Sir Roland Wilson’s abiding legacy was his insistence on the application of reason to public policy. For him, advice must be based on intellectual rigor. It was not enough for the argument to be logically consistent; the premise upon which the argument was based had to be realistic and practicable, rather than based on ideology or wishful thinking.
He emphasized the importance of character, something he learnt at home and as a student. Professor Douglas Copland, one of his mentors, told the Rhodes selection committee that Wilson had ‘sufficient force of character and capacity for leadership’.

Precision in argument, a focus on what was possible, and courage to express one’s convictions frankly were the hallmarks of his work in government and business. Selwyn Cornish, Sir Roland Wilson, A Biographical Essay.
The Sir Roland Wilson Foundation has a continuing role in the development of the Sir Roland Wilson Scholarships and Prizes for graduates of the National Graduate School Management and the Policy and Governance Program.
The Foundation is also responsible for the promotion and presentation of a series of public lectures — The Sir Roland Wilson Foundation Lectures.

These lectures were inaugurated by Mr David Murray, CEO Commonwealth Bank of Australia. The Foundation Lecture in 2004 was presented by Dr Peter Shergold entitled “Once was Camelot in Canberra”. At the launch of this publication Dr Shergold added a post-script address, which is now included in it.

In 2005, the series was continued by Dennis Trewin, Head, Australian Bureau of Statistics with a lecture entitled “100 Years of Official Statistics, with special reference to the contribution of Sir Roland Wilson”.
The 2006 Sir Roland Wilson Foundation Lecture, entitled Australia’s Qantas: Bold, Brave and Innovative, was presented by Ms Margaret Jackson AC, Chair, Qantas Airways Ltd, on the 2nd August, 2006.

There are three parts to this paper. The core of the paper is a description of a history of official statistics in Australia starting from colonial times. A fuller description is given in ABS (2005).
The next part of the paper highlights the special and very important contributions of Sir Roland Wilson.
Finally, the paper highlights some of the key future challenges facing the ABS.
2. From the Commonwealth Bureau of Census and Statistics to the Australian Bureau of Statistics
(a) The Colonial Years
Each of the colonies had an autonomous statistical office. Indeed some of these were world class and recognised as such. They were originally formed to produce the so-called Blue Books, which were in many ways forerunners of the Year Books of today. The Governors sent these reports on the general state of the colonies back to the United Kingdom.
These Blue Books led to a wider range of statistical collections and statistical outputs. There were several very great statisticians during that time. Foremost among them was W.H. Archer, H.H. Heyter and Sir Timothy Coghlan. The latter was the internationally famed New South Wales Statistician for the last part of the 19th Century. Among his many accomplishments was a forerunner of what are now known as the National Accounts.
There was a lot of collaboration between the colonial statistical offices. No doubt there was a lot of sharing of good practice as well as attempts to develop common statistical standards. New Zealand also participated in these collaborations.
(b) Federation
The new Constitution provided for a Commonwealth statistical function. But a statistics agency was not created immediately. It was expected that there could be reliance on cooperation among the new States of Australia. Unfortunately this did not work as well as it might have with its first major test – in the 1901 Population Census. Although at a pre-Census Conference it was agreed to have a number of common topics, with common questions and definitions and common collection methods, this did not happen in practice. The States still tended to use the questions and definitions they had used in previous colonial Censuses.
This led to the creation of the first national Statistical Office, known as the Commonwealth Bureau of Census and Statistics. High among its responsibilities was the coordination of national standards for statistics. It also had responsibility for national collections including the Population Census. It was realised that there was a need for national statistics, not just those that were an aggregation of data provided by the States. In the early days the collection activity was limited – in addition to the Population Census, the collection activities were associated with banking, insurance, cost of living, labour and wages. Most of these statistics were published in what became known as the Year Book. The first was released in 1908 and they have been released in most years since then. But there were only 15 staff in 1908!
Most data collection activity remained with the States and revolved around data collected as part of their administrative systems (e.g. births, deaths, hospital admissions) and special collections (e.g. farm census, factory census). The new Commonwealth Bureau had responsibility for compiling national statistics but was held hostage to the slowest State to provide data.
The operations of the new Bureau were enshrined in the Census and Statistics Act, which was proclaimed on 8 December 1905. This Act still exists. Although it has been revised several times over the years, the original Act is still recognisable. Many of the key elements, particularly those dealing with confidentiality, still exist.
The Commonwealth Bureau of Census and Statistics (CBCS) was instituted under the leadership of George Handley Knibbs, the first Commonwealth Statistician. He was a surveyor by training but had many other skills. He became a demographer of international reputation and his treatise on population was used as a reference for many years.
(c) The Integration of Tasmania into the Commonwealth Bureau of Census and Statistics (1905-1924)
Uniform statistics were not easily achieved. States were often reluctant to change the way they had been doing things for many years. Some States were also slow to provide data to the CBCS. Partly for these reasons, but also because of the increasing demand for national statistics, the CBCS took on more responsibility and slowly grew in size. The Commonwealth Statisticians during this period were Knibbs and Charles Wickens.

The State Statistical Bureaux are now endeavouring, under the authority of the Census and Statistics Act, to collect and arrange all information under a common method and according to uniform categories. The State Bureaux will, therefore, have a double function, viz., they will collect – (a) for their immediate requirements as States, and (b) as integral parts of the Commonwealth.

The CBCS appears to have operated quite effectively. It certainly had a strong international reputation mostly due to the efforts of Knibbs. The time it took to travel to Europe and North America meant that he could not travel often, but when he did travel to international congresses he was one of the more influential participants.
There were (mostly) annual Conferences of Statisticians in Australia between the Commonwealth Statistician and his State peers. The head of the New Zealand statistical service also attended these meetings. Although they agreed on many things, there was still a lot of dissatisfaction. Knibbs in particular was frustrated by the slow rate of progress. This led to proposals for greater integration of the Commonwealth and State statistical services to the extent that there be one national statistical office. In fact, the Conferences of State Premiers of 1906, 1918 and 1923 passed resolutions in favour of combining Commonwealth and State Bureaus. But little happened until 1924. Tasmania agreed to join the Commonwealth Bureau in 1924, mainly for budget reasons. This led to a succession of Commonwealth Statisticians of Tasmanian origin, such as L.F. Giblin, E.T. McPhee, R. Wilson, K.M. Archer and J.P. O’Neill, whose combined length of service in that role was about 50 years. The other States retained their independent Bureaus.

(d) Integration of the Remaining States into the CBCS (1924-1956)
During this period, the role of the CBCS expanded although mostly in economic statistics. Apart from Population Censuses, the States retained major responsibility for social statistics. The Statisticians during this period were Wickens, L.F. Giblin, Edward McPhee, Roland Wilson and Sir Stanley Carver. Some of the developments in economic statistics were world class. Australia was one of the first countries to bring data from different data sources together in a statistical framework that enabled a more coherent economy wide picture to be presented. These included estimates of national income and the world’s first set of balance of payments statistics (published in the 1934 Year Book).

The demand for statisticians with analytical/conceptual skills and university training was growing and this was a problem that Wilson addressed. But more about this later.

Carver’s appointment as acting Commonwealth Statistician was pivotal in obtaining agreement to integration. He was initially appointed as acting Statistician and whilst Wilson was seconded to other roles during the Second World War and in the immediate post-war period. At the same time Carver was head of the New South Wales statistical service. So he had a foot in both camps! Even after Wilson was appointed Secretary of Treasury, Carver retained these dual roles. But it did considerably reduce the suspicion between the Commonwealth and State Governments and this was pivotal in the decision to provide an integrated national statistical service. The Integration Agreements for the remaining five States (Tasmania signed in 1924) were all signed in 1956.

The Commonwealth Government bears most of the cost of the integrated statistical service and it has provided a huge benefit to Australia. It has reduced redundancy of effort by data suppliers as well as assisting to create a more efficient and coherent statistical service.
(e) The Creation of the ABS (1956-1975)
This was a period of rapid growth in statistical outputs and a correspondingly rapid increase in the size of the CBCS. This required a significant increase in the CBCS research and development capacity. Finding the staff was not going to be easy – they were certainly not available in Canberra so had to be obtained from elsewhere. This led to the introduction of the Statistics Cadetship scheme. It attracted some of Australia’s brightest students. Many went on to significant careers in the ABS or other parts of Government, including State Governments. It was one of Keith Archer’s finest legacies. The additional skills available to the ABS through this scheme helped it to get through this period of very substantial growth and laid the foundations for it being regarded as one of the finest statistical offices in the world.

During this period there was widespread introduction of sample surveys under the leadership of Ken Foreman who had benefited from being able to study under the guidance of the top statisticians of the time at the US Bureau of the Census. In the early days these were mainly business surveys but the 1960s saw the introduction of household surveys, again under the leadership of Foreman. These surveys provided the foundation for what became an explosion of social statistics from the late 1970s.

It was also a period when computing was introduced and the CBCS was one of the first organisations in Australia to install a mainframe computer. Given the infancy of the industry in Australia, many of the early Information Technology staff were recruited from the United Kingdom. Statistics and computers were made for each other. The efficiency of statistical operations certainly benefited from computing systems. Importantly, it made statistical outputs much easier to produce so the amount of statistical information available to governments and the community began to expand rapidly. The latter part of this period saw the introduction of systems that recognised end users as the persons who should drive application systems. The systems were designed accordingly.
Statistics such as national accounts, input-output tables and balance of payments were increasing in importance. The advent of large-scale econometric models was one of the drivers. But the more these statistics were used, the more the quality deficiencies became apparent. One key deficiency was inconsistency across data sources. Business collections had been developed more or less independently often using different registers of businesses. Businesses were not necessarily classified consistently across registers. These problems led to a large-scale development project to integrate the CBCS’s business statistics, using a single business register that carried information about each business such as its industry classification and employment size. This took place in the late 1960s. The fundamentals of this program remain in place today. The ABS does have a very cohesive infrastructure to support its business statistics and the statistics derived from them such as the national accounts.
The Statisticians of this time were Carver, Archer and Jack O’Neill. Although integration of statistics in Australia had taken place in 1956, there was still some resistance in the States to the new way. Some continued to operate semi-autonomously but this diminished over time, particularly with the turnover of senior managers.
A very important event in the ABS’s history occurred in 1974. The Whitlam Government was concerned about (a) discrepancies in statistics published by different Departments, and (b) some statistical gaps particularly in economic statistics. It established a review with broad-ranging Terms of Reference. The group established to do the Review, under the leadership of Professor L.F. Crisp, was known as the Committee on Integration of Data Systems.
In 1974 the Committee recommended wide-ranging reforms, and this led to the Australian Bureau of Statistics (ABS) Act 1975 and subsequently to the creation of the ABS itself. Thus the ABS was born from the CBCS in 1975 as an independent statutory authority with well-defined functions, responsibilities and rights. In particular, the ABS was given a political autonomy, which has so far been respected, by both major political parties, but not enshrined in legislation. The Act also created the position of Australian Statistician to become head of the organisation, replacing the former position of Commonwealth Statistician. It also created the Australian Statistics Advisory Council. One important function allocated to the ABS was the coordination of statistics across Commonwealth Government agencies.
(f) Some Significant Developments Since 1975
A lot has happened since that time and the volume of statistical output has increased rapidly even though staff numbers have declined somewhat. It is only possible to report a few highlights.
The way statistics are disseminated has changed dramatically. In the early period, publications were the norm with the occasional release of detailed data on magnetic tape (e.g. Population Census) or microfiche. CD ROMs with supporting analysis software were introduced following the 1986 Census and online data bases were first introduced around the same time. But it is the availability of the web site that has really transformed the dissemination of statistics. Although barely 10 years old, it is now the web site that provides the most common way of accessing statistics. The demand for printed publications is diminishing and many statistical releases are now electronic only. Furthermore the content on the web site has grown rapidly as has the functionality available through the web site. It has revolutionised the availability of statistics. All data on the web site is now available without charge, further enhancing its use.
There have also been important developments in the way we can support research use. The web site has been important because it has enabled more detailed data to be published. Technology developments have also enabled us to respond more quickly to special requests, particularly the Population Census. But it was changes in legislation during the early 1980s that provided the most fundamental change. It enabled us to release unidentifiable micro data for research purposes under specific conditions. Their use is widespread and accelerated quickly after a special agreement was made with the Australian Vice-Chancellors Committee on funding.
The early to mid 1980s were an important period of change for the Bureau – there was a much stronger focus on the way the ABS was managed. Under Roy Cameron, the Bureau subjected itself to rigorous external examination, in the form of the Joint Management Review, which was convened to examine the effectiveness of the top management structure of central office and the state offices, with the aim of guiding the Bureau soundly through current and future challenges, properly using state and central office resources and adequately addressing client needs. The review identified a number of key areas for improvement. With Cameron’s guidance, these recommendations shaped subsequent management planning, contributed to the integration and modernisation of the Bureau and made it more outward looking.
Cameron set up three-year-forward work plans, and brought discipline to development projects. Prompted by the findings of the review, he also implemented a form of matrix management. The document outlining this strategy paved the way for a strong corporate focus that complemented and enhanced the effectiveness of the forward work program. His reforms are still largely in place today.
With the new corporate identity came a questioning and reassessment of the purpose of the Bureau, and an acknowledgement that it was more than a factory for publications. Out of this process emerged the mission statement, the concept of corporate objectives and a commitment to statistical coordination and analysis. These
were conveyed in the Bureau’s first corporate plan, developed under the guidance of Bill McLennan while he was Deputy Statistician. Throughout this period the Bureau increasingly focused on efficiency, producing more with the same or fewer resources. In the 1988-89 Annual Report this issue was clearly enunciated with the paper ‘A quart out of a pint pot’ which spelt out the significant gains in the range and quality of statistical output, which had occurred in the Bureau over the previous decade while resources had remained static.
Of particular importance was the rapid expansion of social statistics on the full range of social concerns. This occurred mainly as a consequence of the expansion of the household survey program. The ABS also started producing publications that brought together a range of statistics on all areas of social concern (e.g. education) or on population groups (e.g. Indigenous people). But it is the annual Social Trends publications, which have become a special feature of the statistical landscape.
Integration of economic statistics continued apace. The Framework of National Accounts was important but reviews showed the key problem was with source data rather than the conceptual framework. A particular problem was the inconsistent methods used across different data sources. Many collections had been developed independently and had their own idiosyncrasies. New economy wide data collections have been introduced, both quarterly and annual, replacing previous collections. Furthermore survey methods have been standardised – this includes the adoption of a common framework for business surveys, common sample and frame maintenance procedures, standardised new business provisions, and the standard treatment of businesses that had been sold, merged or ceased activities. Supply-use tables were used to confront data inconsistencies and to resolve them in a structured way. As a consequence the quality of ABS macroeconomic statistics has improved considerably. OECD now regards them as arguably the best in the world.
The Bureau’s first marketing plan was released in 1989. It followed a government decision that part of the Bureau’s budget should be funded through cost recovery. The plan outlined a major rethink in the way the Bureau regarded its products and its clients. The second marketing plan in 1992 maintained the emphasis on products and the importance of establishing that they were really needed, and out of this emerged the concept of focusing on clients’ needs. The third plan in 1996 placed greater emphasis on clients. In this way marketing became an important plank in the orientation of the Bureau as an outwardly focused, forward-looking agency.
Important in this process was the growth in the Bureau’s analytical capacity. Here the influence of Ian Castles (Australian Statistician 1986-1994) was paramount. He strongly believed that the Bureau could add considerable value to its statistics, but remain objective, by judicious use of analysis and analytical methods.
The Bureau started to devote more publication space to the analysis of its statistics. Authority for this came from the Australian Bureau of Statistics Act, and the emphasis on ‘informed decision-making’ in the mission statement in 1987 highlighted the need for greater effort in this area. In 1995 an analysis unit was established, recognising the potential of statistical methods and models for producing official statistics, improving methods or better understanding statistical relationships.
3. Wilson’s Career in Statistics
Sir Roland Wilson graduated from the University of Tasmania with a Bachelor of Commerce. From there he went to Oxford as a Rhodes Scholar. Although his early intentions may have been for an academic career, he was persuaded by L.F. Giblin to come to Canberra to work on economic statistics.
In 1936 he was appointed as Commonwealth Statistician and Economic Adviser to the Treasury at the remarkably young age of 32.
The outbreak of World War II interrupted Wilson’s plans for the Bureau, and in 1940 he was appointed head of the newly created Department of Labour and National Service. (It was during this period, with the rationing of petrol, that Wilson famously built himself an electric car, which he continued to drive until the late 1940s). At the end of the war he requested a return to the Bureau and this happened in 1946.
In the immediate post-war period, Wilson also served as a member, then Vice-Chairman and later Chairman of the United Nations Economic and Employment Commission. Other service included membership of the board of the International Monetary Fund and World Bank. These positions resulted in him working in the USA for 18 months. Wilson was appointed Secretary of the Treasury in 1951.
During his time at the CBCS, Wilson made some major contributions to statistics. He was instrumental in developing Australia’s first balance of payments statistics (and possibly the first in the world). These were published in the 1934 Year Book.
Wilson took a keen interest in the methodology underpinning the Bureau’s collections. This would often be done by ‘poking’ into the statistics to see how the figures were put together to get a better understanding of the underlying assumptions. He was often horrified at what he saw and this led to a major overhaul of several statistical collections. One such case was a major upgrade of retail price indexes (Series C).
With Giblin’s support he also developed production indexes and a range of business statistics. This work involved a significant research agenda. To assist he created ‘research officers’. This was a first for the Australian Public Service. In an era when virtually all public servants started as messengers or base grade clerks and worked their way up through seniority rather than capability, Wilson employed and promoted university graduates as ‘Research Officers’, encouraged all staff to study at university and helped key staff to gain experience overseas. To do this he initiated overseas scholarships for public service graduates.

4. The Future
The Bureau has provided a very effective national statistical service throughout its rich history. Of course the way statistics are produced and disseminated has changed rapidly, particularly since the advent of electronic dissemination. One of the strengths of the ABS has been its ability to use new methodology and technology to improve its statistical service. This has led to improvements in the statistical product and in the effectiveness with which statistics are collected, processed and disseminated.
But some things have remained constant:
• the core purpose of the Bureau’s work i.e. to provide official statistics for both governments and the community;
• the support of a succession of governments for a strong national statistical office, including the appointment of leaders of the Bureau who were able to put their own stamp on the job (as Commonwealth Statistician until 1975 and then as Australian Statistician); and
• the importance accorded by the staff to the values of integrity, professionalism, equality of access and protection of the confidentiality of information provided to the Bureau, by individuals and organisations.
Our forefathers have created an excellent legacy and foundation for our future development. This is important, as, like all effective organisations, we cannot rest on our achievements. The ABS does and will continue to learn from the experiences of the past in working out how best to position itself for the future.
Bill McLennan, in his article in the 2001 Year Book, articulated a number of important challenges for the future. He concluded:
The basic role of the ABS is unlikely to change, although how it does things and what it does may change significantly.
The rest of this section is devoted to some of the more important challenges for the immediate future. In looking at how the ABS may evolve, it is useful to consider the ABS both as a statistics provider and as a statistical leader.
(a) The ABS as Statistics Provider
The core role for the ABS will of course be to continue to provide a statistical service that is timely, relevant, responsive and respected for its integrity and quality. The users
of this service will continue to expect the most appropriate sets of statistics and that they will be of a quality fit for their needs. But the mix of needs – in terms of statistical content, level of detail, timeliness and presentation media – will evolve. A key challenge for the ABS will be to remain abreast of the changing needs of users and to respond to those needs appropriately. Among other things, the demand for timelier and more detailed statistics is continuing.
It would be foolhardy to try to predict future changes in statistical themes, but they are likely to be substantial. If you look back 15 years, the ABS did not produce State accounts, environment statistics, statistics about most service industries (the only service industries really covered were retail and wholesale trade), information technology statistics, statistics about cultural and recreational activities, or statistics about Indigenous people (except for a few items in the population census). This is a small sample of the many changes in the statistical program over that time. More changes can be expected in the future many of which will pose unique measurement challenges.
Producing statistics about a topic involves many facets. In particular, for the statistics to be meaningful it is necessary to work closely with their users, especially the policy analysts, to better understand the underlying issues. Only then can appropriate statistical frameworks be developed to give context to the statistics and help to make them relatable to others. Such frameworks should not be developed in isolation. Where possible, they should be related to existing international frameworks such as the System of National Accounts. Our colleagues in other national statistical offices are often confronting similar problems. Collaboration with them is important, not just to share knowledge, but to facilitate the comparison of data across countries.
There will be changes in the way the ABS collects data. Until the 1950s most statistics were collected through censuses or from administrative systems such as the Customs system for international trade data or the registration systems to capture births, deaths and marriages data. Sample surveys, introduced midway through the last century, have led to an incredible increase in the amount of statistical information available, particularly on social topics. Whilst sample surveys will continue to be the main source for official statistics, data from administrative systems are making something of a comeback as a source for statistics. They have always been used extensively in the Scandinavian countries, which have a rich history of registers of various forms.
In Australia, because of the evolution of computer technology, data from a range of administrative and transactional data bases are now more readily available. In the past the ABS would normally have been expected to produce official statistics based on those systems, but this is no longer the case – administering agencies are often best placed to compile the statistics themselves but the ABS can provide a Another method for compiling official statistics is likely to increase in prominence – the use of analytical or model-based methods. A number of innovative statistics have been produced in recent years using these methods, such as small area disability statistics, estimates of human capital, and wealth distribution statistics. These methods can be used to produce statistics that are relevant and responsive to user needs. The challenge is to be able to validate the underlying models, and to describe them and their assumptions clearly so that users understand what is behind the statistics. It is also important to be able to describe data quality in a meaningful way. Subject to these preconditions, there is no reason why analytical methods should not become more prominent in the production of the official statistics.
There is another likely change. What is increasingly possible is the ability to link data sets to make them much richer for statistical purposes. One such example is the linking of successive population censuses to provide a longitudinal data file and a rich source for medical research, lifecycle analysis and so forth. Our sister agencies in many other countries have started down this path of using linked data sets for statistical purposes. We have been more careful. There are privacy issues that need to be carefully managed. We would not do anything that would threaten the confidentiality of those who provided the data, nor are we allowed to by law. Also, we do not want to go further than what the Australian public regards as reasonable. But the potential benefits are significant, so linked data sets are definitely a development worth pursuing.
The other big change in data collection will be the use of the Internet. Over the last 20 years, technology has changed the way in which data is collected and captured. Some of the more prominent developments have been Computer Assisted Telephone Interviewing (CATI) for the monthly retail survey, Computer Assisted Personal Interviewing (CAPI) for the monthly labour force survey and a range of social surveys, Optical Mark Reading (OMR) for population surveys or Intelligent Character Recognition (ICR) for the processing of the population census. An e-form is being designed for the 2006 population census. The take-up rate is expected to be about 10%, but this will surely increase over time. Businesses are more interested in reporting over the Internet, especially if statistical returns can be automatically extracted from their own accounting systems. Computer languages such as XML make this increasingly feasible. Some national statistical agencies are examining these possibilities aggressively; we will watch these developments with interest, and pursue them at the appropriate stage.
Technology has facilitated continued improvement in the quality and efficiency of statistical processing. This will continue to be the case. We are increasingly moving to a single architecture across Bureau computer systems for the collection and processing of statistical collections. Among other benefits this will make technological change much easier to manage.
With regard to statistical outputs, it is hard to believe it is only a decade since the ABS first established its website. Now, apart from information provided through the media, accessing our website is how most statistical users obtain ABS data. This trend will continue with the rapid increase in the demand for data. The Internet has provided the means by which the ABS might be able to satisfy this demand without undue impact on its resources. It enables ‘self-help’ facilities, which allow users to generate more of their own statistical outputs without relying on the services of ABS staff.
One important ‘self help’ initiative is the ABS’s Remote Access Data Laboratory (RADL). This facility enables registered users to submit statistical processing requests direct to the ABS’s microdata bases. It also enforces a range of confidentiality checks, supported by manual checks and audits. The use of these remote access facilities, and the types of services that are available, will continue to grow and enable much better access to detail for our more sophisticated users.
More generally, our sophisticated users are looking for improved access to microdata for research and policy analysis purposes. While fully understanding this need and trying to support it, the ABS will continue to ensure that it maintains the trust and confidence of respondents. Without that trust, cooperation in our surveys would be much lower and the quality of the resulting statistics would suffer. This involves finding a continuing balance. While our legislation sets limits on disclosure, it may not always be sensible to go to the limits it allows. This is an issue being debated internationally by national statistical offices, and the ABS has taken a leadership role in trying to get to an agreed position. More sophisticated ways of supporting the research community can be expected.
Statistical users are also looking for increasing detail (but not Microdata) to support regional analysis, industry studies and the like. The ABS will take advantage of automated geocoding facilities to support the need for finer levels of geographical data. This is one of the reasons we have introduced the ‘mesh block’ as the finest level geographical unit from the 2006 population census. It is a much smaller building block than a Census Collectors District so provides users with greater flexibility in defining their own areas. This provides a number of challenges with the protection of microdata being paramount.
Increasingly, we find that users want to compare statistics for Australia with those of other countries, to provide a context for Australian figures. Inter-country differences can often be very illuminating in evaluating the effectiveness of current policy or for assessing alternative policy options. This is one of the main reasons why the ABS is an active contributor to official statistics internationally. To quote from the Corporate Plan, it is an ABS strategy:
To increase the availability of comparable statistical data with other countries through:
• leadership and support in the development of international statistical standards where they are important to us and where we have a position of influence;
• expanding international data available in our publications and other statistical services.
International comparability of economic statistics is relatively well developed. Moves are now afoot to do more for the comparability of social statistics. The approach may be different – international surveys are likely to be a key source of internationally comparable data for social statistics.
Given the growing importance of international comparability, it is arguable whether the international statistical system is sufficiently well set up to play a leadership role. The OECD provides an excellent range of comparable data for member countries (and increasingly other large countries). The United Nations Statistical Office produces useful data on international trade, and population. Also, the mechanisms for setting new international standards for economic statistics have worked reasonably well. But there are some doubts as to the ability of the international statistical system in its current state to play a more expansionary role.
In my view the international scene is not in good shape at the moment and would not be capable of facing such challenges, being at its best chaotic and at its worst very ineffective.
Unfortunately the statisticians of the countries of the world have not played an important enough role in the management and development of international statistics policy.
Undoubtedly changes will be needed in the future if the international statistical system is to be as effective as it needs to be. The national statistical offices have always cooperated with each other and with international bodies in the development of frameworks, standards and systems. But there are now clear signs of increasing cooperation among the international agencies. The United Nations Statistical Commission needs to play a leadership role in the management and development of international statistical policy.
(b) The ABS as Statistical Leader
The ABS has a responsibility for the coordination of official statistics. It is fair to say that, over its history since the ABS Act was passed, the Bureau has struggled to decide how to best fulfil this responsibility. The need for the ABS to take a leadership role is becoming clearer and is likely to become more so in the future. Not surprisingly, this has become more apparent at the same time as the amount of statistical activity outside the ABS has increased. There are other providers of sample survey services, which are utilised from time to time by government agencies. But most importantly, many Australian and state government agencies are custodians of administrative data which are potentially a very valuable source of statistics. Australia is evolving towards a National Statistical Service where the ABS is only one of the providers, albeit a very important one.
There is support for the ABS taking a leadership role in the development of national statistics. Why is this the case?
• Government agencies increasingly need to work in a ‘connected’ way. This will only happen if they are prepared to share information, including statistical information.
• It is important that this information can be related – that is, we are using the same concepts and definitions to the extent possible. This requires leadership on standards and classifications, a role which the ABS is well suited to play.
• It is important that the range of statistics be of good quality – sound statistical methods should be used. Again the ABS has a constructive role to play.
What is meant by statistical leadership? One way of illustrating is to summarise the type of activities that might be pursued. They could include:
• Developing standard classifications (eg an industry classification) and making them available to other providers of statistics. Additional support could be provided through training, coding manuals, software systems to support coding etc.
• Developing and promulgating statistical frameworks, statistical standards and definitions for use by all providers of statistics. Publications of modules of standard questions on particular topics (e.g. employment) can help improve coherence across different sources of related information. These frameworks should be based on international frameworks where they exist.
• Disseminating Manuals of Good Practice and providing training programs to support them. More generally, providing good guidance on other aspects of statistical production.
• Maintaining active networks among the key personnel involved in statistical activities through newsletters, seminars, social gatherings, etc.
• Developing agreed protocols for the National Statistical System.
• Developing meta data standards for describing collections, particularly the quality of the statistics derived from these collections. Data about collections are an important input into the Directory described below.
• Within the structure of the ABS, developing statistical centres of expertise for particular subject matters. These statistical centres should have good knowledge of all statistics produced in particular fields, not just those produced by the ABS. For example, a statistical centre for agriculture would be familiar with agriculture statistics produced by the ABS, the Department of Agriculture, Forests and Fisheries, by State bodies involved in agriculture, producer boards and research institutes involved in agriculture. The statistical centre should also be interested in links with other fields of statistics (e.g. agriculture and the environment, agriculture and the household sector). In recent years, the ABS has set up a number of such statistical centres.
• The production of Information Development Plans, in collaboration with key stakeholders, that describe the availability of existing statistics, the major gaps in these statistics or the major improvements required, and a plan for further development in the field of statistics.
• A Directory of statistical sources, from both the ABS and elsewhere. If electronic this could provide a window on available statistics. Given the technology of today, this may not be a physical directory but rather a virtual directory using ‘Google search’ type facilities across the released data bases. Standards for describing meta data become very important in this context. Links to the actual data, not just the meta data, would be a very valuable extension.
In addition to making advances in this range of activities, the ABS is looking at a range of other new initiatives to improve statistical leadership.
One of the most important is the ABS’s National Data Network initiative. The Network will create a distributed library of data holdings relevant to policy analysis and research. These data holdings will remain held and controlled by their custodian organisations. Whilst data will be held by each custodian, the National Data Network will provide a complete catalogue of available data sources to allow users to easily search for, and access data holdings which have been published. In effect, it will provide a portal to official statistics.
The National Data Network will also provide access to a range of services to support the creation, management, integration and analysis of data.
5. Concluding Remarks
The ABS does have a fine history and, I believe, has served Australia well. It plays a vital role in an Australian democracy – not just because it provides information which provides a mirror on society – but because that information is trusted.
This trust has proven important to governments as well. Because of this trust, discussions can focus on what the statistics mean for policy rather than on the integrity of the statistics themselves.
Our history has provided a fine shoulder on which to stand as we address the challenges of the future. There must be changes if we are to remain relevant and provide value for the money appropriated to us. But more than anything else we have to be careful that we do not lose that trust – it is our comparative advantage. If we lost trust, we risk becoming just another information provider.
6. References
ABS (2001), Year Book.
ABS (2005), Informing a Nation — The Evolution of the Australian Bureau of Statistics.

Articles, Library Science
Commonwealth statistical function, Sir Roland Wilson