-----Original Message-----
From: [log in to unmask]
[mailto:[log in to unmask]]
Sent: Friday, July 16, 1999 11:15 AM
To: [log in to unmask]
Subject: More Misleading ONS Figures
MORE MISLEADING ONS FIGURES; THREAT TO HOUSE OF COMMONS LIBRARY
PARLIAMENTARY CONSTITUENCY UNEMPLOYMENT RATES
British official unemployment statistics continue to go from bad to
worse. ONS has now started to publish, in Labour Market Trends and I
assume in NOMIS, what it untruthfully calls claimant unemployment
"rates" for parliamentary constituencies and for NUTS3 areas. These
are compiled on the "workforce" basis already used for TTWAs and local
authorities, which grossly misrepresent unemployment in many areas. A
"workforce" unemployment "rate" is the claimant unemployed resident in
the area divided by the sum of the number of people with a workplace
in the area plus the resident unemployed. Areas with net incommuting
have their rate underestimated (denominator too big) while those with
net outcommuting have their rate overestimated (denominator too
small). The figure is not a "rate" because many, and frequently most,
individuals appearing in the denominator are not at risk of appearing
in the numerator. It is NOT the rate of unemployment among residents
of the area; it is also NOT the rate of unemployment of those who
work in the area since if they become unemployed they will be ascribed
to their area of residence. It is NOT the rate of unemployment for
any meaningful aggregate and to express unemployment as a percentage,
say, of the number of starlings nesting in the area would be just as
useful, and statistically valid.
As explained below, this action by ONS has created a threat to the
survival of the truthful and very valuable parliamentary constituency
estimated unemployment rates (which really are rates) published by the
House of Commons Library. I hope members of this list will take such
action as is open to them to help ensure the continuation of this
important House of Commons work.
To illustrate how bad the new ONS figures are, we can compare them
with the House of Commons figures for the same date (May 1999). The
worst affected are constituencies which happen to contain a city's
central business district (CBD). Thus we have the following, with the
approximately true House of Commons figure shown first and the false
ONS figure second: Birmingham Ladywood 17.7%, 3.7%; Liverpool
Riverside 14.6%, 4.7%; Manchester Central 13.9%, 3.1%; Sheffield
Central 12.1%, 3.9%; Leeds Central 11.1%, 2.4%; Glasgow Kelvin 7.5%,
1.4%; Edinburgh Central 5.4%, 2.0%; Cardiff Central 5.5%, 2.5%;
Belfast North 11.3%, 4.8%; and, in London, Cities of London &
Westminster 5.4%, 0.4%; Bethnal Green & Bow 13.7%, 7.2%; Hackney S.
& Shoreditch 14.5%, 8.4%; Islington S. & Finsbury 10.1%, 3.5%;
Holborn & St Pancras 11.2%, 2.3%; N.Southwark & Bermondsey 11.1%,
4.1%; Vauxhall 11.2%, 6.6%. These figures are grotesque. They are
not telling you anything about unemployment, merely that these
constituencies coincide with the CBD.
The House of Commons Library states that 148 constituencies (out of
659) had ONS so-called "rates" differing by more than 2 percentage
points from the true rates in May 1999. However even this understates
how disastrously misleading the figures are, since there is a
systematic bias. It is mainly areas whose true unemployment is high
which have their unemployment badly underestimated - in other words
those where it is most important to know the true rate. Of the 25
constituencies with the highest true claimant rates at May 1999, 16
had their rate underestimated by more than 2 percentage points, with 5
actually overestimated by this amount and none correct even to within
a half percentage point; of the 25 with the lowest true rates, none
had errors of this size and 6 were actually shown correctly to one
decimal place.
It is less easy to find the size of the errors in the NUTS3 figures
except where these correspond exactly to local authority areas.
However, these errors also will be large in many cases. For instance,
the Glasgow NUTS3 area is the same as the current local authority area
and therefore the error is the same as for the "unitary authority
workforce rate" already published: whereas ONS shows 6.5%, the true
rate is about 9.3%. I have already posted details of these
corrections to this website.
The threat to the House of Commons Library figures arises because the
Library itself appears to feel threatened by the ONS figures. Its
research paper 99/62 carrying the May 1999 figures, available on the
website at www.parliament.uk, refers to the new ONS figures and states
that "For the time being, this research paper will continue to follow
the methodology used previously and will contain only the Library's
residence-based unemployment rates. A decision on the longer-term
future of the methodology.....will be taken later in the year."
Regrettably, it goes on to understate its own case by saying "The
Library believes that for a minority of constituencies - where there
is a high level of net commuting - these residence-based unemployment
rates provide a more intuitive (sic) measure of the incidence of
unemployment." The concept of "intuitiveness" is not relevant. What
the Library's figures show, foursquare and honestly, is simply the
rate of unemployment among the residents of each constituency. The
ONS figures do not show the rate of unemployment for the residents, or
for the workers, or for any other meaningful aggregate.
Unfortunately, the Library also fails to challenge ONS's description
of its figures as "rates" and appears to concede that they might have
some valid purpose (while not saying what that might be). It states
"The two sets of unemployment rates (sic) have different
interpretations. If one is principally interested in the proportion
of residents in a constituency who are unemployed" (which one would
indeed think would be a Member of Parliament's interest) "then the
Library's residence-based unemployment rate is the most appropriate
measure. If, however, one wants to express the number of unemployed
claimants as a percentage of the total number of jobs in a
constituency, then the ONS unemployment rate is the appropriate
measure." Why would anyone want to express the number of unemployed
claimants as a percentage of the total number of jobs in a
constituency? There is no proper purpose for which such a figure can
be used and I have never encountered anyone who was able to think of
one. Incidentally, I am grateful to the Library for noting a point
which even I had not realised, that if a person has two part-time jobs
they get counted twice in the "workforce" by ONS!
In the three years since ONS started publishing these dreadful figures
I have often wondered what could be the explanation for their
behaviour. I believe that it is due to three factors:-
1. Shortage of money. The chairman of the ONS statistics users group
(a businessman) was recently reported in the FT as saying that it
would cost an extra ?50m per year to give the UK a good quality set of
labour market statistics. These new "workforce" rates use the database
already created for the obsolete TTWA rates, and can be produced at
minimal cost simply by relaxing the (already rudimentary) quality
controls in respect of commuting self-containment hitherto applied to
TTWAs.
2. Commercialism. Ray Thomas has commented on this list about the
shift of ONS away from a focus on effective public policymaking
towards a purely commercial approach. ONS appears to be genuinely not
bothered about distorting the process of public policymaking as long
as "customers" are prepared to use its figures. As a matter of fact
there always are "customers" who are too ignorant to know that the
figures are misleading (even the FT regularly uses the LA workforce
figures without qualification) and other "customers" who have an
interest in dishonestly claiming that their area has a relatively
worse (or, sometimes, better) relative unemployment rate than it
actually has.
3. Ignorance. Some of those in ONS dealing with these figures do not
actually understand the problems with them, especially if they have
not come from a background in labour market statistics. The proposal
to extend the TTWA "workforce" concept to other areas while dropping
the commuting self-containment quality controls came from the former
Department of Employment, not from the CSO or OPCS.
The views expressed above are my own, not necessarily those of Glasgow
City Council. The Council has however formally objected to the
publication by ONS of "workforce" rates for Glasgow.
David Webster - Glasgow City Housing 16 July 1999
[log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|