Revision of the fertility model applied in the national ...€¦ · 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

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

http://www.plan.be/

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


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



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)


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 :


STEP 1 : assumption on the global fertility level in Belgium in the long run (FEC_LT)


Expert opinions :


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)


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)



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)


STEP 2 : Projection of the TFR based on a Error Correction Model (ECM)


• Method -> Shmertmann (2003)

• Based on 3 demographic parameters (TFR, P and H), the ASFR can beestimated with a quadratic splines function


STEP 3 : From the TFR to the Age Specific Fertility Rate (ASFR)


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.)


16

----------- 1991

----------- 2070

Illustration - Belgium


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….


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…

• …


Thank you for your attention

Questions? [email protected] , [email protected]

FEDERAL PLANN ING BUREAU

mailto:[email protected]

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

.

..


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


Adaptation : k1 is estimated


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


Documents

Revision of the fertility model applied in the national ...€¦ · 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