Examination of Witnesses (Questions 120-139)
PROFESSOR DAVID
RHIND, SIR
DEREK WANLESS
AND MR
RICHARD ALLDRITT
7 JUNE 2006
Q120 Chairman: On the specific issue
of the board, you recommended having a statutory commission and
you did not want to see the Commission and the ONS board merged
into one body. Why do you think the Government rejected that
recommendation?
Professor Rhind: I think they
felt there would be a risk of real tensions between the two different
bodies and that there might be warfare break out between them.
I think our concern is the obvious one, that the board being responsible
in some sense for the operations of ONS, but at the same time
being responsible for scrutiny across the whole system, including
ONS, seems to us a very real tension. If I may say so, we are
completely at one with Ivan Fellegi about the need for a very
clear and careful and independent scrutiny across the whole system,
not just ONS. Our experience has been that many of the problems
that have arisen—not all of them but many of them—and
the newspaper comments and the unhappiness in many parts of the
public, arise not solely from within ONS but, in many cases, much
more frequently from other parts of the system. So whatever comes
out has to address the totality of the system, in whatever form,
recognising that we have a very devolved system, where statisticians
and different government departments work closely with economists,
other analysts and even policy people—which is, I think,
a little different from the situation in Canada.
Q121 Chairman: Your point is that,
if you have a single board which is both provider and supervisor,
that cannot enhance the perception of independence.
Professor Rhind: I think you could
set up various Chinese walls and subsets within there. I think
it would not be straightforward to sell the idea to the public—and
perceptions are very important to all of this. We think it could
be made to work, but it would not be easy and it would certainly
take a long time. The Canadian National Statistician I think pointed
out that he had had legislation and this had been operating since
1918 or thereabouts. Certainly there were periods in the 1970s
when the reputation of his fine organisation was lower than now,
but the important thing is that nothing is very short term in
this particular area and the arrangements we have just been discussing
seem to us to require quite a long time to be accepted.
Q122 Mr Todd: As you have already
said, your preference is for arrangements to cover all official
statistics rather than just the ONS's. Does one not then get into
a definitional problem of what is an official statistic anyway,
bearing in mind, as you have said, the embedding of statisticians
in the normal functioning activities of a government department?
Professor Rhind: I think that
is absolutely right. That is particularly acute, of course, when
increasingly a number of statistics are coming from administrative
systems which are embedded in particular departments. We think
there are a number of ways you can ameliorate that problem. The
publication of statistics by the ONS and by the National Statistician,
when they have been looked at and in some sense quality assured
and approved, and publication through a central channel, might
be a central means of buttressing confidence, I think. But I do
not deny that the administrative system complication is one that
no one, I think, has a good solution to yet.
Q123 Mr Todd: So you do not endorse
the star rating system—or perhaps you were, in that remark
you made: passing it through ONS means that you get a little imprimatur
which says this is okay.
Professor Rhind: The crucial thing,
I think, is to have a code of practice about how these things
are produced and, more than that, how they are disseminated and
a process for conveying information fairly.
Q124 Mr Todd: You would prefer a
code of practice to a more robust legislative framework.
Professor Rhind: We would seek
a statutory code of practice. The present code of practice is
in many respects rather aspirational and difficult to judge whether
things match it or not. If I were to make a couple of quotations
from it, the sort of thing that is in there is "the value
of administrative data in producing national statistics will be
recognised". Another one is "statistical data will be
treated as valuable and irreplaceable." There are some better
bits, but essentially what we do not have at the moment—
Q125 Mr Todd: Honeyed words.
Professor Rhind: —is any
set of criteria which will give confidence that what is being
produced is being done in a way which is . . . .
Q126 Mr Todd: How would you imagine
the statutory code of practice would be drawn up? Would it be
based around—striking out, perhaps, some of those rather
softer phrases that you are talking about—the code of practice
that is there now?
Professor Rhind: In an attempt
to be helpful, the Commission will be making some proposals in
the next few months. If a board is set up, we would hope that
the board would hit the ground running rather than having to create
a new code of practice over many months. But the principles are
pretty clear—indeed, it has to start with certain principles.
There are a number of things we can draw on from the existing
code of practice, but we have to make it much clearer, somewhat
more prescriptive, written in plain English, but, equally, it
has to apply not just to statisticians but to the entirety of
departments and agencies producing official statistics.
Q127 Mr Todd: Lastly, is there an
argument for saying that, if this is not aimed at the wrong target,
it is aimed at a rather too narrow a target, and that the issue
is how government agencies and their possession of data, gained
normally from citizens, relates to citizens and the world outside,
and ensuring that people are given proper access to data? Because
I am sure you are familiar with the arguments of people over exclusive
arrangements in government, and third party suppliers who are
favoured and so on. Is that opportunity being missed, perhaps,
in our obsession of whether a particular minister is given a certain
amount of time to worry about statistics or not?
Professor Rhind: I have been speaking
rather a lot. May I invite my colleagues if they have particular
views on that?
Sir Derek Wanless: The things
that are important can be classified but there are not that many
opportunities to put a structure in place which seeks to address
the problem and therefore you have to address the issues you touched
on as part of what we are now trying to do—and that is what
our statutory commission as a preference tried to do. It tried
to address directly the issue of trust in the whole range of official
statistics. I do not think it is that difficult to recognise if
information produced by a government department is to quality
as a statistical output or not. If that is the extent of the worries,
then we would have succeeded in getting the rest across.
Q128 Mr Todd: Not quite. It is more
the exclusive arrangements that government departments and government
agencies have of controlling their own data and how they present
it to us as citizens rather than permitting others to interpret.
The classic example is the United States, which has a much more
liberal framework for the possession of government data and its
use by citizens and third parties.
Professor Rhind: For the federal
government. It is not always true in the states.
Sir Derek Wanless: The important
thing, I think, is the text that accompanies any statistical series.
Certainly, as one of our proposals, the text would be something
that the statisticians produce. When figures were produced and
released to the world, including to ministers, the text would
not be a political text. It would be a text about the numbers,
the reliability of numbers, what purposes they are fit for and
so on.
Q129 Mr Todd: I had understood that
point. I am more thinking about the appropriate relationships
between statisticians and those who provide statistics and the
world outside. If I wish to buy statistics for the use of developing
some product of my own, having an appropriate relationship with
which I can do that—and at the moment that does not exist—
Professor Rhind: Forgive me, Chairman.
My sense is that it is relatively easy to do that in the statistics
world, not always in some others. But I think the general point
we would make—and I think it is an important one—is
that we see statistics being produced by government not just for
government—important as that is, for helping to set policy
and to monitor success—but as in the public interest, to
give the public confidence that things are open, transparent and
are going well.
Q130 John Thurso: Ruth Lea, in her
evidence to us, said there was a very, very strong case for the
ONS to be under the Cabinet Office rather than the Treasury. I
know that you have certainly suggested that the Government should
consider this. Could you set out for me the advantages that you
see in that arrangement?
Professor Rhind: This arrangement
or something akin to it was what pertained in the 1960s when Lord
Moser was the Chief Statistician. I think your Committee, Chairman,
is going to see Lord Moser next week. I think he will be extremely
eloquent on this particular topic and we have been somewhat influenced
by his particular view. Indeed, Ivan Fellegi's comment to the
effect that every time a new minister comes into office he receives
a letter from the Prime Minister, I think was a nice example of
prime ministerial patronage, if you like, which could easily be
replicated. We are really saying: What are the disadvantages of
this being sighted under the Treasury? There are a number of those,
some of which are serious at some moments in time and some of
which are less serious. The obvious one is, of course, that economic
statistics are of great interest to the Treasury but many other
statistics are nothing like as interesting to the Treasury, so
I think their ambit is somewhat more restricted. The Treasury
has a great advantage in having somewhat more control over resources,
perhaps, than other departments, so that is an advantage. But,
I wonder, could I turn to the Chief Executive and ask if he can
you recall the other arguments that we have made.
Mr Alldritt: One of our concerns
is to ensure that there is effective planning of statistical work
and to meet future statistical requirements right across government.
Many of those issues involve not just a single department but
cross-departmental concerns, and the Cabinet Office in many ways
seems a more natural coordinator of statistical planning than
the Treasury does. That is perhaps one issue.
Professor Rhind: A prime example
of something that flows across government departments are migration
statistics, which have huge impacts upon the nature of population
census, the numbers we use for many purposes. That can only be
rectified, at present problems, by actions by the Home Office,
by ONS and by a variety of other government departments as well.
Those were the sorts of reasons we thought commanded a move of
that sort.
Q131 John Thurso: It would be fair
to say that it is more a question of the best home for it rather
than a specific concern that being in the Treasury is bad for
it.
Professor Rhind: Yes.
Q132 John Thurso: One of the things
you highlighted is the difficulty of removing resourcing decisions
from ministerial control, given the need for an independent statistical
office to respond to the Government's changing data demands. What
method of funding would best support independence and consequently
boost public confidence?
Mr Alldritt: I think the big constraint
here is what is possible. Dr Fellegi was describing an arrangement
whereby there might be a central statistical budget. It is hard
to see how that would work within the Whitehall funding structure.
We assume that the Home Office will continue to fund Home Office
statistics, and that, with that constraint, the role that the
independent board would have to play would be to ensure there
was transparency in the decision making within departments such
as the Home Office and between departments, and that something
that is not available now, which is the rationale for either expanding
statistical activities or limiting them, was in the public domain
and subject to parliamentary scrutiny.
Q133 John Thurso: It seems to me
that there are some statistics that we all agree need to be delivered
at a national level. Those surely can be delivered on a fairly
long-term basis and will not regularly be changed. There are quite
clearly other statistics which may be of interest to a particular
department or minister at a particular time when facing a particular
problem and may have a requirement. Do you think—slightly
reflecting on the evidence we heard from Canada, where Professor
Fellegi said that they get money from other sources—that
there might be a model, whether it be in-house or a part of ONS,
whereby there was a purchase arrangement rather than a funding
arrangement?
Professor Rhind: There are indeed
at the moment, I think I am right in saying, cases where ONS is
commissioned by bodies to do work of that sort, so it does exist.
There is, of course, another source of money that the Canadians
have and that is that they actually sell some of their statistics,
going back to Mr Todd's comment earlier. But I do not think that
is a very large fraction of their income.
Q134 John Thurso: The Royal Statistical
Society told us that the funding arrangements for the census should
remain part of the Spending Review progress. What is your view
on funding arrangements for the census?
Professor Rhind: The census is
a huge, lumpy part of statistical activity in the UK. I think
the last time it was of the order of £250 million; this time
it is likely to be substantially more, for a whole variety of
reasons. Accommodating that one-off blip within standard funding
arrangements is clearly not easy, not least because, whilst a
census occurs every 10 years, the preparations for it extend quite
a long time beforehand and quite a long time thereafter. It seems
to me and to my colleagues to be rather a difficult one to accommodate
within, say, a three-year spending horizon, and it really has
to be thought of as an enterprise from beginning to end rather
than the first three years, middle three years or couple of years
at the end.
Q135 John Thurso: We heard evidence
at our last session regarding the impact of devolution on statistics,
and, in particular, the fact that many of the devolved nations
have different priorities according to the ministerial demands
that they have, which is absolutely appropriate. The point was
made to us that with regard to the national census there were
differences. Is funding an issue in that at all, or is that a
red herring?
Professor Rhind: I do not know
whether funding is an issue in that but I understand you are talking
to the people who would know soon afterwards. But, if I might
just talk about the devolved elements of all of this: statistics
is a devolved function but clearly there are some things one would
want to have to do and make some comparisons across the piece.
It seems to be rather difficult to make comparisons on a number
of health variables across the UK. There are indexes of multiple
deprivation for each and every one of the countries, but they
are different, slightly different. There are some areas where
you can argue of public interest and having something consistent
across the piece. How that is best done, I think there are a number
of ways you could do it. At the moment, I think I am correct in
saying, other than informal internal working groups, the only
pan-UK enterprise in statistics is the Statistics Commission,
which is the commission for each of the countries. We hope very
much that, whatever the arrangements are coming out of the Treasury
activities, there will be something akin to that. The board, for
example, we think, should have some kind of similar function across
the piece.
Q136 Peter Viggers: I would like
to ask about breaches of the code and leaks and spin, which must
no doubt be the bane of your lives. There have been occasions
of breaches of the code, and the conclusion you put in your last
annual report was that most breaches of code were due to "accidental
premature release of data or lack of awareness of the Code by
non-statistician within a department . . ." Do you think
that is a robust judgment or were you being easily pleased?
Professor Rhind: I do not know
we were pleased about it. We felt this was a situation which was
somewhat too commonplace and we were not utterly convinced that
the Statistics Commission saying rude things to people would make
the situation better, because many of the problems that have arisen
have been early releases of quiet conversations to newspapers
and so on. I am sorry, I have lost my thread.
Mr Alldritt: The reason that piece
you quote is rather restrained is, at least in part, because the
code itself only has a small number of bits which are clear-cut
in terms of what the requirements on departments is. For example,
the code says that consulting users is a good thing but there
is no specific requirement on what consultation is. On some of
these very important issues like consultation and dissemination
of statistics there is not a clear requirement on departments,
so for us to say publicly that a named department has failed to
adhere to the code is extremely difficult when the department
in many cases says it has acted entirely in compliance with its
own interpretation of the code. That is why we attach considerable
importance to a code that is more prescriptive.
Professor Rhind: That is a point
I was seeking to make. At the moment everyone can make their own
interpretation of the code, which, as we said earlier, was somewhat
ambiguous, so saying this is a clear breach of the code and this
is unacceptable is a bit difficult.
Q137 Peter Viggers: Is the method
of investigating breaches such that it is likely that there will
be clearance of the procedure that was adopted? Is the investigation
procedure itself sufficiently robust?
Professor Rhind: Not at the moment.
We have found at times past some difficulty in getting the responses
from departments, and I think if you do not get the information
you require then it is somewhat difficult to be certain, especially
in this highly ambiguous situation of whether something is a breach
of an ambiguous code.
Q138 Peter Viggers: The Phillis Review
argued that there was no need for ministers to receive 40-hours
advance notice of national statistics. How realistic is it to
assume that statistics can be released by departments without
any interference by ministers? I just wonder what people think
political advisers are for.
Professor Rhind: There are clearly
some data sets where considerable care must be taken, and the
market sensitive ones to which Ivan Fellegi referred I think come
in a slightly different category from some others. Our preference
would be to have no pre-release, but we could envisage various
different models, where, for example, statistics might be embargoed
for a number of hours beforehand but made available not only to
ministers but perhaps also to opposition spokesman and even to
newspapers. There are a variety of different options on that,
and what is critical, I think, is how long they are released in
advance and how they are embargoed and what the penalties are.
As I say, our preference would be for no pre-release, but we recognise
that that may have some difficulties for the operations of government.
Q139 Mr Newmark: I would like to
go back to Professor Rhind's answer, when you were losing your
train of thought.
Professor Rhind: I am sorry.
| 
