Australia’s National Data Infrastructure: Opportunities and Risks
CEDA Trustees, Adelaide, 13 November 2018
Information and evidence is key to good decision making
I came to the role of Australian Statistician as an economist with many years using data. I have seen the value of data to identify key issues and draw inferences, primarily within a public policy context.
Our national statistics are available to inform key decisions made by governments, business and the broader community.
Official data and other information sources shape fiscal and monetary policy settings, providing information about the state of our economy, current risks and key influences.
Our national data provides evidence about many pertinent public policy issues, such as jobs and unemployment, housing affordability, income and wealth inequality, cost of living, energy prices, population change, the quality of life in our cities and regions, education and health outcomes, needs-based school funding, outcomes for migrants, and so much more.
Government decision making draws upon this data, shaping policy and helping design services across health, education, social services, infrastructure and environment sectors. Without good evidence, government decisions rely on anecdotes and ideology.
Official population estimates and other data are used to distribute billions of dollars of financial transfers between the Commonwealth and the states, and maintain fair electoral boundaries. Our statistics support our democratic institutions, informing choices by voters and political aspirants.
Businesses use official data and other information to guide important investment and operational decisions. Similarly, households may draw upon official data when they make life choices. Weather forecasts are critical to protect community safety, and complex data systems contribute to the safety and efficiency of road, air and rail transport.
Data is pervasive in our modern lifestyles. It helps us, our communities, our governments and business make more informed choices and function effectively. Our national statistics help us prepare for the future, through leading indicators and drawing upon a good understanding of the past.
Data is an essential component of our national infrastructure
The ABS is the primary source of official statistics in Australia, last year releasing over 600 statistical series spanning economic, population, social and environmental information. ABS provides this statistical service for all governments in Australia, and the entire community.
Other government organisations contribute to Australia’s data ecosystem, including the Australian Institute of Health and Welfare, Bureau of Meteorology, CSIRO, Australian Bureau of Agricultural and Resource Economics, Australian Institute of Family Studies, Bureau of Infrastructure, Transport and Regional Economics, to name a few.
Many State governments have been expanding their provision and use of data. Commercial sources and the university sector are also contributors to our national information resources. Collaborations across the data ecosystem are leading to improved information for the community.
Our national data should be viewed alongside our transport systems, energy and communications services as part of our essential national infrastructure. Effective infrastructure planning requires accurate population estimates and draws on a wide range of economic, industry, social and environmental statistics.
Recent years have seen greater attention to enhancing physical infrastructure that assists our economic performance and supports our quality of life.
Commonwealth and state government public expenditure on national infrastructure (transport, telecommunications, energy and water) has increased by over 30 per cent over the past decade, with some variability across years and sectors. Telecommunications, roads and bridges have seen the greatest increases.
By contrast, operational funding for Australia’s national statistical office, another part of our key national infrastructure, has been reduced by 30 percent in real terms over the past decade by successive governments.
Governments, the business sector and others across the community do value the national statistics that are produced each and every day. These statistics are keenly anticipated and used extensively to inform key decisions and public debates.
Unfortunately, widespread reliance on our national statistics for many purposes hasn’t translated into secure funding for our national statistics over the past decade. And ABS funding is not a magic pudding that enables us to produce all of the essential statistics our key data users would like from a reduced funding base.
While ABS is now more efficient than it was a decade ago and technology now enables more opportunities to deliver some statistics, ultimately we do need to make choices about what we do, and what we can no longer do, constraining our national information infrastructure.
Recent developments with our national data infrastructure
Our national data system has been changing over recent years.
Ensuring the quality of our national statistics
When I first became Australian Statistician in late 2014, key stakeholders could name the one statistical incident the ABS had about every five years (Labour Force Survey August 2014, National Accounts 2008, Retail Trade 2004, etc). To put this in context, one major statistical error in about 2500 statistical releases over a five year period.
There is an enduring expectation that our national statistics will be right first time every time. Given the importance of our national statistics and how they are used for many critical purposes, they do need to be reliable.
However, this zero tolerance approach to errors with our national statistics contrasts with reality that every service faces occasional human errors and computer system issues.
ABS takes on the challenge of putting together data from multiple sources, which can and do change over time, to create some of our key statistics. We expertly design and implement key household and business surveys, achieving response rates over 90 per cent. Meeting the expectation of perfect statistics is becoming increasingly difficult in the context of reduced funding and fragile, ageing statistical infrastructure.
Over recent years, greater attention has been given to ensuring reliability of our more important key economic and population estimates in the first instance, now moving through to other statistical collections. As a result, our labour market estimates are more robust than they were a number of years ago.
We have taken the opportunity to improve our national population estimates following the removal of outgoing passenger cards. Continuous improvements are also being made to a number of our key statistics, as we adopt new international statistical standards, improve our data sources and keep improving our statistical methods. Our government statistics have been updated for new international standards and we are well down the path of improving our finance statistics, working with APRA, RBA and the finance sector.
However, improving the reliability of our most important national statistics does generally come at a cost, or at least an opportunity cost reducing our scope to deliver new statistics or maintain some other statistical series.
- Improving the relevance and utility of our national data resource
Our nation not only relies upon quality, reliable statistics, but also requires our statistical program to be updated to measure a changing economy, society and environment. ABS has delivered a number of enhancements over recent years.
Reflecting keen policy interest, the ABS now produces more comprehensive information on the labour market. We now have monthly publication of underemployment and underutilisation estimates (trend, seasonally adjusted and original) alongside 40 years of monthly employment and unemployment estimates. This has been complemented by introduction of the ABS Labour Account, drawing together the best information on employment, jobs, wages and hours of work from our multiple sources, delivering more reliable information on employment by industry and multiple job holding. The ABS has also recently published small area employment information from linked employer-employee data.
ABS now provides greater and more accurate insights about our complex and dynamic labour market.
The ABS has improved the accuracy of the CPI and reduced data collection costs, with annual re-weighting of CPI spending patterns and use of scanner data and web-scraped data to update price information. Extended measures of productivity have been released, to help analysts better understand the key influences of productivity at an industry and state level.
Increased attention is being given to improving estimates of the non-market sector of the economy, reflecting the greater importance of the services sector in the economy. There has also been some expansion of environmental accounts information, with new estimates released on the Great Barrier Reef and our land resources. 2018 Budget funding will enable us to improve housing related data.
Australia has been a very active contributor to the international effort to ensure economic statistics are reliable in the face of increased globalisation and digitalisation. The bottom line from the IMF, OECD, and others is that international statistical standards remain suitable and Australia’s official statistics are reliable, with this subject to ongoing vigilance.
Some statistics, especially social statistics, have continued to be produced through user funding. Inevitably, there have been reductions to the national statistical work program (cessation or scaling back of statistics) announced in 2008, 2009, 2013, 2014, 2017 and 2018. Some important surveys have not been undertaken for over a decade now.
Data integration – the new statistical frontier
National statistical agencies across the world are combining data to deliver new information. Improved technology and statistical techniques are now enabling safe and efficient use of very large amalgamated data sets.
ABS has been developing its data integration expertise and capability now for over a decade. We worked across government to develop person and business centred integrated data. ABS encouraged and contributed to the establishment of the Data Integration Partnership for Australia (DIPA), funded from 2017-18 to 2019-20.
Data integration is already delivering new statistical insights such as improved estimates of Indigenous life expectancy, understanding outcomes for respective migrant cohorts and the importance of SMEs for job creation. Data integration has enabled the government to progress needs based school funding using reliable information on parental income rather than the previous method based on the geographic location of the school. 2016 Census data combined with road network information is providing accurate information on journeys to work.
There is still considerable untapped potential in Australia to use data integration to assess the performance of government policies and programs and target budget funding to programs that are assessed as more likely to be effective, as already happens in New Zealand.
Australia does not have a strong tradition of rigorously evaluating outcomes of government programs and policies. DIPA provides an avenue to rectify this deficiency. Comparatively modest data integration capability can potentially improve the efficiency and effectiveness of government policies and programs costing taxpayers many billions of dollars.
Enabling use of our national data resource while keeping sensitive data safe
Our national statistics are a national resource. Taxpayers have paid for our national statistics, and they should be used to benefit the community.
Over recent years, the ABS has been improving access to its data resource for statistical and research purposes, while still protecting the secrecy of sensitive personal and business information that we receive.
ABS is using the Five Safes Framework and our Virtual DataLab to enable greater use of our valuable microdata, building on our strong data secrecy legislative requirements, statistical disclosure controls, organisational expertise and integrity. Resource constraints and ensuring community acceptance are the main limitations on access and use of ABS’ microdata.
The Productivity Commission has recently highlighted opportunities for greater use of public data in their report on Data Availability and Use (PC 2017). The Commonwealth Government has since responded to the report, including agreement to progress the consumer data right and improve data use through a number of mechanisms.
Ensuring effective and safe use of our national data resource inevitably takes place in a community and social context. While those who undertake research and use data can point to the benefits of using data for public policy or other purposes, there also needs to be strong community support for data collection and data use.
In this contemporary environment, there appears to be less community trust that organisations will effectively protect your information. The community is regularly informed of large data breaches, predominantly reported by private organisations, such as Verizon, Sony, Red Cross, Equifax, Cambridge Analytica/Facebook. At the same time, we are witnessing reduced trust in governments, business and the media (Edelman Global Trust Survey), while also hearing about more cyber attacks on governments, businesses and individuals.
The social license for data producers and data users is becoming more problematic, at the same time as technology is opening up new data use opportunities. Data agencies and data users will need to pay greater attention to this social license dimension over coming years.
In this context, a number of strategies support safe and effective use of data, including:
- Transparency about how data is being used;
- Building capability around good data practices;
- Investing in data protection and disclosure controls;
- Efficient processes to enable access to data; and
- Actively engaging with the community to understand their expectations of how information should be used and explain the benefits to the community from greater use of data.
Improving data collection for our national statistics
ABS is an extensive user of government administrative data and commercial data to produce, for example, our trade statistics, population estimates and Consumer Price Index. New information sources are emerging that enable us to either substitute or complement existing survey data. These sources are increasingly being used to construct housing-related statistics, we see more opportunities for our production of agricultural statistics, and in other instances they can improve the quality of our direct information collections.
The ABS has reduced the survey response burden on business by more than 40 per cent over the past 20 years, through increased e-form collection, improved survey design and sampling methods.
Establishment of a consolidated National Data Acquisition Centre in Geelong in 2016 is further professionalising our data collection processes, aided by the gradual introduction of new data acquisition tools and an emerging partnership with expertise at Deakin University.
Collecting information from households through surveys has become more difficult and costly over time, against the backdrop of a geographically dispersed and a culturally and linguistically diverse population, more secure apartment buildings and busy lifestyles. We are still seeing considerable demand for survey information, to complement data from other sources.
We have recently introduced a Household Innovation Panel to conduct Randomised Control Trials to test ways to improve our approach and improve self-initiated response from those selected for our Monthly Population Survey. Methods are increasingly drawing upon behavioural economics and human centred design principles. This Innovation Panel provides a safe space to experiment and improve our statistical processes, without compromising statistical risk for this high profile survey.
Ensuring sufficient capability to deliver our national data, now and into the future
National statistical offices should deliver as much public value as we can from the resources we receive.
ABS has been pursuing a comprehensive transformation program since 2015 to improve its operational effectiveness. This is a wide-ranging and ambitious transformation program, seeking to enhance our organisational effectiveness through improved partnerships, strategy, governance, people, culture and infrastructure.
The ABS is now more engaged with a range of data users and data suppliers as we have increasingly prioritised our work program. We receive greater insights around which statistics are more important than others, the quality of key data inputs, how statistics are used, and anticipated future information requirements.
Risk management has become a more important consideration across our decision making, influencing choices about resource allocations, organisational structures, governance forums, internal processes and the allocation of senior staff. Improved internal collaboration and increased access to external expertise are also contributing to improved risk management.
ABS continues to have a professional workforce, building on our traditionally strong graduate recruitment and professional development. Over recent years, we have broadened our specialist expertise and drawn in staff with a broader range of expertise and organisational experiences, complementing those with deep ABS experience. Flexible work arrangements and positive workplace culture assist with staff attraction and retention at the ABS, especially as data skills become more highly valued across the economy.
In 2013, my predecessor pointed to the fragile ageing statistical infrastructure that the ABS was using and the risks this introduced to our national statistics (ABS 2013). In late 2014, Government agreed to provide most of the necessary funding over a five year period. We are in the process of implementing new systems, while also paying attention to the security and stability of our core data collection, processing and dissemination systems.
While considerable progress has been achieved with the ABS transformation over recent years, more progress will be necessary as our operating environment further evolves.
A Precarious Future for our National Data Infrastructure?
While the ABS has been able to deliver considerable public value to the nation over recent years, I am concerned about the ability of the ABS to continue delivering the range of reliable official statistics expected by our major users over future years.
I see particular challenges for delivery of our national statistics in 2020-21 and beyond, on the basis of the published forward estimates for our non-Census statistical program.
Independent benchmarking, undertaken in 2016, of government funding to the ABS found it was about half the funding provided to Statistics Canada to deliver broadly equivalent statistical programs, and below that provided in New Zealand for their statistical services.
Further funding reductions in Australia will increase the likelihood of the unavailability of key statistics, statistical errors, or reduced ability by the ABS to provide safe access to data. The consequences could be severe for the downstream decisions then made by governments, business and the broader community.
There could also be implications beyond future availability of essential statistics, including the future development of data expertise required across our information-based economy.
ABS has traditionally played a very significant role in training data experts and analysts, and this has continued over recent years. Further constraints on ABS efforts to deliver our national statistics over coming years would more than likely reduce the development of data expertise at the ABS and data proficiency that would be more broadly available for the Australian workforce.
Our national data supports a growing advanced economy and vibrant, informed society. As a nation, we cannot afford to be complacent or indifferent to future availability and productive use of our essential national data resources.
So, to conclude, what can you do? It's clear that data is a key part of Australia's infrastructure, as much as the built or natural environments. As such you can be both good users and good custodians of data.
This translates to a few things:
· that you actively seek out available data from ABS and elsewhere to inform the decisions you're faced with;
· that you build capability to manage and use data within your organisations;
· that you safely share data, usually aggregated data, with others where you can; and
· that you become champions for the generation and use of data amongst your networks.
Thank you.