Memorandum submitted by the Statistics
Commission
EXECUTIVE SUMMARY
1. The Sub-Committee's invitation to submit
written evidence indicates that the inquiry will focus on the
Government's current consultation paper. We believe there needs
to be clarity about the problem that the government proposals
are intended to address. Our own formulation is that Government
and Parliament must act to give the public much greater confidence
that official statistics are being managed in a way that reflects
the public interest.
2. This is not primarily a question of the
accuracy of the statistics nor of the competence of government
statisticians. It rather relates to concerns about the decisions—often
taken by ministers and senior departmental officials—on
matters such as resources, priorities, presentation, release practices
and communication arrangements. We need to be sure that these
things are decided in the public interest (recognising that there
will always be trade-offs between cost and provision), without
partisan pressure to select, shape or delay the product.
3. We have welcomed the Government's commitment
to legislation. We believe that a statutory structure—which
most countries already have—is essential to provide a solid
foundation on which trusted arrangements can be built.
4. We have previously set out a number of
principles against which the detailed proposals should be judged.
On the basis of the consultation document, we think it is unclear
whether the Government's plans meet those principles. In some
important respects, we believe that more development is needed:
— All official statistics should be
included within the arrangements.
— The nature and scope of the Code
of Practice—it must be a binding, statutory code agreed
by the Board that defines the practices which government departments
must follow.
— Transparency in funding and planning
arrangements.
— The scrutiny role of the Board and
its powers of audit—it should be able to commission investigations
and to audit compliance with the Code of Practice anywhere in
government and report to Parliament at its discretion.
— The role of the Board in decisions
about the scope of National Statistics—it should not only
have the right to decide that any output is inadequate in relation
to the Code but also a role in suggesting to ministers additional
outputs that should be within the scope of National Statistics.
— The power of the Board to approve
or reject statistical plans—it will need the authority to
approve or reject annual plans covering all the statistical activities
of government.
— The need to draw in the devolved
administrations to the new arrangements.
— The role of the National Statistician
in the development and deployment of all professional staff—legislation
will need to underline the importance and centrality of the National
Statistician's personal role, including her role in relation to
all professional statistical staff in government.
INTRODUCTION
5. In developing our views, we have kept
in mind the arguments in our report Legislation to Build Trust
in Statistics (2004), and the messages from our 2005 report,
Official Statistics: Perceptions and Trust. We have also
drawn on our experience of carrying out the remit given to us
in the Framework for National Statistics (June 2000) and
ongoing discussions with Treasury, ONS, professional and international
bodies and users of statistics. On 17 May 2006, together with
the Financial Times and the Royal Statistical Society, we held
a major seminar for 150 people to discuss the issues. This valuable
event informed the views expressed here.
6. There is substantial evidence that people
do not much trust official figures or the use that is made of
them. This lack of trust can weaken both public administration
and confidence in it. Government statistics are used in allocating
vast amounts of money to public bodies, judging their performance
and shaping policy. If the people in the system do not trust the
figures, they may not use them effectively. If the general public
do not trust the figures, they will not trust the decisions made
on their behalf.
7. Of course, once figures are in the public
domain, political interpretation and argument will follow. But
when statistics first appear we want to see statisticians comment
openly on their reliability and what can safely be drawn from
them.
8. The current proposals are close to one
of the three models in our report Legislation to Build Trust
in Statistics but not the particular one we recommended. We
believe that the differences in emphasis—which relate primarily
to the role of the proposed Governing Board—are not yet
adequately addressed in the consultation document.
CRITERIA FOR
NEW ARRANGEMENTS
9. The Chancellor's announcement in November
2005 made clear the Government's intention to replace the existing
arrangements—under which there are non-executive members
on the ONS board, as well as independent members of the Statistics
Commission—with a single Governing Board that embraces both
sets of responsibilities. We wrote to the Treasury in January
2006 setting out some principles against which the detailed proposals
should be assessed. These included:
— The new arrangements must relate
to all official statistics, not just those that are currently
the responsibility of ONS. Many sensitive and controversial statistics
are produced by the major Whitehall departments and based on administrative
data—crime, education, health, etc—not on surveys.
The new arrangements must recognise this or risk public confidence
in such statistics being reduced rather than enhanced.
— In order to engage with all the statistical
work carried out by largely autonomous government bodies, a key
component of the new arrangements must be a binding, statutory
Code of Practice agreed by the Board that defines the practices
which those bodies must follow.
— Given the existing spread of statistical
work across government, ONS must continue to work in co-operation
with other government departments that produce statistics. This
may require the new ONS to remain a government department—as
indeed the consultation document recognises. However, ONS funding
arrangements will need to be more transparent than currently;
and auditable in relation to their adequacy to meet agreed plans.
— We see the role, responsibilities,
accountability and constitution of the Board as crucial. The Commission
believes that to bolster public trust and work effectively, the
Board should be able to commission investigations and to audit
compliance with the Code of Practice as it sees fit anywhere in
government and report to Parliament at its discretion.
— In addition, the Board will need
the authority, subject to Parliamentary accountability, to approve
or reject annual plans covering all the statistical activities
of government.
— The Statistics Commission has a UK-wide
remit and we would expect the Board to have such a remit. We recognise
that some special provisions may need to be worked out with the
devolved administrations.
THE GOVERNMENT
PROPOSALS
10. There are a number of aspects of the
consultation document that we regard as unclear or unsatisfactory.
We have drawn the Treasury's attention to some of these and sought
clarification. Until these are resolved, we cannot fully support
the proposals.
11. Whilst we welcome the fact that legislation
is central to the consultation document, it needs also to be recognised
that much of the management of statistical work will properly
remain part of normal civil service activity, governed by legislation
but not described in it. For example, legislation is unlikely
to say anything about the committee structures that co-ordinate
"cross-departmental" statistical activities. The proposed
Code of Practice may need to address a range of such issues that
are not appropriate for inclusion in the legislation itself and
are not currently dealt with adequately in existing Codes and
agreements.
12. Some more specific points:
(a) The role of the National Statistician
(or Chief Statistician) is not fully described in the document.
This runs counter to international practice. The Commission believes
that the legislation will need to underline the importance and
centrality of the National Statistician's role. We are clear that
the National Statistician must remain the Government's, and the
country's, top adviser on statistical matters; analogous in some
respects to the Chief Medical Officer. The powers and composition
of the Board should be such as to enhance that duty.
(b) The Code of Practice must be drafted
to provide a framework against which the actions of government
departments and their agencies can be judged unambiguously. There
is a lack of clarity in the consultation document about the status
and scope of the Code. We want to see a revised Code drafted by
the National Statistician and approved by the Governing Board.
It will need to cover issues that the current (non-statutory)
Code does not fully address. These include the planning arrangements
for the statistical service as a whole, user consultation, data
release practices and the professional oversight of government
statisticians. Development of the current Code should be a priority.
(More specific views on development of the Code are set out in
our report Legislation to Build Trust in Statistics, 2004.)
(c) The role and powers of the National Statistician
and of the Board vis-a"-vis the statistical activities
of government departments are key. The National Statistician must
have authority to match his or her responsibility in being the
government's senior adviser. The Board must have sufficient authority
to be able adequately to assess the actions of government departments
and agencies against the Code and to fulfil its broader strategic
role. The discussion of these matters in the consultation document
is scant and ambiguous. The Commission believes that the Board
should:
— report directly, and be accountable,
to Parliament;
— establish and maintain the Code of
Practice;
— have the authority and responsibility
for determining compliance with the Code of Practice; and
— offer strategic advice on the future
direction of, and priorities for, statistical work.
(d) All normal managerial decisions within
ONS are properly the responsibility of the National Statistician,
as chief executive. The Board's additional role in relation to
ONS (over and above the scrutiny role that it should undertake
in relation to all government statistical activity) needs to be
defined in such a way as to support the National Statistician
in this capacity.
(e) We note the intention in the consultation
document to establish ONS as a "non-ministerial" department,
though we understand that phrase to mean "no specific minister"
rather than "no role for ministers". We assume that
the Government intends ONS to remain a Treasury department. But
the document is vague about the part that ministers and Treasury
officials will play in future decisions about ONS activities.
If ONS is to continue to come under a measure of ministerial control,
we would want to see consideration given to a Cabinet Office location,
rather than a Treasury one.
(f) We assume that ministers and Treasury
officials will be involved in setting an overall budget for ONS.
That will require them to take a view on the appropriate scale
of ONS activities. Moreover, government policy may have direct
implications for ONS priorities. Looking at recent examples, analysis
of options on pensions policy required the collection of better
statistics from the pensions industry. New policy on measuring
public sector productivity required major statistical effort and
resources. So there is a real question about the extent to which
statistical plans and priorities can be separated from ministerial
decisions. We do not want to see an environment in which a highly
independent ONS is sidelined and new statistical work is instead
directed by ministers to other departments. In this sense, "independence",
in terms of governance, should not be seen as the ultimate goal,
though assured independence from interference in professional
and interpretational matters is obviously important for public
confidence.
(g) It is essential that the accountability
of ministers, Treasury, the Board and the National Statistician
are aligned with their actual authority under new arrangements.
If the Treasury and ministers retain final authority over various
decisions, then they should properly remain accountable for them;
the Board's role should be to support Parliament in holding them
to account. These matters of authority and accountability are
not dealt with clearly in the consultation document which variously
presents the Board as being itself accountable for the quality
of official statistics and, elsewhere, holding others to account.
(h) The phrase "the statistical system"
is used in the consultation document. The Commission's own use
of this term embraces the functions of planning, funding, creation,
dissemination and communication with users—across all of
government. The commitment to transparency which the document
makes should, we believe, relate to the entire system. It is not
clear to us whether that is actually the Government's intention.
(i) The document refers to "assessing
against the code those statistics already produced in line with
the Code of Practice and designated as National Statistics".
We are sure that the Board must be free to assess any statistical
output against the code and that only those that fully meet its
requirements should be labelled as "National Statistics".
The Board should not only have the right to decide that any output
is inadequate in relation to the Code but also it should have
a role in suggesting to ministers (and publicising) output which
should be within the scope of National Statistics depending on
user requirements.
(j) Throughout the consultation document
there are references to assessing "statistics" against
the Code of Practice. However, much of the existing Code does
not refer to the properties of statistics as such but rather to
the actions and processes of bodies that produce statistics. References
to "assessing statistics" must be interpreted as "assessing
the relevant activities of all those departments and agencies
to which the provisions of the Code of Practice apply". This
is a very important distinction.
(k) The consultation document has a heading
"Non-executive majority", referring to the Board, but
this is not reflected in the text which merely refers to "a
strong presence of external members". Clearly the balance
of the board is critical to its role and we believe that a strong
non-executive majority is essential.
(l) The document proposes that ONS be funded
much as for other government departments but with a longer interval
between reviews. We are not sure that this is necessarily the
best way to respond to demand for an increasingly "evidence-based"
approach to policy and decision-making. Ministers, quite properly,
will continue to have some authority over priorities. ONS could,
however, be placed in the position of not having the resources
to meet the Government's own demands whilst facing a period of
several years before a funding review. Other parts of the Government
Statistical Service will continue to be funded by their departments
on a normal funding cycle. We would want to see the Board looking
closely at the adequacy of funding across departmental boundaries.
STATISTICS COMMISSION
13. The Statistics Commission has been in
existence since 2000. It published in February 2006 an assessment
of its own work largely based on a study by MORI (Perceptions
of the Statistics Commission). The main conclusions were that
a large majority of stakeholders thought that the Commission had
a positive impact and that its recommendations were useful. But
there was also a large majority who thought the Commission should
have more impact. These conclusions are consistent with our own
view and underlie our belief that the current scrutiny arrangements
are not strong enough to command wide public support.
GOVERNMENT STATISTICIANS
OUTSIDE ONS
14. We would expect the devolved administrations
to work closely with the Board, just as they do now with the Statistics
Commission. We recognise that some special provisions may need
to be adopted to reflect local autonomy but the principles adopted
should be the same.
15. We want to see government statisticians
in all departments working together, and sharing resources and
data, to address statistical issues that cross departmental boundaries.
There are many of these cross-departmental issues—concern
about statistics of migration into the UK and between regions
is a current example. It should be part of the role of the Board
to assess how well these issues are managed.
16. As far as practicable we would want
to see National Statistics, from whichever government department
they come, released first by the Office for National Statistics
with a comprehensive statistical commentary. We recognise that
this might not be applicable to statistics from the devolved administrations.
17. We believe that the phrase "professional
accountability to the Chief Statistician", as used in the
consultation document, has little real substance unless backed
up by the scrutiny role of the Board. The concept ignores the
fact that much important statistical activity is carried out by
departmental staff who are not professional statisticians.
LESSONS FROM
OTHER COUNTRIES
18. Most countries have overarching statistical
legislation though the models vary considerably. It is not necessarily
the case however, that models adopted elsewhere could readily
be adopted in the UK. In particular, it is common for statistical
legislation elsewhere to be restricted to the work of a single
national office. In the 21st Century, an increasing proportion
of statistical data is extracted from administrative sources—NHS
records, police records, social security records etc. This means
that authority over statistical work tends to concentrate where
the large holdings of administrative data are to be found. Thus,
an effective legislative framework for a country where statistical
authority is diffuse must start from that reality and not seek
to impose a model that may work elsewhere but sits awkwardly in
the Whitehall context.
19. One other lesson is that there is no
quick fix. Those countries, like Canada, where the statistical
arrangements are widely respected have had robust legislation
for a long time, 90 years in Canada's case. In the UK, it is now
a matter of creating a sound legal framework and then letting
the subsidiary and supporting arrangements, many of them non-statutory
in nature, gather strength and win confidence over time.
May 2006
|
