Transcript Slide 1

Why do we forecast?
Ludi Simpson, University of Manchester, 21/11/2014
“The survey reveals widespread and growing interest
in demographic prospects, particularly strong in the
Health Service and Local Government. Only a
minority of users looked very far ahead.
“While customers are on the whole satisfied with
official projections, the need for better local
projections was a recurrent theme. Information on
internal migration, and variant assumptions about its
course were in strong demand, as were better and
more up-to-date baseline data on the present.
“Improved methodology for local projections and
training in it is called for.”
Summary
• 3 case studies
• “Predict what you cannot control, plan what is
under your control”
– Which plans need forecasts?
– What is it about population that is not under the
planner’s control?
– Planning scenarios – what is possible?
– How to deal with intrinsic uncertainty in the
prediction?
Case Study: National Parks in Britain
• Zero total net migration, but young adults replaced by older
families
Cairngorms National Park
Peak District National Park
(Source: Marshall and Simpson, 2009)
Peak District National Park
• A reduced and elderly population, fewer of working age,
requiring more housing,
Case Study: South East Wales
Welsh government 2011-based population projections
• Greater growth
• National centre Cardiff
attracts from deindustrialised valleys
Case Study: South East Wales
Welsh government 2011-based household projections with variant population
assumptions - Cardiff
• OAN is based
on satisfying
the
continuation
of recent
trends
• Alternative
scenarios
envisage
more
balanced
development
Case study: Birmingham ethnic
composition
• Population change 2001-2026
700%
600%
(e) Constraint to ONS total
500%
(d) Migration overseas
400%
(c) Migration within the UK
300%
(b) Fertility Impact
200%
(a) Age momentum
100%
0%
-100%
• Change can be
decomposed into
each component
• A young age
structure has
intrinsic momentum
for increase
• Increasing ethnic
diversity outside
the control of
planners
Source: BCC report 2006
“Predict what you cannot control, plan what is
under your control” RJS Baker
• Why do we forecast?
– Which plans need forecasts?
– What is it about population that is not under the
planner’s control?
– Planning scenarios – what is possible?
– What if the prediction is uncertain?
Which plans need forecasts? (source: Joshi and Diamond, 1990)
% of sector using forecasts for
each reason (column=100%)
Central
govt
Local Health
Consult/
Govt
Market Res
Private
Recruitment/ personnel
33
22
33
18
60
Marketing
11
14
6
47
52
Investment
17
15
10
24
31
Location
11
19
27
41
38
Transport
28
63
2
53
6
Planning
56
97
59
65
2
Housing
28
78
4
41
10
Education
22
73
0
18
10
Welfare/Health
28
71
100
18
4
Elderly
39
75
41
6
4
Children
33
75
41
12
2
Research
39
54
51
53
21
Teaching
6
10
14
6
2
44
31
20
29
8
Other uses
What is it about population that is not under the (local)
planner’s control, and therefore should be predicted?
• Fertility and mortality?
– Almost always considered not locally plannable
– Therefore a ‘trend’, or ‘business as usual’ projection is used as a
prediction, informed by expected long-term national trend
– Usually predicted locally by applying trends from national projection
– The official projection is usually used
– Occasionally local data establish the recent experience
• International migration?
– Debatable influence of planners at national scale
– Taken as not under planners’ control at local scale (?)
– The official projection is usually used: ‘trend’, or ‘business as usual’
Can planners control internal migration (within UK)?
• National town planning policy assumes that internal migration
is affected by the local strategic plan
– Extra (or restricted) housing brings extra (or fewer) residents
– Extra (or restricted) jobs bring extra (or fewer) residents
• For other planners (adult care, schools, youth service, store
location, recruitment and personnel, ...), internal migration is
not under their influence
– Accept the ‘business as usual’ projection
– Or the local plan once adopted
• For the projectionist, a tale of many guvnors
– Scenarios, projections, forecasts, predictions
Planning scenarios – what is possible?
• Extra (or restricted) housing brings extra (or fewer) residents
– How many extra or fewer residents?
– Answer not provided by the official projections (“business-as-usual”)
– Software provides answers (POPGROUP, Chelmer, ...)
• How do households translate into need for housing?
– Vacant housing, second or holiday homes
• How does the labour force translate into filling jobs
– Commuting, unemployment
• Migration flows are adjusted to fill the extra housing or jobs
www.ccsr.ac.uk/popgroup
Current National Planning Framework Policy (NPPF)
• Objectively assessed housing need
– Based on the official projection of population and
households (trend, business as usual)
• updated with the most recent demographic information
– Various possible upward adjustments (See PAS June 2014)
• Suppressed housing supply in past; special circumstances;
expected jobs development
• Disputed at examinations
Current National Planning Framework Policy (NPPF)...
continued
• Housing targets
– Planning adjustments to Objectively Assessed Housing
Need
• To achieve national policies such as Green Belt, environmental
protection and local aspirational additional jobs growth
• To achieve local policies that may increase housing provision
– All downward adjustments must be compensated for in
other (usually neighbouring) districts
• Duty to co-operate
• “At the heart of the National Planning Policy Framework is a
presumption in favour of sustainable development, which should
be seen as a golden thread running through both plan-making and
decision-taking”
What if the prediction is uncertain?
• The prediction is intrinsically uncertain
• SNPP average error measured against 2006 MYE:
Years ahead projected:
2
2
3
6
8
ONS 2008 Subnational population projections accuracy report
• For LAs, average error about 0.5% per year of projection
• But it does not show the error for different sizes and
characteristics of LA
10
Recommendation to local projectionists: sensitivity test
for plausible uncertainty
• Penultimate extreme values from past decade
–
–
–
–
Fertility
Mortality
Migration within UK
Migration with overseas
• Other plausible alternatives to the prediction
– ONS high and low fertility and mortality variants
– Migration from early 2000s
ONS 2010-based NPP, publication on variants (2012)
ONS 2012-based NPP: variants
Dealing with uncertainty
• Policy managers won’t like you
• Cost analysis of errors
– Over-projections and under-projections have different costs
• Wasted idle resources vs poorer quality of provision
– These costs may be assessed differently by different people
• School staff vs educational finance officer
• Developer vs resident vs planner
– Political assessment of costs is necessary
• The point of considering uncertainty is to affect a plan
– Weigh costs of over- and under-projection: affect the plan
– Contingency plans for alternative plausible predictions
Summary
• Many different planners require demographic
projections
– Dissemination to suit many guvnors
• Some things not under the planner’s control
– Predict the firm trends that all plans need to cater to
– Measure the prediction’s uncertainty and declare it
• Planners have responsibility
– To plan
– To cope with uncertainty
So what?
• What could help? (collaboration/ government statistics
offices/ LGA etc / LARIA / University research)
–
–
–
–
Databank of indicators
Projection scenarios
Accuracy studies
Training in demographic modelling
• Authoritative research on key debates
– An end to ever-smaller households: a new long-term trend
or a short pause?
– OAN and housing plans: creating a common language
– Unattributable Population Change: monitoring the 2010s