revision of the fertility model applied in the national ...€¦ · expert 1b expert 2 et expert 8...
TRANSCRIPT
Revision of the fertility model applied in the national population projection for BelgiumE U R O STAT/ UN E C E W O R K S E S S I O N O N D E M O G R A P H I C P R O J E C T I O N S , B E L G R A D E , 2 4 - 2 7 N O V E M B E R 2 0 1 9
F E D E R A L P L A N N I N G B U R E A U ( B E L G I U M )
M A R I E V A N D R E S S E
Table of contents
2
• Context
• Old methodology in a nutshell
• New methodology
• Conclusion
Context• Federal Planning Bureau (FPB):
✓ Independent public agency.
✓ Draws up studies and projections on economic, social and environmental policy issues
✓ www.plan.be
Need for demographic projection
• FPB publishes the official demographic projection for Belgium
• Projection up to 2070 – deterministic cohort component model
• Dimensions of the population projection: age, gender, districts (and nationality)
3CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
Old methodologyIN A NUTSHELL
4
Old methodology in a nutshell
5
• Based on the Age Specific Fertility Rates (ASFR) => TFR = σ𝑥 𝐴𝑆𝐹𝑅
• Age Specific Fertility Rates 2030-2070
➢ Constant
➢ Fixed at the average computed on the period 2008-2009
Total fertility rate [2030-2070] ̴ Total fertility rate [2008-2009]
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
1991 2001 2011 2021 2031 2041 2051 2061
Total Fertility Rate - Belgium 1.88 in 2070
CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
Old methodology in a nutshell
6
• Based on the Age Specific Fertility Rates (ASFR) => ICF = σ𝑥 𝐴𝑆𝐹𝑅
• Age Specific Fertility Rates 2030-2070
➢ Constant
➢ Fixed at the average computed on the period 2008-2009
Total fertility rate [2030-2070] ̴ Total fertility rate [2008-2009]
• Age Specific Fertility Rate up to 2030
➢ Linear interpolation
➢ Implicit assumption : recuperation effect
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
1991 2001 2011 2021 2031 2041 2051 2061
Total Fertility Rate - Belgium 1.88 in 2070
CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
Old methodology in a nutshell
7
Weaknesses/points to be improved:
1. Total recovery after the sharp decrease of the TFR since 2008
• Doesn’t seem so plausible anymore
• Mix of structural and conjonctural effects
2. Fertility schedule :
• Is fixed in the long run : ̴ schedule 2008-2009
• Unlikely trend in the short/medium term
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
1991 2001 2011 2021 2031 2041 2051 2061
Total Fertility Rate - Belgium
26.5
27
27.5
28
28.5
29
29.5
30
30.5
31
Mean Age at Birth (MAC)
CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
New methodology
8
New methodology
9
3 STEPS :
1. Assumption on the global fertility level in Belgium (trend) in the long run (FER_LT)
2. Projection of the annual Total fertility Rate (TFR)
3. Age Specific Fertility Rate (ASFR)
New methodology - STEP 1/3
Expert opinions :
10CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
STEP 1 : assumption on the global fertility level in Belgium in the long run (FEC_LT)
New methodology - STEP 1/3
Expert opinions :
11CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
0.50.60.70.80.9
11.11.21.31.41.51.61.71.81.9
22.12.22.32.42.52.62.7
1960 1970 1980 1990 2000 2010 2018 2020 2030 2040 2050 2060 2070
obs
expert 1a
expert 1b
expert 2 et expert 8
expert 3
expert 4
expert 5 et expert 6
expert 7
expert 8a
expert 8b
expert 9
expert 10
long term level
STEP 1 : assumption on the global fertility level in Belgium in the long run (FEC_LT)
New methodology - STEP 1/3
Level ( ̴1.7) based on (implicit) structural determinants of fertility :
1. Number of years spent in education will continue to increase
2. Share of women on the labour market will continue to increase
3. New context : « worldwide » social/political/economic/climate insecurity
This level starts from 2030:
• Most of the experts agree that the current Total Fertility rate is Particularly low
• We should observe a PARTIAL recuperation in the coming years.
12
1.4
1.5
1.6
1.7
1.8
1.9
2001 2011 2021 2031 2041 2051 2061
Belgium
Fertility trend (obs)
Fertility trend (proj)
TFR (obs)
TFR (PROJ)
CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
New methodology - STEP 2/3
dlnt(TFR)= β1*dlnt(UR)+ β2 ∗ [ln 𝑇𝐹𝑅 − ln(𝐹𝐸𝐶_𝐿𝑇)]t-1
Where
• UR = Unemployment rate (conjoncturel effect)
• TFR_LT = Fertility trend (from STEP1) (structural effect)
13
1.4
1.5
1.6
1.7
1.8
1.9
2001 2011 2021 2031 2041 2051 2061
Belgium
Fertility trend (obs)
Fertility trend (proj)
TFR (obs)
TFR (PROJ)
CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
STEP 2 : Projection of the TFR based on a Error Correction Model (ECM)
New methodology - STEP 3/3
• Method -> Shmertmann (2003)
• Based on 3 demographic parameters (TFR, P and H), the ASFR can beestimated with a quadratic splines function
14CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
STEP 3 : From the TFR to the Age Specific Fertility Rate (ASFR)
New methodology - STEP 3/3
From the TFR to the Age Specific Fertility Rate (ASFR)
• Méthode de Shmertmann (2003) :
• Based on 3 parameters (TFR, P and H), the ASFR can be computed witha quadratic splines function
P and H still to be projected : logaritmic trend
15
25
27
29
31
33
35
37
39
1991 2001 2011 2021 2031 2041 2051 2061
Parameters P and H - Belgium
P (obs.) P (proj.)
H (obs.) H (proj.)
New methodology - STEP 3/3
16
----------- 1991
----------- 2070
Illustration - Belgium
CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
Conclusion
17
ConclusionsThe model :
• Takes into account expert opinions to determine the long run fertility trend (structural component) (FEC_LT)
• Includes some conjunctural determinants of fertility (unemployment rate) to project the TFR
• Models also the fertility schedule (ASFR) -> Schmertmann model applied in projection
• Is applicable at the district level (not shown in this presentation)
• Is transparent….
18CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
ConclusionsRoom for improvement:
• Long run fertility trend:
• Take into account birth order?
• Migration background?
• Conjunctural determinants:
• Is unemployment rate relevant enough? (precarity on the labour market, uncertainty…=> but need a projection)
• Should we make age groups ? Determinants of fertility depend on age of the mother…
• …
19CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
Thank you for your attention
Questions? [email protected] , [email protected]
FEDERAL PLANN ING BUREAU
MODELE DE SCHMERTMANN (2003)
21
Estimation of the density function Alpha The youngest age at which fertility rises above zero
P The age at which fertility reaches its peak level
H The youngest age above P at which fertility falls to half of its peak level
k0
k1
k2 k3
k4
.
.
Density function is defined by 5 intervalsjoined at knot values k0 to k4.
kx are computed with alpha, P et H
.
..
CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
MODELE DE SCHMERTMANN (2003) - principes
22
Estimation of the density function quadratic SPLINES function:
f(x) = R ∗ Φ 𝑥 =
𝑘=0
4
Θ𝑘 𝑥 − 𝑘𝑘2
• k0 : α• k1 : (1-W)* α + W*P
où W = min [0.75,0.25+0.025(P- α)](plus P est tardif, plus k1 est proche de P)
• k2 : P• k3 : (P+H) /2• k4 : (H+50) /2• f(x) « normalized » => TFR = R * Φ(x) (R fixes the
level of fertility)• slope at ages P and 50 = 0.
k0
k1
k2 k3
k4
.
..
..
Density function is defined by 5 intervalsjoined at knot values k0 to k4.
kx are computed with alpha, P et H
CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION
Adaptation : k1 is estimated
New methodology - STEP 3/3
From the Total Fertility Rate (TFR) to the Age Specific Fertility Rate (ASFR)
• Méthode de Shmertmann (2003) :
◦ Based on 3 parameters (TFR, P and H), the ASFR can be computed (quadratic splinesfunction)
◦ P and H are projected with a logaritmic trend
◦ In Shmertmann (2003) k1 is computed
◦ Adaptation : k1 is estimated
• Allows a better fit, in particular when fertility is concentraded
23CONTEXT. OLD METHODOLOGY IN A NUTSCHELL. NEW METHODOLOGY. CONCLUSION