paper presented at research methods festival 2008, st catherine’s college, oxford, 2 july 2008...

34
Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions Tony Champion and Mike Coombes [email protected] [email protected] Acknowledgements: Simon Raybould and Colin Wymer (colleagues

Upload: hollie-spencer

Post on 27-Dec-2015

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008

Migration and socio-economic polarisation within city regions

Tony Champion and Mike Coombes [email protected] [email protected]

Acknowledgements: Simon Raybould and Colin Wymer (colleagues at CURDS)

and Joseph Rowntree Foundation (project funder)

Page 2: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Background to the research that is the challenge for data/methods here

• Debate about socio-spatial polarisation in cities and its dynamics

• Two main spatial scales:

- ‘City’ versus ‘suburb’: long-standing decentralisation of better-off

- More localised segregation of rich and poor (‘gentrification’, ‘low-demand area’)

• Migration (residential sorting) seen as important in both: is this still the case?

Page 3: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Aims and scope of paper

• Examine migration’s role in altering the socio-economic patterning of city regions

• Divide task into three separate research questions with distinctive methodologies

• Give a flavour of our result

AND IN SO DOING…• Emphasise new and/or unusual elements of the

research methods used• Demonstrate value (& limitations) of the 2001

Census Special Migration Statistics Set 1, especially NS-SeC table

Page 4: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Data: highlighting innovations in 2001

Data from the 2001 Census on change of address in the pre-Census year

Specifically Special Migration Statistics Set 1, on movement between LAs (lowest level local authority areas in 2001 = unitaries/districts)

Table MG109’s counts of Moving Group Representative Persons (MGRPs) by NS-SeC

‘Moving Group’: One or more persons living in a household on Census night, after being together at a different address one year ago

‘NS-SeC’: National Statistics Socio-economic Classification (replaced SEG in 2001 Census), with some aggregation of categories to raise size of counts and reduce SCAM effect …

Page 5: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Groupings of NS-SeC for this study (4 main socio-economic groups)

1.1 Large employers and higher managerial occupations

Higher M&P

1.2 Higher professional occupations

2 Lower managerial and professional occupations Lower M&P

3 Intermediate occupations Intermediate

4 Small employers and own account workers

5 Lower supervisory and technical occupations Lower

6 Semi-routine occupations

7 Routine occupations

L15 Full-time students Full-time students

L14.1 Never worked Other unclassified

L14.2 Long term unemployed

L17 Not classifiable for other reasons

Page 6: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

The three research questions

Q1 Migration between a city and its region: is thecity’s migration balance less favourable forpeople of higher occupational status?

Q2 Migration for zones of a city region: do thewithin-region moves reinforce the existingsocio-economic geography?

Q3 Migration between pairs of zones in a cityregion: does the net flow always move people towards the better-off of the two zones?

Page 7: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

The analytical framework

• Analysis of one-year changes of address within City Regions (CR)

• For Q1, the city/suburb distinction is recast as Primary Urban Area (PUA) versus Rest of CR (RCR)

- study covers 27 of Britain’s largest cities • For Q2 and Q3 on more localised segregation,

we use a zone breakdown of the whole CRs of 3 contrasting cities

- London, Birmingham, Bristol

Page 8: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

27 large cities* (PUAs in grey) & their RCRs

(in white)

* = largest PUAs that are centres of

City Regions

Page 9: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Q1 Migration between city and its region: is the city’s migration balance less

positive for people of higher status?

Focus on migration flows between ‘cities’ and the rest of their ‘city regions’ (RCR)

Flows expressed as RATIOS of inflow to outflow (cf. NET flows, or migration rates), viz: in/out ratios for broad NS-SeC type of MGRP

‘Cities’ defined as best-fits* of LAs to Primary Urban Areas (PUAs = main built-up areas).

‘City regions’ are best fits* of LAs to GB City Regions as defined by CURDS (1995)

*To allow use of SMS Set 1

Page 10: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

In/out ratio for 27 Cities’

exchanges with their RCRs: Higher M&P

MGRPs

inflow to city greater for just 7:

Norwich Reading

Plymouth Glasgow Portsmouth Bristol

Northampton

27 cities ranked on their in/out ratios for exchanges of Higher Managerial & Professional MGRPs with their hinterlands

-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

Norwich

Reading

Plymouth

Glasgow

Portsmouth

Bristol

Northampton

Newcastle

Southampton

Liverpool

Preston

Cardiff

London

Manchester

Leeds

Sheffield

Edinburgh

Bradford

Brighton

Derby

Nottingham

Stoke

Leicester

Middlesbrough

Coventry

Birmingham

Hull

deviation from unity

OUTFLOW TO HINTERLAND GREATER THAN INFLOW TO CITY

INFLOW TO CITY GREATER

Page 11: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

3 main types of relationship between in/out ratio and all 4 broad NS-SeCs,

as exemplified by 3 cities

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Birmingham

Bristol

London

in/out ratio

Higher M&P

Lower M&P

Intermediate

Low1

1

1

32

4

2

2

3

3

4

4

In/out ratio=1.0

Page 12: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

So, 3 contrasting CRs on the basis of Cities’ migration exchanges with RCRs

by broad NS-SeC

• Bristol: in/out ratio is around 1.0 (in balance) for all 4 NS-SeC and no clear ranking (NS-SeC rank 2 3 1 4)

• London: in/out ratio is below 1.0 (more moving out of city), but positive association between ratio and social status (NS-SeC rank 1 2 3 4)

• Birmingham: in/out ratio is below 1.0 (more moving out of city), and negative link between ratio and social status (NS-SeC rank 4 3 2 1)

Page 13: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Data/method ‘lessons’ derived from our attempt at answering Q1

• Value of NS-SeC data, especially with the flexibility to group categories to suit research topic

• Value of MGRP over HRP, though as yet there is little research to help interpret this concept

• Value of using in/out ratio rather than migration rate, due to lack of denominator for MGRPs (tests for all persons show high correlation)

• Value of PUA/City Region frame to represent the inner and outer parts of the 21st century city

Page 14: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Q2 Migration for zones of a city region: do the within-CR exchanges

reinforce the existing socio-economic patterns?

• Migration measured in terms of in/out ratio (as for answering previous question)

• For each zone based on its exchanges with all the other CR zones combined, for each of the 4 NS-SeC groupings

• Compared with the presence of that NS-SeC among All Classified residents, by mapping and correlation

• Each of the three case study CRs separately …

Page 15: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

3 case study City Regions and their constituent zones

London

Bristol

Birmingham

Page 16: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Example of London (46 zones)

Page 17: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

In/out ratio for Higher M&Ps

Page 18: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Higher M&P % All Classified residents

Page 19: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

London CR: Scatterplot for Higher M&P Log in/out ratio (x-axis), % classified residents (y-axis)

r = +0.006

0.0

5.0

10.0

15.0

20.0

25.0

30.0

-0.20 -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20

Log in/out ratio for HMP

HM

P %

Cla

ssif

ied

Re

sid

en

ts

N = 46

Page 20: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Correlations (r) between Log in/out ratio and % classified residents

NS-SeC grouping

London (N=46)

Birmingham (N=60)

Bristol (N=34)

Higher M&P +0.006 +0.314 +0.045

Lower M&P -0.466 +0.304 +0.037

Intermediate +0.566 -0.012 +0.043

Low +0.538 +0.081 +0.268

N = Number of zones. Significance level: 1%, 5%

Page 21: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

So, do the within-CR exchanges reinforce the existing socio-economic

geography?

• Overall picture from this analysis: within-CR migration generally in conformity with existing social differences between zones, but not significant in over half the 12 (3 x 4) cases

• Socio-economic patterns in 2001 also result from previous years of within-CR migration, migration with rest of UK, international migration and in-situ change

• Within-CR migration also subject to other influences besides social composition

Page 22: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Data/method ‘lessons’ derived from our attempt at answering Q2

• Value of case studies to reveal different CRs’ contrasting patterns

• Useful demonstration of the heterogeneity of both inner and outer parts of cities, even using these large zones

• Maps, scatterplots and correlations are useful exploratory devices, prompting specific questions about the association between migration balance and socio-economic complexion

Page 23: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Q3 Migration between pairs of zones in a city region: does it move people towards the better-off of the two?

• Dependent variable: net flow of migrants between all pairings of zones in CR (e.g. for London CR, 45 x 46/2 = 1035 cases) for (1) All migrants; (2) Higher M&P MGRPs

• Independent variables: difference between the two zones in each pairing on 15 indicators of zone character (listed below)

• Modelling: backward regression with each case weighted by sum of the two flows

Page 24: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

More successful of the modelling resultsvariables measuring change all individual migrants HM&P

variables positively correlated with IMD London Birmingham Bristol Bristol

under 16 + + -

students + - - +

no religion + -

ethnic diversification - - +

down-skilling - -

household income + -

employment rate

employment rate change + -

local job growth -

commuting 10km(+) -

semi-detached price + + -

semi-detached price change -

unoccupied dwellings - + -

green space -

crime + -

Adjusted r2 0.160 0.245 0.523 0.427

Page 25: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Modelling results: how far can the size and direction of net flows between zone pairs be explained by zone differences?

• About half the variance in the case of Bristol CR – for both All and HMPs – but much lower for London and Birmingham especially for HMPs (results not shown)

• Perhaps unsurprising that the multi-dimensional nature of migration is not readily reduceable to a set of high/low parameters (cf. experience of MIGMOD analysis of 1990-91 flows between similar zones)

Page 26: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Modelling results: which indicators of zone differences are significant influences

on net flows between zone pairs?

• 15 indicators in model, covering 5 types of influence: demographic, social, labour market, housing, environmental (reduced from 42 on the basis of inter-correlation)

• % students included in all 4 models, but signs vary; 2001 employment rate is in none; 13 are in 1, 2 or 3 models but not always with same sign

Page 27: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Commentary on modelling results: the example of all migrants in London CR

• Net migration gain is associated with the zones with more students, higher crime, higher incomes and upskilling

• Suggestive of net flows being positively associated with young professionals and upward trajectories

• But low R2 and substantial residuals, notably between Outer S&E and some Inner London zones – see blue lines …

Page 28: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Residuals from London CR’s all-migrants model

(red=smallest, i.e. best-fit; blue=largest)

Page 29: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Commentary on modelling results: the example of all migrants in Birmingham CR

• In contrast to London, the net flow tends to be away from more deprived areas

• Migration gain is negatively associated with IMD-type variables (but also with income and commuting 10+km)

• Gain positively associated with <16s, house price, employment rate change

• But again low R2 & substantial residuals (especially radially from inner Birmingham)

Page 30: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Residuals from Birmingham’s CR’s all-migrants model

(red=smallest, i.e. best-fit; blue=largest)

Page 31: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Data/method ‘lessons’ derived from our attempt at answering Q3

• Innovative analysis of net flow between all pairs of zones to test universality of ‘cascade’ to better-off of the two zones

• Somewhat un clear results from regression analysis; could consider developing composite socio-economic indices

• Mapping of residuals can help identify unexplained patterns (not so successful in prompting other explanatory variables for inclusion)

• Low R2 echoes previous findings on the complexity of migration patterns; could reduce by narrower specification of migrant type, but limited by SMS tables and sample size

Page 32: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Summary of approach and main findings• Paper’s main emphasis on methodology: how to study

migration and socio-economic polarisation in City Regions using 2001 Census data

• Focus on direction of net flow and degree of imbalance between inflow and outflow

• 20/27 Cities lose more HMPs to RCRs than they gain from them; 3 main types of relationship between migration ratio and skill level

• Within-CR migration tends to conform to existing social differences between zones, but ‘r’ is not significant (at 5%) in 7/12* tests (* 4SEC x 3CR)

• Net migration between pairs of zones in London tends to favour the zone with more crime etc, but in B’ham shifts people from more deprived areas

• Greater appreciation of the complexity of migration dynamics: some association with social patterns, but other drivers too

Page 33: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Summary of data/method lessons

• Study illustrates the value of Census SMS data but it is for only one year in every 10

• Many innovations in 2001 Census SMS data, but with varying benefit: + NS-SeC 100%? Moving Group, Reference Person, - SCAM, students

• New measure of in/out ratio needed to deal with lack of denominator for calculating rates for MGRPs

• Greater grasp of applying analytical tools: ranking, mapping, scatterplots, correlation, multiple regression, analysis of residuals

Page 34: Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008 Migration and socio-economic polarisation within city regions

Paper presented at Research Methods Festival 2008, St Catherine’s College, Oxford, 2 July 2008

Migration and socio-economic polarisation within city regions

Tony Champion and Mike Coombes [email protected] [email protected]

Acknowledgements: Simon Raybould and Colin Wymer (colleagues at CURDS)

and Joseph Rowntree Foundation (project funder)