Examination of Witnesses (Questions 280
THURSDAY 16 NOVEMBER 2000
JOHNSON, MP AND
280. So where do you think the United Kingdom
would sit in an international league table of statistics offices
in the developed world?
(Mr Cook) I think it was about six or seven out of
ten OECD countries that were in a study, which was last done approximately
1994, maybe 1995. I would have thought since then that we should
have changed a little more positively. We have had some really
strong developments. For example, in the system of national accounts,
the European system of accounts, we are pretty much a strong leader
there. There are areas of statistics, such as population and demographic
work of national accounts, where we would be up in the handful
of countries that actually lead what is going on in the world
generally, and provide a very large share of the basic research
which countries generally draw on. In the operation of social
surveys we have a pretty lean and flexible operation. There are
other areas where that is less the case.
281. According to a table we have here, published
by the Netherlands Official Statistics Office, we are right down
at the bottom of the league on a number of different indicators:
Canada, Austria, Finland, France, Netherlands, Germany, Sweden,
Denmark and the United Kingdom. The United Kingdom looks pretty
impoverished compared with other national statistics services
or offices. This is not quite what you are saying.
(Mr Cook) One would generally recognise that Canada,
Austria and Norway are countries where the resourcing for official
statistics has been reasonably well maintained over the last three
decades. The strength of a statistical office now reflects as
much the investment that was put into systems and processes in
the previous decade and the decade before. In making those comparisons,
we are clearly in a league that is different from those three
countries because of their continuity and strong investment. When
it comes to other country comparisons, there are some difficulties
at the margin, as you know. Comparing public spending, in terms
of centralised and decentralised operations, we still have a lot
of activity in places like DETR, DfEE. So there are differences
in what is in the starting base.
282. If I could take up that first point you
made, in Germany and France they have similarly decentralised
statistical services, yet still perform very much better than
the United Kingdom.
(Mr Cook) If you take Germany and France, for example,
Germany has a large part of its official statistical system, in
fact, devolved to states. What you would want to do there is to
discover whether the state spending does not include what, for
example, the Greater London Authority would spend on statistics.
So at a state level you are seeing a much larger amount of regional
statistics included in the official statistics. If, for example,
one looked at things like national accounts and broader national
level statistics, one would not put Germany particularly close
to the United Kingdom. If one takes France, there is a larger
amount of analytical work of what we would regard in the United
Kingdom as bordering on policy, done in the French National Statistical
Office. For example, the French Statistical Office trains most
of the economists in the French Government, which is a great thing,
of course. It means that Secretaries of Treasury, or the equivalent
in France, have normally worked in the Statistical Office. So
there is a very significant difference in the scope of activities.
I am not diminishing the value of the comparison but I think it
is important to understand what individual countries do. If you
take the Dutch, for example, because of a particular situation
they have decided to invest disproportionately more than any other
country in environmental statistics. Quite obviously, if a high
tide is going to come and flood your town, you might worry about
it. The Dutch have, in their official statistics, areas of activities
that in the United Kingdom might be seen as part of the scientific
activity in the state. I am learning about what I can say about
the United Kingdom. But in the other direction we separate out
the Registration role. So it is not easy to make quick comparisons
of the funding bases of countries.
283. Thank you for that very full answer. Minister,
in relation to National Statistics, which is more important: the
requirements of Government or other users?
(Miss Johnson) There is a balance to be struck but
it is a balance that we are really interested in, in engaging
with users, and actually looking much more widely at the user
base. The so-called Rayner Doctrine of the 1980s, which basically
said that the purpose of having statistics was for Government
purposes principally, and mainly and really to the detriment of
looking more widely: in the response that Len Cook has recently
made on the PAT 18 initiative, you can see that, in fact, there
are quite a lot of those who are going to benefit as well, so
sometimes there is not a distinction, of course, between what
Government would like to do and those others who are going to
benefit very widely from that activity. I think that with time,
as the PAT 18 sets are developed, a lot of local authority, local
health authority and other users, will find that information as
useful at their level of policy and service provision decisions,
as central Government will find it to have it broken down at a
local level as well. So it is a balance. We are trying to achieve
a balance with a much greater input from these users. The website
is a good example. There are places where people on the website
can record their views, their interest in things, comments on
things, where we are seeking to engage much more with users, much
more widely than Government Department users, of National Statistics
and other statistics. So one thing is a balance between those
different users, and the other thing is not always to assume that
they are necessarily conflicting requirements. Hopefully, through
dialogue, we can arrive at an increasingly better balance between
the needs of Government, which obviously do have to be met, and
some of these key indicators which we all know are very central
in managing the economy or social policy areas, where we have
got to have certain things provided for the Government to make
key decisions; and many of the other areas, where there can be
dialogue and discussion about performance or statistics or the
nature of the statistics that Government is collecting and using;
where I believe increasingly a wide range of statistical users
have a role to play with ONS and National Statistics.
(Mr Cook) When it comes to the undertaking of a large
survey, it is generally initiated because there is a major public
policy need, a departmental policy interest, that will justify
the funding of it. Having undertaken the survey, then the results
of that survey, the way they are presented, fixed, blended with
other information and delivered, is very, very much a significant
matter of public interest to users generally. For example, many
of the surveys we use that make up the information on Social Trends
have been justified for a particular public policy need, whether
it is a family expenditure survey or crime survey. Social Trends
is a very important public statement and its accessibility is
a very important part of confidence in the statistical system.
I guess the most important public survey we have is, of course,
the Census. The success of the Census exists because the public
own it. They see it as a very important part of their involvement
with the Government. It shapes their identity. It is also part
of their way of seeing their communities measured. It is also
part of the way in which we hold Government to account as citizens.
In many ways the Census fits the broader model you are talking
about much more effectively than the great raft of survey-taking
activity we do. But certainly the statistical results we produce
are very much part of it; presented as part of that broader user
284. What about business interest? Businesses
will often say to me that it is one-way traffic. That they are
expected to fill in an awful lot of data but do not necessarily
get a lot of useful information back. Are there mechanisms in
place to ensure that their needs are also considered by the National
(Miss Johnson) We need to engage more with business
in the use of statistics and the statistics that are provided.
We have been developing a database of users. Obviously not all
businesses are linked up by website and what have you, but those
who are can actually register their interests and join targeted
e-mailing lists as well, to get back whatever statistics they
are interested in receiving. We have also got a number of theme
groups looking at different areas of statistics. Those have been
reviewing the arrangements in each area. The key theme group papers
are also available on the website. I know this sounds like an
advertisement for the website but we are very proud of the website,
I have to say. The theme groups have also put their draft work
plans on the website, so hopefully increasingly we will be able
to enable those who would like to engage more with us, who are
helping often to provide the data for the statistics, (in actually
finding out what you say you would want to make of that data),
how we can help them to access it and to make a better use of
it. I certainly do think that does include the business community
because they are very key players in supplying a lot of the data
in the first place.
(Mr Cook) When it comes to very large firms who, in
effect, have the largest burdens, they also tend to be major users
of statistics. Very, very much of the imbalance is in smaller
firms who have a larger respondent load, not often in terms of
the total group because of sampling, but where people are included
in samples, they tend to regard themselves as personally having
a very large load. There are two things here which are important.
One is that we support user groups or representative bodies in
terms of their use of statistics. So, for example, the Business
Statistics User Group actually represents not only individuals
but a lot of bodies such as the various manufacturers' associations,
who themselves use the statistics on behalf of their members and
often their members are not aware of it. Secondly, and I think
this is much more progressive, as we get better at seeing the
internet used for the receiving of dataand this is one
area where ONS, may I say, has been fairly progressive and we
have a lot of work ahead to do thatwe will be able to send
back to people with their enquiries useful summaries of the results
of these surveys. I would imagine that if you were a butcher's
shop, there is no reason why we could not send back to you quite
cheaply a little analysis of butcher's shop sales in your areas.
They are the sort of innovations we need to do. A third thing,
an important element as part of our strategyand I do not
believe we do it well enoughis to explain to people, who
fill in individual questionnaires, how the data is used. Often
you will find statisticiansand I do not think that ONS,
to be frank, is excessively different from anyone elsewe
think that terms like the data is being used for national accounts
is extremely helpful, but that it is an area where I see us doing
better, particularly with the continued pressure on compliance
cost reduction which we certainly get.
285. You mentioned the Census just now, Mr Cook.
Is the Census for next year on track?
(Mr Cook) Yes, in terms of the basic preparation of
the forms; the research to ensure that this will be done well;
the development of the field side; the development of the communications
programme; the system to collect the information once it is in.
We are still doing work on the processes to validate the data
but that is all on track to be completed months before they will
be needed. So basically the answer is yes.
286. So you are within the timeframe which was
(Mr Cook) We are hard up against it because it is
such a huge operation but the answer is yes.
287. Have you had time yet to have a look at
the job vacancy statistics? Can you report your initial findings?
(Mr Cook) I have only become aware that there is an
issue in terms of both the significance that applies to the United
Kingdom of these vacancies and how, in fact, we rate-up and obtain
national estimates. What we are doing in ONS is moving to measure
vacancies in our usual business surveys, so that there will be
an on-going process allowing us to rate-up sample results in a
way which differs from what exists now, with a rating-up process
depending upon historical information.
288. Could you help us with the phrase "rating-up"?
(Mr Cook) The information that we get on vacancies,
whether it is the administrative process we currently have or
any statistical survey that we will have, will only be from a
sample; will only represent part of the activity that is taking
place, or part of the vacancies that exist. So we have to have
some way of knowing what is happening in total. With the sample
survey that we have, we will survey a sample of firms. We will
know how far that sample represents the whole population and we
will, therefore, know how much we can multiply the results of
the sample to obtain what is happening in total.
289. Before you do that, is it safe to construct
any kind of meaningful ratio between unemployment and job vacancies?
(Mr Cook) I think the value in some of these measures
is not that the specific ratio statistically is of the same reliability
as statistical measures that we usually depend upon, such as the
RPI, but being able to see side-by-side what is happening with
employment, unemployment and vacancies, which is of particular
value. We know there are so many contextual things, which differ
in any region, that actually make exact comparisons or creation
of ratios less meaningful.
290. So they have not been meaningful, the ratios
that have been chucked around up till now.
(Mr Cook) I think what is being used is the best estimate
of something which is important and very difficult to measure.
What is most important about what is being used, is being very
clear about how it is being produced and what the limits are.
One of the difficulties with the measures we have is that we cannot,
of course, as statisticians, defineas we can from information
with statistical sample surveysany particularly sound measure
291. I am asking this because when we last considered
this, we were told by the ONS that it was all a "little bit
of thumb in the air" multiplication. There might be three
times as many vacancies as those reported to job centres.
(Mr Cook) ONS does not publish figures that are grossed-up.
It would not publish figures that are grossed-up, unless their
grossing-up had the same degree of reliability as an official
It does not mean to say that people will not find it useful to
multiply out and create some kind of gross figure, but we certainly
would not expect it to have the same kind of reliability as National
Statistics. It certainly would not be published as one. It is
not published now as a National Statistic.
292. If I can explore that a little further
and follow Mr Fallon's questions. We understood there was a problem
when statistics on vacancies nationally were being derived. It
was not just that you took a sample and then you had to scale
that down for the whole population. It was that when you had done
that, that the figure for vacancies was very much lower than those
you saw around as a check for other means. Therefore, some fiddle
factor was applied to the original figure to scale it up. What
gave us that unease was the derivation of that sort of fiddle
factor. This is the point that Mr Fallon is getting at. It raises
the question of how many of the National Statistics are the material
coming from surveys direct, from whoever is replying, and how
many have been scaled up in this sort of way.
(Mr Cook) First, can I say that whatever the measures
are, which are produced in National Statistics, there will be
clear knowledge available of the nature of any rate-ups. We publish
information on all surveys which are produced in ONS, so we know
not only the extent to which surveys are rated-up, but we also
produce sample estimates of the reliability of what has been rated-up
as representative of the population. Where we do have adjustments
which you referred toyou defined some of those as fiddle
factorswhat is important to recognise is that in those
areas, where they have a basis in some past survey or series of
surveys, then that would be quite clear. So the basis of any adjustment
factor should be known. In essence, you are talking about a judgment
and how to apply it to what may be a measurement, which has not
entirely come about from the same process of the survey itself,
but we should still be able to vouch for its reliability.
293. That will be made available? Where that
process has been part of the derivation of the series, that will
(Mr Cook) Yes. If the adjustment factors do not have
the same reliability as National Statistics, then if there is
a significant result the measurement itself would be unlikely
to be a National Statistic. One has to recognise that when you
are compiling some statistics, such as balance of payments, where
what we are attempting to do is to provide a comprehensive measure
of the ownership of the United Kingdom economy or the other direction,
then there are some things we measure well and some things less
well. But those adjustments are always made transparent.
294. Perhaps you would like to comment on this
because it was your document in February this year, The Goal
of Full Employment, that stated: "Moreover, for every
vacancy in job centres there are around two others in the wider
economy." In view of what you have heard from Mr Cook, do
you think that is still robust?
(Miss Johnson) I think it is robust because the comment
is that there are "around two". I believe that is entirely
consistent, in fact, with what Mr Cook has been saying to you
on this count. Obviously we have the evidence that we have a million
new jobs created in the economy, and we do have record levels
of vacancies at job centres as well. So there is plenty of other
evidence to indicate that the sort of points that we are making
in that document are entirely the right ones in relation to this
295. Your instinct is that the multiplier is
(Miss Johnson) We put "around two" because,
as you are quoting there, we try to be cautious and never put
something out that we are not able to be pretty confident that
we can substantiate. The indicator was about a third. I believe
the discussion before was on the basis of a factor of three in
the measures. Because of some degree of uncertainty about it,
we have made sure that we have not used the "three"
figure in there but "around two".
296. Thank you very much. We may want to pursue
our own inquiries in this issue of the Prime Ministerial access,
which you told us about earlier. I am informed that the Chief
Scientist, for instance, does have the formal words attached to
his job, which says "direct access". It is not just
perhaps questions of access. It may be some indication of the
weight attached to Whitehall of these functions, so we may want
to pursue that in our own inquiry.
(Miss Johnson) It is entirely proper for you to pursue
things. Could I make one comment in relation to what I was saying
earlier on about the Framework Document. SomebodyI am afraid
I cannot remember, Mr Fallon or Mr Beardwas asking me about
the Framework Document and the reviews of the Framework Document.
I think it is important to say that whilst we are not about to
undertake anything now in relation to it, and we have got this
big work programme, as it does say in the Framework, there is
a review anticipated in five years' time. That would be five years
from the document itself. I would just like to draw that to your
Chairman: Thank you very much, Minister.
2 Note by witness: "The vacancy figures
currently published by ONS are compiled from the administrative
records of all vacancies notified to the Employment Service (ES).
On a number of occasions during the 1980s and 1990s special surveys
were undertaken by ES to try to estimate what proportion of total
vacancies across the economy these ES notified vacancies represented.
The figure of one third is the best approximation that can be
made from this work-but it is insufficiently robust for ONS to
incorporate into an adjustment procedure for its published vacancy
data. The way forward on this-which ONS is currently adopting-is
to develop a new survey approach designed to estimate numbers
of vacancies in total directly." Back