Title: Labour Force Survey LFS and Labour Market Statistics LMS
1Labour Force Survey (LFS) and Labour Market
Statistics (LMS)
Adrian Jones - Deputy Director Regional and
Economic Analysis Division
2What is this session about
- An introduction to ONSs LFS and Annual
Population Survey as sources of regional and
local labour market data - To increase your knowledge of the Labour Force
Survey how to use it and how it relates to
other Labour Market Statistics
3Agenda
- Labour Market Framework and its sources
- Accessing the LFS/APS Products
- LFS survey design and processing
- Weighting
- Strengths, Limitation and Sampling variability
- Changes
4Key customers
- Bank of England
- Treasury and other Government Departments
- European organisations (ILO, OECD, Eurostat)
- Local Government health authorities
- RDAs, GO, Trade bodies
- Members of Parliament (PQs)
- The media
- Researchers/academics
- Private companies
5Labour Market Framework
- On the supply side of the labour market we have
people who supply their labour - They can be
- employed
- unemployed
- economically inactive (potential supply)
- On the demand side we have employers who need
people to fill their jobs
6Basic framework structure
- Labour Supply Working age employment rate
- Activity rate
- Inactivity rate
- Unemployment rate
- Benefit claimants Claimant count rate
- Incapacity benefit rate
- Other benefits rates
- Total working age benefit rate
- Labour Demand Jobs/jobs density
- Vacancies/vacancies density
7Sources for filling the framework
- Labour Supply
- Labour Force Survey
- Annual Population Survey
- Modelled unemployment estimates
- Population census
8Labour Force Survey
- Quarterly Survey
- Use the quarterly LFS for labour supply estimates
nationally and regionally (GORs) - But also need a way of using LFS data for smaller
areas if we are to have a local area Framework
for LMS
9Annual Population Survey
- Annual version of LFS
- Includes responses from everyone who responded
to the quarterly LFS in a year (but only one
record per respondent) - Plus sample boosts respondents not included
in the quarterly LFS
10Model-based unemployment estimates
- Even with boosted annual LFS can publish
reliable unemployment data for only about 1/4
of local authorities - Developed a model which uses annual LFS data and
a (very accurate) covariate the claimant count - The model produces estimates of unemployment
levels and rates for local authorities
11Sources for filling the framework
- Labour Demand
- Workforce Jobs
- Annual Business Inquiry
- Jobs Densities
- Vacancies
12Workforce Jobs
Estimates of jobs based on a number of sources
STES LFS DASA Admin sources govt trainees/
Public Sector ABI (for benchmarking) Estimates
published for Regions
13Annual Business Inquiry (ABI)
Annual Survey of Employers Provides estimates
of numbers of employees for detailed industrial
and geographical breakdowns (though sample unit
is the reporting unit not the local unit)
14Jobs Density
- Total jobs in an area divided by the resident
working age population - Numerator similar to WFJ, but estimated annually
rather than quarterly - Available for small areas (local authorities,
parliamentary constituencies)
15Vacancies
- Two sources for vacancies data
- Vacancies survey
- Jobcentre vacancies
- Vacancy survey provides good estimates but only
nationally - Jobcentre vacancies available for small areas,
but administrative data and problems with using
them as statistical data
16Benefits Data
- Dont fit into either labour supply or labour
demand but are relevant to the labour market - Main benefit of interest is Jobseekers Allowance
a benefit associated with unemployment (though
NOT a measure of unemployment) - Other relevant benefits are those associated with
economic inactivity, for example incapacity
benefit and disability living allowance
17Recap
- Labour Supply Working age employment rate
- Activity rate
- Inactivity rate
- Unemployment rate
- Benefit claimants Claimant count rate
- Incapacity benefit rate
- Other benefits rates
- Total working age benefit rate
- Labour Demand Jobs/jobs density
- Vacancies/vacancies density
18Accessing Labour Market Statistics
- National First Release
- Eleven country and regional First Releases
published Quarterly - on paper and as pdf on National Statistics
web site - data also available from NS web site (excel
file) - Local area data also published quarterly on NS
web site (Local labour markets statistical
indicators) - downloadable pdf and excel spreadsheets
19Labour Market Statistics Integrated First Release
20Regional Releases
21Periodicity Rolling months and publication
dates
Levels and change
22Regional Releases
23Seasonal Adjustment
24Regional Releases
25Confidence intervals interpretation
26APS based tables
27Accessing Labour Market Statistics
- Nomis
- Data stored in databases which can be
interrogated in many dimensions - time
- geography
- variable, etc.
- Can download just the cells of interest not
necessary to download a whole spreadsheet
28Nomis
http//www.nomisweb.co.uk/reports/lmp/la/contents.
aspx
29Access to LFS/APS data
- NS website SA/ONS
- Printed publications ELMR, Regional Trends
- NOMIS
- Neighbourhood Statistics Developments, profiles
and training - Ad hoc analysis of LFS databases
- Data Archive
- LFS Dataservice
30LFS databases
- Quarterly databases
- for each seasonal quarter from spring 1992 to
date - Longitudinal databases
- Link data from individual respondents
consecutive interviews - Mainly used to analyse numbers of individuals
moving between employment, unemployment and
inactivity - Household databases
- Used for analyses at family unit or household
level - (eg. workless households lone parent families)
- Experimental APS household databases
31Unit level data
- Eurostat Regulations
- Member states are required to conduct an LFS
- Two-thirds of LFS is for Eurostat
32History of LFS
- Great Britain
- 1973 - 1983 (biennial)
- 1984 - 1991 (annual)
- 1992 - present (quarterly)
- Northern Ireland
- 1973 - 1983 (biennial)
- 1983 - 1994 (annual)
- 1994 - present (quarterly)
33Enhancements- ALALFS (2001 to 2003) and APS
- England ELLFS(sponsored DWP/DfES)
- Spring 2000 (annual) 875 per LEA (450 London
Borough) - Wales WLLFS(sponsored NAW)
- Spring 2001 (annual)
- Scotland SLLFS(sponsored SE)
- Spring 2003 (annual)
- APS boost (ONS) -2004 to 2005 (875 UALAD)
- APS datasets rolling basis 2004 onwards
34LFS Survey Design - The Sample
- 57,000 UK households every quarter
- 55,000 from GB
- 2,000 from NI
- 135,000 individuals every quarter
35Survey Design Communal Establishments
- LFS estimates relate almost exclusively to the
population living in private households (97 of
population). - Missing 81,000 employed and 8,600 unemployed
- Most people living in communal establishments are
excluded - Exception of NHS accommodation
- And students in halls of residence
36What are communal establishments
37Survey Design How is the LFS Representative?
- Stratified sample (whole of UK sampled)
- Random selection within stratified areas
- Distributed between 13 stints in each area in
each quarter - Random sample
- systematic sample of postcodes with random start
- Postcode Address File
- NHS Sampling Frame
- Telephone Directory/North of the Caledonian canal
- Weighting of data
- Helps correct for non response
38Wave structure An Addresss Passage in the LFS
- Drawn in the sample
- Receive a letter informing them they have been
selected - 1st interview
- 2nd interview (3 months later)
- 3rd interview (3 months later)
- etc
39Wave structure of the LFS
Households are interviewed in five successive
quarters
40Fieldwork Interviewing Face-to-Face
- Almost all wave 1 interviews are Face-to Face
- LFS has 250 field interviewers
41Fieldwork Interviewing - Telephone
- Generally waves 2-5 are all done by telephone
- Scotland North of Caledonian Canal
- 300 Interviewers
- Titchfield
42Fieldwork Proxy Interviews
- Not always possible to interview everyone in the
household - In these cases the LFS interviewer will collect
some information about these individuals from
another adult person in the household. - Approximately 30 of LFS data is collected in
this way.
43Fieldwork The LFS Questionnaire
44Fieldwork Blaise
45Data Validation
- Most data checking takes place during the
interview because of checks built into the
questionnaire (for example income checks) - Other checks take place in the office
46Derived variables
- Derived variables combine answers to two or more
questions. - eg. the variable WORKAGE combines AGE and SEX
to obtain the working age population (men aged
16-64 and women aged 16-59). - e.g. ILO unemployment estimates are derived from
questions on - whether worked in previous week
- whether had a job they were on holiday from
- whether looked for work in previous 4 weeks
- whether available to start work in next 2 weeks
- whether waiting to take up a job
47ILO employment status
Population aged 16
Economically active
Economically inactive
Wants a job
Does not want a job
Employed
Unemployed
Not seeking
Not available
Government supported training
Self - employed
Unpaid family workers
Employees
Why?
Why?
48Employment (LFS)
- Aged 16 or over
- at least one hours paid work in a week
- temporarily away from a job
- unpaid work for a family business
- Government-supported training programmes
49Economic inactivity
- neither employed nor unemployed
- dont want a job
- want a job but arent looking
- want a job but not available to start
50Weighting
51Weighting up to the population
52Weighting
53Weighting
- LFS weighting is a 3 stage process.
- Multi-stage weighting
- at local area level
- by age sex (single age, national data)
- by region, age sex.
- Population is split into subgroups (cells) by
region, age and sex. - Weight for each cell population in the subgroup
divided by sample size in the subgroup.
54(No Transcript)
55(No Transcript)
56Strengths
- Large sample size
- Comprehensive
- Meets international standards
- Timeliness (monthly estimates of key economic
indicators) - Response rate (about 70 wave 1)
57Limitations
- Coverage (most communal establishments not
included) - Sampling variability
- LFS weighting (uses population estimates)
- Discontinuities
- Self-classification (Industry an issue)
- Proxy responses (Qualifications, income etc)
58Sample errors Quarterly LFS
59Averaging over quarters
60APS - More Complex survey
- Guidance in user guides 1 and 6
- Design effect
- People in different areas do not have even
probability of being in survey - The bigger the sampling fraction the more
reliable the estimate - Clustering
- Collecting via household but, measuring
individuals
61Thresholds
- Thresholds crude assumption of common design
effect (cv20 and effect of 1) - Qtr LFS 10,000
- LADB 2,000 and 6,000
- APS 2,000 4,000 and 6,000 per LAD
- Variables affected by clustering
- Eg Ethnicity has a multiplier for the thresholds
of 2.5 for England
62Changes to come - Re-weighting
- Population estimates and projections are used to
weight the survey results. - Re-weighting is the process of moving from one
set of population data to more up to date
population data. - The last re-weighting exercises were in April
2000, April 2002 and February 2003 . - The 2003 re-weighting is based on the 2001 Census
results. - Microdata on 2001 levels and projections
- Aggregates interim reweighted to latest
63Changes to come - IHS
- Full implementation of the survey is expected to
be completed in 2009 by when a new survey module
will have replaced the Labour Force Survey
64Changes to come- IHS
65(No Transcript)
66(No Transcript)
67Links to sources
- Regional First Releases
- http//nswebcopy/onlineproducts/lms_regional.asp
- Nomis
- http//www.nomisweb.co.uk/reports/lmp/la/contents.
aspx - Neighbourhood Statistics
- http//neighbourhood.statistics.gov.uk/disseminati
on/ - Guide to LMS
- http//www.statistics.gov.uk/about/data/guides/Lab
ourMarket/ - User Guides
- http//www.statistics.gov.uk/statbase/Product.asp?
vlnk1537