Allowing for the effect of migration in calculating future needs of affordable housing in rural areas.

Download Report

Transcript Allowing for the effect of migration in calculating future needs of affordable housing in rural areas.

BSPS Annual Conference, University of St Andrews,
11-13 September 2007
Allowing for the effect of migration in
calculating future needs of affordable
housing in rural areas
Tony Champion
[email protected]
Acknowledgements:
Trevor Cherrett and Commission for Rural Communities
Census data: Crown copyright
NB: The author alone is responsible for the findings.
Allowing for the effect of migration in
calculating future needs of affordable
housing in rural areas
•
•
•
•
•
Background to policy issue
Context of the specific research question
Calculating rural England’s housing needs
Observed net (within-UK) migration rates
How to allow for the effect of observed
migration
• The needs of ‘forced’ and ‘frustrated’
migrants
• Conclusions
Background to policy issue
• Lack of affordable housing is a critical issue for rural
England (Affordable Rural Housing Commission, 2006;
Best & Shucksmith for JRF, 2006)
• Soaring demand by non-locals, while affordable
housing squeezed by Right to Buy sales, reduction in
tied housing and new building focussed on towns
• Means that ‘people have to move out of the countryside
to find a decent home’, while ‘lack of housing is stifling
rural enterprise: employers say they find it difficult to
find workers’ (ARHC, 2006, p14)
• Backlog of 42,299 homes in rural England, according to
CRC (2006, p16)
• Looking forward, ca 45% of newly-forming households
aged 16-34 (the ‘critical cohorts’) are unable to access
housing in their home ward = 22,840 households each
year (CRC, 2006, p18)
Context of specific research question
• But this estimated need of 23k new affordable homes
each year ‘does not take into account the causes and
effects of migration’ (ARHC, 2006, p3)
• The CRC methodology (Roger Tym & Partners, with
David Jordan Research, 2006) uses only mortalitybased cohort survival method for first 15 years of
population projections for ‘critical cohorts’ of 16-34
• Yet it is well known that rural areas experience much
net out-migration of young adults, much of which is for
other reasons besides housing (e.g. further education,
training, work)
• This suggests the 23k is an overestimate of needs, but
by how much? ‘Further work is needed (on) urban-rural
migration patterns and motivations, and their impact on
needs of rural affordable housing.’ (CRC, 2006, p12)
Calculating rural England’s housing needs
The current CRC model involves the following steps:
• Extract 2001 census population data by age/sex
• Run cohort-survival model for 5-year age groups
• Apply Household Reference Person rate projections
• Breakdown by housing tenure
• Allow for housing re-lets
• Apply affordability thresholds:
i) Renting unable to afford
ii) Owning unable to afford, adjusted for access to capital
(i) + (ii) = Total unable to afford market housing
Definitions of Rural England:
R1: ‘small villages and dispersed dwellings’ (settlements of <1,500)
R2: ‘small villages and market towns’ (including ‘town/ fringe’ in
‘sparse areas’)
R3: ‘all rural areas’ (including ‘town/fringe’ in ‘non-sparse areas’)
HRP projections for South West R1 areas
Source: calculated from Roger Tym & Partners’ Report to CRC, March 2006
Age
2001
2006
2011
<20
589
544
476
20-24
3814
5847
6725
25-29
11692
10907
16444
30-34
23303
16773
15870
35-39
32533
25767
21596
40-44
34600
33947
26901
…
…
…
3893
3288
3801
371769
365419
364872
39398
34071
39515
….
90+
All ages
Aged <35
NB. These do not allow for the effect of migration on change in cohort size over time
Observed net (within-UK) migration rates
• What scale of net migration loss of young
adults from rural England?
• Ward-level data from the 2001 Census on net
within-UK migration (Standard Table S003)
• R1 (‘small villages & dispersed dwellings’)
record net loss of 9% of 20-24s in one year,
and net loss of 5% for 16-19s
• R3 (‘all rural areas’) register -6% and -5%
respectively for pre-Census year
• Pretty consistent across 8 Government Office
Regions
see next two slides…
Net migration rate (%), 2000-01, for England’s
R1, R2 and R3 areas, by broad age group
Source: calculated from 2001 Census ST008
R1
R2
R3
4.0
% residents in age group 2001
2.0
0.0
-2.0
-4.0
-6.0
-8.0
-10.0
0-15
16-19
20-24
25-29
30-44
45-59
60-74
75+
Net migration rate (%), 2000-01, for Government
Office Regions’ R3 areas, selected age groups
Source: calculated from 2001 Census ST008
NE
NW
YH
EM
WM
EE
2.0
% residents in age group 2001
0.0
-2.0
-4.0
-6.0
-8.0
-10.0
16-19
20-24
25-29
SE
SW
Need to allow for the effect of migration?
• Clearly, there is substantial net out-migration from
Rural England (however defined) by young adults,
especially 16-24
BUT
• This exodus may not significantly reduce the need for
affordable housing here, because few are leaving for
housing reasons:
- DTZ Pieda (1998): 13% housing reasons in highest of 5 study
areas (2 in 5 of the ca 33% involuntary movers)
- Findlay et al (1999): 6% left households for housing reasons,
18% education, 22% jobs, 47% personal (e.g. partnership)
- Rugg & Jones (1999): few HE students expected to return
BUT
• CRC model applies observed HRRs for the (remaining)
residents to the whole mortality-survived cohort
• so it is necessary to deduct the net migration effect
How to allow for the effect of migration?
Within the context of current CRC model, there
would seem to be three options:
• Adjust the 23k outcome of the model process,
i.e. by a factor based on the migration effects
• Insert an additional ‘cohort-survival’ component
alongside mortality-based one, using evidence
about the migration effect
• Replace the mortality-based cohort-survival
model with estimates of aggregate change in
age/sex cohort size
Look at each of these options in turn…
Adjust the final outcome of the model?
• Has the advantage of not needing to modify the current CRC
model and repeat all the runs
• Simplest method is to apply the census-based one-year net
migration rates for the ‘critical cohorts’
• Ideally, at level of individual GOR or even LA/UA, but broad effect
(national rate for R3) will be to apply -5.2% annual rate for 16-19s,
6.5% 20-24s, -0.5% 25-29
• Use compound rate, e.g. for 20-24s, 93.5% left after year 1, 87.4%
after 2 years…71.5% after 5 years when aged 25-29
• Ditto for 16-19s, 76.6% left after 5 years; for 25-29s 97.5%
• In terms of cohort survival through 15 ‘critical’ years, 16-year old
cohort has reduced by 46.6% by age 30 due to net (within-UK)
migration effect
• More severe depletion for R1 (small villages and dispersed
dwellings), e.g. at -9.3% for 20-24s = 61.4% become 25-29 after 5
years
• But also need to allow for migration effect on cohorts entering the
‘critical cohorts’, e.g. aged under 16 in year 0, and international
migration
Insert an additional ‘cohort-survival’ component?
• The CRC projection model begins with cohort-survival
stage, based on ward-level data on mortality rates by
5-year age group
• Would seem logical to insert an additional cohortsurvival component, based on ward-level migration
rates by 5-year age group
• As previously, could apply the one-year migration rates
from the 2001 Census, but would be better if data for
several (more recent) years could be derived from the
Patient Register (ward type aggregated to LA or GOR)
• This would produce ‘realistic’ populations to which the
observed HRR, tenure, etc., rates directly relate
• Would still need an estimate of international migration:
traditionally not a major component for rural areas, but
quite significant now with A8 etc labour migration
Use estimates of aggregate change in
age/sex cohort size?
• Replace the CRC model’s current mortality-based
cohort-survival model with estimates of aggregate
change in age/sex cohort size (e.g. by comparing size
of 20-24 group at time t with size of 15-19 at t-5)
• Next slide shows example based on 5-year age groups
of 2004 compared in size with projected 5-year-older
age groups of 2009, for England’s most rural districts
(‘Rural-80’) and all types (England total)
• This obviously allows for all ‘survival’ components
(deaths, internal migration & international migration)
• But to feed into model, this LA-level data would need to
be split into wards & re-aggregated to ward types
• Seems unsophisticated compared to having separate
models for each component, but fit for this purpose?
Projected % change in size of age cohorts,
2004-09, for Rural-80 LAs compared to all types
Source: calculated from ONS 2004-based projections for England
20
International
immigration
15
All types
% change for cohort
10
Rural-80
Mainly counterurbanization
5
0
-5
Exodus of
young adults
-10
-15
Mainly mortality effects
-20
-25
0-4
5-9
1014
1519
2024
2529
3034
3539
4044
4549
5054
5559
Age group in 2004 (add 5 for 2009)
6064
6569
7074
7579
But still the needs of ‘forced’ and
‘frustrated’ migrants
• The 3 options are mainly based on observed migration,
i.e. rates that have led to current backlog of 43k
affordable homes in rural England
• So, need to allow for those leaving ‘involuntarily’ for
housing reasons, also for some for job etc reasons in
any policy of rural economic regeneration
• Also need to allow for those denied the chance of
moving into rural settlements because of housing
availability/cost: rural workers, closer to family, lifestyle Frost (2006): rural workers commuting from towns & cities (e.g. 51% of
those working in ‘less sparse’ England villages): how sustainable?
- DTZ Pieda (1998): ca 9% households reported people wanting to live
near them, of whom only half were likely to manage this: should the rest?
- CRC (2006): England surveys regularly report strong preferences for
living in the countryside: should this desire be entirely market selected?
• But very little hard evidence on size of these 2 groups
Allowing for the effect of migration in
calculating future needs of affordable
housing in rural areas: conclusions
• Shortage of affordable rural housing is a key
policy issue
• CRC projections need to be adjusted for
migration
• 3 options suggested for allowing for observed
migration, with varying levels of sophistication
• Still need to allow for aspects of ‘involuntary’
out-migration and ‘frustrated’ in-migration
• Yet still only a weak evidence base for gauging
both observed migration and extra unmet needs
BSPS Annual Conference, University of St Andrews,
11-13 September 2007
Allowing for the effect of migration in
calculating future needs of affordable
housing in rural areas
Tony Champion
[email protected]
Acknowledgements:
Trevor Cherrett and Commission for Rural Communities
Census data: Crown copyright
NB: The author alone is responsible for the findings.