Examination of Witnesses (Questions 100-119)|
WEDNESDAY 30 OCTOBER 2002
100. What is a "common information management
(Mr Cook) A common database with a common set of tools
that is used in the same way regardless of the statistical data
set that we are dealing with so that we have one basic way of
managing all our information and a single set of tools that we
use to get at it. At the moment we have several hundred quite
different survey processes that can be on different computers,
different software, operated with quite different programming
languages, and we have huge difficulty bringing together the data
that we have in the Office.
101. Is that described as standardised data?
(Mr Cook) It will be standardised processes that we
would be using.
102. That will be a bit clearer than some of
the verbiage that is in this report.
(Mr Cook) When you read it out it is quite obscure.
Mr Plaskitt: It certainly is.
Chairman: We will have to leave it there and
suspend the sitting for ten minutes because there is a division
in the House.
The Committee suspended from 6.04 pm to 6.14
103. I think the only point I would like to
make on the languages is if you will undertake to look at clearer
English when you do your next report. There are bits of it which
are incomprehensible. Now moving on to productivity, you have
been carrying out an extensive programme, you say, to improve
the quality of statistics on productivity. In your view how bad
were our productivity statistics before you reviewed them?
(Mr Cook) I think it is not so much a matter of how
bad they were but it is how much do statistics on productivity
in all countries need to adapt quickly for what is essentially
the unravelling of a new economy which is driven by shifts, for
example, to e.commerce, a huge increase in the services sector
and the creation of very different infrastructures such as the
internet than what has traditionally been the business. What we
are seeing, for example, with new information technologies is
the writing off of capital much more quickly, huge difficulties
in measuring the price change of computers. So we have a collection
of projects, most of which are pretty well linked in to the international
statistical community, often in the European Union but many in
the United States and Canada, which are part of an international
move to adjust official statistics as quickly as possible
104. Productivity statistics which are lagging
behind changes in the real economy will be bad productivity statistics,
will they not, because they are not measuring the right things?
(Mr Cook) One of the areas where we have got a very
major project under way, for example, is the measurement of capital
stock so that we can measure productivity from the point of view
of not only labour productivity but the impact of capital changes
on our measures of labour productivity. In the past measures of
value added worked out simply according to per head of the labour
force, for example, were relatively crude compared with the more
refined ways that we can measure now the output, adjust it for
capital stock changes which we expect to but also to have more
refined measures of the labour input which is one thing that we
have been successful in improving.
105. Is this an ongoing project or have you
made some changes in the way productivity is measured and those
changes are in the system now or are you midstream with all this?
(Mr Cook) This is going to be a project which is going
to dominate a lot of the things we are doing for at least the
next five years because it involves not only macro economic based
estimates of productivity but a lot more micro economic studies
of productivity. For example we have a major project under way
in our office with a number of academics to take our business
surveys and link them over quite a long period of time to measure
the impact of technology change on productivity and doing it on
a firm by firm basis.
106. Where you have put the changes in methodology
into effect so far, what impact has it had on productivity data?
Has it changed the story or the picture of UK productivity?
(Mr Cook) Really I am not able to give you a particularly
thoughtful response to that off the cuff of how we have changed
107. Can anyone?
(Mr Lynch) I think we have yet to see the benefits.
It is an ongoing developmental thing, yes.
108. This is important. So you do not think
it has changed the picture of productivity yet?
(Mr Lynch) I am not qualified to judge either.
(Mr Cook) Can I do you a note on that? For example,
if we take hedonic indexes, looking at the way we measure the
prices of computers, what we know when we try and reproduce work
ourselves which is done elsewhere, in fact the option costing
methods that we have used for quite a long time in the UK RPI
have actually resembled the methods of hedonic adjustment more
than we thought. In fact there will not be as much of a shift
in price change and hence output changes as we might have assumed
say two years ago. That is an example of where we would see less
change. Where we refine measures of labour, and we are working
much more to measure labour in terms of hours rather than days,
then clearly the increased use of part-time workers is going to
mean that now we will be measuring more exactly measures of labour
input in terms of measuring it by the hour rather than the day
so we would have ended up with a bias on how we measured labour
input if we continued to simply measure by the day because of
that trend in part-time work.
109. Who are you working with in this review
of the way we measure productivity? In particular, how closely
are you working with officials in the Bank of England?
(Mr Cook) We work mostly with the Bank of England,
the Treasury and the Department of Trade and Industry. They are
involved in the steering group of the project which we have employed
collectively a number of British academics to work on.
110. Is there hope for us in terms of ever getting
to the point where we think we have reliable accurate productivity
statistics? I tell you why I ask that because when we have the
Bank of England in front of us and members of the Monetary Policy
Committee, especially the Bank Governor and his staff, one of
their constant refrains is "We do not know the true story
of what is happening to productivity in the UK economy because
we do not know how to measure it". That is a pretty significant
confession because it is a major variable that you need to get
right in order to know where you should set monetary policy and
all the other things. It is really important that we get to a
point where we do think we know what is happening to productivity,
we know how to measure it and we can trust in the data. Now do
you think we are getting nearer to that point?
(Mr Cook) My first answer to that is my expectation
is within a very short number of years the work that we do in
the United Kingdom should be very little different from the best
work done in the United States and other countries. For example,
we had a conference recently of some 300 people in the United
Kingdoma lot of Americans, including Dale Jorgenson who
is the key American thinker on productivitypart of which
was providing a much clearer focus on the work that we are doing
with respect to the new economy itself. What was quite important
there, I think, was the huge amount of work that we are doing
in the UK that is keeping us very much at the forefront of the
thinking elsewhere. I guess my best answer to you is I would expect
that over the next two to three years our key achievement will
be that we will be doing nothing less than was done elsewhere.
111. That is not an answer to my question.
(Mr Cook) It is not quite the answer because productivity
is a very difficult thing.
112. I understand that.
(Mr Cook) We come at it both from a macro sense in
terms of the economy wide measures that are derived out of the
production aggregates that we produce and the aggregate measures
of labour input and capital stock which are refining both those
measures and also micro economic studies. Our ability to carry
out micro economic studies is improving quite considerably. By
the end of this year or early next year when this project is over
we will have a very rich database and be in a position to inform
productivity estimates based on micro economic studies much more
effectively than we have ever been.
113. Is it a case of running to stand still
because the structure of the real economy out there is changing
fast? Are we able to catch up with those changes as we review
the way we measure productivity or are we always lagging behind
and will there always be doubts about our productivity statistics
telling us the real story?
(Mr Cook) I think there is a huge impetus coming particularly
from the Bank but also from the Treasury and DTI to give priority
to work in measuring productivity which I suspect may not have
been the case in the previous decade.
114. My question is do you think we are going
to get to the point where we will have the Bank coming in front
of us and saying "We are now pretty sure we know how to measure
productivity and we are now pretty sure we know what is happening
to productivity in the UK economy". Do you envisage you are
going to get to a position where they can say that on the basis
of the work you are doing?
(Mr Cook) My goal would be that over the next three
years we get to a stage where there are not statistical problems
but more conceptual problems in the measurement of productivity
which would limit the quality of what we do.
115. That is helpful. About three years?
(Mr Cook) That is the sort of time. We have developed
already, for example, a very comprehensive index of services in
the United Kingdom so we have got quite effective direct measures
of pretty much all output in the United Kingdom. We have published
recently independent measures of public sector outputs directly
measuring outputs that cover pretty much two thirds of public
sector output. So we have made quite a few gains in that area.
We have put a lot of work into the measurement of labour, as I
mentioned. We are involved, also, in measures of labour price.
For example, in the European Union we are doing a study for them.
The new economy itself, most of our major statistical surveys
have been modified and are continuing to be modified for our understanding.
We have produced recently two large surveys on the impact of information
technology for example.
116. I think I would welcome particularly a
note from you itemising the particular changes you have made in
measurements of productivity already.
(Mr Cook) Yes.
117. And the bits which are ongoing.
(Mr Cook) Yes.
118. If you have it, also, an approximate timetable
as to when that might be complete. I do not want you to go into
huge detail and I would like you to keep the English clear but
if you could do that for me I would welcome that and I suspect
other Members of the Committee might as well. Can you do that
(Mr Cook) Certainly we can do that, yes.
119. Moving to pensions. Your annual report
refers to problems with the dataset on self-administered pension
funds where there was some doubt about the validity of a revision
in 1999 and when a revision of £105 billion had been made
to the dataset. You went to investigate it so could you tell us
about it? How did a revision of that magnitude come about in the
first place and how did it get knocked out in the second place?
(Mr Cook) The revision in the first place was quite
a significant and serious error in my office. It came about because
the statistics in the first year were recorded in two separate
environments, one was the environment where it was actually calculated
and prepared, all the individual results were aggregated, and
secondly in our statistical database where we record economic
time series and store them. When the second year was produced
the first year that was produced in 1999 was found to differ in
the database from the calculation which was prepared and the person
responsible for that assumed it was wrong and simply changed it.
That person did not go through our processes for managing revisions
otherwise there would have been a process of validating a change,
particularly a change of that magnitude. It did not happen, the
person thought they were correcting an error in a publication.
Let me come back to the basic cause. Firstly, our systems in that
area simply do not flow together so there was an excessive amount
of manual transcription. Part of our development strategy is to
create standard tools for estimating for delivering data and that
is very much at the heart of the programme that we have been funded
for in this year's spending review. We have changed our processes
for validating revisions in the office to make them extremely
rigid. Secondly, in fact after this change had been made and before
the error was found, we had altered already our methodology for
quality assuring changes to systems. Finally, we store centrally
now all documentation on statistical surveys. One of the problems
which we concluded out of this was in fact the work area had different
practices from other parts of the office. We have a very standard