BSPS Annual Conference, University of St Andrews, 11-13 September 2007 Poles apart? Assessing whether labour migration to England from the A8 countries has a distinctive geography Mike Coombes and Tony Champion mike.coombes@ncl.ac.uk tony.champion@ncl.ac.uk Acknowledgements: Simon Raybould for the maps Poles apart? Assessing whether labour migration to England from the A8 countries has a distinctive geography • Background to A8 migration • Data and approach • Comparison of A8 migration with earlier total international immigration • Comparison of A8 and other labour immigration using NINo data for 2005-06 • Analysis of the ‘drivers’ of A8 and other labour migration • Main findings and conclusions Background to A8 migration • EU enlargement in May 2004: 10 countries comprising Malta, Cyprus and 8 Accession (A8) countries from Central & Eastern Europe • UK, Ireland and Sweden opened borders fully from outset, but transitional arrangements made by the other 12 EU countries • Large numbers have registered for work in UK (>0.6 million under WRS alone), though length of stay not known (so no ‘stock’ counts) • Aim of study: to assess how far this workrelated (and mainly short-term) migration has a geography different from other inflows Data and approach • Analyse data on A8 labour migrants using data from: - Workers Registration Scheme (WRS) set up in UK for employed A8 migrants (but not selfemployed), covering first 12 months of work - National Insurance registrations (NINOs), covering both employed and self-employed • Focus on England - as principal destination of A8 migrants • Use TTWAs (170 best-fits from LA/UAs) - consistent with work-related emphasis - raises likelihood of residence and workplace being in same zone (for multivariate analyses) Comparison of A8 migration with earlier total international immigration Six migration inflows to be compared: • WRSp1: WRS registrations in the first 14 months (May 2004 to June 2005 inclusive) • WRSp2: WRS registrations over the next 18 months (July 2005 to Dec 2006 inclusive) • NI0506A8: NINo registrations by A8 nationals (year ending June 2006) • NI0506all: NINo registrations by all foreign nationals (year ending June 2006) • CensusEA: Census-based counts of economically active residents living outside the UK one year before • IPS0102: IPS-based estimates of immigration from outside the UK and Republic of Ireland for 2001-02 Distribution of England’s immigrant flows between TTWA size groups NB. Bold = overrepresentation cf 2001 pop (ie. LQ>1.0) TTWA size groups Total pop 2001 WRS P1 WRS P2 London 15.3 24.1 13.0 Other 1m+ 10.1 8.8 9.8 0.5-0.7m NI 0506 A8 NI 0506 all Census EA IPS 0102 9.3 23.3 12.5 36.5 13.1 5.6 8.6 7.2 6.6 5.4 6.2 14.2 10.8 13.0 12.4 10.9 12.1 11.1 0.4-0.5m 9.2 5.5 7.2 5.8 5.9 7.4 7.1 0.25-0.4m 13.3 9.7 8.4 7.7 16.3 15.5 10.8 11.2 10.0 <125k 11.7 17.6 18.0 13.4 13.1 125-250k 14.9 18.0 12.2 10.2 6.6 8.2 5.3 100.0 100.0 100.0 100.0 100.0 0.7-1m Total 35.9 36.6 11.6 16.0 100.0 100.0 Location Quotients, by TTWA size group, for NINO registrations 2005-06 2.5 A8 All foreign Location Quotient 2.0 1.5 1.0 0.5 0.0 London Other 1m+ 0.71.0m 0.50.7m 0.40.5m 0.250.4m 125250k <125k Location Quotients for 2005-06 NINOs: A8 compared with All foreign, 170 TTWAs A8 All foreign Location Quotients for 2005-06 NINOs: top 10 TTWAs for All foreign, A8 & Other All foreign LQ A8 LQ Other LQ 1 Boston 3.520 Boston 7.148 London 2.943 2 Peterborough 2.755 Peterborough 5.370 Slough&Woking 2.501 3 London 2.386 Spalding&Holbeach 3.900 Cambridge 1.582 4 Slough&Woking 2.383 Wisbech 3.394 Oxford 1.503 5 Spalding&Holbeach 2.126 Hereford/Leominster 2.913 MiltonKeynes 1.348 6 Cambridge 1.687 Luton 2.388 Brighton 1.256 7 Luton 1.661 Slough&Woking 2.200 Reading 1.255 8 Wisbech 1.602 Kettering&Corby 2.173 Mildenhall 1.245 9 Mildenhall 1.567 Northampton 2.125 Leicester 1.213 10 Oxford 1.401 Mildenhall 2.046 Luton 1.199 Location Quotients (logged) for 2005-06 NINOs: A8 compared with Non-A8 0.5 (O is crossover of LQ=1.0, unlogged) r = 0.557 0.0 Non-A8 O -0.5 -1.0 -1.5 -1.0 -0.5 0.0 A8 0.5 1.0 A8/Poles apart? Correlation analysis • • • • • • Just seen relationship between A8 and all non-A8, r=0.557 How far does the A8’s LQ pattern across 170 TTWAs compare with that for country groups and individual countries? Similarly, how does that for Polish nationals differ from that for the other A8 countries? Correlation analysis, using the publicly available NINO dataset for 2005-06 (data rounded to 10s) Log of recalculated LQs (nb. 10 added to all NINO raw counts (to remove zeros in the rounded raw data) For country groups (all non-A8, EU14, rest) and selected countries with 7,000+ NINO registrations A8 apart? Correlations (r) of logged LQs with country groups, and selected countries (ranked) Country group r with A8 Country r with A8 All non-A8 0.557 Portugal 0.529 EU14 0.624 South Africa 0.486 Rest of world 0.485 India 0.357 Italy 0.346 Spain 0.260 USA 0.236 Australia 0.200 New Zealand 0.182 Bangladesh 0.122 Pakistan 0.120 Poles apart? Correlations (r) with logged LQs of other A8 nationals (ranked) Country group / Country r with Poles All A8 0.946 All A8 excl Poles 0.663 Lithuania 0.508 Latvia 0.504 Slovak Rep 0.484 Czech Rep 0.387 Hungary 0.279 Estonia, Slovenia N <7k A8/Poles apart? Cluster analysis • Cluster analysis to identify communality of log-LQ pattern across the 170 TTWAs • All A8 countries with >7k NINO registering 2005-06, plus ten countries representing southern EU, New World, S Asia • K-mean cluster analysis, requesting 5 clusters, produces: Cluster 1 Cluster 2 Hungary India Portugal Bangladesh Cluster 3 Pakistan Cluster 4 Cluster 5 Czech Rep Australia Latvia Italy Lithuania New Zealand Poland S Africa Slovak Rep Spain USA A8/Poles apart? Explanatory analysis What features of the geographical context are correlated with the NINO-based patterns? Selection of ‘drivers’ to test the effect of: • • • • • • • local economic structure, e.g. % agric, manufacturing, construction/transport, retail/hospitality, other sectors tightness of local labour market, e.g. employment rate (for all, those with degrees, those without quals) population size (& log pop), urbanization index population composition (% non-white, % unqualified) previous migration (net internal migration rate, net international migration rate, % born in E Europe) housing costs (unaffordable housing index) regional location (South vs North) Simple correlation (r) with logged LQs of A8 and Non-A8: significant at 5% or better, ranked A8 Non-A8 % born in Eastern Europe % born in Eastern Europe % working in retail & hospitality % non-white Employment rate, for all Log population 2001 Net international migration rate Urbanization index % non-white Net internal migration rate Employment rate, for no quals Net international migration rate % working in manufacturing Population 2001 South (binary cf North) % working in other sectors % working in agriculture etc % with no qualifications nb.- bold italics denotes negative correlation % working in manufacturing A8/Poles apart? Regression analysis • Regression analysis to identify the separate key ‘drivers’ and their relative importance for the selected migrant groups • Need to omit variables that are highly (r=>0.6) correlated with each other, with labour market emphasised in selection: Selected AGRIC MANUF CONTRAN RETHOSP EMPRATQ0 EMPRATQ4 NOQUAL BORN1EEU NTIM0203 SOUTH Excluded because r=>0.6 with selected variable LOGPOP01 (-), URBINDEX (-), NTIN0203 (+) UNAFFORD(-), OTHIND (-) EMPRATOT (+) EMPRATOT (-), OTHIND (-) MYEPOP01 (+), NONWHITE (+) Regression results for A8 versus NonA8 NB. Bold red = significant at 5% level, N = 170 TTWAs Variable [other variables with r=>0.6] Agriculture etc [-logpop, -urb, +mig] Manufacturing [-unafford, -othind] A8 0.252 0.122 NonA8 -0.216 -0.065 Construction & transport Retail & hospitality No-quals employment rate [+emprat] With degree employment rate 0.117 0.432 0.181 -0.049 0.044 0.236 0.070 0.010 No qualifications [-emprat, -othind] Born in East Europe [+pop, +nonwhite] 0.001 0.496 -0.083 0.510 Net international migration rate 0.220 0.280 0.118 (0.412) 0.120 (0.597) South (Adjusted R2) Regression results for A8 versus NonA8 Standardised (beta) coefficient -0.3 Agric etc Manufacturing Constr & transp Retail & hosp'y No-quals empl rate With degree empl rate No qualifications Born in E Europe Net immigration South -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 A8 NonA8 0.6 Regression results for Poles vs Other A8 NB. Bold red = significant at 5% level, N = 170 TTWAs Variable [other variables with r=>0.6] Agriculture etc [-logpop, -urb, +mig] Manufacturing [-unafford, -othind] Construction & transport Retail & hospitality No-quals employment rate [+emprat] With degree employment rate No qualifications [-emprat, -othind] Born in E Europe [+pop, +nonwhite] Net international migration rate South (Adjusted R2) Poles Other A8 0.275 0.189 0.122 0.261 0.020 0.075 0.469 0.133 -0.076 -0.149 0.468 0.240 0.052 (0.378) 0.271 0.168 -0.050 0.200 0.448 0.155 0.209 (0.338) Regression results for Poles vs Other A8 Standardised (beta) coefficient -0.2 Agric etc Manufacturing Constr & transp Retail & hosp'y No-quals empl rate With degree empl rate No qualifications Born in E Europe Net immigration South -0.1 0.0 0.1 0.2 0.3 0.4 Poland Other A8 0.5 Poles apart? Main findings and conclusions • A8 inflow is less focused on London than total immigration is, but still more than ‘expected’ • More A8s going to smaller TTWAs than for total inflow, but NINO-based share smaller than WRS-based • A8 patterning across 170 TTWAs is closer to EU14 than Rest of World, and most similar to Portugal • 5 of the 6 largest A8 national inflows cluster in one group by themselves – Hungary with just Portugal • Poles parallel Rest-A8 for pull of areas with % born in East Europe, % agric, net immig and retail/hospit’y, but differ on no-quals (-/+) and manufacturing (++/+) • A8 aggregate differs from non-A8 on % agric (+/-), manufacturing (+/-); also pulled more by retail/hospit’y, constr/transp & empl rate among no-quals. But similar response to EEurope-born, South & net immig. • Much ‘unexplained’; check for proxy variables. BSPS Annual Conference, University of St Andrews, 11-13 September 2007 Poles apart? Assessing whether labour migration to England from the A8 countries has a distinctive geography Mike Coombes and Tony Champion mike.coombes@ncl.ac.uk tony.champion@ncl.ac.uk Acknowledgements: Simon Raybould for the maps Annex 1: NINO 2005-06, descriptive stats for LQs of selected country groups and countries, 170 TTWAs De scri ptive Statistics TOTAL A8 POLAND A8NPOL LITH SLOVA K LA TVIA CZECH NONA 8 EU14 NONE UA INDIA SA FRIC OZ PA KI FRANCE GE RMAN CHINA NIGERI PORTUG ITA LY SP AIN IRE LA N US A BA NGLA PHILIP HUNGAR NZ Valid N (lis twis e) N St atist ic 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 170 Minimum St atist ic .105 .090 .107 .000 .000 .000 .000 .000 .108 .000 .070 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 Maximum St atist ic 3.520 7.195 5.874 9.572 15.800 6.760 22.610 4.229 2.943 3.147 2.881 4.003 9.604 4.248 5.967 3.478 3.446 4.745 4.014 16.493 3.888 5.459 2.990 9.620 3.700 3.159 4.757 4.083 Mean St atist ic .65701 .91250 .90613 .92395 .95413 .89037 1.12220 .84597 .49420 .49300 .49457 .51871 .64004 .32716 .45146 .41753 .50842 .62193 .26090 .97688 .37657 .53816 .45522 .56016 .43062 .87631 .72088 .34558 St d. Deviation St atist ic .465795 .808327 .714070 1.111269 2.000129 .770363 2.212179 .664567 .381688 .514167 .367227 .537051 .890535 .451872 .906435 .492320 .523568 .748955 .502624 2.125768 .508221 .716413 .498228 .919616 .489831 .618748 .873818 .534190 Sk ewness St atist ic St d. 2.918 4.442 3.317 4.839 5.172 3.436 6.659 1.602 2.946 2.508 3.117 3.521 6.517 5.426 3.402 3.048 2.565 2.359 4.020 5.477 3.258 3.256 1.891 6.394 3.287 .787 2.470 3.767 E rror .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 .186 Annex 2: List of independent variables Label Description UNAFFORD URBINDEX NONWHITE NOQUALIF EMPRATOT EMPRATQ0 EMPRATQ4 NTIM0203 NTIN0203 BORN1IRE BORN1EEU BORN1CZE BORN1POL BORN1BAL AGRIC MANUF CONTRAN RETHOSP OTHIND MYEPOP01 LOGPOP01 NOSO Unaffordable Housing Index 2003 Urbanization Index 2001 % non-white 2001 % all 16-74 unqualified 2001 % 16-PA employed 2003/4 % unqualified in 16-74 employed 2003/4 (also EMPR0) % with degree in 16-74 employed 2003/4 (also EMPR4) Net total international migration 2002-03 (also NETIM) Net internal migration 2002-03 % born in Republic of Ireland 2001 % born in Eastern Europe 2001 (also BNEEU) % born in Czech Republic 2001 % born in Poland 2001 % born in Baltic States 2001 % employed in agriculture etc 2001 % employed in manufacturing 2001 % employed in construction or transport 2001 (also CONTR) % employed in retail or hospitality 2001 (also RETHP) % employed in other sectors 2001 total population 2001 log of 2001 total population South (cf North) Regression results for 5 largest A8 NINOs NB. Bold red = significant at 5% level, N = 170 TTWAs Variable Poland Lithuania Latvia Slovak Czech Agriculture etc 0.275 0.437 0.415 0.330 0.474 Manufacturing 0.189 -0.014 -0.010 0.182 0.030 Constr & transport 0.122 0.039 0.002 0.130 0.072 Retail & hospitality 0.469 0.137 0.174 0.212 0.052 No-quals empl rate 0.133 0.078 0.130 0.020 0.021 With degree empl rate -0.076 -0.002 -0.037 -0.026 -0.048 No qualifications -0.149 0.337 0.385 -0.216 -0.140 Born in East Europe 0.468 0.351 0.307 0.359 0.349 Net immigration rate 0.240 0.083 0.021 0.113 0.140 South 0.052 0.284 0.186 0.078 -0.027 Adjusted R2 0.378 0.341 0.311 0.183 0.235 Regression results for 5 largest A8 NINOs Standardised (beta) coefficient -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 Agric etc Manufacturing Constr & transp Retail & hosp'y Poland No-quals empl rate With degree empl rate Lithuania Latvia Slovak No qualifications Born in E Europe Net immigration South Czech 0.6