Transcript Looking for a way - Office for National Statistics

Improved Methods for Population and Migration Estimates
ONS Centre for Demography
May 2007
Improved population statistics methods in
2007
• ONS is planning to make improvements in 2007 to
the methods for distribution of the national
population estimates to local areas.
• These principally concern how long term
international migrant numbers are distributed
locally
Improvements being implemented in
Population Statistics in 2007
• Improved distribution of international in-migrants
• Improved distribution of international out-migrants
• Improved assumption setting on proportion of
people who change their intentions on length of
stay (‘switchers’)
Improvements being implemented in
Population Statistics in 2007
• Improved distribution of international in-migrants:
– Between England & Wales and between regions of England
– From regions to local authority level (LAs) using 2 stage process via
improved intermediate level geography
– Improved age distribution of in-migrants
• Improved distribution of international out-migrants:
– Between LAs using 2 stage process from regions to LAs via
improved intermediate level geography
– Improved age distribution of out-migrants
• Improved assumption setting on proportion of people who
change their intentions on length of stay
ONS wider work programme on population
statistics
• This work is part of ONS wider work programme
on population statistics, which aims to
– Improve population estimates between Censuses as
well as the 2011 Census-based estimates,
– minimise the risk of divergence between Censuses
– providing a better understanding of the differences that
remain between the Census and rolled forward
estimates
Further improvements to population
statistics
• Increased sample size of out-migrants in
International Passenger Survey in January 2007
• Reporting the recommendations of the Task Force
on International Migration
• Review of port surveys
• Use of administrative data in population estimation
• Researching feasibility of estimating short term
migration
• Reporting on two of four Local Authority Case
Studies that started at the end of 2005
Timetable for improvements in 2007
• Population estimates, based on the revised
methods will be released in August
– first release of figures for 2006
– revised estimates for the years 2002-5
• Current sub-national population projections, based
on the 2004 estimates, will be revised
(September).
Timetable for further improvements
• Products from this work will be made available as
soon as they are completed.
• No further revisions to existing population
estimates will be made for at least two years.
Quality Assurance and Communication
Methods were:
• subject to rigorous QA during development
• independent internal QA
• discussed at CLIP
• presented at conferences
• external QA panel
• May seminars explaining improvements and impacts
Improved Methods for Estimating
International Migration
Emma Wright, Jonathan Smith & Fiona Aitchison
Overview of Improvements
• Improved geographical distribution of
international in-migrants at:
– Wales/GOR level
– local authority level
Emma Wright
Jonathan Smith
• Improved geographical distribution of
international out-migrants at:
– local authority level
Fiona Aitchison
• Improved estimation of people whose intentions
change with respect to length of stay
Emma Wright
Improved geographical
distribution of
international in-migrants
at country/GOR level
International Passenger Survey (IPS)
• Basis for estimates of long-term international
migration
• Continuous voluntary sample survey
conducted at majority of UK ports
– Airports, Sea Ports & Channel Tunnel
• Migrants interviewed at the start of their stay
– responses based on their intentions
Estimating Total International Migration
Irish Flows
Asylum Seekers
Visitor Switchers
IPS Migrants
Migrant Switchers
Total
International
Migration
(TIM)
Labour Force Survey (LFS)
• Large sample survey of 60,000 households
per quarter; most communal establishments
not sampled
• International in-migrants identified as those
resident overseas one year previously
• Geographical location of in-migrants based
upon actual place of residence
Census
• Coverage of communal establishments as
well as private households
• International in-migrants identified as those
resident overseas one year previously
• Geographical location of in-migrants based
upon actual place of residence
Proportion of UK immigrant inflows by UK
countries/English regions, 2001
N. Ireland
Scotland
Wales
South West
South East
London
East
West Midlands
East Midlands
Yorks & Humb
North West
North East
0%
10%
IPS
20%
30%
LFS
40%
CENSUS
50%
UK Country/English Region level:
Comparison of IPS and LFS
• Higher percentage of international in-
migrants allocated to London in IPS:
– consistent over time
– greatest for young adult age-groups
• Evidence from IPS and Longitudinal Study:
– higher percentage of migrants intending to live
in London state that they are likely to move on
to a different area
– migrants who move on quickly unlikely to be
recorded in internal migration sources
UK Country/GOR level: Methodology
• IPS and LFS data used in combination:
– Distribution of in-migrants in LFS used to
allocate IPS in-migrants at GOR/country level
• Methodology developed:
– worked at IPS-contact level and minimised
changes to weightings
– took account of London/non-London differences
in distributions by age and sex
– used three years’ LFS distributions for
robustness
Change to country/regional distribution of
in-migrants, mid-2004
40%
30%
20%
10%
0%
Wales
North East
North West
Yorks &
Hum b
Existing
East
Midlands
West
Midlands
East
New
London
South East South West
Difference between existing and new
geographical distributions
4%
3%
2%
1%
0%
-1%
-2%
-3%
-4%
-5%
-6%
Wales
North East
North West
Mid-2002
Yorks &
Hum ber
East
Midlands
Mid-2003
West
Midlands
Mid-2004
East
London
South East South West
Mid-2005
Improved geographical
distribution of
international in-migrants
at Local Authority level
Existing Method
Geographic Level
Data/Methods
England and Wales
National
Published IPS
Published IPS
GOR/Wales
North East
Intermediate
Geography:
HA/FHSAs
Local Authority
Newcastle &
N Tyneside
Newcastle upon Tyne
North Tyneside
3 Year IPS
average used to
apportion
GOR/Wales
Distributed to
Local Authority
using Census
immigration
How are we improving distribution to LA?
• ‘Central tendency’ within regions
• Replace intermediate geography
• Harmonise methods estimates/projections
• Age distributions
Local area distribution of in-migrants:
data sources considered
• Surveys
– International Passenger Survey
– Labour Force Survey (LFS)
• Census
• Administrative Sources
– Worker Registration Scheme (WRS)
– National Insurance Numbers
– Patient Registers
IPS Data – Central Tendency
• Example of central
tendency
• Newcastle
• Comparison between IPS
and Census
Identifying Areas with a ‘Positive Central
Tendency’
Percentage of all In migrants to LAs in North East
Upper Confidence Interval
60%
IPS
CENSUS
50%
Lower Confidence Interval
40%
30%
20%
Durham
0%
Sunderland
10%
Newcastle upon Tyne
Percentage of North East In Migrants
70%
Local Authority
New Migration Geography In-migration
(NMGi)
• Positive central tendency LAs
• Thirteen in total
• Grouped with neighbouring LAs
• Remaining LAs grouped
- Sample size
- Neighbouring LAs
New Method Overview – Outside London
Geographic Level
Region
NMGi
Local
Authority
Local
Authority
Data/Methods
North East
NE1
Newcastle
Newcastle
In-Migration Controlled
IPS Data
at GOR
3 Year IPS
average to
NMGi
Census
data to LA
Control to New
Regional Total
In-migration Within London
IPS data
• London – unique as a city and region
• In-migration spread across London
LFS data
• Household survey – not intention based
• Required to group LAs together
• Foreign students in LFS
Migration Geography in London
New Method Overview - London
Geographic Level
London
NMGi
Local Authority
Local Authority
Local Authority
Data/Methods
London Non-Students
London Students
IPS
Data
3 Year LFS
average to
NMGi
LO2
Non-Students
Brent
Non-Students
Brent
Students
Brent
In-Migration
Brent
In-Migration Controlled
Census
data to LA
Combine Students
& Non-Students
Control to New
Regional Total
Age Group
85andover
80-84
75-79
35%
70-74
65-69
60-64
55-59
50-54
45-49
40-44
35-39
30-34
25-29
20-24
15-19
10-14
5-9
0-4
Percentage of All International In Migrants
Age Distribution
• Current Method – National Age Distribution
45%
40%
England & Wales
30%
25%
20%
15%
10%
5%
0%
Improving Age Distribution
Local Areas have different age profiles
• Students
• Workers
• Joining friends/family
IPS data can’t be used directly – group LAs together
Census – Similar in-migrant age distributions
Age Group
85+
80-84
75-79
70-74
65-69
60-64
55-59
50-54
45-49
40-44
35-39
30-34
25-29
20-24
15-19
10-14
5-9
0-4
Percentage of All Internationl In Migrants
Final Age Distribution Examples - Females
45%
40%
35%
30%
25%
Group A
20%
Group B
15%
10%
5%
0%
Improved geographical
distribution of
international out-migrants
at Local Authority level
Existing Method
Geographic Level
National
GOR/Wales
Intermediate
Geography:
HA/FHSAs
Local Authorities
Data/Methods
England and Wales
Published IPS
North East
Published IPS
Newcastle &
North Tyneside
3 Year IPS
average used to
apportion GOR
Newcastle upon Tyne
North Tyneside
Previous year’s
resident population
used to apportion
HA/FHSAs
Limitations with Existing Method
• Method assumes that within the intermediate
geography everyone is equally likely to migrate
• Intermediate geography has too few sample points
in many areas – especially NE and North generally
• Intermediate geography is obsolete – health
geography areas no longer used
Challenges to developing a new approach
• Very limited sources/information on international
out-migrants
• Sample size is a big issue.
– Sample size precludes direct estimation of migrants
from IPS at LA level.
New Method
Geographic Level
England and Wales
National
GOR/Wales
New Migration
Geography for
emigrants (NMGo)
LAs
Data/Methods
Newcastle upon Tyne
Published IPS
North East
Published IPS
NEI1
3 Year IPS
average used to
apportion GOR
North Tyneside
5 Other LAs
Propensity to
Migrate model
used to apportion
NMGo
Examples of New Intermediate Geography:
North East and South East
Reading
North
Tyneside
Newcastle
upon Tyne
Brighton and Hove
Portsmouth
Propensity to Migrate Model
• Used to apportion from intermediate geography to
local authority level.
• Uses model to predict the number of migrants per
head of population.
– Model uses linear regression
– Forward Stepwise entry selection method
• Consists of a number of socio-economic and
demographic factors.
• Model results in a significant improvement
– The percentage of variance explained is increased
– R2 increases from around 40% to over 80%
Example of the Model: 2005
• In 2005 the variables below are used to form the
model, in addition to a constant term.
•
•
•
•
•
Estimated in-migrants
Females aged 10-14
Persons in lower supervisory/technical occupations
Persons in Black ethnic group
Density: persons per hectare
• The model results are then scaled to the total of
the New Migration Geography area (NMGo).
Evaluation of Model
• The Propensity to Migrate Model has been
evaluated to ensure continuity over time
• A number of different models have been evaluated
• The variables included in the model will be re-
assessed each year
Age distribution
• Existing Method:
Applies national age distribution to all LAs
• New Method:
Applies different age distributions to different types
of LAs
– LAs split into 2 groups
• London areas and large towns
• Other areas
– Differences between British and Non-British citizen age
profiles accounted for
– Results constrained to national age distribution from IPS
Improved estimation of people
whose intentions change with
respect to their length of stay
Switchers: Background
Actual length of stay
< 1 year
1 year +
< 1 year
Intended
length of stay
1 year +
IPS Migrants
Switchers: Background
Actual length of stay
< 1 year
1 year +
< 1 year
Intended
length of stay
1 year +
Subtract Migrant
Switchers
IPS Migrants
Switchers: Background
Actual length of stay
< 1 year
1 year +
Add Visitor
Switchers
< 1 year
Intended
length of stay
1 year +
Subtract Migrant
Switchers
IPS Migrants
Current Switcher Adjustments
Migrant Switchers:
• 5 per cent of IPS migrant inflows
• 1 per cent of IPS migrant outflows
Visitor Switchers:
• Fixed proportions of intended long stay visitors
• Different assumptions for EU and non-EU
citizens
New IPS Questions
• New questions have been asked on IPS
since 2004
• Allow actual and intended length of stay to
be compared
• Provide evidence to inform migrant switcher
and visitor switcher adjustments
New IPS Questions – Migrant Switchers
Identifying Inflows of Migrant Switchers
Person leaves UK having stayed less than 12 months
Actual length of stay
Actual length of stay
0-3 months
3-12 months
Migrant switcher question
When you last arrived
in the UK, how long did
you intend to stay for?
Intended length of stay
Intended length of stay
less than 12 months
12 months +
MIGRANT SWITCHER
New IPS Questions – Visitor Switchers
Identifying Inflows of Visitor Switchers
Person stays in UK for 12 months or more
(becomes UK resident)
Person never leaves UK
Person leaves UK
Intends to stay abroad
Intends to stay abroad
less than 3 months
3 months or more
Visitor switcher question
When you last arrived
in the UK, how long did
you intend to stay for?
Intended length of stay
Intended length of stay
12 months +
less than 12 months
FORMER
VISITOR SWITCHER
Analysis of new IPS questions:
Migrant switchers
• 3.1 per cent of IPS in-migrants estimated to be
migrant switchers, lower than the current
adjustment of 5 per cent.
• 4.4 per cent of IPS out-migrants estimated to be
migrant switchers, higher than the current
adjustment of 1 per cent.
Analysis of new IPS questions:
Visitor switchers
• Estimated proportion of intended long stay
visitors that actually stay for a year or more:
Inflows of EU citizens:
Inflows of other citizens:
27%
15%
Outflows of EU citizens:
Outflows of other citizens:
31%
10%
Overall impact of changes to visitor &
migrant switcher estimates, mid-2005
England & Wales
Impacts of new
adjustment
Existing
adjustment
New
adjustment
Inflow
-24,700
-15,100
9,600
Outflow
-2,800
-12,000
-9,200
Net inflow
-21,900
-3,100
18,800
Inflow
49,400
45,500
-3,900
Outflow
19,000
12,500
-6,500
Net inflow
30,400
33,000
2,600
8,500
29,900
21,400
(new adj - existing adj)
Migrant Switchers
Visitor Switchers
Overall
Net inflow
Future Switcher Estimates
• Calculated for each half year period
• Calculation based on data from new IPS
questions for the previous 3 years
• Estimates allocated sub-regionally based
upon new geographical distributions of IPS
in- and out-migrants
National Statistics Website Pages
Improved Methods for Population Statistics Revisions in 2007
http://www.statistics.gov.uk/statbase/product.asp?vlnk=14834
Improving Migration and Population Statistics
http://www.statistics.gov.uk/imps
Population Estimates
http://www.statistics.gov.uk/popest
Sub-National Population Projections
http://www.statistics.gov.uk/statbase/product.asp?vlnk=997
Contact Details
Website: http://www.statistics.gov.uk/imps
E-mail: [email protected]