Upload
indra
View
40
Download
0
Embed Size (px)
DESCRIPTION
Population forecasting of small areas or ethnic groups. Stockholm, 21 st November 2008 Ludi Simpson University of Manchester. www.ccsr.ac.uk/popgroup. Population forecasting: 2 practical dilemmas. Theory without software Cohort component framework - PowerPoint PPT Presentation
Citation preview
Population forecasting of small areas or ethnic
groups
Stockholm, 21st November 2008Ludi Simpson
University of Manchester
www.ccsr.ac.uk/popgroup
Population forecasting: 2 practical dilemmas
• Theory without software– Cohort component framework– Multi-regional, probabilistic, socially disaggregated– Academic research– Project-specific
• Priority problems without stable data– Sub-national and non-standard areas– Ethnic or national groups– Non-cohort, non-component methods un-informative
• Version 1 (1999) – version 3 (2005)
• Local/Regional government concerns– Replicate and develop
national agency work– Population, housing, and
labour force– Impact of local development
policies– Ethnically diverse populations– Large populations and
systems of small populations
• Standard national methods– applied to 1 or more ‘groups’,
named by the user
POPGROUP designPrinciples and practice
• Excel input files
• Excel output files
• Macros do work of structuring files, validating data, projections and most interrogation
• Easy start, then develop– the future is not what it used to be
• Integrate estimates and forecasts
Each component of population change is one input file
Pt+1 = Pt + B – D + IUK – OUK + IOV – OOV
Seven input files, plus:Constraints : population, housing and employmentSpecial populations (students, armed forces)Dwellings (vacancy, second homes, sharing
households)Jobs (commuting, unemployment)
Natural change
MigrationBase popul-ation
Future popul-ation
Each input file represents a collection of assumptions for one component
Mathematical approach to combine the varied availability of demographic rates,
past counts, and targets
Initial population and one age-schedule each for rates of fertility, mortality, migration
The only required inputs
Differentials for groups, years, age-sex groups
Used to create initial projections for year y+1
Counts of births, deaths, migration flows in period
Used to constrain the initial projection
Population, housing or labour force
Used to further constrain the migration flows
Projecting small populations (areas or ethnic groups)
• Few data for non-standard areas• Time series are unstable
– Recent past may not indicate an underlying level of fertility, mortality, migration
• Alternatives provide neither age nor components– Mathematical extrapolation– Shift share– Housing units
Example 1: replicating District population and household projections.
Detailed data available
Household outputs
Household outputsBlinkforthHousehold Types 1991 1996 2001 2006 2011 2016 2021Married couple 23,600 22,750 21,700 20,850 20,200 19,750 19,350Cohabiting couple 2,550 3,100 3,650 4,100 4,500 4,700 4,700Lone parent 1,800 2,050 2,050 2,050 1,950 1,900 1,850Other multi-person 2,300 2,500 2,600 2,650 2,700 2,750 2,700One person 10,000 11,350 12,350 13,200 14,050 14,950 15,700
All Households 40,250 41,750 42,350 42,850 43,450 44,000 44,300
Private household population 100,150 100,500 99,450 98,250 97,250 96,650 96,100Average household size 2.49 2.41 2.35 2.29 2.24 2.20 2.17
Concealed married couple 50 50 50 0 0 0 0Concealed cohabiting couple 50 50 50 100 100 100 100Concealed lone parent 200 200 150 200 200 200 200
All concealed families 250 250 250 300 300 300 300
Decomposition of Household Change
2001-2021Population
EffectHeadship
Effect Change
All groups 6,500 1,350 7,900
Abbafield 4,700 1,150 5,900Blinkforth 1,800 200 2,000
Each figure has been independently rounded to the nearest 50
Blinkforth
Married couple 23,600 22,750 21,700 20,850 20,200 19,750 19,350Cohabiting couple 2,550 3,100 3,650 4,100 4,500 4,700 4,700Lone parent 1,800 2,050 2,050 2,050 1,950 1,900 1,850Other multi-person 2,300 2,500 2,600 2,650 2,700 2,750 2,700One person 10,000 11,350 12,350 13,200 14,050 14,950 15,700
All Households 40,250 41,750 42,350 42,850 43,450 44,000 44,300
Private household population 100,150 100,500 99,450 98,250 97,250 96,650 96,100Average household size 2.49 2.41 2.35 2.29 2.24 2.20 2.17
Concealed married couple 50 50 50 0 0 0 0Concealed cohabiting couple 50 50 50 100 100 100 100Concealed lone parent 200 200 150 200 200 200 200
Decomposition of Household Change
2001-2021Population
EffectHeadship
Effect Change
All groups 6,500 1,350 7,900
Abbafield 4,700 1,150 5,900Blinkforth 1,800 200 2,000
Each figure has been independently rounded to the nearest 50
Example 2: Wards of Bradford.Births, deaths and population available for recent years
• Each area’s counts of births indicate its past level of fertility.
• Trends that are expected to affect all areas are entered on the ‘All-groups’ sheet
• Expressed as a differential to a standard schedule, the area level is continued to the future
• Local trends may be identified and used in assumptions for the future
The impact of fertility file options on the Total Fertility Rate
chart
Total Fertility Rate
1.00
1.50
2.00
2.50
3.00
3.50
1991 1996 2001 2006 2011 2016
BradMoor
Eccleshi
Ilkley
Osdal
Toller
Univrsty
Forecasts, anchored in the average training phase experience
Training phase including birth counts
4. Unusual trends may be continued in future
Forecasts
Training phase
Total Fertility Rate
1. Each area’scounts of births
2. Age-trend on All-areas sheet, from GAD projections
3. Each area’s differential, maintained in future
The impact of migration file and constraint options on migration
Net migration in year after June 30th
-400
-300
-200
-100
0
+100
+200
+300
1991 1996 2001 2006 2011 2016
BradMoor
Eccleshi
Ilkley
Osdal
Toller
Univrsty
Training phase, including population estimates at 1996 and 2000
Forecasts, anchored in the experience of the training phase
A cohort component projection allows understanding of population dynamics
from incomplete information
Percentage change since 2001
Birm-ingham White
Carib-bean African Indian
Pak-istani
Bangla-deshi Chinese Other
Population 2001 (000s) 985 690 48 7 56 106 21 5 51 Total population change 2001-2026 +12% -23% -15% +599% +11% +119% +125% +155% +164%
Impact by 2026 of each factor:
(a) Age momentum +16% +6% +17% +39% +31% +44% +49% +31% +50%
(b) Fertility Impact -4% -9% -18% +7% -12% +23% +27% -27% +21%
(c) Migration within the UK -16% -22% -8% +206% -16% -1% -16% +7% -3%
(d) Migration overseas +9% -1% -11% +280% +1% +31% +41% +105% +59% (e) Constraint to ONS total +8% +2% +5% +67% +7% +21% +23% +40% +37%
• Birmingham city: eight ethnic groups– Estimate components of change for each group
– Estimate population change due to each component
– Present the decomposition of expected population change
Example 3: The impact of housing plans on population
• Population and housing change, a two-way relationship
• Regional planners are interested– Large areas: how much land should be released for
house-building?– Small areas: what is the impact on services of
planned house-building? • Greater numbers of houses built may result in…
– More in-migration– Less out-migration– Lower household size– Higher proportion of vacancies, or of holiday-homes
POPGROUP alters migration to meet housing constraints
(running HOUSEGROUP in the background)
60/65 -74 21,274 21,361 21,484 21,554 21,183
75-84 9,484 9,363 9,155 8,951 9,187
85+ 2,611 2,751 2,892 2,926 3,039
Total 196,269 196,970 197,610 197,886 198,029
Population impact of constraintNumber of persons +56 +38 -267 -351
HousingNumber of households 76,554 77,240 77,958 78,479 79,067
Change over previous year +686 +718 +521 +588
Concealed families 800 786 771 754 737
What’s the impact on population of planned housing developments?
What’s the change in number of households and dwellings each year?
National Parks: ageing populations
Decomposition of household change 2001-2016
Population
size
Population age
structure HeadshipTotal
change
Peak District -896 1090 109 303
Cairngorms 478 732 149 1359
Peak District National Park: new housing will not create a workforce
(unless the migration age structure changes)
Population change – Dwelling led projections (2001-25)
Scenario% Population change % Working age pop
change
Recent migration continues
-14% -35%
48 dwellings p/a -6.3% -29%
95 dwellings p/a 1.1% -22%
150 dwellings p/a 9.9% -13%
Milton Keynes spill-over to Aylesbury Vale (2 wards selected)
0
5,000
10,000
15,000
20,000
25,000
30,000
Populationin 1991
Populationin 2001
Populationin 2006
Populationin 2026
KEY
Green: migration as estimated for 2001-2006. (‘Trend-based’)
Red: Zero new dwellings.
Black: alternative planned house-building
Discussion
• www.ccsr.ac.uk/popgroup
• Good forecasts incorporate good estimates
• Detailed small area forecasts are possible with few data, if they are relevant data