Examination of Witnesses (Questions 1-19)
MR KEITH
DUGMORE, PROFESSOR
DAVID RHIND
AND PROFESSOR
DAVID MARTIN
16 JANUARY 2008
Q1 Chairman: Professor Rhind, welcome
back to the sub-committee. Perhaps you could introduce yourself
and your colleagues, please.
Professor Rhind: Thank you, Sir.
It is a pleasure to be back. I am Chairman of the Statistics Commission
for another few weeks before the Statistics Board takes over.
Sitting next to me is Professor David Martin from Southampton
University, who is here representing, I think, the ESRC and the
Royal Statistical Society, who have put submissions in, and on
the left is our man from the private sector, Keith Dugmore, who
runs a firm called Demographic Decisions.
Q2 Chairman: Thank you very much.
Thank you for assisting us this afternoon. Please do not feel
that all three of you have to answer all the questions; you could
try and rotate it and take a stab at it. Can I ask, first of all,
about emigration. The ONS released figures estimating that around
400,000 people emigrated last year, up from 359,000 in 2005. Are
there any indications that emigration rates will continue to increase?
Professor Rhind: That sounds like
one for David Martin, I think.
Professor Martin: I think our
biggest difficulty with the emigration information is the fact
that all of the estimates are based on small samples of information
about what we know about where people are going, and so we do
not have any systematic way of collecting that information, and
it is one of the biggest gaps in the system that we have for estimating
who has left and who has gone. While we continue not to capture
that information systematically, it is very hard to see what frameworks
you would use for it. It remains for us the greatest uncertainty
in the system, where we try to match the flows of people in and
out of the country and what the stock that remains now is. I would
be extremely hesitant to try to estimate whether those figures
are right, because we are always judging them on the basis of
a very small sample, and quite rapidly changing trends can lead
to quite significant changes in the direction of flow.
Q3 Chairman: Are you telling us that
the emigration statistics just are not accurate?
Professor Martin: We have very
little information on emigration. We can conduct surveys to ask
people where they are going, but we do not know where they have
gone and how long they stay and we do not know a lot about the
extent to which quite a few people have left the country and how
well they do what they have told us they were likely to do when
they return—whether they stay away or change their minds
once they have gone.
Q4 Chairman: If they do leave, how
long is it before their absence is reflected in the population
estimate?
Professor Martin: The mid-year
estimates build in information which is both demographic, to do
with what we understand about the change in local populations
and what we understand about the flows of migrants both internal
to the country and out of the country, and so, on a rolling basis,
the Office of National Statistics is trying to put all that together.
It is not a precise art, in the sense that you know exactly where
the groups have gone. Often we know something about where people
have gone out of the country; we know a lot less about which part
of the country they have gone from; so it is then very hard to
then build it back to migrants from Manchester or migrants from
London. We are actually an art of estimating back which parts
of the country are losing the population according to their demographic
characteristics in quite small sample proportions.
Q5 Chairman: If the ONS tell us,
as they told us in October, that the overall population of the
UK is likely to increase from 61 million last year to 65 million
in 2016, how much accuracy should we attach to that projection?
Professor Rhind: Do I not recall
that there were several scenarios? Was that the central measure
that they were suggesting? I do not think anyone would claim a
great deal of precision for these anticipated population increases.
Q6 Chairman: Obviously there are
assumptions built in here about life expectancy, future fertility,
emigration, immigration, and so on, but why would you expect trend
growth to have increased since the projections made in 2004?
Professor Martin: I do not think
that you can focus only on the emigration part of that as being
the explanation for the figures you have got there. As I recall,
a number of scenarios were presented and they are sensitive to
future decisions, future policy changes, which affect the way
in which people make decisions about whether they go out of the
country short-term and return, and the emigration component, where
we began, is actually the least certain part of that, whereas
the internal movements and the demographic change and the changing
size of different groups who come in is more robust. One of our
biggest uncertainties relates to people's future decisions to
stay away, and I think that is reflected in the scenarios which
we presented.
Q7 Chairman: Which of all these factors
has the most influence on the projections?
Professor Martin: The migration
is the largest component of change, if we look at contemporary
UK population change, and the demographics give us a fairly established
methodology, a way of knowing where we are seeing different ethnic
groups changing at different rates, for example, but the migration
is both the biggest component of actual overall change and also
the most uncertain part.
Mr Dugmore: I think it is also
significant to point out that there is an element of feedback
here, in that depending on the migration and young people coming
in, of course, that can have a knock-on effect on fertility rates
as well, so there is a further complication built in there as
well.
Q8 Peter Viggers: Population estimates
are, of course, critical for three main reasons: resource allocation
and the determination of financial settlements, the second one
is to act as a denominator for statistics on crime, conception,
accidents, deaths, and so on, and the third area would be the
planning of local services. How far do current population estimates
meet the needs associated with their uses?
Professor Rhind: I think, Sir,
if I may start on that one, everyone accepts that there are serious
shortcomings in at least the most important component of the population
estimates, which relates to migration. Ministers have indicated
that they accept that these are not fit for purpose in a number
of cases and, although they work fine in some parts of the country,
they are not good enough in other parts of the country, and I
think every other responsible and expert person says exactly the
same. We have a major task to actually improve the migration statistics
in the UK before the population estimates can be as good as we
need them to be for allocating very large sums of money.
Q9 Peter Viggers: If there are inaccuracies
or inadequacies, how would this impact on government and government
policy?
Professor Rhind: In terms of financial
allocations, some of the allocations, of course, are zero sum
gains. In the case of the local authorities, if some areas are
getting too much and some others are getting too little, that
is not a great problem in the overall sense, but it is certainly
to those local authorities that are either being under resourced
or over resourced.
Q10 Peter Viggers: How useful is
the statistic of "usually resident population"? The
extreme example, of course, is the City of London, where a large
number of people come in to work every day. How do you cope with
that?
Mr Dugmore: I think this highlights
the particular interests of commercial users, because when you
look at the use of population statistics by large retailers, for
example, the population estimates as published are not a great
deal of use because of their level of geography—they are
looking at local authority level rather than store catchment level,
or anything like that—and so a lot of the big retailers
would tend to go back to the Census because that gives them geographical
detail. The two other elements are the fact that the mid-year
estimate for resident population is only people here for more
than 12 months and there is a lot of interest in short-term migrant
flows and people from Eastern Europe as to the numbers of those
coming in and out, and, as you say, the workplace is another definition
of great importance, and one only has to look around Westminster
to see what impact this sort of thing has on the retail world
of people opening coffee shops or in Boots in Victoria Street,
and so on. They are overwhelmingly interested in the numbers of
workers in an area at lunchtime rather than the resident population.
Professor Rhind: But it is fair
to say, if I may continue, that in many urban areas the daytime
population, which imposes some load on the local authorities—whether
it is crime, whether it is street cleaning, whatever it is—is
very different to the usual resident descriptor, and to that extent
there is a major issue about having just one measure of population,
the usual resident.
Q11 John Thurso: Can I return to
the devolved issue? How does the quality and range of the population
statistics in England and Wales compare with that available in
Scotland?
Professor Martin: I think at the
broad level, the information which we have is broadly comparable
across the UK because we are dealing with similar systems which
are strongly related and there is a lot of cross-working between
the statistical agencies in the different parts of the UK. Nevertheless,
there are differences which seem quite subtle which play out in
differences in detail and differences in methodology which are
available. For example, we saw in 2001 different approaches to
statistical disclosure control (which is the protection of confidentiality
in the statistics by a modification of data) used differently
in different parts of the UK—that has an impact—and
there are also some the differences in the way the information
is reported—the timing of mid-year estimates in Scotland
is different to England and Wales—which make it quite difficult.
Academic and commercial organisations mostly have an interest
in the whole of the UK, rather than the local government interested
in one place. I think it would also be true to say that in some
senses there are statistical outputs which are important when
dealing with population. For example, in Scotland and Northern
Ireland there are new longitudinal surveys which have been possible
because of a very joined-up approach to different government departments
working together with the statistical agencies, and in some ways
they are leading England and Wales.
Q12 John Thurso: So there is Scottish
best practice that can be disseminated within the rest of the
United Kingdom.
Professor Martin: Maybe.
Professor Rhind: Perhaps just
two other points. For good reasons, the nature of the Census data
collected in the different parts of the UK differs somewhat. Some
of the questions are different in Scotland, perhaps reflecting
household tenure differences and so on, but there are also some
mechanical aspects underpinning that. For example, an address
database is a necessary condition for running a census, and the
situation for that is somewhat different, say, from Northern Ireland
to England and Wales. Even if the core is relatively consistent,
there are many things, some by design, some by accident, which
differ around that core.
Q13 John Thurso: I know that the
Statistics Commission has given a pretty high priority to ensuring
that there is a good quality of UK-wide statistics, and the idea
should be that there is the UK statistics and then the devolved
administrations can take out further what they want. I also know
there has been a new memorandum of understanding between the various
persons responsible. Do you feel now that enough has been done,
obviously within the bounds of being reasonable, or do you think
there are gaps that need to be addressed?
Professor Rhind: There are some
variations. The Scottish pilot was a year earlier, I think, than
the England and Wales one was going to be. In principle, if the
experience from that feeds into the England and Wales one, it
is actually rather helpful. Most recently, there has been some
change, in that I think the procurement of certain services is
going to be done separately in Scotland from England and Wales;
so one has some concerns about these variations, but I think many
organisations are pressing for the greatest possible consistency
across the piece and will continue to do so.
Mr Dugmore: Can I add to that?
I think that there is a feeling that probably some attention is
now being given to the issue but not enough, and it needs to be
relentless as well, because at the moment we are looking at the
choice of questions for the 2011 Census, and, obviously, that
is a matter of the extent to which things are comparable across
the countries, but even if the questions have much in common,
when it comes down to users getting hold of the outputs, if different
tables are generated and they are called different things, there
can be fairly trivial, practical issues in actually getting hold
of comparable data and putting it together and there is no umbrella
organisation that says, "Here is a set of data, country by
country, all of which is compatible and easy to use."
Q14 John Thurso: How would you like
to see that resolved? What would you need to do?
Mr Dugmore: I think there would
be a case for some sort of central unit to actually oversee this.
Otherwise, I think if it is left to each of the individual offices,
then they will each do their own thing. In theory, a lot of it
can be put together, but it will not happen and that will mean
that usage of the data will be much less than it might otherwise
be.
Q15 John Thurso: You lead me in to
the Scotland Act, which is somewhere we do not want to go today.
Let me go from the devolved on to the international stage. The
Committee visited Sweden, and one of the points that we were told
was that the Swedish population register was updated on a daily
basis. Is there any case for moving to a system that could be
updated more regularly here in the UK?
Professor Rhind: Absolutely. Given
that we know that migration data and, indeed, population census
data in certain respects is not fit for all the purposes we require,
it is appropriate to look around to see what other people are
doing. The Statistics Commission has recently toured parts of
Scandinavia and Europe and also been to the USA, where the approaches
are different. I think we concluded that the traditional census,
of which in some ways the British one is the most traditional
of all now given what has been happening elsewhere, has almost
had its day. We argued very strongly that we should explore other
models and we should explore them now, and I am delighted to see
in the submission to you that the ONS are proposing a study, starting
next year, to look at that. We think that there is probably no
better solution than to actually run something based primarily
upon administrative statistics, administrative data sources, with
information coming from ports about emigration and immigration,
and a whole variety of other administrative data sources used
to triangulate where people are and where they are moving to.
It is certainly made to work extremely effectively in the Scandinavian
and Netherlands and other contexts. It is inexpensive once you
have got the databases, it is very up-to-date—literally
you can detail every day, if you wish to—and the linkage
between population data, households, properties and also businesses
gives a range of outputs that we cannot possibly match, and so
there is a great deal of attraction in having something of this
sort, and the Statistics Commission have suggested that we use
the 2011 Census as an opportunity to use existing administrative
data sources to run in parallel, if you like, an experiment to
see just how the two would stack up, recognising that we need
other administrative data sources as well. I hope that will happen.
There are some disadvantages; there are some difficulties. The
data-sharing culture in British government departments is perhaps
not as good as we need it; clearly there are some confidentiality
constraints, especially in the view of recent events, which we
would need to get over. In Scandinavia this is taken for granted,
but there is clearly a different culture there. Altogether, I
think there is every possible reason to believe that we could
move, over a period of time, to a situation where administrative
data sources were the main source, supplemented by smaller surveys,
rather than having a £500 million census every ten years,
which is a huge logistical operation, which is risky in certain
regards, which is clearly expensive and which produces data some
time after everyone would like to have it.
Q16 John Thurso: Can I quickly ask
each of you to what extent the problems and bad experiences of
2001 have contributed to the public's loss of confidence in official
statistics? Would you like to comment?
Professor Rhind: It is geographically
variable. The 2001 Census, so far as we were able to ascertain,
performed pretty well in many parts of the country. This was,
of course, before the most recent flush of short-term migration,
but we know that there were problems in inner urban areas even
though greater resources were allocated to those. I think there
is no question that some local authorities have grave suspicion
about the population statistics in their areas, for better or
worse; so I think it is variable. It will have contributed in
those areas significantly to mistrust.
Professor Martin: I think that
is very true, and there are specific issues. For example, you
may recall that there was a question about the Welsh identity,
there were various organisations who have used the Census, or
the opportunity of the Census, in order to form particular views
about group identities and group prevalence, and to the extent
to which that happens, there is process of attrition against the
solidarity of the published result and so the Census is partly
suffering from that and partly contributes to it. I guess commentators
internationally would also say in some ways that the Census is
only the same as other sources of public confidence in government,
which we see in many ways under attack, and it is not just the
technicalities of the Census which produce that, the Census in
some way suffers from it too, and 2001 may have had much the same
final result even if it had been technically different.
Q17 John Thurso: What about the users?
Mr Dugmore: Yes, I think from
a user viewpoint we ought to get them more upbeat about this in
that certainly in the commercial world (who are seeking information
for small areas to make decisions) the Census is still very heavily
used and there is very little adverse comment about it. There
might be an awareness that if you are seeking to make decisions
in Central Manchester or Westminster you might need to be a little
careful, but when it comes down to comparing areas across the
United Kingdom as to where we open our next 20 stores, and so
on, you are straight back to the Census again and I do not think
that its reputation is much knocked at all amongst users. To some
extent, I think there is a danger of those who are technical specialists
tending to concentrate on those things that went somewhat wrong—you
know, the glass is three or 4% empty rather than 90% full—but
as far as the decision-making is concerned, the Census is still
very important, even seven years down the line.
Professor Rhind: May I add to
that, Chairman? I think there is some evidence that commercial
sector users (who are, of course, very important) take a slightly
different view to some other users. If you are allocating large
sums of money through the public purse, you would want to be rather
careful about this. Many of the commercial decisions are made
rather quickly and people would recognise that they are made on
the best available evidence and just get on and do it.
Chairman: Certainly we have had an extraordinary
response to the call for written evidence from local authorities
and those who are at either end of the public spending allocation
process.
Q18 Mr Dunne: I would like to pursue
that a little bit further. As MPs we are acutely aware of how
frequently our own electorate changes in various different aspects,
and in my own case over 5% a year appears to be the change in
the Electoral Roll. So, over a 10-year period between censuses,
theoretically, 50% of the population could have moved. Do any
of you have a sense for the average length of time that people
stay in the same house?
Professor Rhind: We did have some
figures recently about mortgages which showed a lengthening period
for staying in the same place before a new person mortgaged it.
As I recall—I think we will have to look at this—the
average longevity used to be about seven years. I think it is
now about ten, or something of that sort.
Q19 Mr Dunne: As a result of price
barriers?
Professor Rhind: I think a multiplicity
of factors.
|
