical techniques.
- I often use the animal farm analogy at ABS, where we understand that not all data is of equal quality and stature. We need to be expert users and consumers of data, assessing whether it is suitable for the purpose it is being put, and at a minimum understanding the constraints.
- Some information can be useful as an initial guide, and some big data can be useful for some purposes. ABS is also using some emerging data as inputs to our national statistics, such as use of retail scanner data in compilation of the CPI. In the chart pack used by the RBA Board to guide the setting of official interest rates, we see extensive use of high quality official statistics produced by the ABS.
Ultimately, it is not just about the data.
Professor Abigail Payne’s keynote address was on the mark with her exhortation that we need to first start with the core question of what we want/need to know, before we jump to the data solutions and data options. There needs to be a greater collaboration between policy advisers and data experts – we need each other, and it would be good to have more people crossing over from one to the other. Collaboration between policy and data expertise is a key to unlocking more effective use of data for policy purposes.
There are many areas in which data can be used to inform public policy, including:
- our understanding of the economy, population, society and environment, and how these are changing;
- the key role that data plays to help inform our democratic institutions; and
- greater analysis of data can inform understandings of policy and service effectiveness – an area where I would suggest there is scope to do much more, drawing on the linked data assets that have been developed over recent years.
Today’s conference highlighted that recent technology, statistical and data developments now enable much more efficient and powerful use of data, with a scale and scope that was not envisaged a number of years ago.
- We can now work efficiently with very large linked data sets and observational data, and envisage that both data and computational power will continue to expand far beyond what we are using today in very short time frames.
- Technology is a wonderful provider of new tools, techniques and capabilities, but this does bring the importance of transparency to the community around what we may seek to do with their data.
- A necessary caution that bigger data does not mean it is of better quality (or representative), as we need to understand the inherent biases that affect many big data sources.
- On the issue of data linkage, we also need to be mindful that not all linked data is of same quality, and ultimately poor linkage techniques can lead to misleading data conclusions.
There were, quite properly, a number of references to the difference between can we
and should we
- Data use and data analysis is not just a technologically driven exercise, but should also draw on sound, ethical judgements.
- Adherence to legislative data and privacy requirements was seen as the minimum, as organisations also considered what was necessary to meet contemporary community standards, maintaining community trust around data practices.
- We also heard that global private sector organisations that face many different regulatory requirements across their markets may choose to adhere to the most stringent domestic requirement across their entire global network.
- There were a number of references to whether practices would satisfy the “pub test”. While this reference to the “pub test” is often made, I would nonetheless caution that we do need to understand what community attitudes really are, and how diverse they may be, rather than us making up what we perceive to be prevailing community attitudes.
One of the useful dimensions of today’s conference was to showcase some of the capabilities and data use within the private sector.
- Business are using their own data, and other data sources, to inform their key business decisions. Business can provide some interesting approaches and practices for the public sector to consider.
- In the other direction, private sector data is increasingly used by the ABS where it is of sufficient quality, such as our compilation of financial statistics, use of retail scanner data for the CPI, and some private sector housing information for a range of housing statistics.
- We should be positioning ourselves to have more collaborations between the public and private sector around the use of data, as well as sharing data expertise.
From today’s discussion, I was surprised one aspect did not receive more attention, and that is the international context.
- Australia is an island, but I thought with today’s technology that we had moved beyond the tyranny of distance.
- There are a number of international connections around data, such as through the ABS, but also in the research community and I would have thought in many of the pertinent policy and analytical environments, we need to find a way to better draw upon the international learnings.
- There was some discussion today of international regulatory requirements such as GBPR but not much more.
- The international scene can be very influential, as I would argue that that there are social license/trust implications from global events (generally negative ones) across private and government sectors.
- In the areas of data production and data use, Australia is a leader in some areas – such as the quality of our labour market, economic and population data, and the insights we are providing around use of big data for official statistical purposes.
- There are other areas where Australia can learn from international data practices and data structures.
So, some final comments from me
1. We need to be cognisant of the need for balance around data use, data governance and transparency, and that all of these dimensions are important as a set.
- Dangers if too cavalier around data use, as ultimately I suspect any poor data practices will lead to reduced community trust and reduced benefit from data in the long run.
- Transparency is useful, but also needs to be accompanied by sound data governance, as this provides the substance that data is being managed properly and effectively.
- Data governance should be an enabler of good, safe data use, not an excuse to stop proper and effective use of data for public purposes.
- Community trust is becoming more challenging – and we need to recognise that this is also likely to change over time.
2. We need to
get better around telling the stories from the data we produce, marketing the data and the insights from the data. This is a challenge for the public sector, but is critical if we are going to maintain community trust and sufficient resources for essential data infrastructure.
3. Going back to one of the early insights from this conference, the Capability challenge is not just now, but I expect that it will become more acute into the future.
- We do currently have significant demand for data specialists, particularly in some locations such as Sydney.
- We also need to build understanding in Agencies when they may be contemplating potentially over-reaching use of data beyond their capability. Again, enhanced collaboration with data specialists may be the way to ensure they make smart decisions.
- We need to manage system wide risks, as the decisions by the “weakest links” can often reverberate and reduce community trust around safe, effective data use across the entire system.
- Collaborations across government, and between governments, the private sector and research community, should be progressed as a means of sharing expertise and insights for mutual benefit.