UK PARALLELS WITH THE US FEDERAL STATISTICAL SYSTEM

There are some interesting and informative parallels with the co-ordination functions that will be needed in the UK under the new framework created by the Statistics and Registration Service Act. This paper looks at some of these parallels.  In doing so it refers to organisations within the US Federal system without describing their roles in detail; our aim is not so much to describe the US system as to identify how the authorities there have sought to deal with issues that correspond to ones that have been identified within the UK system.

Even compared with the UK statistical system – itself fairly complex – the US Federal statistical system is more decentralised and multi-faceted, with some ten major statistical offices[1] and a further 70 statistical agencies embedded in other government bodies. Its development since the beginning of the twentieth century has been evolutionary and mostly without a central plan. However, as it has developed, a number of statutory and informal mechanisms have been created for ensuring co-operation and co-ordination among the different agencies. In this sense the US and UK statistical systems have something in common and are fundamentally different from the more common European and international models in which a single national statistical office was established at an early stage in the evolution of the statistical system and further developments then centred on the powers and the responsibilities of that central office. 

Thus one parallel between the US and UK statistical systems is that their governance superstructure has in each case emerged out of a recognition, by both the legislature and executive, that a system composed of several distinct and largely autonomous agencies requires active central co-ordination and leadership. There are however important differences between the US and UK solutions; in the US, the Chief Statistician, who is a senior official within the Office and Management Budget (part of the Executive Office of the President) has no direct responsibility for the production of official statistics, has only a few staff, but in effect exercises substantial statutory authority over the statistical system. This statutory authority is not vested in her as an individual but in the Director of OMB. Nonetheless, to a large extent, it is exercised at the discretion of the Chief Statistician.  In the UK, the National Statistician has a large and direct responsibility for a wide range of key statistical functions but little direct authority over those parts of the UK statistical system outside the Office for National Statistics.  This will remain the case under the new statutory regime in the UK but a parallel can still be drawn between the statutory role of the Office of Management and Budget on the one hand and the statutory role of the new UK Statistics Board on the other. To the extent that the UK National Statistician is able to exercise authority with the agreement and on behalf of the future Statistics Board, the parallel will be stronger still.

In the US, the statutory authorities derive from several pieces of legislation and executive orders. One important statute is the Paperwork Reduction Act of 1995 though this essentially updated functions that have resided in OMB since the 1930s. OMB has nine main statistical policy and co-ordination functions. In brief these are to:

 

1. Co-ordinate the activities of the Federal statistical system to ensure:

• Efficiency and effectiveness of the system;
• Integrity, objectivity, impartiality, utility, and confidentiality of information collected

 

2. Ensure that budget proposals of agencies are consistent with system-wide priorities ….and prepare an annual report on statistical program funding

 

3. Develop and oversee the implementation of Government-wide policies, principles, standards, and guidelines concerning:

• statistical collection procedures and methods
• data classification
• presentation and dissemination
• timely release of statistical data
• statistical data sources required

 

4. Evaluate statistical program performance and agency compliance with Government-wide policies, principles, standards and guidelines
5. Promote the sharing of information collected for statistical purposes consistent with privacy rights and confidentiality pledges
6. Co-ordinate US participation in international statistical activities, including the development of comparable statistics
7. Appoint a Chief Statistician who is a professional statistician to carry out specified functions
8. Establish an Interagency Council on Statistical Policy to … be headed by the Chief Statistician
9. provide opportunities for training in statistical policy functions to employees of the Federal Government

Allowing for some differences in language, there are some strong similarities with the statutory role of the Statistics Board under the Statistics and Registration Service Act. Whilst the OMB functions do not make explicit reference to a Code of Practice it is clear in the reference to developing and overseeing ‘the implementation of Government-wide policies, principles, standards and guidelines’ that a set of common standards and practices is implicit. A key document in support of this is the Principles and Practices for a Federal Statistical Agency[2], published by the Committee on National Statistics (third edition, 2005).  Similarly, the requirement to ‘evaluate statistical program performance and agency compliance with Government-wide policies, principles, standards and guidelines’ might be seen as analogous, at least in intent, to the statutory Assessment function required under the new UK legislation.

There are however some elements of the OMB functions that do not, yet, have a clear counterpart within the UK system. One of these is the requirement to ensure that budget proposals of statistical agencies are consistent with system-wide priorities and prepare an annual report on statistical program funding. In the US, this report is published annually under the title Statistical Programs of the United States Government[3]. This, among other reports, supports the oversight role of various Committees of the Congress of the United States that take an interest in statistical activities and budgets.  

There are elements of this model which the UK Statistics Board may find valuable. As yet, there has been no announcement about which parliamentary select committee will take the lead responsibility for the work of the Statistics Board but it is likely that several parliamentary committees would have an interest in a full analysis of statistical budgets and plans.  The Statistics Commission has long advocated the need for much greater pan-government planning and reporting of statistical programmes.

Under the US legal framework, the role of the Chief Statistician and her staff can be set out under four headings:

·  Long range planning and budget formulation – the role is to formulate long range plans to improve the performance of the Federal statistical programs so that robust measures are available for use by public and private decision makers. These plans are reflected in a Statistical Programs Crosscut as part of the President’s budget planning process, and a chapter on Strengthening Federal Statistics for the President’s budget documents.

·  Policy and standards setting – these standards include classification systems for industries, products, occupations, geographical units and the collection of data on race and ethnicity; also standards for statistical surveys, maintaining confidentiality and the compilation and release of economic indicators.

·  Statistical programme evaluation and review – under the Paperwork Reduction Act, all information collected from ten or more members of the public must be approved by OMB. This provides the Chief Statistician with a powerful tool for oversight of statistical survey work, promoting new work where appropriate as well as containing burdens on data suppliers. But there is also an important related role for the Committee on National Statistics (CNSTAT) of the National Academies. This body is independent of government and has been in existence since 1972. Its original mandate was to provide an independent, objective resource for evaluating and improving the work of the Federal statistical system. Its work is funded directly by the National Science Foundation (NSF) and the Federal statistical agencies, through a mixture of core subscription funding (25 per cent of total budget) and repayment for individual reviews. CNSTAT has developed a strong reputation for the quality, depth and professionalism of its reports. Each report is produced under the supervision of a panel of relevant experts who do not receive payment. The result appears to be thorough and influential even if the process can be extended and sometimes costly.  A few examples of the many report topics illustrate the range of it’s work[4]:

·  Alternative Census Methodologies

· Confidentiality and Data Access

·  Demographic and Economic Impacts of Immigration

·  Eliminating Health Disparities: Measurement and Data Needs

·  Statistical Issues in Allocating Funds by Formula

·  Measuring Access to Learning Opportunities

· Improving Racial and Ethnic Data on Health

· Re-engineering the 2010 Census: Risks and Challenges

The diversity of the reviews undertaking by CNSTAT and the respect which the reviews are afforded suggest that there may be valuable lessons here for the development of the Assessment function within the UK system. In particular, there are aspects of how CNSTAT assures the quality of its own reports that the UK system may find of relevance.

- Interagency and International Co-ordination – the main vehicle for this is the Interagency Council on Statistical Policy (ICSP) which comprises the heads of fourteen statistical agencies at any one time advises the Chief Statistician on the various co-ordination functions including matters such as the formulation of principles, standards and guidelines.  In the opinion of those to whom we spoke, the ICSP has proved an effective vehicle for co-ordinating statistical work, particularly when activities and issues cut across agencies. One of its initiatives has been the development of FedStats, an interagency website offering access to a wide-range of Federal statistics. There is also a Federal Committee on Statistical Methodology which produces technical working papers on methodological topics. OMB co-ordinates the Federal Government’s participation in international statistical activities drawing on the expertise of ICSP members as required. 

Data sharing and the Census Bureau

The role of the US Census Bureau in the co-ordination of statistical activity has some parallels with the current and prospective role of the Office for National Statistics. The Census Bureau is a large agency with a wide range of outputs beyond the decennial census – including the American Community Survey (ACS – a large rolling survey which replaces the census ‘long form’), small area income and poverty estimates, the National Longitudinal Mortality Study, small area health insurance estimates, quarterly workforce Indicators, detailed local economic statistics, mapping products and many specialised reports. These are widely used within the Federal Government and can be seen in many respects as a central statistical service to other agencies.

To produce these outputs it draws on the Census and ACS data but also on a wide range of administrative records from other government agencies. Indeed it is under a statutory obligation to ‘use administrative records information as extensively as possible in lieu of conducting direct inquiries’. Among the sources on which it draws most heavily are tax records from the Internal Revenue Service; Medicare and Medicaid records; State Unemployment Insurance Files; and Food Stamp participation records. The tax data alone feed into ten Census Bureau programmes. Whilst the Bureau does not have an absolute right to access the administrative records of other agencies – unlike in some countries such as Canada – it does have the support of the Federal Government in doing so and whilst each use of administrative records must be individually justified – and requires the formal agreement of the agency that holds the records – as well as having to comply with various other forms of approval, the process is clearly well established and effective. The Census Bureau also operates strict security arrangements to protect the confidentiality of the records that are provided to it.

Under the Statistics and Registration Service Act, the statutory impediments to the Office for National Statistics making equivalent use of administrative records in the UK will be eased. But there are statutory, cultural and historical differences in the standing of ONS as compared with the Census Bureau that could yet impede full use of UK administrative data for statistical purposes. It may however be possible to draw on the evident success of the US Census Bureau in maintaining the confidence of other agencies and the public, and producing a wide range of valued outputs, to illustrate more forcibly the potential benefits.

 

The Census Bureau and the 2010 Census

It is worth noting that the 2010 US Census will be radically different in certain respects from its predecessor. To that extent, the situation has some parallels with developments in Finland, the Netherlands, Norway and Sweden which have largely dispensed with a traditional census. But the US approach is different in that it relies upon the new ACS (see above) to collect – in effect continuously – detailed information across the country with the decennial Census collecting a much smaller core of information than previously. The next census will make use of hand-held computers and other new developments; the Census cycle will cost over $11 billion and employ 500,000 at its peak.  It is noteworthy that the prime driver for introduction of the ACS was not the saving of money but rather to get better reliability from a continuously employed and better trained if smaller workforce – and to deliver an annually up-dated description of the changing demography and socio-economic characteristics of small areas in the USA.

 

Conclusions

It is not possible in a short paper to offer a balanced view of a statistical system as highly evolved and distinctive as the US one. However, the insights provided by all the people we spoke to have been of real value to us and point to some informative parallels with the evolving UK statistical system.  We would like to thank everyone we met for their immensely helpful and hospitable welcome. A list of some of those we met is appended.

VISIT TO FEDERAL STATISTICAL OFFICES, 21-24 OCTOBER 2007

Among those we met:

Katherine K Wallman , Chief Statistician,

 

Committee on National Statistics  

Constance F Citro, Director, Committee on National Statistics, The National Academies

Miron L Straf, Deputy Director, Division of Behavioural and Social Sciences and Education, The NationalAcademies

US Census Bureau  

Charles Louis Kincannon , Director,

Preston Jay Waite , Deputy Director and Chief Operating Officer,

Susan Schechter Bortner , Chief, American Community Survey office

Richard A Denby , Assistant Division Chief, Data Integration Division

Dawn E Haines , Assistant Division Chief, Data Integration Division

Carole Popoff , Chief of International Relations

 

Others

Peg Young , Bureau of Transportation Statistics