International Migration Using administrative datasets for migration analysis and estimation

Transcription

1 International Migration Using administrative datasets for migration analysis and estimation Peter Boden ONS Centre for Demography Titchfield May 2009 This work is part of ESRC Research Award RES (1/10/07 to 30/9/09) What happens when international migrants settle? Ethnic group population trends and projections for UK local areas 1

2 This presentation 1. Research Context 2. Migration measurement and analysis 3. The development of the New Migrant Databank (NMD) 4. Patterns and trends in immigration evident from alternative sources 5. Alternative methods for estimation of immigration flows 6. The impact upon population estimates and projections 7. Summary 2

3 Research Context 3

4 Research Context What happens when international migrants settle? Ethnic group population trends and projections for UK local areas ESRC Research Award RES , Information on the project: 4

5 Research Team Phil Rees Paul Norman Peter Boden Pia Wohland Information on the project: 5

6 UPTAP research initiative The primary aims of the UPTAP initiative are to: Build capacity in secondary data analysis Promote the use of large-scale social science data sets, both qualitative and quantitative Improve our understanding of demographic trends and processes which affect society and the population Collaborate and communicate with user groups and policy-making individuals and organisations. The initiative is predominantly targeted at early or mid-career researchers wishing to enhance their skills, experience, knowledge and expertise in secondary data analysis. Information on the project: 6

7 Dissemination Mar 2008 UPTAP conference June 2008 GLA stakeholder review Sept 2008 QMSS conference, Southampton Sept 2008 ONS, Titchfield Dec 2008 RSS, UPTAP Dec 2008 GLA stakeholder review Feb 2009 GRO Scotland, UPTAP Mar 2009 UPTAP conference May 2009 RSS, YH 7

8 Migration measurement & analysis 8

9 UK Lindsey Oil refinery 9

10 UK data quality 10

11 Migration Statistics Improvement Programme MSIP New methods Enhanced data collection New data Long-term and short-term migration Reporting methods 11

12 Data & methodology reviews Alternative reviews of migration data sources and estimation methods: Rees, P. and Boden, P. (2006) Estimating London s new migrant population: Stage 1 review of methodology. A Report commissioned by the Greater London Authority for the Mayor of London. Green, A. E., Owen, D. and Adam, D. (2008) A resource guide on local migration statistics. Report prepared for the Local Government Association. 12

13 Alternative Sources Reporting period Coverage Lowest level of geography Immigration Emigration Migrant Population Stocks Flows Age Sex Ethnicity Data sourced from Censuses UK Census Decennial UK OA All ONS Pupil Census Biannual England & Wales LSOA Children attending State Schools DCSF Surveys International Passenger Survey (IPS) Quarterly UK GOR All migrants ONS Labour Force Survey Quarterly UK GOR All (but with survey exclusions) ESDS Composite Total International Migration (TIM) Annual England & Wales LADUA Long-term migrants ONS Administrative sources Immigration Control Annual UK National Non-EEA migrants HO Work Permits Annual UK LADUA Non-EEA migrants HO Points Based System Annual UK tbc Non-EEA migrants HO Seasonal Agricultural Workers (SAW) Annual UK Postal District National Insurance Number (NINo) Registrations Seasonal Workers from Bulgaria & Romania Quarterly UK LADUA Migrant workers DWP Works & Pensions Longitudinal Study (WPLS) n/a UK Postcode UK Labour Force DWP/HMRC HO Worker Registration Scheme (WRS) Quarterly UK LADUA A8 workers excluding selfemployed HO GP Registrations Annual UK LADUA All ONS/NHS Higher Education Statistics Annual UK LADUA Students HESA Electoral Register Annual UK LADUA Voting age population Local Authorities (no central source) 13

14 Migrant impact Substantial number of studies have supplemented existing statistics (NINO, WRS, LFS) with primary data collection, specifically to examine the impact of new migrants upon: Economy The workplace Integration and cohesion Local service provision Yorkshire & Humber 120 studies identified Lewis, H. et al (2009) Refugees, asylum seekers and migrants in Yorkshire and Humber, Some attempts at producing improved migration estimates but generally constrained by sampling difficulties 14

15 New Migrant Databank New Migrant databank 15

16 Objectives of the NMD New Migrant databank Single view of alternative statistics Understand conceptual and measurement differences Framework for analysis of trends and patterns in migration Analysis of short-term and long-term migration measurement Derivation of ethnic-group migration estimates 16

17 Datasets New Migrant databank National Insurance Number Registrations NINo GP registrations Flag 4 ONS components of change MYE ONS components of change SNPP Census Census Higher Education Statistics Authority HESA Workers Registration Scheme WRS Labour Force Survey LFS 17

18 Geography New Migrant databank Countries (England, Wales, Scotland) Regions (9 English GOR) Intermediate geography NMGi (63 in E&W) NMGo (45 in E&W) Local authority District / Unitary Authority / Council Areas (417) 18

19 New Migrant databank Demonstration 19

20 Patterns & Trends evident from alternative sources 20

21 England new migrant trends England 700,000 NINO - All GP Regs Census NINO - non-accession TIM - MYE Immig 600,000 Migration Count 500, , , , All data are Crown copyright. Sources: 100% data extract from the National Insurance Recording System (NIRS): ONS Mid-year estimates; GP registration statistics provided by ONS 21

22 280,000 London 60,000 Yorks & Humb 18,000 North East 60,000 North West 260, ,000 50,000 16,000 14,000 50,000 Migration Count 220, , , ,000 Migration Count 40,000 30,000 20,000 Migration Count 12,000 10,000 8,000 6,000 Migration Count 40,000 30,000 20, , ,000 10,000 4,000 2,000 10, , Migration Count 60,000 50,000 40,000 30,000 20,000 10,000 - West Midlands Migration Count 50,000 45,000 40,000 35,000 30,000 25,000 20,000 15,000 10,000 5,000 - South West Migration Count 70,000 60,000 50,000 40,000 30,000 20,000 10,000 - East 100,000 90,000 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10, Migration Count South East ,000 40,000 35,000 East Midlands 20,000 18,000 16,000 Wales 700,000 NINO - All GP Regs Census England NINO - non-accession TIM - MYE Immig Migration Count 30,000 25,000 20,000 15,000 10,000 5, Migration Count 14,000 12,000 10,000 8,000 6,000 4,000 2, Migration Count 600, , , , , All data are Crown copyright. Sources: 100% data extract from the National Insurance Recording System (NIRS): ONS Mid-year estimates; GP registration statistics provided by ONS

23 West Midlands and Yorkshire & Humber West Midlands Yorkshire & Humber NINO - All GP Regs Census 60,000 NINO - non-accession TIM - MYE Immig NINO - All GP Regs Census 60,000 NINO - non-accession TIM - MYE Immig 50,000 50,000 Migration Count 40,000 30,000 20,000 Migration Count 40,000 30,000 20,000 10,000 10, All data are Crown copyright. Sources: 100% data extract from the National Insurance Recording System (NIRS): ONS Mid-year estimates; GP registration statistics provided by ONS

24 Leeds NINO - All GP Regs NINO - non-accession TIM - MYE Immig Share of Region 16,000 Census TIM - SNPP % 14,000 TIM - MYE 28.0% 12,000 Census 21.1% Migration Count 10,000 8,000 6,000 WRS GP Regs NINO Acc 15.0% 17.4% 22.2% 4,000 NINO non-acc 24.0% 2, % 5% 10% 15% 20% 25% 30% Percentage 24 All data are Crown copyright. Sources: 100% data extract from the National Insurance Recording System (NIRS): ONS Mid-year estimates; GP registration statistics provided by ONS

25 County of Herefordshire Share of Region 3,000 2,500 TIM - SNPP % Migration Count 2,000 1,500 1,000 TIM - MYE Census WRS GP Regs 2.0% 2.4% 2.6% 16.9% 500 NINO Acc NINO non-acc 1.3% 7.2% % 5% 10% 15% 20% Percentage 25 All data are Crown copyright. Sources: 100% data extract from the National Insurance Recording System (NIRS): ONS Mid-year estimates; GP registration statistics provided by ONS

26 Newham 18,000 16,000 TIM - SNPP 2016 Share of Region 3.9% 14,000 TIM - MYE 4.3% Migration Count 12,000 10,000 8,000 6,000 Census WRS GP Regs 1.4% 2.9% 6.2% 4,000 NINO Acc 7.8% 2,000 NINO non-acc 6.2% % 2% 4% 6% 8% 10% Percentage 26 All data are Crown copyright. Sources: 100% data extract from the National Insurance Recording System (NIRS): ONS Mid-year estimates; GP registration statistics provided by ONS

27 GOR immigration rates: TIM Immigration Rate 2006 (per 000 population) East South East Yorks & Humb East Midlands South West North West West Midlands North East All GOR excluding London 27

28 GP Registrations vs TIM estimates South West 16% South East London -13% -9% East of England 1% West Midlands -34% East Midlands 1% Yorkshire and The Humber 16% North West -12% North East 3% England -5% -45% -35% -25% -15% -5% 5% 15% 25% 35% 45% -ve = TIM<GP regs +ve = TIM>GP regs Three year comparison 28

29 Estimation Methods 29

30 Alternative Models Based on allocation of national TIM flows using proportional distribution evident in administrative datasets Do not attempt to use actual counts from administrative datasets Attempt to calculate short-term flows as residual of long-term estimation process using administrative datasets. 30

31 Alternative Models National GOR distribution LADUA distribution Long-term migrants Model A TIM GP regs GP regs 31

32 Alternative Models National GOR distribution LADUA distribution Long-term migrants Model A TIM GP regs GP regs Model B TIM GP regs NINo HESA GP regs Model C TIM GP regs NINo (all) HESA GP regs NINo (all) Model D TIM GP regs NINo (non- Accession) HESA GP regs NINo (non- Accession) Short-term migrants Model E (flows) NINo HESA GP regs NINo Model F (stock) 32

33 Immigration estimation Model B IPS Reason profile, Model B 100% Not stated, 6% 100% Not stated, 6% 90% Other, 12% 90% Other, 12% 80% 80% 70% Formal study, 26% 70% Formal study, 26% 60% 60% 50% Accompany/join, 17% 50% Accompany/join, 17% 40% 30% Looking for work, 13% 40% 30% Looking for work, 13% 20% 10% Definite job, 27% 20% 10% Definite job, 27% 0% 0% HESA NINo GP regs 33

34 Model Estimation - GOR South West -8,083 South West -19% South East -3,513 South East -4% London 20,334 London 12% East of England -13,584 East of England -23% West Midlands 11,057 West Midlands 33% East Midlands -591 East Midlands -2% Yorkshire & Humber -10,292 Yorkshire & Humber -21% North West 3,644 North West 8% North East 1,027 North East 7% -25,000-15,000-5,000 5,000 15,000 25,000 -ve = Model < TIM +ve = Model > TIM -30% -20% -10% 0% 10% 20% 30% 40% -ve = Model < TIM +ve = Model > TIM 34

35 Model estimation: intermediate geography e.g. Yorkshire & Humber and West Midlands Sheffield/Kirklees + -2,339 Sheffield/Kirklees + -15% Leeds/Bradf/Hgate + -7,644 Leeds/Bradf/Hgate + -30% Hull/York Hull/York + -4% Coventry + 1,787 Coventry + 30% Heref/Worcs/Warks Heref/Worcs/Warks + -6% Staffs/Shrops/Wolv + 4,617 Staffs/Shrops/Wolv + 89% Birmingham + 5,008 Birmingham + 32% -10,000-7,500-5,000-2, ,500 5,000 7,500 10,000 -ve = Model < TIM +ve = Model > TIM -100% -75% -50% -25% 0% 25% 50% 75% 100% -ve = Model < TIM +ve = Model > TIM 35

36 Model estimation Yorkshire & Humber North Lincolnshire 491 North Lincolnshire 105% Wakefield 513 Wakefield 48% Selby 86 Selby 38% North East Lincolnshire 121 North East Lincolnshire 25% Rotherham 93 Rotherham 10% Doncaster 111 Doncaster 9% Barnsley 55 Barnsley 8% Kingston upon Hull, City of -149 Kingston upon Hull, City of -5% Bradford -508 Bradford -8% Kirklees -624 Kirklees -21% Scarborough -134 Scarborough -21% York -529 York -21% East Riding of Yorkshire -329 East Riding of Yorkshire -27% Sheffield -2,487 Sheffield -29% Harrogate -706 Harrogate -35% Leeds-4,889 Leeds -36% Calderdale -514 Calderdale -37% Hambleton -200 Hambleton -47% Ryedale -204 Ryedale -48% Craven -205 Craven -50% Richmondshire -284 Richmondshire -58% -5,500-4,500-3,500-2,500-1, ,500 -ve = Model < TIM +ve = Model > TIM -80% -55% -30% -5% 20% 45% 70% 95% 120% -ve = Model < TIM +ve = Model > TIM 36

37 Model estimation: West Midlands East Staffordshire 544 East Staffordshire 252% Telford and Wrekin 895 Telford and Wrekin 188% Sandwell 1,679 Sandwell 161% Stoke-on-Trent 1,250 Stoke-on-Trent 136% Stafford 454 Stafford 118% Redditch 330 Redditch 100% Wolverhampton 1,234 Wolverhampton 88% Lichfield 160 Lichfield 81% Worcester 345 Worcester 73% Tamworth 84 Tamworth 65% Walsall 470 Walsall 60% Shrewsbury and Atcham 185 Shrewsbury and Atcham 56% Herefordshire, County of 366 Herefordshire, County of 51% Oswestry 57 Oswestry 50% Staffordshire Moorlands 38 Staffordshire Moorlands 34% Birmingham 2,931 Birmingham 26% Coventry 1,228 Coventry 24% North Shropshire 39 North Shropshire 21% Wyre Forest 43 Wyre Forest 15% South Shropshire 11 South Shropshire 10% Dudley 8 Dudley 1% ,000 1,500 2,000 2,500 3,000 0% 50% 100% 150% 200% 250% Model 4 vs TIM estimates Model 4 vs TIM estimates (%) Cannock Chase -1 Cannock Chase -1% Bromsgrove -12 Bromsgrove -5% Wychavon -41 Wychavon -6% Newcastle-under-Lyme -40 Newcastle-under-Lyme -7% Stratford-on-Avon -97 Stratford-on-Avon -10% South Staffordshire -23 South Staffordshire -18% Solihull -258 Solihull -26% Bridgnorth -25 Bridgnorth -26% North Warwickshire -54 North Warwickshire -32% Malvern Hills -213 Malvern Hills -35% Warwick -1,086-1,200-1, Model 4 vs TIM estimates Warwick -50% -60% -50% -40% -30% -20% -10% 0% Model 4 vs TIM estimates (%) 37

38 Impact upon local population estimates and projections 38

39 Resource allocation Robust population estimates are a key input to local authority funding, resource allocation and planning plus a wide range of business planning activities. Migration assumptions have been identified as a source of uncertainty for population estimates in a number of local authorities The issue has typically been raised where the immigration estimates are thought to be too low, thus reducing the population estimate Situation is blurred by the difficulty of identifying short-term from long-term migrants Yorkshire & Humber is a region where the immigration estimates may be too high (but there have been no complaints) 39

40 Population estimates & projections 2006-base Average per year Net Migration Area Name Natural Change Internal International Total Change Barnsley 495 1, ,611 Doncaster ,242 Rotherham ,584 Sheffield 2,489-2,153 3,268 3,605 Bradford 5,247-2,721 4,347 6,874 Calderdale ,011 Kirklees 2, ,037 Leeds 5,184-4,368 8,547 9,363 Wakefield 758 1, ,184 Kingston upon Hull, City of 1,442-1,921 2,642 2,163 East Riding of Yorkshire , ,047 North East Lincolnshire North Lincolnshire 284 1, ,668 York ,947 2,400 Craven Hambleton Harrogate ,211 1,832 Richmondshire Ryedale Scarborough Selby South Yorkshire 4, ,468 8,042 West Yorkshire 14,468-5,068 14,068 23,468 Humber 1,779 3,105 3,789 8,674 North Yorkshire 932 3,189 4,500 8,621 Leeds City Region 16,016-2,611 17,426 30,832 Yorkshire & Humber 21,668 1,311 25,826 48,

41 Population Projections Y&H Yorkshire & Humber Scenario: Revised immigration estimates Maintains existing emigration estimates Leaves natural change component unchanged Scenario Population (000) incr % incr 1 Base (SNPP 2006) 5,231 6, % 2 Base corrected 5,231 5, % Difference -220 Result Projected regional population in 2026 is reduced by 220K 41

42 Population Projections Leeds Leeds Scenario: Revised immigration estimates Maintains existing emigration estimates Leaves natural change component unchanged Scenario Population (000) incr % incr 1 Base (SNPP 2006) % 2 Base corrected % Difference -108 Result Projected regional population in 2026 is reduced by 108K 42

43 Summary 43

44 Further investigation required. Are there inconsistencies in the way GP registrations are captured and reported by GOR? Are there regional/local issues in the way (international) student GP registrations are captured? At a sub-regional level, what local issues might be influencing the capture and reporting of registration statistics: Where students live and where they study Distribution of worker registrations Long-term versus short-term migrants 44

45 Estimation and validation National Regional Intermediate Local Local evidence and intelligence is key to local calibration of a national model Sample sources are less robust at smaller spatial scales Census data may not be appropriate for sub-national distribution given the demographic changes since 2001 Administrative datasets provide an alternative for distribution of sub-national immigration at all spatial scales (the Scottish model) For estimation purposes, trying to make maximum use of available data at each geographical level (either for estimation or for validation of estimates) 45

46 Acknowledgements Research Funding ESRC and UPTAP Research Team Phil Rees, Paul Norman, Peter Boden, Pia Wohland Informal Steering Group Chaired by John Hollis at Greater London Authority Datasets ONS (Flag4 data) 46

47 End Link to the paper on immigration estimation: General information on the project: 47