Memorandum submitted by Ruth Lea, Centre
for Policy Studies
OVERVIEW
These are the following key points of my submission:
— There is an overwhelming need to
restore the people's confidence in official data. Official data
have, arguably, been undermined by the current Government's own
behaviour. For example, there is the strong suspicion in the media
and the financial markets that the absence of Network Rail's liabilities
in the public sector balance sheet is because of pressure from
the Treasury on the ONS.
— "Independence for statistics"
is a fine phrase and of course it is a "good thing".
But such is the public's cynicism that fine phrases mean nothing
until they are translated into hard action.
— The Treasury's general proposals
have some merit. The proposal that the ONS should be removed from
Ministerial control and become a NMD answering directly to Parliament
with a governing board should be supported.
— But the relationship between the
ONS (headed by the all-important role of the National Statistician)
and its proposed governing board needs clarifying. The proposals
in the Treasury's consultation document are opaque to say the
least. There needs to be a clear definition of roles. The governing
board should have the non-executive scrutinising and advisory
role and the ONS should be the executive, delivery arm of official
statistics, headed by a statistician of the finest professional
credentials. Clarity of roles is the first and most basic rule
of governance.
— The decentralised nature of the Government
Statistical Service (GSS) needs to be reconsidered. If public
confidence is to be restored in "departmental" data
on, for example, crime, education and health they will probably
have to be taken out of their respective departments (HO, DfES,
DH) and transferred to the ONS, under the National Statistician.
HM TREASURY: INDEPENDENCE FOR STATISTICS:
A CONSULTATION DOCUMENT MARCH 2006: DETAILEd COMMENTS ON OPTIONS
FOR REFORM AS LAID OUT IN CHAPTER 4
Introduction
Objectives (4.1-4.4)
1. The Government's intention is to reinforce
the quality and integrity of statistics produced in Government
should be supported. But it could be argued that the Government
itself has been responsible for the undermining of trust in official
statistics in recent years. For example, the fact that Network
Rail's liabilities (worth about £20 billion) are not counted
towards the national debt is highly contentious. PFI projects
(worth about £24 billion) are similarly not counted towards
the national debt. The suspicion is that the ONS has been pressured
by the Treasury to keep these liabilities off the public sector
balance sheet to aid the Treasury in meeting its fiscal rules.
The recent reclassification of London & Continental Railways
(LCR) from the private to the public sector was an interesting
precedent of what, arguably, should be done with Network Rail.
2. Another area of concern is that on the
ONS's work on public sector output and productivity (covered in
the Atkinson Report, final report, January 2005), the measurement
of which is always problematic. Whilst it is not unreasonable
to try and make adjustments for "quality" of output
instead of just measuring outputs by the inputs (as was the traditional
method), concerns arise when the "quality" adjustments
include, for example, scores of GCSE results, when there is such
concern about "grade inflation" in public examinations.
In addition, given the fact that early ONS estimates of public
sector productivity (and productivity of the health sector) showed
falls for the years of the late 1990s and onwards, leads to the
suspicion is that the Treasury is pressuring the ONS for data
which show a "better" picture. The exercise risks being
dismissed as "fiddling the figures". There is an analogy
here with the redefinitions to the unemployment (claimant count)
data of the 1980s.
3. Another area of disquiet concerns the
data produced by departments (excluding the ONS), where the suspicion
remains one of political interference. Mention has already been
made of the highly suspect data on GCSE results from the DfES—the
same department that has targets for improving pass rates for
GCSEs. Similar suspicions, for example, surround crime data from
the Home Office where redefinitions and twin sources of data (British
Crime Survey and police data) only further serve to muddy the
picture, confuse the public and cast a shadow on the integrity
of the official data.
Structure of legislation (4.5-4.6)
4. The analogy with the Bank of England
is not a good one. In May 1997 a major decision was taken to hand
the function of deciding interest rates over to the Bank from
the Treasury. The Treasury's proposals for the ONS do not involve
such a change of function for the ONS.
5. The consultation document refers to the
proposed board's responsibility for "statistical quality
and integrity". But it is not clear whether this refers to
delivery or scrutiny. The board should be concerned with the non-executive
role of scrutiny (and advice) and National Statistics should be
concerned with the executive role of delivery. This distinction
is a vital (and basic) aspect of good governance.
Benefits of decentralisation (4.7-4.8)
6. Whilst noting the benefits and dis-benefits
for decentralisation, there must be disquiet about having statisticians
in departments under Ministerial control. I've already referred
to education and crime data. If independence of statistics is
to mean anything the same "independence" proposed for
the ONS should be provided for departmental statistics. The question
then becomes whether this is possible. If it is not, and I would
question whether it is possible, then the case for a centralised
service, where all data are compiled independently, becomes overwhelming.
The fact that Ministers are able to comment on data ("spin")
issued by their departments is, in itself, a very good argument
for a centralised service. The further Ministers are away from
data delivery the better.
Accountability to Parliament (4.9)
7. The Treasury's proposals appear acceptable—but
the proof of the pudding is in the eating.
Integrated independence (4.10)
8. This is a ghastly phrase, but I take
it to mean that the proposed board and the ONS ("statistics
office") would be both independent of Ministers and responsible
for the scrutiny and delivery of the data.
Detailed options for reform
NMD (4.11-4.12)
9. This looks acceptable—though it
is unclear what is the proposed relationship between the ONS (and
the National Statistician) and data produced in departments. The
governing board should basically NOT have an executive function—surely
its role should be one of scrutiny and advice.
Civil service status (4.13)
10. Agreed.
Scope (4.14-4.16)
11. There is a clear need for a Code of
Practice. But there is probably a case for the National Statistician
to revise the Code and have it approved by the proposed governing
board and Parliament.
12. It is difficult to see how the proposed
board could fulfil the role of "ensuring quality and integrity
across the system", when departmental data compilation and
delivery is still under departmental Ministers. And it would be
difficult, if not impossible, for Parliament to hold the board
to account for the "system as a whole" (I assume they
are referring to the whole of the Governmental Statistical Service
(GSS).) There is no mention of the National Statistician in this
section—there should be. He/she is crucial to the whole
exercise.
The Board
The Board (4.17-4.23)
13. Under this section, could I draw together
the concerns I have already voiced about the proposals. They are:
firstly, there appears to be a confusion between delivery and
scrutiny; secondly, the board's primary function should NOT be
an executive one (this should be for the ONS); thirdly, the difficulties
of the board's "ensuring quality and integrity" across
the National Statistics system, when the departments (and other
agencies) are responsible for data production outside the ONS.
14. There is a need for a clear separation
of roles between the National Statistician (to be responsible
for delivery of development and programme of the ONS) and the
proposed board (to agree the programme and hold the National Statistician
to account). Perhaps I am misreading the consultation document—but
the tenor of the document seems to be undermining the authority
of the National Statistician.
Minimising business burdens (4.24-4.25)
15. Whilst recognising the need for data
for business, survey forms can undoubtedly be a burden for business—especially
at a time when they have been burdened with extra complicated
business and employment regulations and an increasingly complicated
and arcane tax system. The CBI's remark concerning the reduction
of duplication is wholly to be endorsed. An integrated, centralised
ONS could well reduce the business burden.
Data access and protecting confidentiality (4.26-4.29)
and pre-release (Box 4)
16. There is a strong case for providing
administrative data for statistical purposes, subject to protecting
confidentiality.
17. The practice of pre-release—especially
to ministers who can then manipulate the "news agenda"—should
be curtailed and minimised. The proposal that all data releases
should be released from the ONS has great merit.
Board structure
Board structure (4.30-4.31)
18. There is no need for the chairman to
be a statistician. But whether he/she is or is not, it is vital
that he/she is not seen as a competing voice of executive authority
on the delivery of statistics. That voice should be, unequivocally,
the voice of the National Statistician. The board chairman should
not be in the position to second guess and undermine the authority
of the National Statistician. The consultation document appears
to be curiously silent on this critical issue. There is a very
clear need to sharpen up the proposed demarcation of accountabilities
and responsibilities of the board chairman and the National Statistician.
This is a vital governance issue if there is to be confidence
in the new system.
19. If the board is to have authority it
must have people of stature in key fields (crime, education, health)
who know what data are required in their fields and thus be able
to provide authoritative advice. They do not need to be statisticians—indeed
it may be better if they are not.
Chief Statistician (4.23)
20. Of course this role must be filled by
a statistician of the highest calibre. And I speak as a professional
statistician of many years experience.
21. There is absolutely no reason to change
the title from National Statistician to Chief Statistician, which
has other connotations as surely HMT understands. Of course, the
post-holder should be the Chief Executive of the ONS (as a NMD),
Chief Statistical Adviser to the Government at Permanent Secretary
level, and on the governing board. And the National Statistician
should unequivocally be in charge of the delivery of data.
Independence issues (4.33-4.34)
22. Agreed.
Government statisticians (4.35-4.37)
23. The consultative document's comments
are uncontroversial. If, however, the GSS ceased to be a decentralised
service there would clearly be knock-on effects.
Parliament (4.38-4.41)
24. Broadly agree with the proposals, but
then proof of the pudding is in the eating.
Funding (4.42-4.45)
25. The proposal that extra funding would
be provided for additional government-desired projects but not
for other projects (proposed by the board) would not inspire public
confidence in government statistics. Having decided the allocation
of funding to the ONS, the government should not be seen to be
providing extra funding for its own pet projects.
The devolved administrations (4.46-4.47)
26. I have no expertise in this area.
Additional consequences
The Statistics Commission (4.48)
27. Agreed that the Statistics Commission
can be wound up, when the new arrangements are in place.
The Registrar General (4.49-4.51)
28. The proposals relating to the Registrar
General appear sensible.
Legal ramifications (4.52)
29. The proposals in the consultation paper
appear uncontroversial.
May 2006
| 
