The Nordic Contrast - paper on visit to statistical offices of Finland, Norway, the Netherlands and Sweden - September 2007

During September 2007 the chairman and chief executive of the Statistics Commission visited the national statistics offices of Finland, Norway, the Netherlands and Sweden.  This paper does not seek to describe the work, or particular characteristics, of the four statistical services but rather to draw some comparisons between their approaches to broad statistical issues and the corresponding arrangements in the UK.

In matters of statistical governance, ‘Nordic’ is not so much a geographical reference as a set of characteristics and values that have evolved from the nineteenth century or even earlier. Several countries which are not northern European - New Zealand and Slovenia among them - share elements of the Nordic approach.  In the UK, aspirations to have systems more like those in the Nordic countries are regularly expressed by both statisticians and commentators; and the Statistics and Registration Service Act 2007 may facilitate some steps in that direction.

A European Commission survey[1] in Spring 2007 asked,  in each of the 27 EU member states, whether the public trust official statistics. The Netherlands, Finland and Sweden were in the top five, and we are sure Norway, not covered, would have been alongside these countries. The UK came an uncomfortable 27^th. There can be little doubt that the Nordic countries have something to which we should properly aspire.

Whilst the fundamentals of the Nordic model can be described in different ways, and take different forms in different countries, they typically include:

• A long established statistical office whose public service role and independence is formally recognised in legislation.
• Acceptance of the statistical office as a trusted source by both government and the general population.
• A reliance on registers – of population, households, addresses, businesses etc – rather than sample surveys as the basis for most of the key social and economic statistics. The development of register data varies between countries but in each case their use for statistical purposes has statutory authority and is fully accepted by society – indeed some offices are under a statutory obligation to use such administrative data if it exists and have corresponding powers to access records from any part of government.
• A long standing role for the statistical office to inform the public about society and the work of government.
• A positive relationship with the media in which the statistical office is regarded as a trusted source. The offices are mindful to avoid being the subject of news stories themselves. 
• Well developed arrangements for engagement with users of statistics and other stakeholder groups. However, user engagement continues to present problems that none of the four statistics offices visited believe they have fully resolved.
• Practical dominance in the field of official statistics, in the sense that few other official bodies seek to produce statistics separately from, or without the formal endorsement of, the national statistical office.
• Well developed ethical and professional codes, such as Statistics Finland’s excellent Guidelines on Professional Ethics, Statistics Sweden’s Sufficient Quality Criteria for Official Statistics and Statistics Netherlands’ Code of Practice[2].

It would be wrong to suggest that these characteristics are manifested in the same way in each of the four countries visited. There are some quite deep differences both in how the statistical offices are organised and the extent to which current arrangements are stable, as opposed to rapidly evolving. Finland and Norway have perhaps the most stable statistical regimes, with which both the statistical offices themselves and the respective government and population seem generally content. The challenges they face are about maintaining relevance, setting priorities and containing costs but their position in the system of government does not appear to be in question. In Sweden, the statistical arrangements have changed significantly and continue to evolve, arguably in ways no less fundamental than in the UK. However they have retained and, in some respects, enhanced the essentials of the Nordic model which do not seem to face significant challenge. 

While the four countries each have distinctive statistical arrangements and should not be  regarded as sharing a single philosophy or model, it is nonetheless possible to draw broad contrasts between their circumstances and those in the UK. We consider some of these contrast below.

Practical autonomy and independence

All four countries have trusted statistical offices with substantial independence; but their relationships with other parts of government vary. In Norway and Finland, responsibility for deciding society’s needs for statistics and statistical services rests squarely with the statistical office and, ultimately, the national statistician. Legislation and culture demand that the statistical office gives high priority to identifying emerging needs and responding to those within its resources. There is little chance of other parts of government establishing alternative centres of statistical knowledge and advice. This would run against  prevailing culture and would raise public suspicion about the motives of the body issuing the figures. The challenge for these statistical offices is to maintain the relevance and quality of their services in a changing world in the face of growing demand and constrained resources.

In Sweden the position is more complex. The statistical office provides specific services – sometimes on a contractual basis – to other government bodies. In 1994 the Swedish government decided to transfer responsibility for determining a large part of statistical requirements from Statistics Sweden to 24 other government bodies. However, Statistics Sweden has maintained its capacity and influence, partly through providing other government bodies with statistical services on repayment terms but also by being the single recognised source of statistical expertise and authority in the country.

The Swedish Council for Official Statistics, formed in 2002, assures co-ordination among government bodies. It is chaired by the Director General of the statistical office and comprises six of the heads of the 24 relevant government bodies on a rotating basis. In a manner broadly analogous to the UK position, these bodies have the authority to determine which statistics shall carry the label and symbol for “Official Statistics of Sweden”. This corresponds to the concept of “National Statistics” enshrined in the Statistics and Registration Service Act, although in the Swedish case it appears to be the current and potential Council members who take this decision rather than government ministers. In each case the label implies that the statistics have been produced in accordance with centrally determined and assured guidelines. Arguably, the new UK model has rather stronger central assurance procedures than the Swedish one.

The work of Statistics Netherlands is overseen by a Central Commission for Statistics (CCS) made up of 11 senior figures from commerce, academia and government. The Director General attends meetings but is not a member. The role of the CCS is to evaluate and approve the statistical offices long term work programme, assuring its coherence and relevance to society’s needs. The CCS is also involved drawing up the statistical budget and supervising the director general’s authority in a number of respects. Statistics Netherlands is the dominant producer of official statistics and has complete methodological autonomy, but its work programme is subject to a substantial measure of oversight in the public interest. There is a parallel with the new Statistics Board’s role in the UK and it might be helpful to compare experience in due course.

There are some more general messages here. Centralisation of statistical work within a single authority is seen to have many practical benefits, but it is recognised as not being a perfect solution and is not necessarily a one-way street; the 1994 reversal of central control in Sweden shows that. Nor is there one self-evidently successful approach to the creation of commissions, boards and advisory committees. These vary according to local circumstance and it is not unknown for such committee systems may be re-designed from time to time – the UK is not alone in its taste for reinvention of statistical committee structures. However, one strong message is that a central statistical office whose reputation for independence transcends its day-to-day working relationships with the rest of government is a great asset to both government and society. It can do much to secure trust in official statistics. In the UK, the Office for National Statistics has had neither the history (created in 1996), practical dominance nor, until recently, the statutory independence seen in the four countries we visited. Circumstances are now more favourable and the goal must be to emulate some of the best features of the world’s most respected statistical offices. 

History and legislation

All four countries established their statistical offices in the nineteenth century. All have strong statistical legislation which has been revised and updated over the years and which establishes the statistical office’s place within government and its authority over statistical work.  The legislation also gives the statistical offices power to access and link administrative information from elsewhere in government for statistical purposes. Whilst much has changed, and continues to change, each statistical office has established itself as the recognised and trusted source of statistical information and advice. 

The UK statistical system has a much steeper hill to climb to build public trust than the Nordic countries did 100 years ago. We have numerous encampments of statistical expertise spread through central government and the devolved administrations and, until now, little central authority. Whilst we now have a statutory framework, it is looser, lighter and less easily understood than in the Nordic countries. We also have a history of distrust of official statistics which the Nordic countries seem never to have had. Nonetheless, we have the beginnings of a strong system and perhaps have a few advantages over the more centralised and established statistical offices. The UK system is highly flexible and can readily adapt to changes in government structure including devolution and the distribution of authority to arms length agencies. The new Statistics Board will be able to exercise much greater central authority than was possible previously. The statutory assessment function is a potentially stronger and more structured way to ensure standards than we have seen elsewhere. So whilst the hill is steep, the route up it is now clearer.

The statistical offices recognise that reliance on registers does have some disadvantages. They cannot dictate what information will be held on the registers and they cannot prevent changes being made when deemed necessary for administrative purposes. Nonetheless, they speak with real enthusiasm for the many benefits of linked registers. Most strikingly, periodic Censuses can be replaced by a flow of continuously updated, and generally reliable, information about people, households and businesses, saving a great deal of money in the process. This can be supplemented where necessary with sample survey information that uses the register as a reliable sampling frame.

Given these advantages, we might ask why the United Kingdom, and other large European countries, are not making more rapid progress towards adopting the same approach. There seem to be two main reasons. One is that the task of establishing reliable population and other registers in large countries, with significant migration in and out, is likely to prove more difficult and expensive. Another is that public opinion is much less ready to accept that such registers will be in the public interest and used only in the public interest. Despite these reservations the development of computerised administrative records in the UK has moved on rapidly in recent years and looks set to continue.  The Department of Work and Pensions already has an extremely powerful register of personal information.  The Statistics and Registration Service Act has lowered the hurdles that need to be crossed before such information can be used for statistical purposes and, subject to strong ethical safeguards to protect personal information, we hope that it will be possible to extend the statistical use of such systems.

Relations with the media

If UK official statistics seem at times to have an insecure international reputation, the  reputation of the UK press and media seems to be consistently negative among statisticians worldwide. They are almost universally seen as aggressive, self-interested and cynical. It was striking in each of the four countries we visited that communication with journalists was mostly low key, relaxed, professional and positive – although occasional sharp comment was not unknown. The statistical offices took the view that statistics, not the statistical office, should be the story and generally sought to maintain a low profile whilst doing their best to help journalists understand the statistical material – including running training courses and providing telephone services.

Short of the UK news media itself changing, we are unlikely ever to achieve the comfortable level of constructive communication that exists between the statistics offices and journalists in the Nordic countries. But nor can we give up trying to establish a basis for progress. Defensiveness cannot be the way forward. Journalists must be made to feel more welcome and perhaps they will then listen more too.

Relations with users of statistics

We did not see, in any of the four countries, a perfect model for engagement with users, or even a particularly structured concept of who should be regarded as the important users. All four statistical offices accept the importance of improving focus on the use made of statistics and are doing what they can to ensure effective communication. There are advisory councils, working groups and committees of various kinds, almost always structured around statistical themes - such as education, economics, crime etc - rather than grouped by user type – government, public service, business, voluntary. In its guidance on official statistics[3], the Council for Official Statistics in Sweden says “it shall be clearly stated for what and how the statistics are intended to be used.” One Director General said that “without users, the service has no value”. These are sentiments that we strongly endorse.

One factor that complicates the user dialogue is that centralised statistical offices tend to have a provider-customer relationship with many other parts of government and it is easy for the priorities of those customers – sometimes paying customers -  to become dominant. One discussant said that government departments are expected to speak on behalf of users more generally. That is not a model that we regard as desirable in the UK. Our decentralised system has the potential to get the statistician closer to a range of users as long as all parts of government accept that this is the right thing to do.

We were impressed by the set of indicators of public and user satisfaction that Statistics Finland has established which gives regular updates on how the reliability and objectivity of its work are assessed. The other statistical offices also monitor their public image regularly. In the UK, we may need to decide whether it is the image of the central institution, ONS; the new authority, the Statistics Board; or of the service (the Government Statistical Service) that we should monitor. Given the range of producer bodies in the UK, it might prove most practicable and relevant to concentrate on assessing the reputation of the Statistics Board, whilst recognising that weaknesses in any part of the decentralised system could compromise the standing of the Board.

Pre-release access and the use of embargoes

The slight fuzziness about who should be regarded as important users of statistics is balanced in the Nordic countries by a sharp appreciation of the need to treat all users equally and to respect the needs of the general public. Rules on pre-release access to statistics are very strict. In most circumstances, all four statistical offices regard pre-release access as being in breach of the UN Principles of Official Statistics and the EU Code of Practice. We need to remember this in the UK where those provisions are given a looser interpretation. Our pre-release practices, soon to be statutory, are unlikely to enhance the UK’s statistical reputation abroad. 

The strict Nordic rules often extend to a ban on releasing statistics under embargo to the press – with a few minor exceptions. Embargo arrangements too are seen as inconsistent with the basic principle of equal access. The Statistics Commission shares the Nordic disquiet about both pre-release access and embargo arrangements.

End note

This paper does not do justice to the wide range of valuable insights we were given into the highly developed statistical systems in these four countries. However they point to some important messages for the UK statistical system and the Statistics Commission will take these forward in its advice to government.

 

David Rhind, Chairman & Richard Alldritt, Chief Executive


Visit to Statistical Offices

Among those we met:

Statistics Finland

Ms Heli Jeskanen-Sundström, Director General

Mr Jarmo Hyrkkö, Deputy Director General

Ms Riitta Harala, Director, Social Statistics

Ms Hilkka Vihavainen, Director, International Affairs

 

Statistics Norway

Mr Oystein Olsen, Director General

Ms Per Olav Lovbak

Mr Jan Magnar Byfuglien

 

Statistics Netherlands

Mr Joop van Lunteren, Chairman, Central Commission for Statistics                      

Mr Gosse van der Veen, Director General                                  

Mrs Ada van Krimpen, Director International Relations               

Mr Dick Meuldijk, Senior Adviser Policy Staff                              

Mr Peter Spannenburg, Secretary to the CCS and Senior Adviser Policy Staff

Mr Eric Schulte Nordholt, Senior researcher, Division for Social and Spatial Statistics

Mr Paul van der Laan, Senior policy adviser, Division for Social and Spatial Statistics

 

Statistics Sweden

Mr Kjell Jansson, Director General

Ms Gunilla Lundholm , Senior Adviser, Secretary of the Council for Official Statistics

Mr Claus -Göran Hjelm , Head of Register Coordination and Microdata Unit,

Mr Åke Bruhn, Project Manager, Executive Office

Ms Cathy Kruger, International Affairs