APPENDIX 10
Memorandum from the Royal Statistical
Society (JG 16)
1. The Royal Statistical Society published
a report in 1995 on the measurement of unemployment in the UK.
Three of its recommendations are particularly relevant to the
Committee's question about the success of the official measures,
such as the claimant count area statistics and those provided
by the Labour Force Survey (LFS), in presenting the spatial disparity
in UK unemployment. These recommendations are that:
— the monthly headline count should
be based on the LFS using the International Labour Organisation
(ILO) definition of the unemployed;
— the information contained in the
claimant count should continue to be published for special purposes,
such as indicators of short-term changes nationally and locally;
— a system of national labour market
accounts should be developed, which would enable researchers to
extract, in aggregate form, all changes of employment status.
These recommendations were widely welcomed and
the broad principles endorsed through changes to the presentation
of labour market statistics implemented by the Office for National
Statistics.
2. Labour market status is now generally
recognised as complex to define and difficult to measure. There
is a need for a headline or benchmark measure of unemployment,
which is based upon the LFS and the internationally accepted ILO
definition. Different definitions of "unemployment"
are appropriate to meet various policy and operational requirements.
These requirements must be stated clearly and all concepts must
be well defined before considering methods for estimating them
reliably. For example, in the Society's report we showed, using
LFS data, that around 50 per cent of new employees were previously
classified as not in the labour force. This suggests that the
current definition of the labour force is weak and that the labour
force is not a well defined concept.
3. To present and understand the spatial
disparity of unemployment within the UK requires rates of unemployment
(rather than absolute numbers) to be estimated, in order to standardise
for differences in population size (and perhaps other factors).
There is a problem in the calculation of rates for local areas.
Rates need to be defined for both time and space. Both the numerator
and denominator should refer to the same time period and the same
geographical space. The LFS is internally consistent and designed
to be robust for large areas. However, LFS estimates of unemployment,
employment and economic inactivity are simply too volatile for
meaningful comparisons at the local level and it would not be
cost effective (or even practical) to expand the LFS to cover
every local area to a suitable degree of accuracy.
4. If labour force totals cannot be estimated
reliably then it may be possible to produce rates that can be
reliably estimated from the same data set, for example by producing
rates for the total population of working age. However, the only
solution for many small areas is to look for ways in which claimant
count data, which are accurate locally, can be used in a standard
and transparent way to devise reliable local indicators. Some
of the rates that have been produced are based on mixtures of
figures collected from different sources, as they must be, but
also covering different times and different areas. We strongly
recommend that both numerator and denominator should have the
same time and area bases.
5. This will still leave the likelihood
that the LFS estimate of unemployment, or even of claimant count
unemployment, for a small area will differ from the claimant count
derived from administrative sources. The differences reflect conceptual
and methodological differences between survey results and an administrative
count. The challenge is to produce reliable labour force totals
and components for each small area, using LFS and claimant count
data. Recent work by Harvey and Chung on this topic is now being
quality assured before it can be presented in a paper at a public
meeting organised by the Society and published in the Society's
Journal.
6. Finally, the concept of a "jobs
gap" needs to be defined and measured with care. As the Society
pointed out "falling unemployment (however measured) has
not been matched by a corresponding rise in employment".
This underlines the need for a system of national labour market
accounts, which could also be replicated locally, within which
each individual can be unambiguously placed at any point in time.
This would then account for labour supply, against which labour
demand would need to be measured (nationally and locally) through
surveys of jobs and vacancies from a representative sample of
employers. The Society notes that Statistics Netherlands, the
Netherlands Central Bureau of Statistics, has constructed such
a set of national labour market accounts.
Measurement of Unemployment Working Party of the
Royal Statistical Society
October 1999
| 
